Copyright is owned by the Author of the thesis.  Permission is given for 
a copy to be downloaded by an individual for the purpose of research and 
private study only.  The thesis may not be reproduced elsewhere without 
the permission of the Author. 
 
STRUCTURAL STUDIES ON TH� 
NUCLEAR LAMINS 
AND OTHER 
INTERMEDIATE FILAMENT PROTEINS 
A thesis presented in partial fulfilment 
of the requirements for the degree of 
Doctor of Philosophy in Biophysics at 
Massey University. 
James Frederick CONW AY 
October, 1989 
H.�!;sey University Library. Thesis Copyright Form 
Title of thesis: SfnALfuroJ, S-fuJieJ (h.ilt.£ rV{).C(� kn<{�J 
a!t4 o!frv.- fv,d�(ah Fik� lfuf��J 
(2) 
(3) 
I give permission for my .thesis to be �de ava110Lle to 
readers in the Hassey University Library under condition!; 
determined by the Librarian. 
I do not �ish �J thesis to be �de available to re�dcrs 
�ithout my �ritten consent for' months. 
I agree that my thesis, or a copy, may be sent to another 
institution under conditions determined .. ?y the Librarian. 
I do not �ish my thesis, or a copy, to be sent to '-Inother 
institution �ithout my vritten consent for month!:, 
1 agree that my thesis may be copied for Library u!:e. 
I do not �ish my thesis to be copied for Libr�ry U$C for 
" lDon ths. 
------
The copyright of this t�esis belon gs to the author. Re�der� �U5t 
sign their name in the space belo� to shov thllt they recognise 
this. They are aske.d to add their pe�nent nddrcss. 
NAME }JlD ADDRESS DATE 
page ii 
ABSTRACT 
A number of aspects of IF chain and molecular structure, as well as molecular 
aggregation, have been examined. These include the delineation of periodicities in the 
sequences of structural domains of IF proteins, the distribution of amino acid residues 
within the heptad substructure, the flexibility of the peptide backbone, the extent of 
homology among the IF proteins, the packing of chains in the dimeric molecule, and 
the axial packing of molecules in the IF. Particular focus has been placed on the newly 
sequenced type V IF proteins (the nuclear lamins and the Helix pomotia B protein) and 
on a type m IF protein (peripherin). 
A parallel in-register arrangement of chains in the molecule is predicted for peripherin 
and the type V chains from a consideration of interchain ionic interactions. Also, 
periodicities in the linear distribution of charged residues in the rod domains of these 
proteins are shown to be comparable with periods in other IF chains. Ionic interactions 
between lamin molecules have been used to assess the likely modes of molecular 
aggregation in an in vitro assembly and a model is presented which also satisfies the 
constraints imposed by electron microscope data. In this model, antiparallel arrays of 
molecules are half-staggered and an extended conformation for the carboxy-terminal 
domains is predicted. Simple explanations are given for the transition between 
paracrystalline and lattice structures and for the disassembly of the lamin meshwork 
concommitant with hyperphosphorylation. The method of calculating intermolecular 
ionic interaction profiles is enhanced and a new, three-dimensional method IS 
developed. 
The inhomogeneous distribution of residues in the heptad substructure can be 
correlated to the coiled-coil structure and chain packing in the molecule. In particular, 
the -75% occupancy rate of apolar residues in the internal!! and d heptad positions is 
shown to be a general feature of a-fibrous proteins. Variability of re si dues in the outer 
12, £ and f positions indicates that structural or functional specificity in the rod domain 
may be determined by these parts of the sequence. The predicted flexibilities of IF 
chains have been compared to the underlying structure for the chains. 
Evidence from sequence homology studies suggests that several new sUbtypes are 
appropriate in the classification scheme. For the hard keratins the terms types la and 
Ha are proposed and for the soft keratins, types Ib and lib; the need to separate the 
neurofilaments into the type IV class separate from the type III IF chains is confmned; 
and division of the type V chains into cytoskeletal and karyoskeletal groups is 
indicated. A more detailed delineation is made of regions within the amino- and 
pageiii 
carboxy-terminal domains than has been possible previously. Periodic features of the 
homology profiles for the rod domain are examined and found to be similar to those in 
the linear distribution of residues in the amino acid sequences. Comparison between 
amphibian and mammalian keratins, and also between hard and soft keratins reveals 
that type II chains are maintained at a higher level of fidelity than type I chains. 
Consensus rod domain sequences are derived for the various IF sUbtypes: absolutely 
conserved regions of primary structure identify types or subtypes. 
page iv 
ACKNOWLEDGEMENTS 
It is with great pleasure that I acknowledge my chief supervisor. Prof. D.A.D. Parry. 
for his assistance and commitment to this work. His indefatigable energy and high 
standards have contributed largely to the work described herein and I look forward to 
working with him in the future. Prof. P.T Callaghan and the staff of the Physics and 
Biophysics Department at Massey University have been very supportive during the 
course of my studies and I appreciate their efforts on my behalf. The other doctoral 
and graduate students have provided a friendly atmosphere within which to work and I 
wish them every success in their endeavours. Mr P.M. Ngan. Director of the Massey 
University Image Analysis Unit. has been of great assistance in providing access to 
the Unit' s facilities and in many interesting discussions on a variety of topics. 
I much appreciate the kind hospitality provided by Drs. P.M. Steinen and A.C. Steven 
during the course of a visit to the National Institutes of Health in 1986 and also for the 
extra impetus they have provided in the completion of this work. I look forward to 
continued collaboration with them. In addition. I acknowledge the valuable 
contributions towards the cost of that trip from the Royal Society of New Zealand. the 
Dean of the Science Faculty at Massey. Prof. G.N. Malcolm. and the Fibrous Protein 
Merit Award. 
Thanks also to my parents for their encouragement and assistance during the earlier 
period of my studies. Finally. special thanks to Sharon Wards for her support and 
encouragement during the course of my PhD studies. I am indebted to her for all her 
time spent directly and indirectly in the preparation of this manuscript. 
Some amino acid sequences were kindly provided prior to publication by: 
Or RA. Lazzarini (human NF-M chain) 
Or SSM. Chin (rat NF-M chain) 
Professor GE. Rogers and Or BC. Powell (a type 11 keratin chain from sheep) 
Or PM. Steinen and Or DR. Roop (mouse M50k and M55k chains) 
Electron micrographs were kindly provided by: 
Or PM. Steinen (Figure 1-1 of reconstituted IF) 
Professor GE. Rogers (Figure 1-2 of wool ftlaments in cross-section) 
Or AE. Goldman and Or RD Goldman (Figure 4-2 of lamin paracrystals) 
Or C. Cohen (Figure 4-7 of a tropomyosin lattice/paracrystal structure) 
page v 
TABLE OF CONTENTS 
1. An Overview of Intermediate Filament Structure ................ 1 
1.1. ........ Early Structural Work on Keratin IF ................. ........... . . . . .  2 
1.2 .... ..... Soft Keratin and Other IF .............................................. 7 
1.3 ...... ... Current Models of Intennediate Filament Structure .............. 10 
1.4 ......... Conservation of Amino Acid and Gene Sequences ............... 15 
1.5 ......... Lamin, Peripherin and the Helix A and B proteins .... , .......... 17 
1.6 .........  Function of Intermediate Filaments ................................. 22 
1.7 ......... Structural Fonn of the Thesis .... " ............................ , .... 25 
2. Pr imary S truct u re .................................................. 2 6  
2.1 ......... Fourier Analysis ......................................................  29 
2.1.1 ....... Fourier Analysis - Method ....................................... 30 
2.1.2 .......  Fourier Analysis - Results ............................... . ....... 33 
2.2 . . ....... Residue Distribution in the Heptad . . ..... .... . ....... . . ...... '" ... 37 
2.3 .........  Flexibility ................................................ . .............  41 
2.4 ... . . . . ..  Summary ..... ...... . . ..... ........... . ....................... . . .. . ....  47 
3. Sequ ence H omol ogy ................ ............................... 5 0  
3.1 .........  Homology Statistics .................. . . ...................... ........ 51 
3.1.1. . .....  Amino Acid Homology Score, ha ............................... 52 
3.1.2 . . .....  Residue Homology Score, hr ... .... . ...................... . ..... 52 
3.1.3 . . .....  Segment Homology Score, hs . . . . .... ................... ........ 54 
page vi 
3.2 ......... Homology within the Rod Domain ............. .................... 54 
3.2.1 ....... Coiled-Coil Segments ............................. ............... 56 
3.2.2 ....... Link Segments .................................................... 57 
3.2.3 ... .... Type V Chains ...... .... ..................................... ..... 61 
3.3 ........ . Subtyping on the basis of Homology 
................ ........... ... 61 
3.3.1. ...... Hard and Soft Keratin IF Chains ............................... 61 
3.3.2 ....... Type IV IF Chains ................................................ 62 
3.4 ......... Homology among the 'H' subdomains ............................  63 
3.5 ......... Comparison of Amphibian and Mammalian Scores .............. 66 
3.6 ......... Periodic Features in the Homology Score Distributions ......... 69 
3.7 ......... Consensus Sequences for IF Chain Types ........................ 78 
3.8 ......... Summary .............................................................. 81 
4. Secondary and Tertiary Structure ............. : . . . . . . . • . . . . . . . . . .  8 4 
4.1. ........ ID Ionic Interactions Between Chains ........ . .. . ................. 85 
4.2 ....... ..  ID Ionic Interactions Between Molecules .. ....................... 90 
4.3 .. . . . .... Modelling Lamin ................... ............... ... . . ............... 96 
4.4 ......... 3D Ionic Interactions Between Molecules ....................... .  110 
4.4.1 ....... Generation of Coordinates for a Coiled-Coil Molecule ...... 112 
4.4.2 ....... Determination of Interactions ............... .... . ............... 113 
4.4.3 ....... Analysis of the Interaction Maps ................. . ............. 117 
4.5 ......... Summary .... ..... . ............ ....................... . .. ............. 120 
5. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3  
page vii 
Appendices .......................................................... 1 3 1 
Appendix A : Zero-Filling prior to Fourier Transfonnation .................. 131 
Appendix B : Fourier Transforms ............................................... 135 
Appendix C : Curve Smoothing ................................................. 149 
Appendix D : Intermolecular Ionic Interactions ................................ 150 
B ibl iography .................. . .................................. ... 161 
Publications ........................ .......... .... . . . ............. . .. 189 
page viii 
LIST OF FIGURES 
Figure 1-1 .... Electron micrographs of IF in vitro. . . .. . . . .. .. . .. . .. . . . . .. .. . .. . . . .. 1 
Figure 1-2 . ... Electron micrograph of wool microfibrils in cross-section ....... 3 
Figure 1-3 ....  Schematic representation of the IF protein chain ................... 9 
Figure 1-4 ....  Schematic representation of the heptad substructure .... . ........  12 
Figure 1-5 . . . .  Schematic comparison of IF and lamin chain structures . . ... .... 20 
Figure 2-1 .... Example of applying the baseline correction operation 
prior to the Fourier transfonn ...................................... 32 
Figure 2-2 ....  Comparison of the human and Xenopus lamin A protein 
sequences .............................................................. 35 
Figure 2-3 ....  Flexibility profiles for a selection of IF chains ................... .43 
Figure 3-1 ... . Homology profIles for the IF chain types .................. ....... 55 
Figure 3-2 .... Comparison of the link segments Ll. L12 and L2 ............... 59 
Figure 3-3 .... Fourier transforms of the homology profiles . ..................... 70 
Figure 3-4 . . . .  Consensus sequences of the rod domain . .. . ... ................... 79 
Figure 4-1 . . . .  Intermolecular ionic interaction curves for human lamin A . ... . . 92 
Figure 4-2 ... . Electron micrographs of lamin paracrystals . . . . . ... . ...... . .. . . . .. 98 
Figure 4-3 . . . .  Diagram of staggered arrays of lamin molecules in Model A . .  100 
Figure 4-4 .. . . Diagram of staggered arrays of lamin molecules in Model B.. 101 
Figure 4-5 .. . . The alignment of conserved sequences .......... . ..... . ......... 106 
Figure 4-6 .... Comparison of Models A and B ..... ..................... . ....... 108 
Figure 4-7 . ... Schematic of a lattice collapsing into a paracrystal .... . . . . . . . . . . 109 
pageix 
Figure 4-8 .... Electron micrograph of a tropomyosin lattice/paracrystal 
s tructure ............................................................... 110 
Figure 4-9 .... Schematic of the relative orientations of molecules ............. 112 
Figure 4-10 . . Schematic of a pair of segment IB dimers .......................  116 
Figure 4-11 .. Intensity map of the 30 intermolecular ionic interactions.. . . .. 117 
Figure A-I . .. Rectangle and sinc functions - Fourier pairs .................... 133 
Figure B-1 . ..  Fourier transfonns for peripherin. .. . . . .. . . . .. . . . .. . . . . . . .. .. . .. .. 136 
Figure B-2 ... Fourier transforms for human lamins A and C . .... . .. . .. . ... .. 138 
Figure B-3 . .. Fourier transforms for Xenopus lamin A . . . ........ . . .  '" ...... 141 
Figure B-4 . " Fourier transforms for Xenopus lamin B ..... . . . . . . . . . . .. . . . . . .. 144 
Figure B-5 . . Fourier transforms for Helix pornotia B ......................... 147 
Figure 0-1 ... Intermolecular ionic interaction curves for peripherin .......... 151 
Figure 0-2 ' "  Intermolecular ionic interaction curves for human lamin A . . .. 154 
Figure 0-3 ... Intermolecular ionic interaction curves for Xenopus lamin A. 156 
Figure D-4 . .. Intermolecular ionic interaction curves for Xenopus lamin B. 158 
Figure D-5 . . . Intermolecular ionic interaction curves for Helix pomotia B .. 160 
page x 
LIST OF TABLES 
Table 1-1 ..... Lengths of the rod domain segments for IF types I-N . .........  11 
Table 2-1 .....  IF amino acid sequences ... . .. . ... . ... . ....... . . .... . . .. . ..... 27 
Table 2-2 . . . .  Comparison of periods present in peripherin and other 
type ID IF proteins .................................. ... ......... . .... 33 
Table 2-3 .....  Peaks resulting from multiplying the Fourier transforms 
together ..... ........ . . .......... . ......................................  36 
Table 2-4 .....  Residue distribution in the heptad for IF and myosins ...... ... . .  38 
Table 2-5 .. ... Mean flexibility indices for chain segments from a 
selection of IF chains ............................ .... ............. .. 44 
Table 3-1 . . ...  Look-up table for mixed homology scores ... . . . . . ................  53 
Table 3-2 .. ' "  Look-up tables for acidic homology. basic hOqlology 
and large apolar homology scores ............... ....... ..... . .....  54 
Table 3-3 .....  Regions of high sequence homology (h�90%) .................. 56 
Table 3-4 ... . . Regions of low sequence homology (hr<60%) ................... 57 
Table 3-5 . . . . . Mean segment homology scores ..... . ....... . . ... . . .... ... . ... . ....  58 
Table 3-6 .. . .. Mean segment homology scores for the rod domain 
segments in soft and hard keratins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  62 
Table 3-7 ..... Lengths of the homologous subdomains HI and H2 ...... . . . . . .  64 
Table 3-8 . . . .  Extents of the structural domains in IF chains .................... 67 
Table 3-9 . .. . . Mean segment homology scores for the rod domain 
segments in amphibian and mammalian keratins ......... . ........ 68 
Table 3-10 .... A selection of the most significant periodicities in the 
homology proftles .......... . . . . ..... ......... . .. . . . . . .......... ......  75  
page xi 
Table 3-11. ... Mean residue homology scores for each position in the 
heptad substructme ................................................ ... 78 
Table 3-12 .... Percentage occurrence of highly conserved residues ............ . 81 
Table 4-1 ..... Interchain ionic interactions for peripherin and type V IF ...... . 86 
Table 4-2 .....  Interchain ionic interactions for a selection of IF ................. 89 
Table 4-3 .....  Interchain ionic interactions per dual heptad ........... ............ 89 
Table 4-4 ... . .  Significant intermolecular ionic interactions between 
human lamin A IIlOlecules ...................... ..................... 95 
Table 4-5 ..... Significant ionic interactions between peripherin IIlOlecules .....  96 
Table 4-6 ... .. Volume calculation for the C-tenninal domains of human 
lamins A and C ............................................... ........ 104 
Table 4-7 ..... Ionic interactions used for Models A and B . ....... . . .. ... . ..... 105 
Table 4-8 " ... Values used to derive a set of five pitch lengths for the 
coiled-coil. ............................................................ 111 
Table 4-9 . . . .. Starting set coordinates in an undistorted a-helix ............... 113 
Table 4-10 .... Coordinates for a pair of segment IB dimers . ............. ... . .  114 
Table 4-11.. .. Selection of the highest 3D interaction scores (antiparallel) . .. 118 
Table 4-12 .... Selection of the highest 3D interaction scores (parallel) ........ 119 
Table B-1.. ... Peaks in the Fourier transfonns for peripherin ....... . .. . . . .....  135 
Table B-2 ..... Peaks in the Fourier transforms for human lamins A and C . .. 137 
Table B-3 ..... Peaks in the Fourier transforms for Xenopus lamin A ....... . .  140 
Table B-4 ..... Peaks in the Fourier transfonns for Xenopus lamin B ... . .... . 143 
Table B-S ..... Peaks in the Fourier transfonns for Helix pomotia B ... . . . . .. .  146 
Table D- l ..... Peaks in the ionic interaction curves for peripherin .. . . ..... . . . .  150 
page xii 
Table 0-2 ..... Peaks in the ionic interaction curves for human lamin A . . . . . . .  153 
Table 0-3 ..... Peaks in the ionic interaction curves for Xenopus lamin A . .. .  155 
Table 0-4 ..... Peaks in the ionic interaction curves for Xenopus lamin B . . . .  157 
Table 0-5 ..... Peaks in the ionic interaction curves for Helix pomotia B .....  159 
CHAPTER 1: Introduction page 1 
1. AN OVERVIEW OF INTERMEDIATE 
FILAMENT STRUCTURE 
Intermediate filaments (IF) are long, unbranched structures found in a diverse range of 
cell types (Figure 1 - 1 ). They fonn one of three major classes of cytoskeletal networks 
and are so-called because their diameters (-1 0  run) lie intermediate in size between the 
smaller actin-containing microfilaments and the larger diameter microtubules and 
myosin-containing thick filaments. The recent addition of the nuclear lamin proteins 
into the IF family has shown that IF networks are found not only throughout the 
cytoplasm but also within the cell nucleus where they are thought to provide at least 
part of the structural integrity of the nuclear envelope. 
The IF proteins were initially classified according to the cell types with which they 
were associated. The first of these classes comprised the relatively large group of 
keratins from epithelial cells; the second class contained the desmin protein found in 
myogenic cells; glial fibrillary acidic protein (GF AP) in astroglial cells, and vimentin in 
cells of mesenchymal origin, made up the third and fourth classes respectively. 
Figure 1-1 Examples of the 10 nm-diameter intermediate filament (IF) structures in vitro 
(separate images are not to scale). Branching filaments are evident as are filaments crossing over each 
other. Figure courtesy of Dr P.M. Steinert. 
CHAPTER 1: Introduction page 2 
(Vimentin, however, was also found in cells that expressed other IF). The fifth class 
encompassed the neurofilaments from neuronal cells and was further divisible into 
three sub-classes according to the relative molecular weights of the chains. 
As amino acid sequence data became available for the various IF chains, an altem�tive 
classification system based on sequence homology present in the major a-helix­
containing fraction of the IF proteins largely replaced the cell-specific nomenclature. 
The early sequence studies of Crewther and co-workers on wool keratins revealed two 
kinds of a-helix-rich segment: type I had a net acidic character and type IT was neutral­
basic. Subsequent keratin sequences, primarily from the epidermal keratins, were 
found to fall into these same two groups and consequently the chains from which these 
characteristic sequences were derived were termed type I and type II IF chains. The 
equivalent a-helical regions in desmin, vimentin and GF AP showed greater sequence 
homology with one another than with the keratins and were classified together as type 
III IF chains. Although the neurofilament proteins were initially included within the 
type III grouping, increasing evidence has indicated that they are members of a distinct 
group and they are now referred to as type IV IF chains. Several recent additions to the 
IF family of proteins have been made: peripherin, a type III IF protein; the nuclear 
lamin proteins which are now established as a new IF type-protein - type V; and the 
Helix pomotia B protein which shows structural homology with the nuclear lamins. 
Since the presence of cell-specific IF indicates a diversity of function, it is pertinent to 
ascertain whether this translates into a number of variants on the basic model for IF 
structure. The low-resolution substructures of some IF have been partially defined in 
the last five years or so and this will form the basis of the work described here. This 
thesis, therefore, represents a collection of studies on the structure and aggregation of 
IF molecules with special reference to the more recently characterized members of the 
IF family - the nuclear lamin proteins in particular but also peripherin and the Helix 
pomotia B protein. 
1. 1 Early Structural Work on Keratin IF 
The structure of wool keratin has been a focus for research over the past 50 years due 
largely to the commercial significance of the wool fibre. The keratin IF (originally 
termed microfibrils) in wool, hair and quill are about 10 nm in diameter and are 
embedded in a matrix of proteins of high-sulphur and of high-glycine-tyrosine content 
(Figure 1-2). Keratins from these sources are termed 'hard' to distinguish them from 
the 'soft' (or epidermal) keratins found in the stratum corneum of skin. IF in hard 
keratin are preferentially oriented parallel to the axis of the fibre and have proven to be 
ideal subjects for X-ray diffraction studies. In contrast the orientation of IF in the soft 
CHAYrER 1: Introduction page 3 
keratins is much poorer and consequently few X-ray diffraction data have been 
recorded. Pioneering work in this area was carried out by Astbury and co-workers 
who studied hard keratinized tissues from a variety of sources (Astbury and Street, 
1931; Astbury and Marwick, 1932; Astbury and Woods, 1933). They noted a number 
of distinctive X-ray diffraction patterns: a (from wool keratin), J3 (from stretched 
mammalian hard keratin), the feather-pattern (from hard avian and reptilian tissue), and 
the amorphous pattern (from the cuticle of animal hair). The terms 'a-keratin' and 
'a-fibrous protein' are derived from this nomenclature and refer to proteins that give 
rise to the a-pattern. Astbury and Woods (1933) proposed that the a-pattern was 
generated by a folded structure and that the J3-form of keratin arose from stretching the 
molecular chains into an almost fully extended conformation. Current descriptions of 
the J3-structure are based on the pleated sheet models of Pauling and Corey (1951 b, 
1953b) and will not be discussed further here (see, for example, Fraser et ai, 1972 
and Fraser and MacRae, 1973b). 
A number of polypeptide chain structures were proposed to explain the a-pattern (see 
for example, Astbury and Bell, 1941; Huggins, 1943; Astbury et ai, 1948; Ambrose 
et ai, 1949; Bragg et aI, 1950; Pauling and Corey, 1950). The model of Pauling and 
Corey (elaborated in Pauling and Corey, 1951a and Pauling et ai, 1951) placed the 
residues in a helical arrangement, and two variants were described: one had 3.7 
residues per turn of helix and the other had 5.1 residues per turn. These were termed 
the a-helix and the y-helix respectively (Pauling and Corey, 1951a) and featured 
intrachain hydrogen bonds and planar amide groups. Perutz (1951a,b) eliminated from 
consideration all models but the Pauling and Corey a-helix by observation of a 
meridional reflection of spacing 0.149 nm in the X-ray patterns of poly-y-benzyl-L­
glutamate, keratin, and hremoglobin. Only the a-helix was expected to give rise to this 
spacing which corresponds to the axial rise per residue. (The 0.149 nm meridional 
Figure 1-2 Cross-section of intennediate filaments in wool. Figure courtesy of Professor G.E. 
Rogers. 
CHAPTER 1: Introduction page 4 
reflection was originally noted by MacArthur, 1943, in the X-ray diffraction pattern 
from African porcupine quill tip but was interpreted in terms of repeating sidechains 
along the polypeptide chain. By using the synthetic polypeptide, poly-r-benzyl-L­
glutamate, Perutz demonstrated that the 0.149 nm reflection was independent of the 
specific natures of the sidechains). Cochran et al (1952) calculated the Fourier 
transform of an a-helix and quantitative agreement for this conformation was found 
from X-ray diffraction studies on the synthetic polypeptides poly-y-methyl-L­
glutamate (Bamford et ai, 1952) and poly-L-alanine (Brown and Trotter, 1956). 
The screw sense of the a-helix in vivo was not determined for a number of years. 
Pauling et al (1951) observed that for residues with the L-configuration (which is 
predominantly the case in vivo) the position of the sidechains differed between left­
and right-handed a-helices (except for glycine which has a sidechain consisting of a 
single hydrogen atom). Huggins (1952) showed that for L-amino acids, significant 
steric hindrance was probable between the l3-carbon and the carbonyl-oxygen in the 
same residue for a left-handed a-helix whereas no such hindrance would occur for a 
right-handed one. Subsequent studie� on a-helix-forming synthetic polypeptides (for 
example poly-L-alanine: Elliott and Malcolm, 1956, 1959) showed that the a-helices 
were right-handed, as is the general case, although several left-handed a-helices of 
marginal stability have subsequently been reported (see Fraser et ai, 1972). The first 
direct confirmation of the presence of a-helices in proteins was obtained from the 0.2 
nm Fourier synthesis of myoglobin (Ken drew et ai, 1960). In addition, the screw 
sense of all the a-helical segments in the molecule was shown to be right-handed. 
A prominent feature on the meridian of the X-ray diffraction patterns from a-keratin 
and other a-fibrous proteins (for example a-tropomyosin) is a strong reflection at a 
spacing of 0.515 nm (Astbury and Street, 1931). This is not predicted by an 
undistorted a-helical structure which should instead show off-meridional layer line. 
diffraction at an axial spacing of 0.54 nm (Crick, 1952). An explanation of the strong 
0.515 nm reflection and the missing layer line was made independently by Crick 
(1952) and Pauling and Corey (1953a): in their models the a-helices were distorted so 
as to wrap around one another in a supercoil of opposite sense to that of the a-helix 
and this structure was termed a 'coiled-coil rope'. The 0.515 nm reflection was shown 
to arise from the axial rise per turn of the supercoil. Crick (1953) described a seven 
.residue substructure, later termed the heptapeptide or heptad repeat, in which 
sidechains from two of every seven residues could be made to fit neatly into the spaces 
between the sidechains on the other a-helix or a-helices. This arrangement of side­
chains along the line of contact was termed 'knob-hole' packing and the nature of these 
internalized sidechains was suggested to be hydrophobic. Confirmation of this 
CHAPTER 1: Introduction pageS 
suggestion was provided many years later by Hodges et al ( 1972), Parry ( 1974) and 
Stone et al (1974) who demonstrated such a substructure in the amino acid sequence 
of tropomyosin - apolar residues were indeed successively three and four residues 
apart. McLachlan and Stewart ( 1975) introduced a nomenclature for the positions of 
the residues in the heptad, iL h., k' d, � f and � where it. and d are the internal 
positions of the coiled-coil and are filled predominantly by apolar residues. In IF 
proteins, about 75% of residues in the � and � positions are apolar (Parry and Fraser, 
1 985). 
Hard a-keratin, although ideal for X-ray diffraction studies, is not readily amenable to 
chemical analysis. The high content of covalent disulphide bonds present in wool, for 
example, effectively welds the proteins into a mechanically inert structure (Fraser 
et aI, 1972). Even when the disulphide bonds are reduced, a complex mixture of 
related proteins is revealed (Crewther et ai, 1965). The problems associated with 
sequencing the keratin proteins in wool (using the c lassic protein sequencing 
techniques) were summarized by Hogg et al ( 1978) as follows: the protein chains 
were relatively large at 40-50 kDa; the number of similar chains within the relatively 
small range of molecular weights was also large; and identification of particular chains 
was difficult. Nonetheless, Crewther and Dowling (197 1 )  and Dowling and Crewther 
(1972) have shown by amino acid analysis, peptide mapping and optical rotatory 
dispersion measurements that two types of chain segment, termed I and 11, were 
present in subfractions of helical fragments of S-carboxymethylkerateine-A. Crewther 
and co-workers (Crewther, 1976; Crewther et ai, 1978a; Elleman et aI, 1978; Gough 
et aI, 1978;  Hogg et aI, 1978) sequenced a type I and a type 11 segment of lengths 
103 and 109 residues respectively and showed that a heptad substructure existed with 
apolar residues common in the � and Q positions. Consequently, in a helix of 3.6 
residues per turn these apolar residues generate a stripe running around the outside of 
the undistoned a-helix - a result in accord with the coiled-coil model of Crick ( 1953) 
and the tropomyosin model (see above). They also noted that 32% of the residues 
were identical between the type I and the type IT segments when the sequences were 
aligned to maximize homology. Crewther et al ( 1978b) extended these results by 
comparing the sequences of five fragments (-30 residues each) from different wool 
keratin proteins: of those, three were type I segments and two were type 11 segments. 
As before, the identity between the groups was about 30% but within each type there 
were few differences. More recently, Crewther et al ( 1985) have fully sequenced the 
rod domains of two type I wool keratin proteins (components 8a and 8c- l )  and two 
type 11 proteins (components 7 and 7c) and have reponed that the identity within the 
type I and 11 groupings is 90-95%. 
CHAYfER 1 :  Introduction page 6 
Long standing problems concerning the structure of the coiled-coil in a-keratin 
proteins have been to detennine the pitch length of the coiled-coil and the number and 
relative orientation of polypeptide chains that comprise the rope. Crick ( 1953) 
described coiled-coil ropes with two and three a-helical strands and suggested that the 
three-stranded rope would be an appropriate model for a-keratin. Pauling and Corey 
( 1953a) originally suggested that a-keratin comprised 'a-cables' of six a-helices 
coiled around a seventh with single a-helices in the interstices between the c ables. 
Evidence for a two-strand rope was found by Cohen and Holmes ( 1963) in the X-ray 
diffraction pattern of a highly oriented specimen of paramyosin and they concluded 
that the pitch length of the left-handed coiled-coil was 17 .8± 1  nm. The degree of 
orientation in the X-ray diffraction pattern from a-keratin was insufficient to allow a 
clear choice to be made between two and three strands (Fraser et aI, 1964a, 1964b, 
1965) and in addition the X-ray diffraction pattern was overlaid with an elaborate 
interference pattern associated with higher levels of structure (Fraser et aI, 1 972). 
Estimates of the pitch length for a-keratin ranged from 14 to 17  nm and 21 to 26 nm 
for two- and three-stranded ropes respectively (Fraser et ai, 1965) but it was 
suggested that the degree of similarity between the diffraction patterns from 
tropomyosin, myosin, paramyosin (all of which were known to contain two-stranded 
ropes from physico-chemical data) and a-keratin would tend to favour a two-strand 
rope model for a-keratin. 
In principle, two-stranded coiled-coils can adopt one of two forms: in the first both 
a-helices are similarly directed, or parallel (ie, the polypeptide mainchain sequences, 
-NH-aC-C' -, were oriented in the same direction), whereas in the other the chains are 
oppositely directed, or antiparallel. Parry and Suzuki (1969) calculated the free 
energies of the parallel and antiparallel forms for a two-stranded model of poly-L­
alanine and compared them with the energies of the analogous pairs of straight 
a-helices. The coiled-coils were shown to be significantly more stable than the straight 
a-helix analogues and the anti parallel coiled-coil was favoured over the parallel. A 
feature not considered at that time, however, was the role played by charged 
sidechains in specifying the orientation of the chains in the coiled-coil (Parry, 1974, 
1975; McLachlan and Stewart, 1975, 1976). McLachlan and Stewart (1975) and Parry 
(1975) showed that the distribution of charged residues in a-tropomyosin resulted in a 
largely acidic stripe in the � position of the heptad and a largely basic stripe in the g 
position where both stripes were adjacent to, but on opposite sides of, the 'internal' 
apolar stripe. Parallel chains allowed salt bridges (or ionic interactions) to be made 
between a pair of helices but anti parallel chains placed similarly charged residues in 
close proximity and caused destabilization. Interchain ionic interactions were studied 
for the type I and type 11 chain segments of Crewther et al (described above) in a two-
CHAPTER 1 :  Introduction page 7 
stranded rope (Parry et al, 1977) and a three-stranded rope (Parry, 1979) and all 
combinations of chains, polarity (ie, parallel or antiparallel) and relative stagger were 
examined. While the number of chains in the molecule could not be detennined, a 
parallel, in-register arrangement of a type I chain and a type 11 chain was considered 
among the more likely candidates. A related study by McLachlan (1978) suggested that 
the coiled-coil in a-keratin was more likely to be two-stranded than three-stranded. 
The chemical studies on wool keratin originally appeared to favour a three chain 
subunit (Crewther and Harrap, 1967; Crewther et ai, 1968, Crewther and Dowling, 
1971; Skerrow et ai, 1973; Lee and Baden, 1976; Lotay and Speakman, 1977; 
Steinert, 1978; Steinert et al, 1980a) although in some cases a two chain structure 
could not be excluded. Subsequent studies by Crewther et al ( 1980) showed that type 
I proteins interact specifically with type 11 proteins in a 1: 1 molar ratio although there 
was no evidence to show whether single or multiple chains of each species were 
involved in the molecular structure. However, crosslinking studies by Ahmadi and 
Speakman ( 1978) and Ahmadi et al (1980) provided strong evidence for a four-chain 
subunit. Further studies by Woods and Oruen (1981) and Oruen and Woods (1983) 
showed that the wool microfibril contained equal numbers of type I and IT chains and 
that the fundamental sub-unit was a tetramer of chains configured as a pair of coiled­
coils. More detailed studies by Inglis et al (1983) and Woods and Inglis (1984) 
confirmed that the dimer of chains in the coiled-coil molecule from wool keratin was 
indeed comprised of a type I chain and a type IT chain - a heterodimeric structure - and 
that the chains were parallel. Their data also indicated that the molecules were 
antiparallel and approximately half-staggered so that the N-terminal regions of the rod 
domain were overlapped. 
1.2 Soft Keratin and Other IF 
Until relatively recently, the so-called 'soft' or epidermal keratins from the stratum 
corneum of skin had not been studied as extensively as the hard keratins from wool or 
quill. Certainly the filaments found in the stratum corneum were known to be similar 
in diameter to those from hard keratin-containing tissue (- 10 nm). However, the 
degree of alignment was so much poorer than in the hard keratins that the use of 
physical techniques such as X-ray diffraction and electron microscopy had been very 
limited although specimens prepared from mammalian epidermal extracts had yielded 
an a-pattern of poor quality. A comparison of the amino acid content of hard and soft 
keratins had shown much similarity. Major chemical differences did exist, however, 
between the hard and soft keratins, mainly in the relative contents of the cystine and 
glycine residues (see Fraser et ai, 1972). 
CHAPTER 1:  Introduction pageS 
The hard keratins are very insoluble and are embedded in a matrix of high-sulphur 
proteins and high glycine-tyrosine proteins. A high degree of disulphide crosslinks 
between hard-keratin chains gives rise to this insolubility as well as the 'hardness' of 
the filament. It also gives rise to the degree of orientation so necessary for useful X­
ray diffraction study. The soft keratins, however, have few disulphide cross-links and 
are much more amenable to chemical investigation. 
Chemical studies on all members of the IF family revealed that the proteins were more 
closely related to one another than had been thought initially (for example, Steinert 
et ai, 1978, 1980a,b). Studies on the proteolytic digests of epidennal keratins 
(Woods, 1983; Parry et ai, 1985) produced four-chain and two-chain particles 
analogous to those produced from hard keratins (Woods and Inglis, 1984; and see 
above) and it was concluded that epidennal keratin IF was also a type I-type II 
heterodimer composed of parallel, two-stranded coiled-coils. Quinlan and Franke 
(1982, 1983) induced crosslinks in type III IF chains using 1,10-phenanthroline 
cupric ion complexes and subsequently isolated crosslinked homo- and hetero-dimers: 
they suggested that the crosslink between the single cyste.ine residues in these 
sequences was inter-molecular. Parry et al (1985) showed that the crosslink could 
instead be made intra-molecularly between parallel, in-register chains and used this as 
further evidence for a common molecular structure for all IF proteins. Sequence 
studies of the wool keratins, the epidennal keratins, and the other IF have also shown 
that these proteins, originating from a wide variety of cell types, exhibit a high degree 
of homology in the central, largely a-helical (rod domain) portion of the chains 
(Geisler and Weber, 1982; Weber and Geisler, 1982; Crewther et aI, 1983; Dowling 
et ai, 1983). 
Little was known about the tenninal domains of IF proteins before amino acid 
sequence data became available. Optical rotatory dispersion experiments had indicated 
that the keratin molecule was not all a-helical: Harrap (1963), for example, showed 
that the helical content was only 50-60% (see also Crewther et ai, 1966, 1968). Early 
sequencing studies on wool keratin by O'Donnell and co-workers (for example, 
O'Donnell, 1969) had revealed fragments with no structure that could be related'to the 
regular a-pattern deduced from the X-ray diffraction data. Later studies by Crewther 
et al (1983) on more complete hard a-keratin sequences showed that the N- and 
C-terminal portions of the chains were not likely to be a-helical in confonnation and in 
fact �-bends were largely predicted. 
It is now clear that all IF chains share a common structural plan (a central coiled-coil 
domain bracketed by tenninal domains of non-repetitive secondary structure - see 
CHAPTER 1 :  Introduction page 9 
lA  IB  2A 2B 
El HI H2 V2 E2 
Amino- Carboxy-
terminal I terminal 
domain I�  Segment I � I� Segment 2 "-1 domain 
Figure 1·3 Schematic representation of the IF protein chain. The central rod region is largely of 
coiled-coil structure (shaded) with breaks at the linker regions, L l  and Ll2, neither of which are 
predicted to be coiled-coil or a-helical in conformation, and at link L2 which is predicted to be 
a-helical but not a coiled-coil. The lengths of segments 1 and 2 are each about 20.5 nm and the 
combined length of the rod is predicted to be about 47 nm. The terminal domains vary greatly in size 
among the IF proteins and are largely non-a-helical. 
Figure 1-3: Geisler et aI, 1982a; Crewther et ai, 1983). However, a large range of 
molecular weights is apparent in the IF group. This arises almost entirely from 
variations in the sizes of the terminal domains: extreme examples are the cytokeratin 19 
protein (mol. wt. 44 kDa: Bader et aI, 1986; Eckert, 1 988) which is almost devoid of 
a C-terminal domain (only nine residues in length from the generally accepted 
boundary between the rod and C-terminal domains) and the neurofilament NF-H 
protein (mol. wt. 140 kDa: for example the mouse NF-H chain, Shneidman et aI, 
1988) which has a C-terminal domain of length 658 residues. 
The structural role of the terminal domains is not altogether clear although the amino­
terminal domain does appear to be involved in the formation of at least some IF. 
Certainly the carboxy-terminal domain is not essential for keratin IF formation, as 
exemplified by the tailless cytokeratin 19 protein. Also, removal of a portion of the 
carboxy-terminal domain does not inhibit the development of normal type III IF but 
removal of part of the amino-terminal domain from desmin (Geisler et aI, 1982a: 
Kaufmann et aI, 1 985), vimentin (Traub and Vorgias, 1983) and keratin (Sauk et ai, 
1 984) does appear to limit the filament-forming ability of these proteins. (Lu and 
Johnson, 1 983, have disputed the involvement of the amino-terminal domain in 
filament formation but no corroboration of their evidence has been forthcoming). 
However, Steinert et aI (1983a) have shown that pieces of the non-a-helical terminal 
domains from epidermal keratin chains could be removed enzymatically from intact 
filaments without affecting the structural integrity of the filament, as judged by its 
appearance in the electron microscope. Hence, although portions of the terminal 
domains may be necessary for the assembly of IF, the mature structure seems to be 
stabilized largely by the rod domains and does not necessarily rely on the terminal 
domains for maintenance of the structure. The accessibility of the terminal domains to 
enzymatic cleaving implies that they are at least partly, if not fully, on the exterior 
CHAPTER 1: Introduction page 10 
surface of the ftlament. 
Examination of the N- and C-tenninal domain sequences enables classification into 
subdomains which show homology within each IF type (Steinert and Parry, 1985; 
Steinen et ai, 1985b; see Figure 1 -3). Adjacent to the rod domain are regions of high 
sequence homology, HI and H2. Distal to these are domains which are variable in 
content and size, termed VI  and V2. At the ends of the chains are the E l  and E2 
domains which usually have a high charge content. Keratin chains, for example, often 
exhibit glycine-serine rich motifs in the VI  and V2 domains and are basic in the E l  and 
E2 domains as are the type III chains. This net basic character is in contrast to the 
major coiled-coil regions of the rod domain which are acidic (Steinen and Parry ,  
1985). Some of the N- and C-tenninal subdomains are not evident in particular IF 
groups: type I soft keratins are missing the H2 domain for example, whereas the 
C-terminal domain of type ill chains is all H2 with no V2 or E2 regions. 
1 . 3 Current Models of Intermediate Filament Structure 
Aspects of the currently held model for the structure of the IF molecule have already 
been outlined: a pair of parallel right-handed a-helices aligned in axial register to form 
a left-handed coiled-coil configuration (the rod domain) and bracketed by terminal 
domains of non-repetitive secondary structure. Predictive schemes have indicated that 
the rod domain of the IF molecule is comprised of four coiled-coil regions (segments 
lA: 35 residues, IB: 101  residues, 2A: 19 residues, and 2B: 12 1  residues) separated 
by shon links (link Ll :  7-14 residues; L12: 1 6- 1 7  residues and L2: 8 residues) of 
largely unknown conformation (Crewther et ai, 1983; Steinert et ai, 1983a; Dowling 
et ai, 1983; Steinen and Parry, 1985; see Figure 1 -3 and Table 1- 1 ). The coiled-coil 
regions are readily grouped into two segments, I ( lA-LI-1B) and 2 (2A-L2-2B), each 
about 20.5 nm in length (assuming the average axial rise per residue of 0. 1485 nm 
measured from X-ray diffraction studies of a-keratin). This is very similar to a 
structural repeat deduced from the X-ray diffraction patterns of hard keratin (see Fraser 
et ai, 1980) and visualized in electron microscopy studies of keratins and other IF 
(Milam and Erickson, 1982; Sauk et ai, 1984). If a similar average axial rise per 
residue is assumed for the linker regions as for the coiled-coil segments, the combined 
length of the rod domain would be about 47 nm. The linker regions, however, could 
be arranged as loops external to the coiled-coil and this could result either in a 
minimum length of 20-2 1 nm if the rod domain was folded back on itself around the 
L12 linker region or of 4 1  nm for a linear rod structure. Alternatively, the links could 
be fully extended with an axial rise per residue of 0.334 nm leading to a rod domain as 
long as 53 nm. High-resolution X-ray diffraction studies have revealed an axial period 
CHAPTER 1 :  Inttoduction 
N 
rennmru •• � ________ __ 
domain 
I 35 1 1 -14 101  
II 35 10- 14 101  
III 35 8 101  
IV 35 10  101  
Rod domain 
1 6  
1 7  
16  
17,22 
page 1 1  
c 
__
__
__
__
__ �.� rennmru 
domain 
19  8 121  
19  8 121  
19  8 12 1  
19  8 121  
Table 1 - 1  Comparison of the lengths of structural domains within the rod domain of IF proteins. 
of 47 nm (Fraser et aI, 1976) and measurements of rotary shadowed IF molecules 
from a variety of sources show that the length of the molecule is about 45-50 nm 
(Steinert, 1 981; Geisler et aI, 1982a, 1985b; Quinlan et aI, 1984; Ip et aI, 1985; see 
ruso Steven et aI, 1989) and this is consistent with a partially extended structure of the 
link regions in a linear rod domain of the type described above. 
The regular coiled-coil structure of the rod domain is correlated with the regular 
distribution of apolar residues in a heptad substructure - about 75% of the residues in 
the .it and d positions of the heptad are apolar (Parry and Fraser, 1985). Aggregation of 
pairs of coiled-coil chains into a parallel, in-register molecule is largely due to the basic 
and acidic residues (lysine, arginine and aspartic and glutamic acids) which are also 
distributed in a non-random manner within the rod domain. The � position of the 
heptad has a net basic character while the � position is net acidic and hence ionic 
interactions (ie, salt bridges) can be made between residues in the � and � positions of 
adjacent chains in the molecule. This implies a parallel rather than antiparallel 
arrangement of chains (see Figure 1-4). Ionic interactions are maximised for an in­
register arrangement of parallel chains. 
In addition to regularities involving the � and � positions, highly significant long range 
periods have been found in the linear distributions of the acidic and basic residues in 
the coiled-coil domains (Parry et aI, 1977; Parry and Fraser, 1985). In segment 1B 
CHAPTER 1: Introduction page 12 
(a) 
(b) 
...... 
(c) 
...... 
Figure 1-4 The heptad pattern in coiled-coils can be represented as (a-h-ki1-�-f-i.>n where a and d 
are typically apolar. The coiled-coil structure is stabilized by the knob-hole packing of these apolar 
residues, as shown in (a) where the a and d positions joined by dashed lines are axially staggered 
relative to each other to optimize the meshing of the hydrophobic sidechains. Oppositely charged 
residues in the � and g positions are also able to interact by fonning salt bridges (ionic interactions) 
and, in doing so, specify both the parallel orientations of the chains and the in-register alignment. 
Two stutters are apparent in the heptad substructure of the rod domains. Continuation of the 
undistorted coiled-coil geometry through these points are shown in (b) for the stutter near the midpoint 
of segment 2B and in (c) for the stutter within the link, L2. Figure adapted from Parry and Fraser 
(l985). 
(101 residues) the periods for the acidic and the basic residues are both ",,9.6 residues 
(=1.42 nm) and in segment 2 (=2A-L2-2B) they are ::=9.8 residues (=1.46 nm). The 
periods of the acidic and basic residues differ in phase by about 1800 in each case. 
Despite the heptad stutters in segment 2, the period in the charged residues is 
continuous throughout this section of the rod domain and indicates that segment 2 is a 
single structural unit (Parry, 1989). The slight, but significant difference in periods 
between segments I B  and 2 could prevent the aggregation of these segments via ionic 
interactions: only I B- I B  and 2-2 combinations might be possible. However, 
supercoiling of segments IB and 2 about each other might permit an alignment of the 
CHAPTER 1 :  Introduction page 13 
axial periods of the charged residues and allow IB-2 aggregation to be achieved (Parry 
and Fraser, 1985). 
The possible arrangements of a pair of two-stranded coiled-coils to form the four-chain 
structural unit (tetramer) was investigated by Crewther et al (1983). Only alignments 
that maximised the overlap of the two major coiled-coil domains, segments IB and 2, 
were considered as these provided the maximum opportunity for ionic interactions to 
be made between the molecules. Five classes of arrangements were considered 
possible: parallel molecules either in-register or approximately half-staggered, or anti­
parallel molecules that are either completely overlapped or half-staggered. The 
favoured model was a half-staggered, antiparallel arrangement with the N-tenninal 
segments of the rod overlapped by about 28 nm giving a combined length (excluding 
the terminal domains) of about 60 nm. Support for this model was found in the tryptic 
digestion study of Woods and Inglis ( 1984) on hard a-keratin in which two types of 
helical particles were found: a four-chain fragment from the N-terminal segment of the 
rod domains (previously described by Woods and Oruen, 1 98 1  and Oruen and 
Woods, 1983) and a heterodimeric fragment from the C-terminal segment of the rod 
domains. Since no fragment with both N- and C-terminal segments was found, the 
data strongly implied an antiparallel arrangement rather than a parallel one. (Note that 
an in-register, parallel arrangement of coiled-coils dimers is possible in theory but 
would not give rise to a filament structure). Further evidence for the antiparallel 
arrangement of dimers has come from limited chymotryptic digestion experiments on 
reduced carboxymethylated wool IF which resulted in covalently linked N-terminal 
segments of the rod domain (Sparrow et ai, 1989). Cross-linking peptides were 
isolated and characterized and their locations were compatible only with the dimers 
being approximately half-staggered and antiparallel. 
Support for a half-staggered arrangement of molecules is provided by the electron 
microscopy studies of Steven et al (1989) who visualized particles with a variety of 
lengths: 20-25 nm rods; 45-50 nm rods, some of which were kinked near their centres; 
and 70-80 nm rods which were kinked about one-third along from one end. All of 
these structures are compatible with the -20 nm dimensions of the major coiled-coil 
domains in the IF chain. In addition they show at least two modes of molecular 
aggregation: fully overlapped and half-overlapped. 
The conformations of the terminal domains are currently unknown. The enzyme 
cleavage studies on epidermal keratin IF by Steinert et al ( 1983a, described above) 
indicate that at least part of the terminal domains are located on the outside of the 
filament. Solid state Nuclear Magnetic Resonance (NMR) studies on epidermal keratin 
CHAYI'ER 1: Introduction page 14 
IF (Mack et ai, 1988; Steinert et ai, 1989) and prekeratin IF (Steven et ai, 1 989) 
revealed little order in the structures of the N- and C-terminal domains but showed 
high flexibility about the ubiquitous glycyl peptide bonds, and by implication, of the 
polypeptide backbones in these domains. Properties of the VI and V2 domain s  bear 
some similarities to those of the proposed a-loop (Zhou et ai, 1988; S teven et ai, 
1989) but as yet there is no direct evidence for this element of secondary structure in 
IF chains. 
The reported diameter of the IF has varied from the earlier estimates of 7 nm (electron 
microscopy, Birbeck and Mercer, 1957; Rogers, 1959; X-ray diffraction, Fraser and 
MacRae, 1959; Fraser et ai, 1959, 1973) to 14-1 5  nm (Steven et ai, 1982, 1 985; 
Steven, 1989) and depends to some extent on the criteria used for determining the 
filament's 'edge'. Protofilaments of about 2 nm diameter were described by Fraser 
et al ( 1962) and a '9+2' arrangement of protofilaments, reminiscent in part of the 
organization of tubules in cilia and flagella, was suggested for the filament by Filshie 
and Rogers (1961 )  on the basis of evidence from thin-section electron rnicroscopy. 
Subsequent evidence (for example, Fraser et ai, 1972) suggested instead a ring-core 
substructure. Scanning transmission electron rnicroscopy (STEM) measurements of 
the radial density of unstained freeze-dried IF showed that the filament was composed 
of a uniform density-core of diameter 9- 10 nm surrounded by a diffuse periphery that 
extended to 15- 16  nm (Steinert et ai, 1983b; Steven et ai, 1985, 1989). The ratios of 
core mass to peripheral mass for vimentin and an epidermal keratin were similar to 
those for the rod domain to terminal domain masses, implying that the tenninal 
domains are largely on the external surface of the filament. Mass-per-Iength 
measurements of native unstained filaments of vimentin using STEM have also 
indicated that IF consist of polymorphs: the density of the major component IF was 
consistent with 16 molecules in cross-section (=32 chains) and a minor component 
contained about 1 1  molecules (Steven et ai, 1982). The lower mass variant was 
suggested to be be a breakdown product or an immature filament (Steven ,  1 989). 
Reconstituted keratin IF also showed several mass variants and in each case the axial 
density of the filament scaled with the average density of its constituent subunits 
(Steven et ai, 1983a,b). Neurofilaments reconstituted in vitro also contained 1 6  
molecules in cross-section although polymorphs with fewer or greater numbers of 
chains were present to a lesser degree (Troncoso et ai, 1989). 
X-ray diffraction studies on well-ordered hard keratin tissue have revealed an axial 
period of 47 nm in which there are seven or eight quasi-equivalent units on a helix of 
pitch 22 nm (Fraser and MacRae, 1973a, 1983, 1985; Fraser et ai, 1985,  1986).  A 
discontinuous surface lattice model has been generated from the X-ray diffraction data, 
CHAPTER 1:  Introduction page 15 
the STEM mass data and ionic interaction calculations (Fraser et ai, 1985, 1986, 
1988). Each lattice point is associated with a single tetramer (two antiparallel molecules 
approximately half-staggered). Some studies have suggested that the IF structure is 
organized by increasing levels of aggregation of subfilaments: a single molecule forms 
the 2 nm-diameter protofilament; pairs of these protofilaments form a 4.5 nm-diameter 
protofibril (the tetramer); and protofibrils aggregate in groups of four (typically) as 
variants of the intact IF (Aebi et ai, 1983; see also Steinert and Roop, 1988). 
1 . 4  Conservation of Amino Acid and Gene Sequences 
A comparison of the amino acid sequences of the rod domains of IF chains reveals 
several features common to all IF. Two regions of very highly conserved sequence 
occur in the last 4-5 heptads at the C-terminal end of segment 2B (Geisler and Weber, 
1982; Hanukoglu and Fuchs, 1983) and in heptads 3-4 of segment lA (Weber and 
Geisler, 1982, 1984; Steinen et ai, 1984a; Parry and Fraser, 1985; Steinert and Parry, 
1985). In contrast, the linker segment L1 is variable in length and shows little or no 
homology amongst the various IF chains. Segment Ll2  is more regular in length 
(generally 1 6  or 17 residues depending on the chain type) and in homology (Crewther 
et ai, 1983). The secondary structure of these two link segments is not predicted to be 
a-helical and they do not contain the heptad substructure or regular distribution of 
ionic residues apparent in the coiled-coil regions. The third linker segment, L2, is 
absolutely conserved in length (8 residues) and shows high homology across the IF 
chains. In addition, the regularities in the distributions of ionic residues are maintained 
across L2 although the heptad substructure is disturbed somewhat by the irregular step 
between consecutive apolar residues at that point (see below). 
The heptad substructure of the coiled-coil undergoes a conserved stutter (ie, the heptad 
sequence i!-Q-£-d-�-f-� is broken by the insertion or deletion of several residues) near 
the midpoint of segment 2B (an insertion of four residues: Geisler et ai, 1982a; 
Dowling et ai, 1983; Steinert et ai, 1985c). This general feature has been observed in 
the coiled-coil sequences of all a-fibrous proteins except tropomyosin (see Cohen and 
Parry, 1989). The consequences of this disruption of the regular heptad structure are 
unclear although possible distortions of the coiled-coil have been theorized (McLachlan 
and Karn, 1983; Parry and Fraser, 1985). The effect of this panicular stutter, 
however, is to cause the internal i! and d positions to be rotated by 3600n (=54°) 
relative to the axis of the a-helix (see Figure 1 -4b). Assuming  that the coiled-coil is 
undistorted at some distance remote from the stutter on both sides, the stutter may be 
accommodated by a gradual change in the pitch length of the coiled-coil over an 
extensive region (McLachlan and Karn, 1983). Alternatively, there may be a highly 
CHAPTER 1: Introduction page 16 
localized discontinuity in structure which could result in a kink in the axis of the 
coiled-coil (Parry and Fraser, 1985). A similar feature occurs in the link region, L2, 
where an insertion of five residues (or deletion of two) occurs and as a result the il and 
Q positions are rotated by :::::154° (Figure 1-4c). The requirement for such distortion of 
the regular coiled-coil structure is not known but its importance is reflected in the 
absolute conservation of these stutters in all IF protein sequences. 
Regularities in the dispositions of acidic and basic residues in the rod domain of IF 
proteins have already been described: the charged stripes evident in the � and g 
positions of the heptad specify in part the alignment of chains to form the dimeric IF 
molecule and the next level of structure appears to be determined by the 9.6-residue 
period of the charged residues. It is possible that higher levels of IF structure are also 
dependent to some degree on other regularities in the positioning of these residues. 
The self-assembly of IF in vitro, and possibly in vivo also, must be some function of 
the protein sequences. The dispositions of acidic and basic residues, as well as the 
apolar residues, appear to play an important part in this process. 
Conserved features are also apparent in the nucleotide sequences of the genes  that 
encode IF proteins. In particular, the positions of six introns are generally conserved 
within the rod domain (Quax et ai, 1983, 1985; Lehnen et ai, 1984; Marchuk et ai, 
1984; Krieg et ai, 1985; Rieger et ai, 1985; Tyner et ai, 1985; Lewis and Cowan, 
1986). Desmin, vimentin and GF AP have, in addition to these six introns, two others 
at common positions within the C-terminal domain and several keratin protein genes 
also exhibit one of these (Marchuk et ai, 1984; Tyner et ai, 1985; Lewis and Cowan, 
1986). 
An interesting exception to this pattern is the bovine cytokeratin 19 protein (Bader 
et ai, 1986) which was found to be lacking the intron near the junction of the C­
terminal and rod domains. This protein is also unusual in that the C-terrninal domain is 
small (9 residues in length) and effectively constitutes a continuation of the heptad 
substructure from the rod domain: apparently it is more an extension to the rod than a 
distinct non-a-helical domain. The corresponding human cytokeratin 19 protein 
(Ecken, 1988) also has the same intron structure and the same number of amino acids. 
Furthermore, the two protein sequences are 89% identical. 
Other exceptions to the general IF intron pattern are the neurofilament protein s  which 
have only two introns within the rod domain, both of which are conserved among  the 
NF proteins but which do not correspond to intron positions in the other IF proteins 
(NF-L: Lewis and Cowan, 1986; NF-M: Myers et ai, 1987; NF-H: Lees et ai, 1988). 
Genes for mouse and human NF-L proteins show 90% homology within the exons, 
CHAYrER 1: Introduction page 17 
conservation of the positions and sizes of the introns but considerable differences in 
the nucleotide sequences of the introns (Julien et ai, 1987). 
The purpose of introns has not been established and their positions do not seem to 
coincide with structural domains or functional properties of the proteins in any regular 
way. They do, however, allow some conclusions to be drawn on the evolution of the 
genetic material of which they are a part. The conservation of introns among a protein 
family expressed in such diverse cell types is an important indication of a common 
ancestor gene. In addition, the difference between the intron positions for the types 
I-Ill and type IV groups provides evidence on how long ago they diverged (see, for 
example, Steinert and Roop, 1988). 
1 .  5 Lamin, Peripherin and the Helix A and B proteins 
The nuclear envelope is a double membrane composed of two lipid bilayers separated 
by approximately 20 to 40 nm. Between the external and internal membranes is a 
region known as the perinuclear space. The outer nuclear membrane is continuous 
with the membrane of the endoplasmic reticulum, as is the perinuclear space with the 
lumen of the endoplasmic reticulum. The ribosomes that 'stud' the rough endoplasmic 
reticulum are also apparent on the outer nuclear membrane. The inner and outer nuclear 
membranes are connected at the nuclear pores which allow exchange of material 
between the cytoplasm and nucleus. 
A complex layer of filaments lining the nucleoplasmic surface of the inner nuclear 
membrane has been observed in invertebrate cells, and a similar layer of filaments, 15-
20 nm thick, has also been seen in ultra-thin sections of vertebrate cells (Fawcett, 
1966; Scheer et ai, 1 976). The diameters of the filaments, called the nuclear lamina, 
have been reported as 5- 10 nm (Scheer et ai, 1976). The native lamin meshwork of 
Xenopus oocytes was shown to be an orthogonal network with sides of about 52 nm 
(Aebi et ai, 1986). This filamentous mesh work comprises a scaffolding associated 
with the pore complexes, the inner nuclear membrane and interphase chromatin .  It also 
appears to assist in the structural organization of the nucleus. Gerace et al ( 1 978) 
showed that the nuclear lamina from rat liver cells was comprised mainly of three 
proteins (termed lamins A, B and C by Gerace and Blobel, 1 980) with molecular 
weights of 60-70 kDa and which were located at the periphery of the normal interphase 
nucleus. In addition, lamin B was found to be significantly different from lamins A 
and C on the basis of peptide mapping and immunological studies (Gerace and Blobel, 
1980; Shelton et ai, 1980). However, coincident with the disassembly of the nuclear 
envelope during prophase, antigens to the three nuclear lamin proteins became 
distributed evenly throughout the cell until telophase, at which stage they became 
CHAPTER 1 :  Introduction page 18 
localized around the daughter chromosomes. It was proposed that the lamina was 
reversibly disassembled concomitant with disintegration of the nuclear envelope during 
mitosis. 
The nuclear lamina was shown to be interposed between the nuclear envelope and the 
chromatin (Fawcett, 1966). Lamin B was more resistant to extraction from membranes 
than lamins A and C (Gerace and Blobel, 1982) which suggests that it has a special 
membrane binding role and indeed a lamin B receptor protein has been described 
recently in turkey erythrocytes (Worman et aI, 1988) and in yeast cells (Georgatos 
et ai, 1989). During mitosis, lamins A and C become diffuse throughout the 
cytoplasm but lamin B remains associated with membranes (Gerace and Blobel, 1980; 
Burke and Gerace, 1986). It has been suggested that nuclear membrane fragments may 
be 'labelled' by lamin B for use in re-establishment of the envelope (see Stick et aI, 
1988). A study on chicken nuclear lamin proteins, lamins A, B l and B2, also found 
that lamin A was dispersed throughout the cytoplasm during mitosis and that lamins 
B l and B2 were membrane-bound and hence were considered functional analogues of 
the mammalian lamin B (Stick et aI, 1988). In addition, lamin B2 was found to be 
associated with the endoplasmic reticulum and, during the mitotic disassembly of the 
nuclear envelope, the lamin B2 proteins may become reversibly concentrated in the 
endoplasmic reticulum. Characterization of the different biochemical and structural 
features of the A and B lamins has been described recently by Peter et al ( 1989). 
The integrity of the lamin mesh work is thought to be regulated by the degree of 
phosphorylation of the lamin proteins (Gerace and Blobel, 1 980). S ites for 
phosphorylation in keratin are specific serine and threonine residues in the terminal 
domains (Steinen, 1988). Such residues are abundant in the C-terminal domain of the 
lamin molecules: for example, examination of the protein sequences shows that this 
domain of the human lamins A and C contains 44 and 23 serine residues respectively 
and 20 and 14 threonine residues respectively. However, newly synthesized lamin 
proteins require some alternative to phosphorylation in order to avoid the possibility of 
premature oligomerization while in the cytoplasm. A higher molecular mass precursor 
(7 1 kDa) to lamin A (68 kDa) was studied by Lehner et al ( 1986a) who suggested that 
the precursor form could allow migration of lamin A protein into the nucleus where the 
precursor would be processed into the mature protein and integrated into the lamina 
mesh work in some unspecified manner. Lehner et al ( 1986a) also identified two 
variants of the lamin B protein: a mammalian form, lamin B 1, and an avian form, 
lamin B2. An apparent higher molecular mass precursor for lamin B2 was also 
synthesized but in vitro translation produced equal amounts of the two fonns of the 
protein. The reason for this is unclear. No precursor was found for lamin B 1 .  (No 
CHAPTER 1 :  Introduction page 19 
investigation was made into precursors for lamin C. However, if the lamin proteins 
copolymerize, a non-aggregating fonn of one protein could be sufficient to prevent 
ftlament fonnation). 
Zackroff et al (1984) isolated 'keratin-like' proteins located near the periphery of the 
nucleus and suggested a relationship with the nuclear lamin proteins. This was 
confirmed in further studies by Goldman et al (1986) who also proposed that the 
nuclear lamins formed an IF-like network in the nuclear lamina and that this was 
connected with the cytoplasmic IF network, possibly through the nuclear pores or by 
some trans-membrane mechanism. The possibility of the filamentous networks being 
unrelated was discounted for two reasons: the component proteins were related 
biochemically and the cytoplasmic IF network was known to be closely associated 
with the nuclear surface as well as the plasma membrane. The protein sequences for 
human lamins A and C were determined by McKeon et al (1986) and Fisher et al 
(1986), who showed conclusively that the nuclear lamins were indeed IF-type 
proteins .  Amphibian lamin proteins from Xenopus laevis have been described and 
named lamins A, L., LIl, Lm, LIV (for review, see Wolin et ai, 1987). Invertebrate 
lamins from the surf clam and Drosophilia have also been reported as have minor 
lamins (Lehner et ai, 1986b). 
The primary structures of human lamins A and C bear a remarkable resemblance to 
those of the other IF proteins and common structural properties have been postulated 
for both lamins and IF (Gerace, 1985; McKeon et ai, 1986; Fisher et ai, 1986). 
McKeon et al (1986) demonstrated that human lamins A and C shared the general 
structural plan of IF chains - an a-helix-rich rod domain bracketed by largely non­
a-helical N- and C-terminal domains - and that the rod domain of lamins may be 
subdivided into segments in much the same manner as for IF. In addition, the human 
A and C lamins were seen to be identical from the N-terminus through the rod domain 
and into the C-terminal domain and McKeon et al (1986) suggested that both proteins 
arose from the same gene by differential processing - this may be related to the 
unusual repeat of four histidine residues after which the two polypeptide sequences 
diverge. 
An important difference noted between the rod domains of IF and lamin was a six­
heptad insert (ie, 42 residues) in segment IB of the lamin chains. Weber (1986) also 
noted that the start of the insert occurred at the position of a conserved intron (see also 
Steinert et ai, 1985b). This extension is predicted to result in a length of -52 nm for 
the rod domain of lamin molecules. Dimers of lamins A and C from rat liver reveal a 
-52 nm rod with two globular heads at one end (Aebi et ai, 1986), presumably the 
CHAPTER 1 :  Introduction page 20 
lA lB 2A 2B 
N C 
/ Ll L12 L2 Stutter 
/ 
/ 
N C 
Segment 1 Segment 2 
Figure 1-5 Schematic representations of the chain structures for IF types I-IV (top) and lamin. 
Unlcer segments Ll and Ll2 in typeS I-IV IF are not predicted to be a-helical in structure, as indicated 
by boxes wider than the shaded coiled-coil regions (top). In the type V (lamin) chain, however, the 
entire rod domain is predicted to be a-helical (bottom). Segment 1B  is extended by a 42-residue 
insertion and linlcer segment L l  has been replaced by a stutter in the heptad substructure.Figure 
adapted from Parry et al (1986). 
larger C-tenninal domains. Dimers of lamin B were similar in form although typically 
only one globular head was apparent. Other points noted by Parry et al (1986) in their 
analysis of the sequence data were (i) the 42-residue insert is part of a 70-residue 
piece in human lamins A and C that replaces residues 43-70 inclusive in segment 1B of 
IF chains (numbering from the amino-terminal end of segment 1B), (ii) segment L12 
is predicted to be a-helical in lamin chains but non-a-helical in other IF chains, 
(iii) segment L1 in lamin has a heptad substructure and is also predicted to be a-helical 
and indeed it can be reduced to a single stutter in the heptad phasing for segment 1, 
and (iv) unlike IF type I-IV chains, lamins do not appear to fonn long filamentous 
structures in vitro but instead fonn tactoids or paracrystalline arrays which have 
alternating dark- and light-staining bands when negatively stained and observed by 
electron microscopy (Zackroff et ai, 1984; Goldman et ai, 1986). The resultant model 
for the lamin chain is shown in Figure 1-5 alongside that for IF chains (Parry et ai, 
1986). The rod domain of human lamins A and C is predicted to be entirely a-helical 
with a heptad substructure present throughout segment 1 and in segments 2A and 2B. 
In addition, the ionic and heptad regularities in the rod domain were found to be 
maintained across the linker region L12. Parry et al (1986) suggested that the lamins 
be tenned type V IF proteins since they showed a high degree of homology with IF 
overall but did not fall naturally into the type I-IV classification previously established. 
Indeed, the novel extension to segment 1B and the intranuclear location of these 
proteins together imply that the lamins are fundamentally different  from the 
cytoplasmic IF proteins. 
�Rl: Introduction page 21  
An interesting parallel to the description of the lamin protein structure is  the recent 
characterization of other proteins that have the same longer 1B segment seen first in the 
lamin A and C proteins. Weber et al ( 1988) have sequenced two epithelial proteins 
from the snail Helix pomotia of molecular weights 66 kDa (Helix A) and 52 kDa 
(Helix B). Both form monocomponent IF in vitro indicating that they are not 
analogous to the heteropolymeric keratin IF despite their occurrence in a near equal 
molar ratios in epithelial tissue. Although the homology in structure and sequence 
between the nuclear lamin and Helix proteins might have been taken as evidence for a 
similar location and function, this is not supponed by Weber et al ( 1988) who note 
that a karyophilic motif in the C-terminal domains of the nuclear lamin proteins (the 
sequence Lys-Lys-Arg-Lys-Leu-Glu, Fisher et ai, 1986) is not apparent in the 
sequence of the Helix proteins. However, the departures from the type I-IV IF 
structure described above for human lamin A and C are shared also by the Helix B 
protein. No other IF proteins have demonstrated this variation to the structure of the 
rod domain and so it is appropriate to designate the Helix B protein a type V IF chain. 
The sequence of the Helix A reponed by Weber et al ( 1988) was incomplete and no 
assignment is yet possible for this chain. 
Studies on a neuronal protein of molecular weight -58 kDa by Liem et al ( 1 978), 
Portier et al ( 1984a,b), Franke et al (1986) and Parysek and Goldman ( 1987) had 
shown that it was related to the IF family but was not one of the neurofilament triplet 
proteins, NF-L, NF-M or NF-H. This protein was termed peripherin by Portier et al 
( 1984a). The sequence (Leonard et ai, 1988; Parysek et al , 1988) was shown to 
have greatest homology with the type III IF sequences, desmin and vimentin. 
Although its expression in neuronal tissue might suggest that peripherin is a type IV IF 
protein, the gene structure (Thompson and Ziff, 1989) follows the pattern for the type 
I-In IF proteins (six conserved introns in the rod domain) rather that for the 
neurofilament proteins (two conserved introns in the rod domain) and peripherin is 
therefore conclusively not a type IV IF protein. 
The high content of serine residues in the N-terminal domain was described as being 
"novel" (Leonard et ai, 1988). However the peripherin protein is only average in this 
respect amongst the type ID IF proteins which have 23 (hamster desmin), 25 (hamster 
vimentin) , 7 (mouse GFAP) and 24 (peripherin) serines in this domain (sequences as 
cited by Leonard et ai, 1988). Indeed, it is only the GFAP protein that is unusual in 
having a lower serine content in its N-terminal domain. The serine residues are 
potential sites for phosphorylation, a process which appears to have imponance as a 
regulatory mechanism for the polymerization of IF proteins in general (see later). 
Phosphorylated peripherin, however, is found almost exclusively in the insoluble 
CHAPTER 1 :  Introduction page 22 
cytoskeleton (Aletta et al, 1989), indicating that it may still be an integral part of the 
filament network. Aletta et al (1989) suggest that this is consistent with 
phosphorylation-promoted assembly, an opposite effect to that reported for other IF 
proteins. Vimentin (lnagaki et ai, 1987) and desmin (Geisler and Weber, 1988), for 
example, have a reduced ability to polymerize when phosphorylated and the nuclear 
lamin network disassembles concomitant with phosphorylation of the lamin proteins 
(Gerace and B10bel, 1980). Aletta et al ( 1989) also report that non-detectably 
phosphorylated peripherin is present in the cytoskeleton and so no finn conclusion can 
be drawn from these data as to the effect of phosphorylation on the in situ proteins. 
A truly novel feature of the peripherin protein is its expression in neuronal tissue, 
previously thought to be the exclusive domain of the neurofilament proteins. Leonard 
et al ( 1988) reported that the distributions of the NF proteins and peripherin mRNA 
within the nervous system are different, although there is some overlap, and that the 
regions which express peripherin are all evolutionarily old. The term "peripherin" 
originated from the location of this protein in neurons peripheral to the nervous system 
(Portier et ai, 1984a) although it is also expressed in certain of the central nervous 
system neurons (Leonard et ai, 1988). 
1 . 6  Function of Intermediate Filaments 
Intermediate filaments form a highly insoluble skeleton that extends throughout the 
cytoplasm and is also found in the nucleus. The functions of the IF networks are 
unclear as are the mechanisms by which they are controlled and their associations with 
other elements of the cytoplasm. Their Ubiquity and underlying homology imply some 
fundamental importance for cell physiology that transcends species and cell types, 
although some cultured cell lines apparently grow normally in the absence of IF 
(Venetianer et aI, 1983). The nuclear lamin proteins appear to be especially 
widespread in eukaryotic cells: lamin A and B analogues have been reported in yeast 
cells (Georgatos et aI, 1989) in addition to mammals, amphibia, insects and birds. 
The fIrst steps towards understanding the role of IF have been made by determining 
the anchorage sites of some IF proteins to membranes and it has become increasingly 
apparent that IF constitute some form of connection between the cell surface and the 
nucleus. 
Vimentin IF were shown to be associated with the cell membrane, possibly via head­
on interactions between the non-a-helical amino-terminal domain and at least one 
membrane-anchoring protein, ankyrin (Georgatos et al, 1985; Georgatos and 
Marchesi, 1985). Studies on desmin IF located the plasma membrane binding site to 
within the amino-terminal domain (Georgatos et ai, 1987). Georgatos et al ( 1 985) 
�R l: lntroduction page 23 
showed that ankyrin inhibited IF formation of vimentin beyond a protofilament stage 
(probably the tetrameric subunit) and, since the in vitro assembly process for IF 
appears to require intact amino-terminal domains (vimentin: Traub and Vorgias, 1983; 
desmin: Kaufmann et ai, 1985), this suggests that ankyrin may serve in vivo as a 
"capping site" that effectively terminates the filament at the membrane (Georgatos 
et ai, 1987; Steinert and Roop, 1988). An alternative attachment method was 
suggested by Georgatos et al (1985) whereby the filaments bind side-on to the 
membrane via short branches and loop back into the cytoplasm. However, the 
branches might themselves abut the membrane in an end-on fashion. In addition to 
vimentin and desmin, keratin IF also appear to attach to the cell periphery, possibly to 
desmoplakin proteins which are found in complex regions of cell-to-cell adhesion 
called desmosomes (see, for example, Goldman and Dessev, 1989; Green et ai, 
1989). 
IF are also localized around the exterior of the nuclear surface. In vitro binding of 
desmin to lamin B has been described (Georgatos et ai, 1987) and this association 
apparently involves a portion of the carboxy-terminal domain of desmin adjacent to the 
rod domain. Vimentin has also been observed to associate with lamin B in a similar 
manner to desmin (Georgatos and Blobel, 1987). Indeed, lamin B has been suggested 
as a nucleating centre for IF (see, for example, Gerace and Burke, 1988) since it is 
clear that the ankyrin-mediated attachment to the plasma membrane is not capable of 
performing such a role (Georgatos et ai, 1985). The method by which lamin B might 
interact with IF proteins through the nuclear envelope, if indeed it does so at all, is 
unclear. A connection through the nuclear pores is possible as is a trans-membrane 
interaction mediated by some membrane process. Georgatos and Blobel (1987) have 
suggested that IF networks may be directly anchored to the nuclear lamina at distinct 
locations coinciding with the nuclear pores. 
A feature of the direct connection model for IF to membrane-binding proteins is that a 
certain amount of local polarity is implied for the IF: the amino-terminal domains of the 
IF chains link to the plasma membrane whereas the carboxy-terminals bind at the 
nuclear membrane. However, the structures reconstituted in vitro show no axial 
polarity (see, for example, the GFAP paracrystals of Stewart et ai, 1989a,b and 
Quinlan et ai, 1989). If the filaments are axially apolar (ie, equal numbers of 
oppositely directed chains), why should the amino- and carboxy-terminal domains of 
the molecule associate with separate membranes in the cell? What role then is played 
by the amino- and carboxy-terminal domains in the main body of the filaments? Do 
they bind to some other membranes in the cytoplasmic space? 
CHAPTER 1: Introduction page 24 
In addition to the locale of IF attachment, recent work has examined the role of 
phosphorylation as a post-synthetic modifier of IF structure. Phosphorylation of the 
nuclear lamina has already been described as has the general oligamerization-blocking 
effect of phosphorylation on vimentin and desmin (above) and evidence is now 
accumulating that the cytoplasmic IF networks also undergo structural changes as a 
result of phosphorylation. Major sites for phosphorylation have been detennined for 
the human epidermal keratin 1 protein as well as the turnover rates of phosphate 
isomers of this protein (Steinert, 1988). Sites in the more mobile parts of the tenninal 
domains showed the highest rate of turnover and the rate dropped in sites closer to the 
relatively stiff rod domain. This may be due in part to the accessibility of different 
regions of the chain to external phosphorylating agents. As Steinert ( 1988) points out, 
there is no indication why the keratin IF examined should undergo post-synthetic 
disassembly and reassembly and so the purpose of this phosphorylation remains 
unclear. Other cellular activities may cause the proximity of phosphorylating agents to 
the IF network leading to opportunistic interactions. 
Disassembly of IF networks during mitosis has been studied recently by Chou et al 
( 1989). The amount and sites of phosphorylation on the IF chains (desmin and 
vimentin) were stated to be different during mitosis than during interphase and 
complete disassembly of the in vitro IF network by phosphorylation was described. 
These results indicate that phosphorylation is a regulator of IF structure: 
phosphorylation can disassemble IF networks (and block oligamerization of the 
subunits: Geisler and Weber, 1988) and conversely dephosphorylation will allow self­
assembly of the molecules into IF. 
An interesting development has been the recent discovery of lamins A and B in yeast 
(Georgatos et ai, 1989). These cells divide by closed mitosis where the nuclear 
envelope is not dismantled as in higher eukaryotic cells but instead develops a 
cytoplasmic extension that 'buds off' as a daughter nucleus with a separate membrane. 
There is no apparent requirement for the disassembly/reassembly of the nuclear lamina 
and the role of the lamina must be directed entirely towards maintaining the structural 
integrity of the nuclear envelope (as described above). An investigation into the 
primary structure of such lamins would be of use in identifying regions of the chain 
than are involved in specific functions. Also, the consequence of hyper­
phosphorylation of the yeast lamins, which presumably does not take place d uring 
closed mitosis, would be of interest. 
CHAPTER 1: Introduction 
1 . 7  Structural Form of the Thesis 
page 25 
The detailed structure of IF cannot be detennined from the currently available X-ray 
diffraction data or from electron microscopy studies. However, there are some 
attributes of their conformations that may be elucidated by non-physical techniques. 
The studies described in this thesis are a collection of such indirect methods which are 
nonetheless evaluated on the basis of their ability to conform with such physical data 
as are available. Protein sequence data are the raw material on which these studies are 
based: the ever increasing body of sequences available allows new insights to be made 
into the structural hierarchy of the IF. 
Periodicities are examined in the linear distribution of charged residues within the rod 
domains of a type ID IF protein (peripherin) and of various type V IF chains (the 
nuclear lamins and the Helix pomotia B protein). It is important to establish whether 
differences occur between these proteins and other IF chains, especially with regard to 
the extended IB segment characteristic of the type V chains. 
Studies on the primary sequences of IF in general include investigation of residue 
occupancy in the heptad substructure, calculation of chain flexibilities (Chapter 2) and 
quantification of the degree of homology present amongst the IF proteins (Chapter 3). 
For the latter work, amino acid sequences of the rod domains of IF chains are 
compared in order to describe more accurately the features that are common to IF 
chains as well as to distinguish sub-groups. Consensus sequences for the sUbtypes 
and for all IF chains are derived and, in addition, homology among residues in the 
heptad positions of the coiled-coil are determined. The current data base is sufficiently 
large to allow some interesting conclusions to be drawn that aid in further elucidation 
of IF structure and function. 
The packing of IF chains into the molecule and of the molecules into larger scale 
structures is also studied (Chapter 4). Prediction schemes based on the calculation of 
potential ionic interactions enables the most favourable axial alignments to be found. 
The simple methods employed in investigating the alignment of molecules are 
expanded to include more of the three-dimensional information in the molecular 
structure. 
Models for the aggregation of lamin molecules based on electron microscopy data and 
ionic interaction studies are constructed. These are used to explain some of the more 
unusual features of the nuclear lamina as well as indicating how phosphorylation can 
result in depolymerization of the lamina filaments. 
CHAPTER 2: Primary Structure page 26 
2. PRIMARY STRUCTURE 
The number of IF protein sequences available has increased dramatically over the 
seven or so years since the first full sequence - chicken gizzard desmin - was 
determined by Geisler and Weber (1982). At the present time over 40 sequences have 
been fully or partially completed. Examination of these sequences has yielded valuable 
clues to the likely molecular and filament structure of the proteins. For example, 
secondary structure can be predicted fairly reliably for fibrous proteins where 
extensive regions of fixed conformation exist. On this basis, a model for the structure 
of the IF molecule has been proposed in which there is a central rod-like domain with 
a regular coiled-coil structure that is interrupted by well conserved breaks. Higher 
levels of organization, however, are less easy to determine due to the complexity of 
the information encoded in the protein sequences. Regularities in the disposition of 
residues having a certain character (for example, the mutually attractive acidic and 
basic residues) may be indicators of special modes of aggregation of specific IF 
chains. The likelihood of this feature arising purely by chance is greatly reduced where 
these regularities are shared by many other protein chains .  The characterization of 
regular patterns in the distribution of residues is not sufficient to deduce the packing 
order of the protein chains but does indicate that regular packing of the proteins does 
occur. A list of the proteins used in various studies in this work are given in Table 
2- 1 .  Several other IF protein sequences have been published recently but are not 
included in any of the analyses detailed in this thesis. These include the human 
epidermal type II keratin K5 (Lersch and Fuchs, 1988) and the rat NF-H protein 
(Dautigny et ai, 1988). 
In this Chapter, regularities in the dispositions of the charged residues are examined 
for the rod domains of the newly characterized proteins: the human and Xenopus 
lamin proteins, the Helix pomotia B protein and peripherin. Fourier transforms of the 
distributions of these residues have already been undertaken on a number of type I to 
IV IF proteins and these are compared with the new data presented here. Secondly, 
this study looks at the distribution of residues in the heptad substructure and how it 
relates to the predicted structures of the coiled-coil segments in the rod domain. 
Finally, the flexibility profiles of representatives of each of the classes of IF proteins 
are calculated with a view to comparing known structural features of the chains with 
their predicted mobility. 
CHAPTER 2: Primary Structure 
Abbrev. Full name 
B40K Cow (Bovine) 40K 
B50K Cow (Bovine) 50K 
B54K Cow (Bovine) 54K 
H46K Human 46K 
H50K Human 50K 
H56.5K Human 56.5K 
M47K Mouse 47K 
MSOK Mouse 50K 
M50K Mouse M pkSCC 50K 
MS2K Mouse M pkSCC 52K 
M55K Mouse 
MS9K Mouse 
8a Sheep Component 8a 
8c- l Sheep Component 8c- l 
XL51 Xenopus 
XL70 Xenopus 
XL81 Xenopus 
Table 2-1 IF amino acid sequences. 
page 27 
Type I 
Source 
Bader et al ( 1986) 
Jorcano et al ( 1984) 
Jorcano et al ( 1984), Rieger et al 
(1985) 
Raychaudury et al ( 1986) 
Hanukoglu and Fuchs (1982), 
Marchuk et al ( 1984, 1985) 
Steinert et al (private communication) 
Singer et al (1986) 
Steinert and Roop (private 
communication) 
Knapp et al (1987) 
Knapp et al ( 1987) 
Steinert and Roop (private 
communication) 
Krieg et al ( 1985), Steinert et al 
(1983a) 
Crewther et al ( 1985) 
Crewther et al (1983), Dowling et al 
( 1983, 1986) 
Hoffmann and Franz ( 1984) 
Winkles et al ( 1985) 
Jonas et al ( 1985) 
CHAPTER 2: Primary Structure 
Type 11 
Abbrev. Full name 
H55K Human 55K 
H56K Human 56K 
H67K Human 67K 
M60K Mouse 60K 
M67K Mouse 67K 
5 Sheep Component 5 
7c Sheep Component 7c 
7x Sheep Component 7x (unnamed) 
XL64 23 Xenopus XL 64 (pUF23) 
XL64 164 Xenopus XL 64 (pUFI64) 
Type TII 
Abbrev. Full name 
COD Chicken Gizzard Desmin 
HD Hamster Desmin 
PSD Pig Stomach Desmin 
CV Chicken Vimentin 
HELV Hamster Eye Lens Vimentin 
PELV Pig Eye Lens Vimentin 
MGFAP Mouse Glial Fibrillary Acidic 
Protein 
PGFAP Pig Glial Fibrillary Acidic Protein 
RP Rat Peripherin 
Table 2-1 continued 
page 28 
Source 
Glass et al (1985) 
Hanukoglu and Fuchs (1983), Tyner 
et al (1985) 
Steinert et al (1985a) 
Steinert et al ( 1984a) 
Steinert et al ( 1985a) 
Crewther et al ( 1985) 
Sparrow and Inglis ( 1980), Crewther 
et al (1983), Dowling et al (1983), 
Rogers ( 1984) 
Powell and Rogers (private 
communication) 
Hoffmann et al ( 1985) 
Hoffmann et al ( 1985) 
Source 
Geisler and Weber (1982, 1983), 
Geisler et al ( 1982a) 
Quax et al ( 1984) 
Geisler and Weber (1981),  Geisler 
et al ( 1982b) 
Zehner and Paterson ( 1985) 
Quax et al ( 1983), Quax-Jeuken et al 
(1983) 
Geisler and Weber (1981), Geisler 
et al ( 1982b) 
Lewis et al ( 1984) 
Geisler and Weber ( 1982, 1983) 
Leonard et al (1988), Parysek et al 
(1988), Thompson and Ziff ( 1989) 
CHAPI'ER 2: Primary Structure 
Type IV 
Abbrev. Full name 
HNF-L Human Neurofilament Light 
MNF-L Mouse Neurofilament Light 
PNF-L Pig Neurofilament Light 
RNF-L Rat Neurofilament Light 
HNF-M Human Neurofilament Medium 
MNF-M Mouse Neurofilament Medium 
PNF-M Pig Neurofilament Medium 
RNF-M Rat Neurofilament Medium 
HNF-H Human Neurofilament Heavy 
MNF-H Mouse Neurofilament Heavy 
PNF-H Pig Neurofilament Heavy 
Type V 
Abbrev. Full name 
HLA Human Lamin A 
XLA Xenopus Lamin A 
XLB Xenopus Lamin B 
H1x B  Helix pomotia B 
Table 2-1 continued 
2 . 1  Fourier Analysis 
page 29 
Source 
lullen et al ( 1987) 
Lewis and Cow an ( 1985, 1986) 
Geisler et al ( 1982b, 1983, 1985c) 
I ulien et al ( 1985) 
Myers et al ( 1987) 
Levy et al ( 1987) 
Geisler et al ( 1984) 
Napolitano et al ( 1987) 
Lees et al ( 1988) 
Shneidman et al ( 1988) 
Geisler et al ( 1985a) 
Source 
McKeon et al ( 1986), Fisher et al 
( 1986) 
Krohne et al ( 1987) 
Wolin et al ( 1987) 
Weber et al ( 1988) 
This section will be concerned with an investigation of the regularities in the linear 
disposition of the acidic and the basic residues in the amino acid sequences comprising 
the major heptad-containing segments of human lamins A and e, Xenopus lamins A 
and B, Helix pomotia B protein and peripherin (see Table 2- 1 for sources) . Each of 
these proteins manifests some unique extension to the basic model for IF protein 
structure or location : the lamin proteins are special amongst IF in that they occur 
within the cell nucleus (other than during mitosis); the type III peripherin protein 
occurs in neuronal tissue, previously thought to be the exclusive domain of the type IV 
neurofIlament proteins; the lamin and Helix pomotia proteins have a novel extension to 
the I B  segment of the rod domain that would allow the coiled-coil structure to be 
extended, however unlike lamin the Helix pomotia proteins do not appear to be located 
within the cell nucleus (Weber et ai , 1988). 
Regularities in amino acid sequences are associated either with the structural or the 
CHAPTER 2: Primary Structure page 30 
functional features of proteins, and their evaluation gives important clues in model 
building studies. The heptad substructure characteristic of regions of coiled-coil 
conformation, for example, is largely defined by the apolar residues spaced at intervals 
of three and four residues successively (described in more detail in Chapter 1) .  In 
addition, the charged residues are often found in specific positions within the heptad 
which facilitate either interchain or intennolecular interactions. Patterns such as these 
involving the � and g heptad positions (defined in Chapter 1) allow pairs of such 
coiled-coils to pack together in an optimal way (the 'knob-into-hole' packing 
postulated by Crick, 1952), shielding hydrophobic residues from the aqueous 
environment and bringing oppositely charged residues from both chains into close 
proximity where they can form attractive electrostatic interactions. IF proteins express 
these sequence features clearly and together they are largely responsible for the self­
assembly of chains into the dimeric molecule. Other regular periods are also evident in 
IF protein sequences and these are thought to direct higher levels of aggregation from 
the molecule to the tetrameric unit, and the tetrameric unit to the 10 nm filament. Parry 
and Fraser (1985) have investigated the rod domain sequences of a number of IF 
proteins for regularities in the placement of acidic and basic residues. They described a 
feature common to the major coiled-coil domain sequences of all IF proteins in  the 
type I to IV categories: segment IB shows a -9.54 residue period for these residues 
and segment 2 has a slightly longer period at -9.84 residues. In addition, the periods 
of the acidic and basi� residues were found to lie approximately 1 800 out of phase 
indicating that charged residues were grouped along the chain in clumps of alternating 
charge. 
2 . 1 . 1  Fourier Analysis - Method 
Fourier transform techniques applied to primary sequence data from proteins have 
been described by McLachlan and Stewart (1976) and are used here after minor 
modification. The original method is as follows: residues of interest in the amino acid 
sequence (for example, acidic or basic residues) are represented by unity and all other 
residues are replaced by zero. The data set is then zero-filled to improve resolution of 
the resulting Fourier data and also to meet the modulo 2 size constraints imposed on 
the data set by the Fast Fourier Transform (FFT) algorithm used. Protein sequences 
studied have lengths of either 101  residues (segment IB  for IF types I-IV) , 1 43 
residues (segment IB for IF type V) or 148 residues (segment 2 for all IF types) and 
all are zero-filled to 2048. The discrete Fourier transform D(f) of the zero-filled data 
set d(x) is then calculated: 
N 
D(f) = F{d(x)} = �Jd(X).exPc2�")tb: 
CHAPTER 2: Primary Structure page 31  
The variable x is  the residue position in the linear sequence and f is  its analogue in  the 
Fourier domain. The Fourier intensities are scaled by 
where Nr = size of the unfilled residue data set 
er = count of residues of interest 
10 = Fourier intensity at zero frequency 
The scaled Fourier intensities will thus be given by 
FI{d(x)} = Nr er 1.. ID(f) 12 
Nr - er Io 
where ID(012 are the Fourier intensities of D(t). 
The method outlined above is modified by the introduction of an extra operation prior 
to zero-filling: the linear sequence is baseline corrected, ie, the average value of the 
sequence is subtracted from all data points so that the new average is zero. This simple 
procedure avoids the large zero-frequency spike and its associated side lobes that are 
otherwise evident in the Fourier transforms. A comparison between the original and 
the baseline correction methods is shown in Figure 2- 1 together with the difference 
between the transforms. The large zero-frequency peak (Figure 2- 1b) results from the 
non-zero average value of the data set and would not generally be considered a 
problem. However, in combination with the zero-filling operation, the large spike is  
spread by a significant amount through the Fourier domain and interferes to  varying 
degrees with other peaks as well as producing 'spurious' peaks at low frequencies 
(see Appendix A). These 'spurious' peaks, or side lobes, are only apparent in the 
Fourier data when the central peak is relatively large (the ratio of the central lobe of the 
sinc function to the first side lobe is approximately 20: 1 ). This is generally the case for 
the zero-frequency peak if the baseline is not corrected. As can be seen from Figure 
2- 1c, the effect of the baseline correction is small beyond the zero-frequency sinc 
function. 
The scaling factor described above from McLachlan and Stewart ( 1976) makes use of 
the zero-frequency intensity 10 produced by their method. However, the baseline 
correction method results in 10 being zero and so an alternative derivation for the 
scaling factor must be used. The value of 10 for the original method is also equal to the 
square of the average value of the digitized sequence prior to the baseline correction 
and may be used in place of Io in the calculation of the scaling factor. 
The relationship between frequency in Fourier space and period in the residue 
sequence is: 
'od Nf pen -- frequency where Nf = size of the fIlled data set. 
CHAFfER 2: Primary Structure page 32 
Ca) Peripherin, Segment 1B: 
8 FT of acidic residues - baseline corrected 
>. 
... 
.... en 
6 c 0 
... 
� 
"0 4 0 
cu u Cl) 
2 
0 
Cb) Peripherin, Segment IB 
8 FT of acidic residues - uncorrected >. 
... 
.... en C 6 £ 
� 
"B 4 cu u Cl) 
2 
0 
Cc) Difference between the corrected and uncorrected transforms 
4 
>. 
.... 
.... en C 2 0 
.... 
� 
"B 0 cu u Cl) 
-2 
-4 
0 256 512  768 1024 
Frequency 
Figure 2-1 Comparison of the scaled Fourier intensities of (a) the baseline-corrected and (b) raw 
data. The differences between the Fourier intensities of (a) and (b) are also shown (c). 
CHAPTER 2: Primary SbUCture page 33 
2 . 1 . 2  Fourier Analysis • Results 
The scaled Fourier intensities for acidic and basic residues in rod domain segments of 
peripherin. Xenopus lamins A and B and the Helix pomotia B protein are shown in 
Appendix B, Figures B-1  to B-5, and a selection of the peaks are listed in Tables B-1 
to B-5.  The transform for human lamin A, which has been analysed previously by 
Parry et aI  ( 1986), has been included to provide a comparison with the other 
sequences showing an extended 1B segment. 
The peripherin data display strong peaks corresponding to a 9.6-9.9 residue period in 
both the acidic and the basic residues in segments 1B  and 2. In the case of the 1B  
segment, however, the period for the acidic residues appears to have merged with a 
nearby peak of slightly longer period to produce a single, wider peak (Appendix B, 
Fi.gure B-1 ). Comparison with the results obtained for some other type III proteins 
(chicken gizzard desmin, hamster eye lens vimentin and mouse GFAP - Parry and 
Fraser, 1985) shows a high degree of similarity (Table 2-2) and provides further 
support for the assignment of peripherin as a type ill IF protein. The acidic residues in 
segments 1B  also reveal the second and third orders of the heptad period (3.47 and 
2.33 residues respectively) which is to be expected since charged residues are located 
predominantly in the h. £, k' f and � positions of the heptad but rarely in positions � 
and g. The heptad repeat is interrupted near the middle of segment 2B and also in link 
segment L2 and consequently no significant orders of this period are observed in the 
transforms of either the acidic or basic residues in segment 2. This is generally true of 
all short-range (but not necessarily long-range) periods in sequences containing heptad 
stutters in otherwise regular patterns. 
Peripherin Other Type ID IF 
Period Intensity Period Intensity 
Segment 1B 
Acidics - - 9.63 ± 0.21 5.62 ± 1 .4 1  
Basics 9.62 5.64 9.62 ± 0.05 4.39 ± 0.40 
Segment 2 
Acidics 9.75 7.39 9.76 ± 0.05 4.40 ± 2. 10  
Basics 9.89 7.36 9.82 ± 0.02 6.05 ± 0.34 
Table 2-2 Comparison of the dominant period in the linear distribution of the acidic and the 
basic residues for segments IB and 2 of rat peripherin and some other type III IF proteins (Parry and 
Fraser, 1985). The -9.6 residue period for the acidic residues in segment IB ofperipberin bas merged 
with a nearby peak to produce a single broad peak (Appendix B, Figure B-1) and so no exact value can 
be determined for it 
CHAYI'ER 2: Primary Structure page 34 
Xenopus lamin A and B and Helix pomotia B proteins share the extended IB segment 
which was first observed in the human lamin A and C proteins. The Fourier 
transforms of charged residues from the rod domain segments of these proteins are 
shown in Figures B-3 and B-4 with the human lamin A protein providing a basis for 
comparison (Figure B-2; also Parry et ai, 1986). The similarities between the 
transforms of the human lamin A and Xenopus lamin A proteins are striking. 
Examination of the dispositions of the charged residues in the rod segments of the two 
proteins shows that they are nearly identical although some difference does occur in 
the distribution of the acidic residues in segment 2. Extensive homology is apparent 
throughout the primary sequences (Krohne et ai, 1987): the subdomains of the rod are 
identical in length and 83% of the residues are identical within the rod with a further 
7% conserved in character (see Figure 2-2). 
An interesting feature of the transforms of human lamin A that was not noted 
previously (Parry et ai, 1986) is an apparent -19.86 period which can be inferred 
from peaks corresponding to periods of 9.94 residues (::= 19.86+2) and 6.61 residues 
(.::: 19.86+3). This pair is particularly strong for the acidic residues in segment 1 and in 
the rod domain as a whole but is notably absent from the basic residues in segment 2. 
The Xenopus lamin A protein also shows this feature but not the Xenopus lamin B 
protein. No evidence for a -19.86 period in the distributions of charged residues has 
been reported for other IF proteins in either of the two major rod domain segments 
(Parry and Fraser, 1985). The significance (or otherwise) of this observation i s  not 
apparent. 
Several of the transforms of charged residues in both human and Xenopus lamin A 
show peaks corresponding to orders of the heptad substructure (7 , 3.5, and 2.33 
residue periods). In addition, the basic residues of segment 2 and the rod domain as a 
whole reveal two strong peaks corresponding to the first two orders of a -5.06 residue 
repeat. The segment 1 transforms also show strong peaks corresponding to periods of 
3.10 residues (acidics) and 3.02 residues (basics). Neither the -5.06 or 3. 10  and 3.02 
periods falls on any orders of the -19.86 repeat (19.86+4=4.97 and 1 9.86+6=3.33) 
and no explanation of the significance of these periods is currently possible. 
Xenopus lamin B does not reveal a -19.86 repeat but shows the 9- 10 residue repeat 
common to all IF. In the acidic residues this period is 9.85 residues and in the basic 
residues it varies from 9.23 residues (segment 1 )  to 9.85 (segment 2) . The 3 .03 
residue repeat apparent in the segment 1 basics of the lamin A proteins is also strongly 
represented in Xenopus lamin B and falls close to the third order of the 9 .23 residue 
repeat (9.23+3=3.08). 
CHAPTER 2: Primary Structure page 35 
m..A:l ETPSQRRATR SGAQASSTPL SPTRITRLQE KEDLQELNDR LAVYIDRVRS 
XLA: ETPGQKRATR S----THTPL SPTRITRLQE KEDLQGLNDR LAVYIDKVRS 
• 
• • • 
lll..A:5 1 LETENAGLRL RITESEEVVS REVSGlKAAY EAELGDARKT LDSVAKERAR 
XLA: LELENARLRL RITESEDVIS REVTGIKSAY ETELADARKT LDSVAKERAR 
• 
• 
• 
lll..A: I01 LQLELSKVRE EFKELKARNT KKEGDLlAAQ ARLKDLEALL NSKEAALSTA 
XLA: LQLELSKlRE EHKELKARNA KKESDLLTAQ ARLKDLEALL NSKDAALTTA 
• • • 
lll..A: 15 1  LSEKRTLEGE LHDLRGQVAK LEAALGEAKK QLQDEMLRRV DAENRLQTMK 
XLA: LGEKRNLENE lRELKAHIAK LEASLADTKK QLQDEMLRRV DTENRNQTLK 
• 
lll..A:201 EELDFQKNIY SEELRETKRR HETRLVEIDN GKQREFESRL ADALQELRAQ 
XLA: EELEFQKSIY NEEMRETKRR HETRLVEVDN GRQREFESKL ADALHELRAQ 
lll..A:251 HEDQVEQYKK ELEKTYSAKL DNARQSAERN SNLVGAAHEE LQQSRIRIDS 
XLA: HEGQIGLYKE ELGKTYNAKL ENAKQSAERN SSLVGEAQEE IQQSRIRIDS 
• • 
lll..A:301 LSAQLSQLQK QLAAKEAKLR DLEDSLARER DTSRRLLAEK EREMAEMRAR 
XLA: LSAQLSQLQK QLAAREAKLR DLEDAYARER DSSRRLLADK DREMAEMRAR 
• 
lll..A:35 1 MQQQLDEYQE LLDIKLALDM ElHAYRKLLE GEEERLRLSP SPTSQRSRGR 
XLA: MQQQLDEYQE LLDIKLALDM EINAYRKLLE GEEERLRLSP SPNTQKRSAR 
lll..A:401 ASSHSSQTQG GGSVTKKRKL ESTESRSS-F SQHARTSGRV AVEEVDEEGK 
XLA: TIASHSGAHI SSSASKRRRL EEGESRSSSF TQHARTTGKV SVEEVDPEGK 
lll..A:451 FVRLRNKSNE DQSMGNWQIK RQNGDDPLLT YRFPPKFTLK AGQVVTIWAA 
XLA: YVRLRNKSNE DQSLGNWQIK RQIGDETPIV YKFPPRLTLK AGQTVTIWAS 
lll..A:501 GAGATHSPPT DLVWKAQNTW GCGNSLRTAL INSTGEEVAM RKLVRSVTVV 
XLA: GAGATNSPPS DLVWKAQSSW GTGDSIRTAL LTS SNEEVAM RKLVRTVVIN 
lll..A:551 EDDEDEDGDD LLHHHHGSHC S----SSGDP AEYNLRSRTV LCGTCGQPAD 
XLA: DEDDEDNDDM EHHHHHHHHH HDGQNSSGDP GEYNLRSRTI VCTSCGRPAE 
lll..A:601 KASASGSGAQ VGGPISSGSS ASSVTVTRSY RSVGG-SGGG SFGDNLVTRS 
XLA: KSVLASQGSG LVTG-SSGSS SSSVTLTRTY RSTGGTSGGS GLGESPVTRN 
lll..A:65 1 YLLGNSSPRT QSPQNCSIM 
XLA: FIVGNGQRAQ VAPQNCSIM 
Figure 2-2 Protein sequences for human lam in A (HLA) and Xenopus lam in A (XLA) - for 
sources, see Table 2-1 .  The heptad-containing regions of the rod domain are indicated above the 
sequences by a line. Several blanks have been inserted to maintain optimal alignment of the 
sequences. The internal a and d positions of the heptad are marked above the sequences by '.' . 
CHAPTER 2: Primary Structure page 36 
The Helix pomotia B protein reveals a very strong peak corresponding to a 9.23 
residue repeat in the disposition of basic residues in the rod domain (Figure B-5). This 
is slightly less than the 9.31  and 9.35 residue periods of segments 1 and 2 
respectively. The acidic residues of the rod domain also show a dominant peak 
corresponding to a 9.27 residue period. Another strongly represented period in the 
basic residues is a 12. 19 residue repeat which is apparent in both segment 1 and the 
whole rod domain. 
The dominant 9.23-9.35 residue repeats reported above for the Xenopus lamin B and 
Helix B proteins are similar to the period resolved in the acidic residues of segment IB  
of the type I keratins (9.28±o.02: Parry and Fraser, 1985). These period s  are 
significantly less than the 9.5- 10  residue periods found in all other transfonns 
undertaken here and by Parry and Fraser (1985). Weber et al (1988) located the Helix 
pomotia A and B proteins in cells that also expressed keratin IF and suggested that the 
A and B proteins were not localized in the nucleus as are the structurally similar lamin 
proteins. If the Helix B protein coexists in the cytoplasm with keratin, then it is 
possible that the common .... 9.23 period in charged residues may allow co aggregation 
although there is no evidence yet to support this idea. Similarly, the lamin B protein, 
although sited within the nuclear envelope, is postulated as a nuclear anchorage site for 
the cytoplasmic IF network (Gerace and Burke, 1988). It may be of significanc e  that 
the -9.25 residue period is present in the distribution of acidic residues of segment IB  
in keratins and in the basic residues of segment 1 i n  the Xenopus lamin B protein .  
The six transforms for each of the type V proteins have been multiplied together to 
highlight periods that are commonly represented in each. The results are summarized 
in Table 2-3. A low intensity period of 1 1 .64 residues is revealed for the human and 
Xenopus lamin A proteins in addition to the dominant 9.94 residue period already 
discussed in the human and Xenopus lamins. A similar operation on the four 
Human lamin A Xenopus lamin A Xenopus lamin B Helix pomotia B 
1 1 .64 (452) 1 1 .64 ( 136) 12. 19  ( 166) 
10. 14  (228) 
9.94 (2938) 9.94 (1702) 9.89 (3 198) 9.25 ( 1 1932) 
2.77 ( 1 17) 2.84 ( 1 1 3) 
2.52 (142) 2.54 (103) 
2.47 (290) 
2.022 (100) 2.022 (320) 
Table 2-3 Periods corresponding to major peaks (ie, scaled intensities greater than 100) resulting 
from multiplying the six Fourier transforms for each protein together. The intensities are shown in 
brackets. 
CHAPfER 2: Primary Structure page 37 
transforms of peripherin yields a single peak (intensity 1 1 30) corresponding to a 
repeat of 9.85 residues. 
2 . 2  Residue Distribution in the Heptad 
The seven residue quasi-repeat, or heptad substructure, that is found in major portions 
of the primary sequences of a-fibrous proteins is fundamental to the coiled-coil 
structure. Apolar residues are commonly found at spacings of three and four residues 
successively and give rise to a hydrophobic stripe that winds in a left-handed sense 
around the axis of a right-handed a-helix. Pairs of a-helices suffer a small distortion 
that allows the stripes to be aligned along the interior axis of a supercoil thus shielding 
them from the aqueous environment. The apolar residues interlock in a 'knob-into­
hole' packing that provides much stability to the coiled-coil structure (see Chapter 1 ) .  
These internal positions of the coiled-coil in IF are, on average, occupied at a rate of 
75% by apolar residues (Parry and Fraser, 1985). The adjacent � and g positions are 
frequently occupied by charged residues: in IF proteins the � positions are more 
commonly basic, or positively charged and the g positions are often acidic, or 
negatively charged. This arrangement specifies the relative alignment of a-helices prior 
to the precise docking along the apolar stripes: the oppositely charged � and g stripes 
from different a-helices are adjacent when the chains are parallel. 
The distribution of residues in the major heptad-containing portions of IF protein 
sequences is presented in Table 2-4a. The data reveal that (i) acidic residues are very 
uncommon in position a and relatively uncommon in d, and that such residues occur 
with approximately equal frequency in positions 12, £., � and g, (ii) the basic residues 
occur very rarely in d though they may occur in a, and are favoured in � relative to g, 
and (iii) the apolar residues are most highly maintained in positions a and d, are found 
comparably in positions �, f and g and are least common in positions 12 and £. Table 
2-4a shows other points of interest, some of which have been noted previously 
(Parry, 1982; Parry and Fraser, 1985). For example, certain residues occur with 
greatest frequency in the outer positions of the coiled-coil Ch, £ and f); these include 
glycine (80%), serine (76%), aspartic acid (73%), cysteine (63%), alanine (61 %) and 
asparagine (56%). In general, such residues (excluding aspartic acid) are not normally 
associated with the ability to specify intennolecular association. 
Residues occurring with greatest frequency in the innennost � and g positions are 
tyrosine (86%), leucine (74%), isoleucine (72%), tryptophan (70%), valine (52%), 
methionine (50%) and histidine (48%). In some cases there is  an asymmetric 
distribution between the a and d positions thus emphasizing the structural uniqueness 
of each position. Examples include isoleucine �-81%;  d.- 19%), valine �-64% ;  
Residue count Percentage Occurence 
� b � d � f g Totals i! 11 � d � f g 
Ala 76 263 171  123 66 200 143 1042 Ala 4.8 17.0 10.4 7.5 4.0 12.2 8.9 
Cys 9 0 23 8 13  34  3 90 Cys 0.6 0.0 1 .4 0.5 0.8 2.1 0.2 
Asp 5 148 198 2 66 154 108 681 Asp 0.3 9.5 12. 1  0. 1 4.0 9.4 6.7 
GIu 9 294 282 1 18 348 200 382 1633 GIu 0.6 19.0 17.2 7.2 2 1 .2 12.2 23.9 
Phe 60 1 47 79 13 0 4 204 Phe 3.8 0. 1 2.9 4.8 0.8 0.0 0.2 
Gly 16  38 121 17  20 78 6 296 GIy 1 .0 2.5 7.4 1 .0 1 .2 4.8 0.4 
His 27 17 24 42 3 1 8  1 2  143 His 1 .7 1 . 1  1 .5 2.6 0.2 1 . 1  0.7 
ne 334 14 12 80 55 51 3 1  577 ne 20.9 0.9 0.7 4.9 3.4 3.1  1 .9 
Lys 95 79 50 8 226 1 16  1 8 1  755 Lys 6.0 5. 1 3 .1  0.5 13.8 7.1 1 1 .3 
Leu 406 41 33 717 1 17 86 1 12 1512 Leu 25.5 2.6 2.0 43.6 7 .1  5.2 7.0 
Met 76 31  8 48 34 30 23 250 Met 4.8 2.0 0.5 2.9 2. 1 1 .8 1 .4 
Asn 63 69 1 1 1  3 1  101 1 17 35 527 Asn 3.9 4.5 6.8 1 .9 6.2 7 . 1  2.2 
Pro 2 9 3 0 0 0 1 15  Pro 0. 1 0.6 0.2 0.0 0.0 0.0 0. 1 
Gin 8 230 91 43 1 1 5  121 228 836 Gin 0.5 14.8 5.6 2.6 7.0 7.4 14.2 
Arg 86 137 142 9 262 167 1 14 917 Arg 5.4 8.8 8.7 0.5 16.0 10.2 7.1  
Ser 24 95 193 19  54 150 44 579 Ser 1 .5 6. 1 1 1 .8 1 .2 3.3 9. 1 2.7 
Thr 15  51 81 37 104 61 95 444 Thr 0.9 3.3 4.9 2.3 6.3 3.7 5.9 
Val 147 30 30 81  29 5 1  67 435 Val 9.2 1 .9 1 .8 4.9 1 .8 3 .1  4.2 
Trp 6 0 0 8 0 6 0 20 Trp 0.4 0.0 0.0 0.5 0.0 0.4 0.0 
Tyr 1 3 1  3 19  174 15  2 12 356 Tyr 8.2 0.2 1 .2 10.6 0.9 0.1 0.7 
Totals 1595 1550 1639 1644 1641  1642 1601 1 13 12  Totals 100.0 100.0 100.0 100.0 100.0 100.0 100.0 
Apolar 1 154 120 149 1 1 79 263 220 249 3334 Apolar 72.4 7.7 9. 1 7 1 .7 16.0 13.4 15.6 
Basic 18 1  216 192 17 488 283 295 1672 Basic 1 1 .3 13.9 1 1 .7 1 .0 29.7 17.2 18.4 
Acidic 14  442 480 120 414 354 490 23 14 Acidic 0.9 28.5 29.3 7.3 25.2 21 .6 30.6 
Table 2-4a The distribution of residues in the heptad-containing regions of IF proteins combined. The symbols H represent residue positions 
within the heptad. Large apolar (phe, lIe, Leu, Met, Val, Trp, Tyr), basic (Lys, Arg) and acidic (Asp, GIu) residues are summarized at the bottom 
of the table. 
Residue count Percentage Occurence 
a b c d e f SI Totals a b � d e f SI 
Ala 32 36 22 75 1 7  36 35 253 Ala 10.2 1 1 .9 7.0 24.0 5.4 1 1 .5 1 1 .2 
Cys 2 0 0 1 0 0 0 3 Cys 0.6 0.0 0.0 0.3 0.0 0.0 0.0 
Asp 0 42 45 3 14  27 23 154 Asp 0.0 13.9 14.2 1 .0 4.5 8.6 7.3 
Glu 1 63 69 21 91 43 60 348 GIu 0.3 20.8 2 1 .8 6.7 29.1 13 .7 1 9.2 
Phe 6 2 1 5 1 2 0 17 Phe 1 .9 0.7 0.3 1 .6 0.3 0.6 0.0 
Gly 2 6 8 4 6 14 4 44 Gly 0.6 2.0 2.5 1 .3 1 .9 4.5 1 .3 
His 4 12  5 3 4 10 3 41  His 1 .3 4.0 1 .6 1 .0 1 .3 3.2 1 .0 
TIe 35 3 6 16 8 5 7 80 TIe 1 1 . 1  1 .0 1 .9 5 . 1  2.6 1 .6 2.2 
Lys 3 1  50 48 3 20 37 36 225 Lys 9.9 16.5 15 .2 1 .0 6.4 1 1 .8 1 1 .5 
Leu 104 5 1 1  109 1 7  9 20 275 Leu 33 .1  1 .7 3 .5 34.8 5.4 2.9 6.4 
Met 8 2 2 8 2 3 0 25 Met 2.5 0.7 0.6 2.6 0.6 1 .0 0.0 
Asn 14 16  1 3  4 20 1 1  10 88 Asn 4.5 5.3 4.1 1 .3 6.4 3 .5 3.2 
Pro 0 0 0 0 0 0 0 0 Pro 0.0 0.0 0.0 0.0 0.0 0.0 0.0 
GIn 4 20 29 14  59  18  48 192 GIn 1 .3 6.6 9.2 4.5 18.8 5.8 15.3 
Arg 22 22 27 4 15  47 32 1 69 Arg 7.0 7.3 8.5 1 .3 4.8 15 .0 10.2 
Ser 7 10 19 6 14  27 16  99 Ser 2.2 3 .3 6.0 1 .9 4.5 8.6 5 . 1  
Thr 5 9 8 7 17  1 1  10  67 Thr 1 .6 3 .0 2.5 2.2 5.4 3 .5 3.2 
Val 3 1  3 3 18  8 12 9 84 Val 9.9 1 .0 0.9 5.8 2.6 3.8 2.9 
TIp 0 0 0 3 0 0 0 3 TIp 0.0 0.0 0.0 1 .0 0.0 0.0 0.0 
Tyr 6 2 0 9 0 1 0 18  Tyr 1 .9 0.7 0.0 2.9 0.0 0.3 0.0 
Totals 314  303 3 16  3 1 3  3 1 3  3 1 3  3 1 3  2185 Totals 100.0 100.0 100.0 100.0 100.0 100.0 100.0 
Apolar 190 17 23 165 36 32 36 499 Apolar 60.5 5.6 7.3 52.7 1 1 .5 10.2 1 1 .5 
Basic 53 72 75 7 35 84 68 394 Basic 16.9 23.8 23.7 2.2 1 1 .2 26.8 2 1 .7 
Acidic 1 105 1 14 24 105 70 83 502 Acidic 0.3 34.7 36. 1 7.7 33.5 22.4 26.5 
Table 2·4b The distribution of residues in the heptad-containing regions of pararnyosin, myosin and tropomyosin combined. Large apolar (Phe, 
TIe, Leu, Met, Val, TIp, Tyr), basic (Lys, Arg) and acidic (Asp, Glu) residues are summarized at the bottom of the table. 
Total Counts Average Percentage Occurance 
g 12 £ d � f � Totals .a 12 k d � f � 
A1a 108 299 193 198 83 236 178 1 295 Ala 7.5 14.4 8.7 15.7 4.7 " 1 1 .8 10. 1 
Cys 1 1  0 23 9 1 3  34 3 93 Cys 0.6 0.0 0.7 0.4 0.4 1 .0 0. 1 
Asp 5 190 243 5 80 181  131  835 Asp 0.2 1 1 .7 13.2 0.5 4.2 9.0 7.0 
Glu 10 357 351 1 39 439 243 442 198 1  Glu 0.4 19.9 19.5 6.9 25. 1  13 .0 21 .5 
Phe 66 3 48 84 14 2 4 221 Phe 2.8 0.4 1 .6 3.2 0.6 0.3 0. 1 
Gly 1 8  44 129 21 26 92 10 340 Gly 0.8 2.2 5.0 1 .2 1 .6 4.6 0.8 
His 3 1 29 29 45 7 28 15 184 His 1 .5 2.5 1 .5  1 .8 0.7 2 .1  0.9 
ne 369 17 18  96 63 56 38 657 ne 16.0 0.9 1 .3 5.0 3.0 2.4 2.1 
Lys 126 129 98 1 1  246 153 217 980 Lys 7.9 10.8 9. 1 0.7 10. 1 9.4 1 1 .4 
Leu 510 46 44 826 134 95 132 1787 Leu 29.3 2.1 2.7 39.2 6.3 4.1 6.7 
Met 84 33 10 56 36 33 23 275 Met 3.7 1 .3 0.6 2.7 1 .4 1 .4 0.7 
Asn 77 85 124 35 121 128 45 615 Asn 4.2 4.9 5.4 1 .6 6.3 5.3 2.7 
Pro 2 9 3 0 0 0 1 15  Pro 0. 1 0.3 0. 1 0.0 0.0 0.0 0.0 
GIn 1 2  250 120 57 174 139 276 1028 GIn 0.9 10.7 7.4 3.5 12.9 6.6 14.8 
Arg 108 159 169 13  277 214 146 1086 Arg 6.2 8.0 8.6 0.9 10.4 12.6 8.7 
Ser 3 1  105 212 25 68 177 60 678 Ser 1 .9 4.7 8.9 1.5 3.9 8.9 3.9 
Thr 20 60 89 44 121 72 105 5 1 1  Thr 1 .3 3 .1  3.7 2.2 5.9 3.6 4.6 
Val 178 33 33 99 37 63 76 519 Val 9.5 1 .5 1 .4 5.3 2.2 3.5 3.5 
Trp 6 0 0 1 1  0 6 0 23 Trp 0.2 0.0 0.0 0.7 0.0 0.2 0.0 
Tyr 137 5 19  183 1 5  3 12 374 Tyr 5. 1 0.4 0.6 6.7 0.5 0.2 0.4 
Totals 1909 1853 1 955 1957 1954 1955 1 914 1 3497 Totals 100.0 100.0 100.0 100.0 100.0 100.0 100.0 
Apolar 1344 137 172 1344 299 252 285 3833 Apolar 66.4 6.7 8.2 62.2 13.8 1 1 .8 13.5 
Basic 234 288 267 24 523 367 363 2066 Basic 14. 1  1 8.8 17.7 1.6 20.5 22.0 20. 1 
Acidic 15  547 594 144 519 424 573 2816  Acidic 0.6 3 1 .6 32.7 7.5 29.4 22.0 28.6 
Table 2-4c (Left) The distribution ofresidues in the heptad containing regions of the IF and myosin-type proteins used in Tables 2-1a and 2-1b. 
(Right) Average percentage distribution for the IF and myosin-type proteins used in Tables 2-1a  and 2-1b. Large apolar (phe. De. Leu. Met. Val. 
Trp. Tyr). basic (Lys. Arg) and acidic (Asp. Glu) residues are summarized at the bottom of the table. 
CHAPTER 2: Primary Structure page 41 
d-36%), methionine (a-61 %; £1-39%), histidine (a-39%; £1-6 1  %) and leucine (a-36%; 
d-64%). 
Another feature of Table 2-4a is the preponderance of leucine residues in the d position 
where leucine occurs four times as often as any other residue and makes up 44% of 
the total. Leucine and isoleucine contribute 26% and 21  % respectively to the residues 
found in the a position. The average occurrence of large apolar residues in the A and d 
positions is 72%, a figure in close agreement with that determined by Parry and Fraser 
(1985). Including the alanines in the apolar group raises this figure to 78%. 
The coiled-coil rod regions of several myosin-type sequences (myosin, paramyosin 
and tropomyosin) have also been analysed (Table 2-4b) for comparison with the IF 
proteins. The occurrence of large apolar residues in the a and d positions is not as 
great as for IF proteins (57%) but this figure becomes comparable when alanines are 
included (74%). Interestingly, acidic residues clearly dominate the � position for the 
myosin-type proteins (33.5% acidic: 1 1 .2% basic) but the & position for the IF 
proteins (29.8% : 18.6%). 
When the "myosins" and IF proteins are combined into a single table (Table 2-4c) the 
dominance of leucine is again apparent in the d position of the heptad and together 
with isoleucine it dominates the A position. Acidics and basics are now more evenly 
matched in � but acidics remain more common in &. It is interesting to note that some 
of the less common residues are excluded from certain positions within the heptad: 
cysteine is never found in 12; proline (an a-helix-disrupting residue) is never found in 
do � or f; and tryptophan is only found in 11, do and f. 
This study emphasizes the importance of apolar residues in specifying the heptad 
substructure and hence the stability of the coiled-coil. In particular, isoleucine and 
leucine are especially common in the a position of the heptad and leucine is dominant 
in the d position. Charged residues are found fairly evenly amongst the remaining, 
more external positions: those in the � and & positions are involved in specifying the 
relative orientations of the chains that form the dimeric molecule while the others are 
. on the exterior of the coiled-coil and specify higher orders of aggregation. 
2 . 3  Flex ibi l i ty 
Prediction schemes are used to characterize the attributes of protein chains where no 
direct method of measurement is readily available. These schemes generally apply a 
statistically weighted scoring system to a region of protein sequence in an effort to 
establish the potential for a given characteristic such as hydrophobicity (Rose, 1978; 
CHAPI'ER 2: Primary Structure page 42 
Kyte and Doolittle, 1982), antigenicity (Hopp et ai, 1981)  or the probability of 
adopting a particular element of secondary structure (Chou and Fasman, 1974; Gamier 
et ai, 1978). A scheme for calculating the relative axial stagger of coiled-coil proteins 
has met with particular success in predicting the stagger of collagen molecules (and 
also myosin and paramyosin) and is used in Chapter 4 in a modified form. 
Karplus and Schulz (1985) introduced a method for predicting the flexibility of protein 
chains by using temperature factor data deduced from refined crystallograpbic 
structures. Their prime aim was to provide an improved tool for selecting peptide 
antigens and cross-reacting peptides based on the link between segmental flexibility 
and antigenic determinacy (Westhof et ai, 1984; Tainer et aI, 1984). This method was 
used by Conway et al ( 1989) and Con way and Parry ( 1989) to compare the 
flexibilities of different IF chain types, and also the different segments of the IF 
chains. However, some care must be taken in interpreting the results from IF protein 
chains as the method of Karplus and Schulz is based on data collected from globular 
proteins - the rod domain of IF is clearly not globular although the N- and C-tenninal 
domains may be. 
The data base of Karplus and Schulz comprised the temperature factors of the Ca 
atoms of residues from 3 1  globular proteins. The averages and ranges of these factors 
were noted to vary between proteins and this was assumed to be a result of differences 
in structure refinement methods rather than natural variances. As a consequence, the 
temperature factors were normalized and the root-mean-square deviation was made 
constant The average normalized temperature factor was determined for each residue 
and the residues were grouped into two classes: average nonnalized temperature factor 
> 1  ( ,flexible') or <1  ('rigid'). Separate flexibility indices were determined from the 
average normalized temperature factors according to whether a residue had zero, one 
or two 'rigid' neighbours. These indices were used to generate a profile of chain 
flexibility which was then smoothed with a triangular weighting profile (ie weights 
1/ 16 , 2/16, 3/16, 4/16, 3/16, 2/16, 1/1 6). 
Conway et aI ( 1989) and Conway and Parry ( 1989) assumed that the same method 
could be applied to the IF chains even though they are not globular proteins. The 
weighting scheme was modified in order not to disguise effects arising from the 
relatively short 7 and 10  residue periods known to be present in the rod domain 
sequences (ie, the heptad substructure and the charged residue periods respectively) 
and the weights used were 1/4, 2/4, 1/4. Flexibility profiles from a selection of IF 
chains from all chain types are shown in Figure 2-3 and the average scores (± s.d.) for 
the chain segments are listed in Table 2-5. 
CHAPTER 2: Primary Structure 
� 1 .2 
Cl) 
� 1 .0 
:.s .� 
page 43 
ti: O.S � ID H3i 2B @ (a) Component Sc- l 
�---- �====��==��------������--� � 1 .2 
� 1 .0 
:E 
� O.S 
� 1 .2 
Cl) 
� 1 .0 
P-I< 
� ID H3i 2D @ (b) Component 7 
d: 0.8 � ID H3i 28 @ (c) Mouse M59K 
�--� �====��==��--�--������--� 
� 1 .2 8 Cl) 
� 1 .0 
:.E .� ti: 0.8 � ID H3i 2B @ (d) Mouse M67K 
�---- �====�====��--�--������--� � 1 .2 
Cl) 
>. � 1 .0 
:.E .� ti: 0.8 � IB H3i 2B @ (e) Chicken Gizzard Desmin �---- �====�====��--�--��---�--�--� 
� 1 .2 
(.) Cl) 
,q 1 .0 
:.E 
.>< d: 0.8 
� 1 .2 
Cl) 
� 1 .0 
:.E 
.>< d: 0.8 
� 1 .2 
(.) Cl) 
.q 1 .0 
:.E 
.>< 
� ID H3i 
� ID H3i 
2B @) 
2B @ 
(f) Pig NF-L 
(g) Human NF-M 
G: 0.8 0 lA I 18 H3l 2B @) (h) Human Lamin A L-
__ �========�====�� __ � __ ����� ____ � 
Figure 2-3 Flexibility profiles for a selection of IF chains. nle horizontal axis is divided into units 
of 1 00 residues. Scores above 1 .0 on the vertical axis indicate greater than average flexibility and 
scores below 1 .0 indicate more rigid regions of sequence. Sequences are aligned by the N-tenninus of 
segment 2A. 
Segment Type la Type ITa Type lb Type lIb Type ID Type IV-L Type IV_M Type V 
Comp. 8c- 1 Comp. 7 M59K M67K COD PNF-L HNF-M LaminA 
El and V I  ? ? 1 .064±O.082 1 .061±O.081 1 .037±O.074 - - -
HI ? ? 1 .06O±O.078 1 .03 1±O.071 0.995±O.080 1 .016iO.073 1 .044iO.083 1 .053±O.O61 
lA 1'()04iO.064 I .OO2iO.087 1 .000±0.O70 1 .009±0.078 0.982±O.074 0.989±O.077 0.998±O.080 1 .007±O.073 
LI  1 .062iO.107 1 .014iO'(>92 1 .072iO.070 1 .050i0.053 1 .064iO.074 1 .058±O.074 1 .003±<>.085 -
IB  1 .01 0i0.073 0.999±O.074 I .OO3iO.068 1 .012±<>.068 1 .001iO.068 1 .005±O.072 1 .000±0.07 1 1 .01 1±<>.067 
L12 0.990±0.049 1 .01 1iO.089 1 .002iO.068 1 .014iO.070 0.983iO.070 0.976±O.079 0.988±<>.074 1 .041iO.075 
2A 0.987iO.061 0.984±O.07 1 l .OO5iO.052 0.989±O.074 0.979±O.072 0.986±O.074 1 .038±<>.059 0.987±<>.050 
L2 1 .032±<>.061 1 .027±O.077 1 .046±O.061 1 .033±O.08 1 0.976±O.060 1 .002±<>.046 0.967±<>.073 1 .040i0.052 
2B 1 .003iO.074 1 .008iO.075 1 .024iO.069 1 .015iO.069 1 .00liO.070 1 .006±0.072 I .OO7iO.068 1 .007±O.071 
H2 ? ? - 1 .017iO.056 1 .015iO.OS5 1 .030±0.067 1 .02IiO.069 1 .025iO.083 
V2 and E2 ? ? 1 . 1 1  OiO.064 1 . 105iO.067 - 1 .079±O.044 1 .062±<>.066 1 .041±<>.081 
N 1 .003±O.OS4 I .OO2±<>.076 I .063iO.OS2 1 .05 1iO.079 1 .028iO.077 1 .016±O.073 1 .044±O.083 1 .053±O.061 
Rod I .OO7±O.072 I .OO4±O.077 1 .0I 5±O.069 1 .013iO.072 0.99S±O.070 1 .003±O.073 1 .003iO.071 1 .0 1 0i0. 069 
C 0.998±O.079 1 .000±0.074 1 . 1 1OiO.064 1.094±O.072 1 .015±O.085 1 .047±O.064 I .055iO.068 1 .025±O.082 
Table 2-5 Mean flexibility indices (± s.d.) for chain segments from all IF Types. Sources are shown in Table 3- 1 .  Abbreviations are as follows: 
Comp. 8c- I ,  Component Se- I ;  Comp. 7, Component 7; M59K, Mouse 59K; M67K, Mouse 67K; COD, Chicken Gizzard Desmin;  PNF-L, Pig 
neurofilament light chain; HNF-M, Human neurofilament medium chain; ?, extent of the segment has not been determined (see Chapter 3); -, 
segment is not present in the sequence. 
CHAPfER 2: Primary Structure page 45 
A striking feature of the flexibility profiles is the regular pattern shown in the terminal 
domains of the epidermal keratins M59K and M67K and also in pans of the 
C-terminal domain of human NF-M. The calculated flexibility values are particularly 
high for these 'soft' keratin chains in the VI and V2 subdomains and this is consistent 
with these regions being flexible, interactive and having an external location on the 
surface of the IF. Indeed a high degree of flexibility as well as strength is required in 
the epidermis which forms the barrier between an animal and its environment. The 
C-terminal domains of neurofilament chains are part of an extended bridge between IF 
and microtubules and may also be seen to require a degree of flexibility. 
The periodicities present in the flexibility proftles of the V 1 and V2 subdomains of the 
epidermal keratins include a 5-residue period and a quasi-halved 9-residue period in 
the V I  subdomain of M59K, a to-residue period in the V2 subdomain of M59K, a 
5-residue period in the VI su1:xlomain of M67K, and a 28-residue period in the V2 
subdomain ofM67K. These periods and others (see Steinert et aI, 1983a, 1985a) are 
in each case associated with a glycine-serine rich structural motif. The regular pattern 
observed in the middle of the C-terminal domain of human NF-M is due to a 
l 3-residue repeat (KSPVEEKGKSPVP). 
Average flexibility scores for the segments are shown in Table 2-5. Segments lA, lB,  
and 2B are typically close to unity as is  the rod domain as a whole. The HI  and H2 
domains are generally more flexible than the rod domains (with the exception of HI  in 
the chicken gizzard desmin sequence) and the End and Variable domains combined (ie 
E l  and V I ,  E2 and V2) are the most flexible. The hard a-keratin chains, however, 
. have mean flexibility indices close to unity over their entire N- and C-terminal domains 
and this may be an indication that these regions display a considerable degree of intra­
sUbtype homology (see also Crewther et aI, 1985, and Sparrow et aI,  1 989). 
Furthermore, their particular scores are comparable to those for the entire rod domains 
of all the chains studied and suggest that the terminal domains of the hard a-keratins 
are better defined structurally than most other IF chains. 
The rod domain segment predicted to have the highest mean flexibility is generally the 
link, Lt .  This is consistent with the variable length of this link which can differ even 
within a single chain type and which is actually non-existent in type V chains. Almost 
certainly the structure ofLl  is poorly defined. 
Segment 2A is predicted to have (marginally) the lowest flexibility while the adjacent 
link segment L2 is predicted to have the second highest within the rod domain (except 
for component 7 from wool a-keratin where it is the highest). Interestingly, the pig 
NF-L sequence shows the reverse of this trend: segment 2A has the highest mean 
CHAPTER 2: Primary Structure page 46 
flexibility score and segment L2 has the lowest. These segments are of particular 
interest as they represent the major portion of a region where the coiled-coil undergoes 
a rearrangement caused by the discontinuity, or stutter, in the heptad phasing that 
occurs in segment L2. This latter segment, although predicted to be a-helical (see, for 
example, Geisler and Weber, 1982; and Parry and Fraser, 1985), lacks the heptad 
repeat characteristic of coiled-coil conformations found in the a-fibrous protein class 
of structures. The flexibility scores indicate that both chains of the IF molecule are 
flexible in the region of L2 and it is possible that such flexibility may provide a means 
by which the integrity of the coiled-coil structure in segments 2A and 2B can be 
maintained whilst allowing the coherence of segment 2 across the stutter to be re­
established in a gradual manner. The differences between flexibility proflles in  this 
region may indicate alternative conformations of the two chains around the stutter. 
The flexibility profiles of chains comprising a hetenxiimeric keratin molecule are not in 
phase in the vicinity of the second heptad stutter found close to the centre of segment 
2B in all IF chains. This may indicate that the heptad phasing is re-established at this 
point by a relatively sharp discontinuity in the molecular structure, possibly involving 
a kink in the axis of the coiled-coil. The nature of the discontinuity suggested here is in 
contrast to that postulated above for the stutter in segment L2. 
A feature of the flexibility proftles is that regions in the rod domain of high flexibility 
are frequently bounded by regions of particularly low flexibility. It follows that a high 
local flexibility score should not be interpreted in terms of a breakdown in coiled-coil 
structure but instead as a shon region of structure with a greater propensity for 
interaction with other IF molecules or perhaps intermediate filament associated 
proteins (IFAPs). Indeed many of the peaks can be correlated with the presence of 
basic residues (predominantly lysine) and acidic residues. These charged residues are 
intimately involved in the interactions which specify both molecular assembly and 
aggregation in vivo (see Chapter 1). 
The lamin chain shows a higher flexibility score for the N-terminal domain than the 
C-terminal domain, unlike most of the other sequences studied. The N-terminal 
domain of lamin is amongst the smallest of any of the IF chains whereas the 
C-terminal domain of lamin A is one of the largest after the NF-H chains. 
In summary, some of the features of the flexibility profiles can be related to the 
sequences of the chains, such as the glycine-serine rich area in the terminal domains of 
the epidermal keratins, and a tentative link to function can also be proposed, viz. the 
flexibility of skin. Generalizations of the proflle data made above always have 
exceptions and these may be the result of a link between predicted flexibility and the 
CHAPI'ER 2: Primary Structure page 47 
function of individual IF. Indeed, although IF share a common plan for the rod 
domain, they are found in a wide range of cell types and hence it is to be expected that 
the variability present in the terminal domains in particular, as well as in the less well 
defined regions of the rod domain, is a result of differing function. 
2 . 4  Summary 
A variety of structural studies based on IF protein sequences have been presented in 
this Chapter. These include an examination of the periodicities in the rod domains, the 
distribution of amino acid residues in the heptad substructure and the flexibility of the 
peptide backbone. Fourier techniques were used to investigate regularities in the 
disposition of charged residues in the rod domains of peripherin, the nuclear lamin 
proteins, and the Helix pomotia B protein. The Fourier data for peripherin showed a 
high degree of similarity with that from other type ill IF chains and confinn the 
assignment of peripherin to the type ID grouping. The close homology between the 
human and Xenopus lamin A proteins is reflected in the Fourier transfonns of their 
charged residues. A novel feature of these transfonns is an apparent doubling of the 
9- 10  residue period found commonly in IF chains; this is associated with the larger 
scale packing of the molecules (compared to the short range heptad structure which is 
related directly to the mode of chain packing). 
The transforms of sequences from the Xenopus lamin B rod domain segments reveal a 
period of -9.3 residues in the disposition of basic residues in segment IB .  This is a 
little shorter than generally found in the same segments in other IF chains although it is 
characteristic of the period found in acidic residues in segment I B of type I keratins. 
The Helix protein also reveals high intensity peaks corresponding to a similar period. 
Lamin B proteins are thought to be nuclear anchorage sites for the cytoplasmic IF 
network (Georgatos and Blobel, 1987; Gerace and Burke, 1988) and this common 
shorter period allows the possibility of a direct linkage between the cytoplasmic and 
nuclear IF networks. Helix B type V IF are apparently coexpressed with keratin IF in 
the cytoplasm (Weber et aI, 1988) and it is possible that the common period might 
also allow coaggregation of the two networks although it must be emphasized that 
there is no evidence yet available to suggest that this actually takes place. 
An analysis of the distribution of residues within the heptad substructure shows the 
preference of certain types of residues in certain positions. Apolar residues are 
confmned to occupy -75% of the i! and si positions though the distribution of residues 
between the two positions is highly non-uniform. For example, leucine is found much 
more frequently in the d position than the i! position but for isoleucine the il position is 
more commonly occupied than the d. This emphasizes that, although the i! and si. 
CHAPI'ER 2: Primary Structure page 48 
positions are similar in that they are internal to the coiled-coil structure and crucial for 
the formation of the hydrophobic core that stabilizes the molecule, they are nonetheless 
stereochemically non-equivalent. This 'difference had also been observed by Phillips 
et al ( 1986) who correlated the occurrence of apolar residues having branched 
sidechains in the d. position with more flexible regions of coiled-coil or even "hinge" 
regions in the case of myosin (see also Cohen and Parry, 1989). 
The charged residues are also unevenly distributed between the il and d. and the � and " 
positions although of course there are relatively few non-apolar residues in the internal 
il and d. heptad positions. The asymmetric distribution found, however, does enable 
basic residues in the il and � positions of the same chain to interact with acidic residues 
in the d. and , positions of an adjacent chain (see Figure 1-4) to stabilize the two-chain 
molecular structure. 
An interesting comparison between two families of a-fibrous proteins reveals that for 
IF proteins the acidic residues are more common than basic residues in the " position 
of the heptad substructure but that for myosins the acidics dominate in the structurally 
similar � position. These two heptad positions are important for specifying the 
(parallel) orientation of the chains in the molecule. In the case of tropomyosin, which 
has a more pronounced division of charged residues among the two positions (parry, 
1975), the � and , stripes on adjacent chains will attract if the chains are parallel or 
repel if antiparallel. For IF and the myosin-type proteins in general, the net charge of 
each position is more even. 
The flexibility proftles of representative IF chains has revealed some features that may 
be identified with structural elements of the molecule. The terminal domains are 
generally more flexible than the rod domains, especially for the soft keratins and the 
neurofilament chains studied. This flexibility may be a functional requirement of these 
domains (or at least parts of them). The epidermis, for example, is tough yet yielding 
and this 'soft' character may be attributable in part to the terminal domains of the 
constituent IF chains and the interactions these make with other IF or IF-associated 
proteins. In contrast, the hard keratin chains studied are predicted to be less flexible 
than their epidermal counterparts. Regular patterns of residues in parts of the amino 
acid sequence are also reflected in the flexibility profiles. 
For most IF chains the link region Ll  is generally predicted to have the greatest 
flexibility of the rod domain segments: this is also one of the rod domain regions with 
the lowest degree of sequence homology. Another interesting observation is the 
flexibility apparent in the link L2 which also contains a stutter in the phasing of the 
heptad substructure. The apparent freedom allowed in the conformations of the chains 
CHAYfER 2: Primary SbUCture page 49 
in this region may allow the integrity of the coiled-coil structures to be maintained over 
the discontinuity. The stutter towards the centre of segment 2B, however, appears to 
be·accommodated by a more abrupt conformational rearrangement of the coiled-coils. 
CHAPrER 3: Sequence Homology page 50 
I 3.  SEQUENCE HOMOLOGY I 
In general terms, homology may be described as a measure of correspondence or 
sameness among objects. This concept is useful for identifying groupings of objects 
with high degrees of homology and may also show the features that are common to a 
grouping as well as those that differentiate one group from another. In this chapter, 
homology is defined in terms of the degree of similarity between the linear distribution 
of the amino acid residues constituting the coiled-coil rod domain of IF protein 
molecules, although the techniques used here may be applied to the sequences of any 
family of proteins. The structme of the coiled-coil rod domain is relatively well defined 
and constant in length compared to the terminal domains of the IF proteins and so 
sequence comparisons amongst the proteins are more easily carried out in this domain. 
Sequence data from a selectioQ of IF proteins (Table 2- 1)  are used to generate three 
homology statistics (ha, hr. hs)  that describe, in large part, the degree of 
correspondence amongst groupings of the proteins. In general, mammalian sequences 
are examined although some pans of this study will compare non-mammalian (usually 
amphibian proteins) and mammalian proteins. The homology statistics are also used to 
generate a consensus, or archetypal, sequence for the various IF chain types I-V. The 
alignment of incomplete rod domain sequences is maintained by using blank positions 
which are otherwise ignored in the calculation of homology. Blanks are also used in 
the variable length segments L1 and L12 to maximize homology within these segments 
as well as to maintain alignment. The positions and numbers of blanks are somewhat 
arbitrary in L1 and L12 as the degree of homology is relatively low. The total number 
of residue positions for the rod domain of types I-IV IF has been set to 327; for type V 
IF chains this has been increased to 359 as a result of the 42 residue insert in segment 
1B  and the replacement of the link L1 by extensions to segments lA  and 1B for which 
no padding is necessary. 
Two sequences exhibit anomalous lengths for some rod domain segments. The Endo 
B cytokeratin (M47K - Singer et ai, 1986) contains 17 residues in link L12  instead of 
the usual 16 for type I IF chains (Table 1-1).  Also, an extra residue (acidic) is insened 
within the last heptad of segment 2B which is strictly conserved in length in all other 
IF chains. For these reasons, this protein has been excluded from the analyses 
described in this section. The human NF-L chain (Julien et ai, 1987) has a length of 
103 residues for segment 1B,  unlike all other type I-IV IF chains. The two extra 
CHAFfER 3: Sequence Homology page 51 
residues (a glutamic acid and a threonine residue) are easily detected by comparison 
with other NF-L chains and in both cases represent a repeat of the residue immediately 
preceding it. Also, segment L12 is only 16 residues in length unlike all other NF-L and 
NF-M chains for which the length is 17 residues. Because relatively few NF-L 
sequences available, a modified sequence has been used where these two repeated 
residues in segment 1B have been deleted. 
3 .  1 Homology Statistics 
Suppose that N sequences are to be compared and that the padded lengths of the 
sequences is n (n=327 for type I-IV chains; n=369 for type V chains) where the origin 
is at the amino-terminus of segment lA. Let A(iJ) denote the amino acid within protein 
sequence i ( lSiSN) at a linear position j (1�j�n). The various segments of the rod 
domain for types I-IV IF are defined as follows: 
segment lA A(i, l )  to A(i,35) 
segment Ll A(i,36) to A(i,56) 
segment 1B A(i,57) to A(i, 157) 
segment L12 A(i, 158) to A(i, 179) 
segment 2A A(i, 1 80) to A(i, 198) 
segment L2 A(i, 199) to A(i,206) 
segment 2B A (i,207) to A(i,327) 
For the type V sequences, the segment boundaries are: 
segment lA A(i, l )  to A(i,41)  
segment 1B A(i,43) to A(i, 189) 
segment L12 A(i, 190) to A(i,21 1 )  
segment 2A A(i,212) to A(i,230) 
segment L2 A(i,23 1)  to A(i,238) 
35 residues 
21  residues 
10 1 residues 
22 residues 
19  residues 
8 residues 
121 residues 
41  residues 
147 residues 
22 residues 
19 residues 
8 residues 
segment 2B A(i,239) to A(i,359) 121  residues 
The 42-residue insert in segment lB is part of a larger 70-residue piece at positions 90 
to 1 59 inclusive in the type V chains which replaces residues 99 to 126 of the types 
I-IV chains (Parry et ai, 1986). Although this causes misalignment of the type V 
chains with the type I-IV chains, only homology within chain types is calculated and 
so the relative alignment of chains of different types is not important. 
The stutter in the phasing of the heptad substructure that occurs near the middle of 
segment 2B necessitates further subdivision of this segment into 2B 1 and 2B2 which 
are bounded for types I-IV IF as follows: 
segment 2B 1 A (i,207) to A(i,264) 
segment 2B2 A(i,266) to A(i,327) 
58 residues 
62 residues 
CHAPTER 3: Sequence Homology 
and for type V:  
segment 2Bl A(i,239) to A (i,296) 
segment 2B2 A(i,298) to A(i,359) 
58 residues 
62 residues 
page 52 
A single residue has been designated as the 'locus' of this heptad stutter (residue 265 
in types I-IV IF and residue 297 in type V IF). This is a somewhat arbitrary choice as 
in fact the heptad pattern can be continued through the 'locus'  from either direction for 
a short way and indeed it is not at all apparent whether the stutter i s  manifested as a 
sharp discontinuity or as a more extensive but gradual re-ordering of the coiled-coil 
structure. A similar situation arises for the link Ll in type V chains which is reduced to 
a stutter in the heptad phasing: in type I-IV chains it is a short non-ex-helical segment. 
As with the segment 2B stutter, a single residue position has been chosen in the type V 
chains as the centre of the stutter - position 42. 
The homology statistics ha, hr and hs are defined below. Note that the homology 
scores are given as percentages where 100% homology indicates identical sequences. 
3 . 1 . 1 Amino Acid Homology Score, ha 
Each amino acid is  compared in turn with each amino acid in the N sequences sharing 
a common value of j,  including itself. For every comparison a look-up table (for 
example, Table 3- 1 )  is used to score the degree of homology between the pairs of 
residues. The basis of the scoring system is as follows: residue pairs having  no 
relationship are scored zero; residue pairs of conservative size and character (eg. Leu­
Val, Met-Phe, etc.) are scored 0.5; residue pairs of conservative charge (eg. Lys-Arg 
or Asp-Glu) are scored 0.75; pairs of identical residues are scored unity. Other look­
up tables are used for scoring homology among specific types of residues and are 
listed in Table 3-2. For each comparison made the scores are added and normalized to 
generate the homology statistic ha(iJ). This may be formulated mathematically as: 
ha(iJ) = � �S[A(iJ) :A(kJ)] 
k=1 
where S[A(iJ):A(kJ)] is the look-up score for the residue pair A(iJ) and A(kJ) .  Note 
that this defmition implies a minimum homology score of 
� S [A(iJ):A(iJ)] 
3 . 1 .2 Residue Homology Score, hr  
This i s  defined as the largest value of ha(iJ) at some residue position j, ie. 
hr(J) = max[ha(iJ)] for 19� 
Asp GIu Arg Lys His ne Leu Met Phe Tyr Val 
Asp 1 
GIu 0.75 1 
Arg 1 
Lys 0.75 1 
His 0.5 0.5 1 
TIe 1 
Leu 0.5 1 
Met 0.5 0.5 1 
Phe 0.5 0.5 0.5 1 
Var 0.5 0.5 0.5 0.75 1 0.5 0.5 0.5 0.5 0.5 
Ala 0.25 0.25 0.25 0.25 0.25 
Asn 0.5 
Cys 
GIn 0.5 
GIy 
Pro 
Ser 
Thr 
Trp 0.5 0.5 
A� Glu Arg Lys His ne Leu Met Phe Tyr 
Table 3-1 Look-up table for mixed homology scores S [ A(iJ):A(k.j)J 
1 
0.25 
Val 
Ala Asn Cys GIn Gly Pro 
1 
1 
1 
1 
1 
1 
Ala Asn Cys Gin Gly Pro 
Ser Thr Trp 
1 
0.5 1 
1 
Ser Thr Trp 
Asp 
GIu 
Arg 
Lys 
His 
TIe 
Leu 
Met 
Phe 
Tyr 
Val 
Ala 
Asn 
Cr;s 
G n  
Gly 
Pro 
Ser 
Thr 
Trp 
CHAPrER 3: Sequence Homology 
Asp Glu Arg Lys ne Leu Met 
Acidic Asp 1 
Homology Table Glu 1 1 
Basic Arg 1 
Homology Table Ly_s 1 1 
Large Apolar ne 1 
Homology Table Leu 1 1 
Met 1 1 1 
Phe 1 1 1 
Tyr 1 1 1 
Val 1 1 1 
Asp Glu Arg Lys ne Leu Met 
Table 3-2 Look-up tables for Acidic, Basic and Large Apolar homology 
page 54 
Phe Tyr Val 
1 
1 1 
1 1 1 
Phe Tyr Val 
and is a reflection of the most common feature of the various sequences being 
compared at that position. It leads naturally to the idea of a consensus residue and, by 
extension, to a consensus sequence. 
3 . 1 . 3  Segment Homology Score, hs 
The value of hs(i) is  the average ha(iJ) value within an entire sequence or some 
subsequence (often a coiled-coil or link segment). If ml and m2 are the positions of the 
fIrst and last residues respectively in some sequence of interest then the hs(O score for 
that subsequence is defmed as 
1 m2 hs(i) = (m 2 - m 1 + 1 ) ''2jza(iJ) 
j=ml 
where 1�1<m2�n. The value (m2 - ml + 1 ) is the number of residues over which 
the average was taken. This statistic provides an indication of the divergence of IF 
proteins and of their appropriate classifIcation within the types I-V categories. This 
may also be assessed using other homology techniques (see, for example, Crewther 
et ai, 1983 and Parry and Fraser, 1 985). 
3 . 2  Homology within the Rod Domain 
The residue homology scores (hr) have been calculated using the look-up table 
specifIed in Table 3- 1 for the mammalian type I-IV sequences and all the available type 
V sequences (Table 2-1 ). Homology profiles for each type are shown in Figure 3- 1 .  
Prior to graphing, the raw scores have been smoothed by convolution with a Gaussian 
function of half width two residues (see Appendix C). Regions of these smoothed 
curves where three or more consecutive residues have residue homology scores 
CHAPrER 3: Sequence Homology 
lOO 
e 80 
8 en 
i 
60 
� 40 
20 
(a) Type I O �--------------------------------------------------� 
lOO 
20 
(b) Type IT O �--------------------------------------________________ � 
lOO 
20 
(c) Type ill 
o �--------------------------------------------------� 
lOO 
u 80 
8 en � 60 
� 40 ::I: 
20 
(d) Type IV 
page 55 
O �------------------------------------------------____ � ____ � 
lOO 
u 80 
8 en � 60 
� 40 ::I: 
20 
(e) Type V 
o 50 lOO 150 200 250 300 350 400 
Residue Translation 
Figure 3-1  Homology profiles for mammalian types I-IV chains and all available type V IF 
chains. The residue homology scores (hr) have been smoothed with a Gaussian function of half width 
two residues. Regions of highly conserved sequence are apparent in segment l A  and at the C-terminal 
end of segment 2B in the type I-IV profiles. 
CHAYrER 3: Sequence Homology page 56 
Segment Type I Type 11 Type ill TypeN Type V 
lA 4-20 1-4 7-12 8-17 8-15  
Ll - - - - -
1B 6>-65 99-101 64-68 6>-65 -
85-97 1 1 1-1 14 80-87 72-76 
1 10-124 1 17-120 129-132 
140-150 1 35-145 
150-154 
L12 - - - - -
2A - 187-192 1 82-187 1 86-189 -
1.2 199-202 200-204 204-206 - -
2B 234-236 237-241 207-210 298-324 359-367 
244-248 3 10-323 2 17-229 
242-246 
248-262 
290-327 
Table 3-3 Regions in the homology profiles (hr scores) where the scores are greater than or equal 
to 90%. The regions listed are at least three residues in length and indicate highly conserved primary 
structure. 
h�90% or hr<60% are listed in Tables 3-3 and 3-4 respectively. In addition, the 
segment homology scores (hs) have been calculated for the 7 subdomains of the rod 
region and for the rod domain as a whole. The scores for the chains belonging to the 
various types were then averaged and are shown in Table 3-5. The lamins have also 
been analysed as a subgroup of type V IF to highlight the greater degree of homology 
present than when they are taken in combination with the Helix pomotia B protein. 
3 . 2 . 1  Coiled-Coil Segments 
The homology data indicate that segment lA is generally highly conserved within each 
chain type (Figure 3- 1 and Table 3-4) and also in the type I-IV IF chains a s  a whole 
(Table 3-5) compared to the other coiled-coil segments. The observation of highly 
conserved sequences within segment lA  and at the C-terminal end of segment 2B has 
been noted previously (see Chapter 1) and is confmned by this analysis. Indeed the 
degree of homology is very high in the C-terminal portion of segment 2B for the type 
III and type IV chains and over the last 30 residues it is close to 100% (Figure 3-1 ;  
Table 3-3). The type V chains reveal a similar highly conserved region in segment lA 
but no corresponding region at  the C-terminus of segment 2B. This is due to the 
overall low homology of the Helix pomotia B protein with the other type V rod 
CHAPrER 3: Sequence Homology 
Se nt 
lA 
L1  47-50 
1B  74-77 
L12 161-165 176-179 
174-176 
2A 
L2 
2B 
ID 
34-35 
36-41 
45-50 
72-76 
174-179 17 1-174 
page S7 
v 
35-41 
43-55 
154-156 
267-270 
Table 3-4 Regions in the homology profiles (hr scores) where the scores are less than 60%. The 
regions listed are at least three residues in length and indicate variability in the conservation of primary 
structure. 
domains despite other obvious structural similarities (see Table 3-5): the lamins taken 
alone do exhibit this feature. 
The mean hs scores (Table 3-5) show an interesting difference between coiled-coil 
segments for the types I-IV IF. The mean hs score for segment IB of type I chains is 
about 5% higher than that for segment 2B (and indeed for segment 2 as a whole). For 
types IT, ill and IV chains the mean values of hs for segment IB are lower by 6, 9 and 
12% respectively than those for segment 2B. The type V lamin chains reveal similar 
mean scores for the IB  and 2B segments (82.3% and 86.7% respectively). 
Consequently for keratin molecules, which are believed to be obligate heteropolymers 
containing a type I chain and a type 11 chain (see Chapter 1 ) ,  the sequences of 
molecular segments IB and 2 are preserved at comparable levels of significance 
whereas for the homopolymeric type III and type IV molecules, the sequence of 
segment 2B has been maintained with greater fidelity than has segment lB .  This 
indicates that segment 2 in the type ID and the type IV molecules plays a slightly 
different structural role than in the keratin molecules and again raises the possibility 
that IF may exhibit more than one structural fonn. 
3 . 2.2 Link Segments 
The a-helical link segment L2 exhibits a higher degree of homology within and across 
all chain types than the other link segments (Table 3-3, Table 3-5, Figure 3- 1 ,  Figure 
3-2). The hs scores for L2 range from 83-89% for types I, II and III chains and these 
values are very similar to those noted previously for coiled-coil segment lA. The type 
IV and V hs scores for segment L2 are a little lower at 68.4% and 66.3% respectively 
CHAYI'ER 3 :  Sequence Homology page 58 
Segment Type I Typell Types I and IT 
lA 85.8±4.4 (8) 86.3±1 .9 (8) 69.9±2.2 ( 16) 
L1 42. 1±6.8 (10) 42.5±3.6 (8) 37.6±5.4 ( 1 8) 
1B 80.0±3.7 (to) 72.5±1 . 8  . (8) 58.8±4.8 ( 1 8) 
L12 54.4±4.8 (12) 62.9± 1 .7 (8) 48.9±4. 1 (20) 
2A 74.2±4.7 ( 13) 78 .4±4.3 (8) 63.8±4.7 (21 )  
U 88.2±3. 1 ( 13) 89 .6±3 .8 (8) 72. 1±5.5 (21 )  
2B 74.8±4.2 (13) 78.3±2.4 (8) 59.7±4.2 (21 )  
Rod 70.4±4.3 (8) 72.5±1 . 8  (8) 57.5±3.3 ( 16) 
Segment Type ill Type IV Types ill and IV 
lA 85.0±2.7 (5) 8 1 .8±3 .3 (9) 76.4±3.4 ( 14) 
L1 3 1 .8±1 .5 (5) 38.7±2.4 (9) 34.6±4.6 ( 14) 
1B  78.0±4. 1 (4) 70.4±3.6 (9) 65.5±2.9 ( 13) 
L12 56.3±3.7 (6) 5 1 .3±0.9 (10) 46.8±4.4 ( 16) 
2A 88.6±4.9 (7) 75 .2±0.6 ( 1 1)  73.4±4.9 ( 18) 
U 86.0±3.7 (7) 68.8±7.6 ( 1 1 )  7 1 . 1±7.3 ( 1 8) 
2B 86.9±2.5 (8) 82.0±2.2 (10) 75.4±1 .8  ( 1 8) 
Rod 7 1 .8±2.9 (4) 66.7±2.5 (9) 62.5±2.9 ( 1 3) 
Segment Types I, IT, Type V Type V 
ill and IV (Lamins only) (all) 
lA 66.3±2.3 (30) 86.7±3.2 (3) 72.5±8.5 (4) 
L1 32.6±6.0 (32) 
1B  52.7±2.9 (3 1)  82.3±2.9 (3) 64.5±1 1 (4) 
L12 38.7±5 . 1  (36) 78 .0±2.6 (3) 57.3±19 (4) 
2A 60.2±5. 1  (39) 75 .7±4.0 (3) 60.0±8.9 (4) 
U 61 .3±7.3  (39) 83.0±0.0 (3) 66.3±10 (4) 
2B 56.7±3.8 (39) 86.7±4.0 (3) 67 .3±12  (4) 
Rod 5 1 .7±2.9 (29) 70.0±2.6 (3) 55 .0±10 (4) 
Table 3-5 Mean segment homology scores (hs) for the rod domain segments in IF proteins (see 
Figure 3-1). Values are quoted as mean ± s.d. (number of complete sequences in group). Segments L l  
and L12 are padded to allow optimum alignment of the sequences - the values correspond to the 
maximum possible score for those segments and cannot be directly compared to other entries in the 
table. The type V sequences have been examined as two groups: one without the Helix pomotia B 
protein (lamins only) and one with it (all). 
CHAPI'ER 3: Sequence Homology 
8e-l 
HS<IC 
MS2K 
H47K 
MSIK 
MSSK 
MSIK 
B4(IC 
MS9K 
BS4K 
HS6.5K 
7c 
7x 
HS6K 
HSSK 
M6IK 
H67K 
M67K 
XL64%1 
XL64,  .. 
HNF-L 
MNF-L 
PNF-L 
RNF-L 
HNF-M 
MNF-M 
PNF-M 
RNF-M 
HNF-H 
MNF-H 
PNF-H 
HLaminA 
XLaminA 
XLaminB 
Helix B 
page 59 
Figure 3-2 Comparison of the amino acid sequences for IF link segments LI,  LI2 and L2. The 
sequence abbreviations are defined in Table 2-1 . LI and LI2 have both been padded with blanks in 
order to maximize the homology present and to maintain alignment within the rod domains. Those 
residues which are conserved in character (ie, conservatively substituted) are boxed and those that are 
absolutely conserved within a chain type or subtype are shaded. The symbol '? '  is used to represent an 
incomplete sequence. No link LI segment is defmed for the type V chains. 
although the lamins alone have a rather higher mean score of 83.0%. The differences 
in the sequence of segment L2 between IF types are also of significance as they clearly 
distinguish one type from another. For example, the first four residues in type I and 
type Il chains are N-R-basic-D and S-R-A-E respectively and each is maintained 
virtually without exception (Figure 3-2). For type Ill, IV-L and IV-M chains, the 
second residue in segment L2 is apolar (cf. arginine or lysine in type I and IT chains) 
and for type IV-H chains it is threonine. The fourth residue is non-acidic in type IV 
chains, unlike all other IF chains except the type V Helix pomotia B chain. Type ill 
CHAYfER 3 :  Sequence Homology page 60 
and N chains are also similar in that link L2 ends with the three residue motif E-E-W. 
The lamins can be distinguished by their sequence in this region which may be 
generalised as Y-K-·-E-L-E-·-T. The Helix B protein does not easily fit into this or 
any other IF pattern and remains somewhat anomalous. 
The 01u-6 in segment L2 is conserved in all IF chains studied except for M47K and 
Xenopus lamin A (Figure 3-2) and appears to represent a residue crucial to structure or 
function. The fact that it occurs in a region where the heptad regularity is temporarily 
disrupted is almost certainly significant. The implication of highly conserved features 
in segment L2 is that the IF chains have a highly specific confonnation here that cannot 
readily tolerate sequence changes and the effect of this may be to allow the integrity of 
the coiled-coil structure to be maintained either side of the discontinuity. It is 
interesting to compare the homology here with that of the other heptad discontinuity 
that occurs in a region close to the centre of the coiled-coil segment 2B (residue 265; 
see Figure 3- 1 ). In this case the homology scores vary somewhat amongst the types 
(Figure 3- 1 )  but are neither very high (�90%) nor very low « 60%).  The M47K 
sequence is somewhat different from those of other type I keratins, as detailed in the 
introduction to this Chapter, and is only listed in Figure 3-2 for completeness.  The 
Xenopus lamin B chain has a glycine residue in this position and may well adopt an 
alternative structure around the discontinuity. 
Segments LI and LI2 present a problem in the calculation of their homology statistics 
due to their variable lengths even within types. Some homology clearly exists within 
both segments as can be seen in Figure 3-2 where they have been padded to allow 
alignment of the sequences. Overall, however, the LI and L12 segments show the 
least conserved primary structure of any rod domain segment (Table 3-3, Table 3-4 
and Table 3-5). Segment LI2, which can be from 15  to 22 residues in length, has been 
padded with up to seven blank residues to maximize the reasonable degree of 
homology in the amino acid sequences of all IF chains (Figure 3-2). Segment LI  is 
even more variable in length, and requires about twice as many blanks as segment 
LI2: even then the maximization of homology shown in Figure 3-2 is relatively poor. 
It is clearly inappropriate that the hs homology scores from these padded segments be 
compared with those from non-padded segments. As this and other studies have 
shown that IF chains have evolved from a common ancestor (Section 1 .4) , it is 
tempting to speculate that the precursors to segments L1 and L12 were both somewhat 
longer than they now appear and that their conformations have permitted variations 
(primarily deletions) to be tolerated with only limited structural restrictions. For the 
case of the type V proteins, the link LI appears to be replaced by small extensions to 
the coiled-coil segments lA  and 1B. 
CHAPTER 3 :  Sequence Homology page 61 
3 . 2 . 3  Type V Chains 
The type V chains require special considerations since their structure is unique 
compared with other IF proteins. For example, the link L1 is effectively replaced by 
extensions of the coiled-coil segments lA  and 1B:  consequently it does not require any 
padding. Also, the relatively recent establishment of this IF type has prevented any 
sensible comparisons of the chains that constitute the group. Until recently only the 
sequences of human lamins A and C were available and these shared a common rod 
domain sequence. A number of type V sequences from non-mammalian sources have 
now been completed and some comparison amongst the chains is possible. 
An observation made in Chapter 2 was that the Xenopus and human lamin A chains 
were very similar (83% of residues in the rod domains were identical). The mean 
homology scores of Table 3-5 for the lamins (human lamin A and Xenopus lamins A 
and B) reflect this although the Xenopus lamin B sequence, which is not so similar as 
the others, tends to reduce the scores somewhat. The Helix pomotia B protein was 
included in a separate calculation and the resulting scores were significantly lower than 
for the lamin chains alone. This raises the possibility of further sub-typing within the 
type V classification although more sequence data from a variety of sources would be 
required to establish the validity of this suggestion. 
3.3 SUbtyping on the basis of Homology 
3.3. 1 Hard and Soft Keratin IF Chains 
The division of keratin protein chains into the type I and type II categories is 
straightforward on the basis of homology within the rod domain segments and also the 
N- and C- terminal domain segments (see Chapter 1 ). However, analysis of the hard 
CL-keratin sequences available (Table 2- 1 )  show that significant differences occur 
between each of them and the epidermal keratin chains from the same type 
classification. For example, the hard keratin chains have a higher cysteine content than 
their epidennal counterparts and the majority of these residues fonn disulphide 
linkages which render the keratin complex highly insoluble (Chapter 1) ;  the type IT 
hard CL-keratin chains have 10 residues in segment L1 (Fraser et al , 1988) compared 
to 1 1  or 1 2  for the epidermal type II chains and these are the shortest L I  segments 
among the protein chains included in this study; when combinations of hard and soft 
keratin are compared directly, the numbers of identical residues are notably lower than 
between combinations of the soft epidermal chains or between combinations of the 
hard keratin chains. Furthermore, Crewther et al (1985) have reported that there is a 
CHAPrER 3: Sequence Homology page 62 
Type I Type II 
Segment Hard Soft Hard Soft 
lA 79.0** 88.0±1 .7 85.3* 86.8± 1 . 1  
L1 32.0** 44.6±4.9 39.0* 44.6±2.9 
1B 74.0** 8 1 .5±1 .9 72.7t 72.4±2.3 
L12 49.0* 55.5±4.5 63.3t 62.6±2.2 
2A 66.0** 75.7±3.3 79.3t 77 .8±5.5 
12 84.0* 89.0±2.6 93.0*t 87.6±3.4 
2B 68.0** 76. 1±3.2 79.0t 77 .8±3. 1 
Rod 64.0** 72.5±2.0 72.7 72.4±2.3 
Table 3·6 Mean segment homology scores (hs) for the rod domain segments in hard and soft 
mammalian keratins. Those mean hs scores for the hard keratins that differ by at least 1 s.d. and 2 s.d. 
from the soft keratin scores are indicated by • and .* respectively. t marks the hard keratin hs scores 
that are higher than the epidermal equivalents. 
90-95% identity between the rod domains of components 8c-1 and 8a (both type I hard 
chains) and also between the rod domains of components 7c and 5 (both type II hard 
chains). 
The hs homology scores for the hard and soft keratins are listed in Table 3-6. The ha 
scores, from which these hs values were derived, originate from an analysis 
encompassing all mammalian keratins of a particular type. The mean hs score (±S.d.) 
for the epidennal keratin chains was then calculated and, because the contribution from 
the hard and soft keratins has been separated, the values quoted differ a little from 
those listed in Table 3-5. The results indicate that the hs scores for the type I hard 
keratins are generally at least two standard deviations less than those derived for the 
type I soft keratins. The differences between the type IT hard and soft keratins are less 
marked, indicating that the sequence changes which have occurred have been of a 
more conservative nature. 
As proposed in Conway and Parry ( 1988), these differences suggest that i t  is 
appropriate for type I and type IT IF chains to be subdivided into type la and ITa (hard 
a-keratins) and type Ib and llb (epidermal keratins). This classification implies that 
type la and type ITa chains could coaggregate and that type Ib and type lib chains could 
coaggregate but that cross-combinations of hard and soft are not likely. Further data is 
still required to substantiate this proposal. 
3 . 3 . 2  Type IV IF Chains 
The type IV IF neurbfilament chains can be subdivided into three groups on the basis 
CHAPrER 3: Sequence Homology page 63 
of molecular weight: NF-L (light), NF-M (medium) and NF-H (heavy). Analysis of the 
rod domains of the neurofIlament chains included in this study shows that the chain 
designations L, M and H may also be assigned on the basis of sequence homology. 
Indeed, the NF-L rod domain sequences show a remarkable 98% identity within the 
rod domain. The NF-M and NF-H sequences also show high levels of identity at 95% 
and 96% respectively (see also Myers et ai, 1987 and Napolitano et ai, 1987 for 
identity throughout the NF-M chains). In almost all cases the non-identical residues are 
conservative substitutes. In contrast, the degree of identity between the groups is much 
lower. for NF-L and NF-M rod domain sequences it is about 53%; NF-L and NF-H are 
42%;  NF-M and NF-H are 53%; and all NF rod domain sequences together show 
about 35% identity. It is interesting that the heavy and light chains are least identical 
but show similar conservation with the medium chains. This is an indication that the 
three neurofilament subtypes may have evolved from a common IF-type protein. The 
greatest evolutionary changes to this group are apparent in the differing extents of the 
carboxy-terminal domains but the rod domains also reveal the effects of divergence. 
Section 3 .4 will show that the sequences comprising the HI and H2 subdomains in the 
NF-L and NF-M chains are maintained at a similar level to those in the rod domain. 
The extraordinarily high degree of sequence identity within the NF-L, NF-M and NF-H 
subtypes is a strong indication that only minor sequence variations can be tolerated 
without losing structural and/or functional viability. The amino acid sequences of 
neurofIlament chains clearly provide an alternative means of chain classification (L, M, 
or H) without reference to molecular weight data. 
3 . 4  Homology among the 'H' subdomains 
Analysis of the sequences of the N- and C-terminal domains in keratin IF chains has 
shown that both may be subdivided into segments of special character (Steinen et ai, 
1984b; Steinen and Parry, 1985; Steinen et ai, 1985a; see Figure 1 -3). These 
subdomains exhibit bilateral symmetry with respect to the IF coiled-coil rod domain. 
Segments HI and H2 are defmed as those subdomains with a high degree of sequence 
homology within a chain type grouping which are located immediately adjacent to the 
ends of the rod domain (ie. at the N- and C-terminal ends respectively). Distal to HI  
and H2 are the so-called VI  and V2 'variable' subdomains. These differ greatly in 
both sequence and length and frequently have high contents of glycine and serine 
residues. Finally, distal to VI and V2 and hence at the N- and C-tenninal ends of the 
chain are the El  and E2 'end' domains respectively (Parry et ai, 1987b). The 
increased sequence data available has allowed the character and size of the HI and H2 
subdomains to be reassessed. It should be noted that the precise N-terminal boundary 
CHAPrER 3: Sequence Homology page 64 
for HI  and the C-tenninal boundary for H2 are often not well-defmed and the limits 
may lie a few residues displaced from those stated. The presence of the HI and H2 
subdomains implies directly that these regions adopt a specific confonnation in vivo, 
though their role has yet to be elucidated. 
(i) Type I. Steinen and Pany (1985) and Steinen et al (1985a) have reponed that the 
HI and H2 subdomains are absent in type I chains. This study confIrms that H2 is 
indeed absent for type Ib chains but no conclusions can be made for the type la chains 
as there is only one sequence, component 8c-I ,  available for this subdomain. There is 
no evidence of an H2 subdomain in the non-mammalian type I chains (Xenopus) nor 
is there sign of any signifIcant homology between the individual Xenopus chains in 
this region. 
For subdomain HI  there is evidence of a seven residue region adjacent to the rod 
domain that is homologous in all mammalian type I IF chains (Table 3-7 and see also 
Figure 3-4). A longer HI subdomain may exist for the type la chains but, as above, 
this cannot be determined without funher sequence data. The limited Xenopus 
sequences do not appear to have an HI subdomain nor do they exhibit any signifIcant 
interchain homology. 
Some sequences included in this analysis exhibit extensive homology between 
themselves far in excess of that specifIed by the lengths of HI  and H2 (see, for 
Chain Type HI (residues) H2 (residues) 
I (all) 7 0 
la �7 �O 
Ib 7 0 
IT (all) 36 20 
ITa �32 � IO  
ITb 36 20 
ITI (all) 20 58 
IV (all) 9 1 3  
IV-L 89 160 
IV-M 102 83 
IV-H 98 75 
V (all) 5 0 
Lamins 1 3  6 
Table 3·7 Lengths of the homologous subdomains HI  and H2 for mammalian type I-IV chains 
and for all type V chains. Greater than symbols �) are used where there is insufficient sequence data to 
delineate precisely the extent of a subdomain. 
CHAPrER 3: Sequence Homology page 65 
example, HI :  H46K, H50K and H56.5K; H2: H50K and M50K and also B50K and 
B54K). It may be concluded that such sequences have similar conformations and 
functions in vivo. 
(ii) Type IT. The extents of HI  and H2 in type IT chains as a whole are 36 and 20 
residues respectively as originally reported by Steinert and Parry (1985) and Steinert 
et al (1985a) (Table 3-7). The lIb grouping also follows this pattern but as there is 
only limited data available for the ITa subtype no definitive sizes for the HI  and H2 
subdomains can be established beyond the minimum lengths of 32 and 10 residues 
respectively. There are no Xenopus data pertaining to H I  but in H2 the extent of 
homology for Xenopus chains appears to be 20 residues, the same as that noted for 
mammalian type llb chains. No homology beyond this limit is apparent among the 
Xenopus chains in this region. 
As noted, for the type I analysis, individual combinations of type IT chains can reveal a 
higher degree of homology over a much longer piece of the sequence than  that 
specified by the HI and H2 subdomains (see, for example, HI :  M60K and M67K; 
H2: M67K and H67K and also M60K and H56K). 
(iii) Type m. The sizes of the HI and H2 subdomains as reported by Steinen and 
Parry ( 1985) and Steinert et al ( 1985a) are confirmed for type III chains as being 20 
and 58 residues respectively (Table 3-7). The latter value represents the entire 
C-terminal domain of type ill chains and includes three blanks inserted to maximize 
homology among the chains. 
(iv) Type IV. Insufficient neurofllament sequence data have prevented an assessment 
of the lengths of the HI and H2 subdomains prior to Conway and Parry ( 1 988).  
Sequences of four NF-L, four NF-M and three NF-H chains are available and the 
results of the analysis are listed in Table 3-7. As a whole, the type IV chains reveal 
short HI  and H2 subdomains of 9 and 13  residues in length respectively. However an 
extensive pattern of homology is evident within each of the subtypes and it is apparent 
that the H, V, and E subdivisions do not apply as well to the type IV chains as to the 
type I and IT chains. The entire N-terminal domains of of NF-L chains, NF-M chains 
and of NF-H chains (89, 102 and 98 residues respectively, with inserted blanks where 
appropriate) are highly homologous and may be designated HI -L, HI -M and H I-H 
respectively. The proportions of identical residues in this domain are 92%, 98% and 
86% for the light, medium and heavy chains respectively. These values are comparable 
to the identity within the rod domain of these groups (Section 3.3.2). However, in the 
C-terminal domain two homologous regions are evident which are separated by a 
variable region. The first homologous region, which is adjacent to the rod domain, is 
CHAPrER 3: Sequence Homology page 66 
160 residues long for NF-L chains, 83 residues for NF-M chains and for NF-H chains 
it is 75 residues. This region encompasses all of the C-tenninal domain for NF-L 
chains. The second homologous region is 1 1 1  residues for the NF-M chains and 75 
for the NF-H chains. The intervening region in the NF-M and NF-H chains contains a 
glutamic acid-rich region followed by a repeating run of between 6 to 13  residues that 
commonly includes the sequence LVS-SER-PRO. (These repeats can be observed, 
for example, as regular patterns in the flexibility proftle for the human NF-M chain in 
Figure 2-3 :  see Geisler et aI, 1987, and Shneidman et ai, 1988, for discussions on 
the structure and function of this terminal domain). The small data base of 
neurofilament proteins limits the reliability of the analysis and the extents of the 
subdomains may need revising as further data becomes available. 
(iv) Type V. The lamin HI and H2 subdomains have lengths different to those found 
in type I-N IF chains (Table 3-7).  Extensive homology is present in the N- and 
C-terminal domains of the human lamins A and C and the Xenopus lamin A sequence 
(see Figure 2-2) but inclusion of the Xenopus lamin B chain reduces the lengths of HI 
and H2 to 13 and 6 residues respectively. The Helix pomotia B chain further reduces 
the extents of these regions. As for the type N chains, the subdivision scheme is not 
entirely appropriate for the type V chains. More protein sequences are required to 
define the nature and boundaries of the HI and H2 subdomains and also to detennine 
whether further sUbtyping of the type V proteins is appropriate. 
Table 3-8 shows the extents of all the IF chain domains and represents an update from 
that given in Chapter I (Table 1 - 1 ). It incorporates the new data presented in this 
section on the HI  and H2 domains and on the type IV and V chains in general. Several 
chain types still require further sequence data to allow a complete analysis of their 
terminal domain boundaries including the hard keratins, the neurofilaments and the 
type V chains. 
3 . 5  Comparison of Amphibian and Mammalian Scores 
Complete or partial sequences of three type I, two type IT and two type V amphibian 
(Xenopus) IF proteins have been used in this study (fable 2- 1 ). Much of the analysis 
of homologous features has focused on mammalian proteins but it is also important to 
assess the degree of relatedness that exists between IF chains from these two diverse 
sources. 
The classification of the amphibian keratin chains in types I and IT is unequivocal using 
the standard criterion of sequence homology in the rod domain where the differences 
between the mammalian and amphibian chains are generally small. In particular, the 
CHAPrER 3: Sequence Homology page 67 
N-tenninal 
domain ......  ----- Rod domain ----... � 
C-tenninal 
domain 
��� 
I 
� )) 
El VI :!!mJ LI 
t1! 
L12� L2b! 
':,,!J 
V2 E2 
la ? ? ?7 35 1 1  101 16 19 8 121 ?O ? ? 
Ib 0-10 65- 140 7 35 1 1-14 101 16 19 8 121 0 0-1 10 5-30 
ITa ? ? �32 35 10 101 17 19 8 121 ? 1O ? ? 
IIb 5-70 20- 160 36 35 1 1 - 14 101 17 19 8 121 20 25- 125 2 1  
ill 19-81 20 35 8 101 16 19 8 121  58 0 
N-L 0 89 35 10 101 17 19 8 121 160 0 
N-M 0 102 35 10 101 17 19 8 121 83 237 1 1 1  
N-H 0 98 35 10 101 22 19 8 121 75 ?493 200 
V 25-64 5 41 0 148 15,19 19 8 121  0 185-287 
Table 3-8 Extents of the structural domains for IF chains. '?' indicates insufficient sequence data 
are available for the boundaries to be evaluated. In the case of the type III and V chains no clear 
division into E and V domains is possible on the same basis as for type I and IT chains. The same is 
true for the type IV chains although a domain of variable length does exist in the carboxy-terminal of 
the chains. This domain is bracketed by regions which show a high degree of homology within each 
sUbtype. 
type II comparisons show that the segment homology scores for the Xenopus chains 
are comparable with those for the mammalian epidermal chains (Table 3-9), ie, the 
Xenopus sequences show little variation from the consensus. A point of minor 
divergence lies in segment L2 (which is generally highly conserved, see Section 3 .2.2) 
where the Xenopus sequences SKLEAEAL differ in several respects from the 
mammalian consensus sequence SRAEAES{v (Figure 3-2). The lengths of segments 
L2 (8 residues) and L12 ( 17  residues) are strictly conserved in the mammalian and 
amphibian chains, However, the length of segment L1 ,  which is either 1 0  or 1 2  
residues long i n  mammalian chains, is 14 residues long in the amphibian chains. In 
general there is little homology between any of the type IT sequences in this segment. 
In contrast to the type II comparisons, the mean segment homology scores for the 
three type I Xenopus keratins are less than those for the equivalent segments in the 
CHAPTER 3 :  Sequence Homology page 68 
Type I TypeD 
Segment Xenopusa Mallillla1ian Xenopusb MallililaHan 
lA 8 1 .0** 89.2±1 .8  - 9 1 .4± l . l 
Ll 32.5** 44.3±5.9 55.0*t 47.4±4.2 
IB 75.0** 8 1 .9± 1. 8 77.5  77.8±3.3 
L12 48.0* 56.0±4.9 65.0t 64.4±2.6  
2A 67.0** 76.8±3.2 82.0t 80.8±5.8  
L2 66.5** 86.2±2.2 80.0* 82.8±2.3  
2B 66.5** 75.7±3.0 75.0* 78.0±2.3  
Rod 61 .5** 72.7±1 .5  - 74.0±2.0 
Table 3-9 Mean segment homology scores (hs) for the rod domain segments in amphibian 
(Xenopus) and mammalian epidermal keratins. Those mean hs scores for the Xenopus keratins that 
differ by at least 1 s.d. and 2 s.d. form the mammalian scores are indicated by • and •• respectively. t 
marks the Xenopus hs scores that are higher than the epidermal equivalents. (a) Xenopus XL51,  XL 70 
and XL81.  (b) Xenopus XL64 23 and XL64 164 (See Table 2-1 for sources). 
mammalian chains, usually by at least one standard deviation. In this respect the 
Xenopus type I and type IT keratins parallel the hard type I and type IT a-keratins in 
that the variations in homology between the type I chains (Xenopus, hard, and 
epidermal keratins) are greater than between the type IT chains from the same sources 
(Tables 3-6 and 3-9). Apparently type IT chains as a whole seem less able to permit 
variation in primary structure than type I chains. This feature can also be seen in the 
smaller standard deviations for the segment homology s cores for type 11 chains 
compared with those for type I chains (Table 3-5). 
The lengths of segments L12 (17 residues) and L2 (8 residues) for Xenopus type I 
chains are the same as for the mammalian type I chains but segment LI,  which is either 
1 1  or 14 residues long in mammalian chains, is 1 1  or 15  residues long in Xenopus 
chains. Once again, relatively little homology is present in this segment when 
mammalian and amphibian chains are compared. Segment L2 in the Xenopus chains 
shows the most variation in any segment when compared to the consensus type I 
sequence : NReAcApAc" in Xenopus and NRKDAEAcW in mammals (Figure 
3-2). 
Insufficient sequence data for type V IF chains are currently available to allow a similar 
analysis of the degree of homology amongst mammalian and amphibian lamin 
sequences. However a preliminary examination of their sequences reveals some 
pronounced similarities between several mammalian and amphibian members of this 
group. The human lamins A and C have identical amino-tenninal and rod domains, 
and a portion of their carboxy-terrninal domains are also identical. The Xenopus lamin 
CHAPfER 3: Sequence Homology page 69 
A rod domain sequence is 83% identical with these two human lamin chains and a 
further 7% of the sequences are maintained in chemical character (see Figme 2-2). The 
similarities extend throughout the protein sequences of the three lamin proteins: 23  of 
the 30 residues are identical in the N-tenninal domains (which include 4 blanks to 
maintain alignment) and 166 of the 283 residues are identical in the C-terminal 
domains of the lamin A sequences (which also include 4 blanks to maintain alignment; 
human lamin C is excluded because it has a shorter C-tenninal domain). A comparison 
of human and Xenopus lamin B chains would be of great interest but as yet no 
sequence data are available for the human lamin B protein. 
3 . 6 Periodic Features in the Homology Score Distributions 
Fourier transforms of the residue homology scores hr derived separately for the large 
apolar residues, the acidic residues and basic residues have been calculated for the 
coiled-coil segments IB and 2 using the look-up tables in Table 3-2. The methOd of 
performing the transform is similar to that described in Chapter 2 (Section 2 . 1 . 1 ) 
where the input data record is baseline corrected and zero-filled, although for this 
study the filled sequences were only 1024 elements in length and the resulting Fourier 
intensities have not been scaled. The Fourier intensities for types I to V chains are 
shown in Figure 3-3 and a selection of the most significant periods are listed in Table 
3-10. Note that the stutter in heptad phasing located towards the middle of segment 2B 
(residue 265) was taken into account by calculating separately the Fourier transforms 
of the apolar homology profiles either side of the stutter, ie. segments 2B 1 and 2B2 
(defined in Section 3. 1). 
The distribution of apolar residues in segment 2B shows no periodicity longer than the 
heptad repeat In segment I B  some less well defmed periods are revealed in addition to 
the heptad repeat. These other periods are also seen in the acidic and basic homology 
distributions described below. 
Periods common in the homology profiles of the acidic and the basic residues (orders 
marked with a in Table 3-10) are almost identical to those previously reported for the 
distribution of the same residue types in individual IF chains (-9.55 in segment IB 
and -9.85 in  segment 2 - see Chapters 1 and 2). This conservation of  charged 
residues is a further indication of their importance in specifying modes of molecular 
aggregation in the IF. Another period revealed by the transforms for segment IB  is 
equal to three times that of the -9.55 period. This longer period (28-30 residues or 
4.3 nm) can be considered as arising from the -9.5 residue repeat beating with the 
heptad period, ie. 3x9.5 "" 4x7.0 residues, and is an indication of the fact that charged 
residues occur frequently in the outer 12, £ and f positions of the coiled-coil structure 
CHAPrER 3: SeQuence Homology page 70 
(Parry. 1982; Parry and Fraser, 1985; also see Section 2.2 and Table 2-4a). Such 
positions have an inherent seven residue repeat. In addition to the 28-30 residue period 
in segment IB  there is a less well defIned period of about 10.5- 1 1 .0 residues in the 
distribution of homologous acidic residues in all but type n chains. This period is also 
present in the distribution of the homologous basic residues in type V chains. The - 1 1  
and -28 residue periods lie almost exactly i n  the simple ratio 3 : 8  (ie, 0.375 :  
type 1-0.372;  type III-0.377; type IV -0.375) which implies that they are both 
suborders of a longer regularity. For type 1 chains this segment IB period is 84 
residues (ie. 3x28.2 or 8xI0.5) and for type rn, IV and V chains it is 88 residues (ie. 
3x29.2 or 8x I 1 .0). Note that the 84 residue period (type I) is an integral multiple of 
heptads whereas the 88 residue period (types rn, IV and V) is not. The significance of 
this observation is presently unclear. 
1 .0 
0.5 
0.0 
?;-
J 
1 .0 
"E 0.5 
] Cl) 
0.0 
1 .0 
0.5 
0.0 
0 128 
(a) IB acidics 
256 384 
Frequency 1 .0 
0.5 
0.0 
0 1 28 256 
Frequency 
(b) 2 acidics 
(g) 2B2 apolars 
384 5 12 
Figure 3·3a Fourier transfonns of the homology arrays for type I IF proteins derived from the 
acidic residues in (a) segment IB and (b) segment 2, the basic residues in (c) segment 1B and (d) 
segment 2, and the large apolar residues in (e) segment 1B, (f) segment 2B 1 and (g) segment 2B2. The 
frequency axes are related to periods in the homology profIles by the expression: 
.od 1024 pen = frequency 
CHAPTER 3: Sequence Homology page 71  
As for the -9.5 residue period in segment 1B, the -9.8 period in segment 2 represents 
the third order of a longer repeat that is consistent among the chain types (29.5-29.8 
residues). The profiles of the acidic homologies in segment 2 for type I-IV chains 
show a period of -10.9 residues as do the type 11, III and IV basic homology profiles. 
This period is related to the -29.7 period in the ratio 4: 1 1 . The type I basic homology 
curve shows a slightly higher period of 1 1 .7 residues which is related to 29.7 in the 
ratio 4: 10. Neither of the -10.9 or 1 1 .7 periods occur in the type V basic homology 
distribution of segment 2. 
The three main periods found can beat together to show a longer range period of -1 1 8  
residues (ie. 4x29.7 = 1 1 8.8; lOx 1 1 .7 = 1 17.0; 1 1x lO.8 = 1 19.4). This period is 
considerably less than the length of segment 2 (143 residues) and thus cannot be 
attributed to an artefact of the edge effects arising from computation. 
1 .0 
0.5 
0.0 
>. 
... 
. � 1 .0 
.s 
� 0.5 ca u Cl) 
0.0 
1 .0 
0.5 
Frequency 
(a) IB acidics (b) 2 acidics 
(c) 1B basics (d) 2 basics 
0.0 -t-�-�..:.....I.-�--=---r-�L.--..---,.L...;..---l-�---l 
o 128 256 
Frequency 
384 5 12  
Figure 3-3b Fourier transforms of the homology arrays for type I I  IF proteins derived from the 
acidic residues in (a) segment IB and (b) segment 2, the basic residues in (c) segment 1B  and (d) 
segment 2. and the large apolar residues in (e) segment 1B,  (f) segment 2B1 and (g) segment 2B2. The 
frequency axes are related to periods in the homology profIles by the expression: 
.od 1024 pen = frequency 
CHAFfER 3: Sequence Homology page 72 
The main conclusions from this section are summarized as follows: 
(1) The homology profiles for large apolar residues in segment IB and segment 
2 in all chain types are dominated by the heptad repeat. This emphasizes the 
importance of these residues in specifying the heptad repeat and, by implication, the 
integrity of the coiled-coil structure of the IF molecule. 
(2) The highly conserved periodic distribution of acidic and basic residues in 
both segment IB and segment 2 implies that these residues play an important role in 
the self-assembly of IF molecules into IF through the formation of specific ionic 
interactions. 
(3) The degree of regularity in the distribution of the homologous ionic 
residues, as determined by the magnitude of the Fourier intensities and the numbers of 
orders present, varies between chain types. This indicates the possibility that 
alternative modes of assembly via ionic interactions may occur occur in the IF 
?;-
·a u 
... 
.El 
"3 
<a 0 Cl) 
1 .0 
05 
0.0 
1 .0 
05 
0.0 
1 .0 
0.5 
0.0 
0 128 256 
Frequency 
(a) IB acidics (b) 2 acidics 
(c) IB basics (d) 2 basics 
(e) IB apolars (f) 2B 1 apolars 
384 (g) 2B2 apolars 
1 .0 
05 
O.O-+-�-�r--"'�<......,.---.-�---+-��.....-j 
o 128 256 
Frequency 
384 5 12 
Figure 3-3c Fourier transforms of the homology arrays for type III IF proteins derived from the 
acidic residues in (a) segment 1B and (b) segment 2, the basic residues in (c) segment IB  and (d) 
segment 2, and the large apolar residues in (e) segment lB, (1) segment 2Bl and (g) segment 2B2. The 
frequency axes are related to periods in the homology profiles by the expression: 
.od 1024 pen = 
frequency 
CHAPfER 3: Sequence Homology page 73 
corresponding to the different chain types present. 
(4) Several periodicities occur in the distribution of homologous ionic residues 
in addition to the 9.3-9.9 residue repeat and such periods can be related to a repeat that 
is comparable (although shorter) than the segment length in both segment IB  and 
segment 2. 
0 
'� � 
.El 
'3 
(;l (.) Cl) 
1 .0 
05 
0.0 
1 .0 
05 
0.0 
1 .0 
05 
0.0 
0 128 256 
Frequency 
(a) IB acidics (b) 2 acidics 
(c) IB  basics (d) 2 basics 
(e) IB apolars (f) 2B 1 apolars 
384 (g) 2B2 apo1ars 
1 .0 
0.5 
0.04---+-,.:........,�-....:::::..�---1'--..--,,---;.--� 
o 128 256 
Frequency 
384 5 12  
Figure 3-3d Fourier transforms of the homology arrays for type IV  IF proteins derived from the 
acidic residues in (a) segment 1B and (b) segment 2. the basic residues in (c) segment IB  and (d) 
segment 2, and the large apolar residues in (e) segment lB.  (f) segment 2B 1 and (g) segment 2B2. The 
frequency axes are related to periods in the homology profIles by the expression: 
'od 1024 pen - frequency 
CHAPrER 3: Sequence Homology page 74 
(a) IB acidics (b) 2 acidics 
l .0 
05 
O.O�--�--��----�--------�r-�--------���------�� 
� 
J 
l .0 
� 05 
u Cl) 
O.O��-----L--�--���----�+-L-�--�----��--------� 
l .0 
05 
Frequency 
O.O�"""':''';''''':'''':'''''�...J..-----r--�....-;r-:-��-i 
o 128 256 
Frequency 
384 5 12  
Figure 3-3e Fourier transforms of the homology arrays for type V IF proteins derived from the 
acidic residues in (a) segment l B  and (b) segment 2, the basic residues in (c) segment IB and (d) 
segment 2, and the large apolar residues in (e) segment lB, (f) segment 2B 1 and (g) segment 2B2. The 
frequency axes are related to periods in the homology proflles by the expression: 
00d 1024 pen = frequency 
Type ! Type 11 
Period 
I B  Acidic .5 P 1) . . 
9.3 1 (P/3)a 1 . 1 8  7.21 (P/4) 0.37 
7. 1 1  (P/4) 0.68 5 .66 (P/5) 0.67 
5.69 (P/5)  0.74 4. 10 (pn) 0.55 
4.06 (PI/2) 0.43 3.57 (P/8) 0.52 
3 .51  (P/8, P1/3) b 0.58 3.22 (P19) 0.53 
2.33 (P/12)b 0.64 2.62 (Pil l )  0.39 
2.32b 0.87 
2. 1 9  (P/1 3) 0.39 
P = 28.7 res 
aSlC . 
0.60 
0.63 
P = 28.5 res (P = 28.7 res) 
Type III 
. Period 
1 1 .01 [p;q) 9.66 ) a 
5.5 1 (P'/2) 
4.85 (P/6) 
3.68 (P/8) 
3.52b 
2.93 (P/IO) 
2.75 (PI/4) 
2.33b 
1 .01 
0.42 
1 .09 
0.50 
0.68 
0.56 
0.82 
0.83 
P = 29.2 res 
P'= 1 1 .0 res 
a 0.69 
4.79 (P/6) 0.45 
3.23 (P19) 0.49 
(P = 28.9 res) 
Type IV 
. 
5.54 (p'n) 0.64 
4.88 (P/6) 0.82 
3.53b 0.74 
2.93 (P/1O) 0.63 
2.74 (PI/4) 0.67 
2.45 (PIl2) 0.50 
2.33b 0.79 
2.26 (P/1 3) 0.69 
P = 29.3 res 
P'= 1 1 .0 res 
. 
4.92 (P/6) 0.50 
(P = 29.8 res) 
Type V 
1 1 .01 (P'/1) 
9.94 (P/3)a 
3.5 1 b 
2.75 (PI/4) 
2.3 3 b  
P = 29.9 res 
P'= 1 1 .O res 
. 
9.23 (P!3)a 0.89 
3.95 (pn) 0.54 
2.76 (P/IO,P'/4) 0.61 
2.32 b 0.54 
P = 27.7 res 
pt= 1 1 .0 res 
Table 3- 10 A selection of the most significant periodicities in the homology scores of IF coiled-coil segments. (a) indicates the domimant 
9.3-9.9 residue period in the homology distributions of the acidic and basic residues and is directly comparable to the period found in the linear 
distribution of the same residues in individual sequences. (b) indicates either the second or third order of the fundamental heptad repeat in 
coiled-coil structures. A period in parenthesies, ego (P=29.6 res), indicates that only one order is present but that by analogy with other data the 
true period is shown. 
� 
VJ 
Vl 
.8 
� 
8 
g: 
9 g. 
� 
Q 
Type I Type n Type ill Type IV Type V 
� 
Period Intensity Period Period Intensit Intensit � 
1B Apolar 1 1 .91 (P'/1) 7.6 (PIl) 1 1 .91 �'Il � .44 . 1 .  5 CIJ .8 5.89 (P'/2) 0.58 5.57 (P/5) 0.82 5.92 p'n. 0.49 3.52b 1 . 17 1 . 1 8 § 3.50b 1 .3 1  4.66 (P/6) 0.50 4.02 (P'/3) 0.70 3.02 (P'/4) 0.77 
2.95 0.76 4.02 (pn) 0.83 3.5 1b 1 .3 1  2.33b 0.86 g ::z= 
2.33b 1 . 14 3.52 (P/8)b 1 .59 3.01 (P'/4) 0.99 0 3 
3.06 (P/9) 0.49 2.33b 0.98 
� 2.78 (P/lO) 0.54 
2.32 (P/12)b 0.66 
2. 12 (P/13) 0.53 
heptad heptad heptad heptad heptad 
P'= 1 1 .8 res P = 27.8 res P'= 12.0 res P'= 12.0 res 
1 . (P'Il) . 1 . (P 1 )  . .7 
9.85 (P!3)a 1 .06 9.85 (P/3) a 1 .2 1  9.75 (P!3) a 1 . 13  1 1 .01 (P'Il) 0.62 0.76 
5.92 (P/5) 0.76 5.89 (P/5) 0.53 7.47 (P/4) 0.85 9.66 (P/3) a 0.94 0.59 
4.27 (pn) 0.64 5.45 (P'/2) 0.74 5.95 (P/5) 0.77 7.53 (P/4) 0.7 1 
3.7 1 (P/8) 1 .00 3.72 (P/8) 1 . 1 5  5.45 (P'n.) 0.33 5.95 (P/5) 1 .01  
2.70 (PIl l ,  P'/4) 0.85 2.98 (P/lO) 0.64 4.25 (pn) 0.39 3.74 (P/8) 1 .02 
2. 17 (P'/5) 0.90 2.7 1 (Pil l ,  P'/4) 0.83 3.7 1 (P/8) 1 . 1 3  3.36 (P!9) 0.86 
2.29 (PIl3) 0.8 1 2.99 (P/1O) 0.7 1 2.99 (P/1O) 0.53 
2. 17 (P'/5) 0.90 2.70 (Pil l ,  P'/4) 0.44 2.33b 0.66 
2.47 (PIl2) 0.35 
2.29 (P/13) 0.54 
2. 12 (PIl4) 0.5 1 
P = 29.7 res P = 29.5 res P = 29.7 res P = 29.8 res P = 29.8 res 
P'= 10.9 res P'= 10.9 res P'= 10.9 res (P'= 1 1 .0 res) 
11 
C§ 
Table 3-10 continued � 
2 Basic 
Type J 
Period 
1 1 .77 (P"/l ) . 
9.85 (p/3)a 0.65 
5.89 (P/5, P"/2) 0.83 
2.92 (P"/4) 0.62 
2.33 (P"/5)b 0.46 
P = 29.5 res pIt= 1 1 .7 res 
2.32 b 
heptad 
. 
Type 11 
Period 
10.78 �'/l )  . 
9.85 /3) a 1 .00 
2.33b 0.48 
(P = 29.5 res) 
(P'= 10.8 res) 
heptad 
Type m 
Period 
10.78 �'/1 ) 
9.85 /3) a 
7.47 (P/4) 
5.39 (P'/2) 
4.27 (pn) 
3.58 (P'/3) 
2.99 (P/lO) 
2.69 (P'/4) 
2.33b 
• b 
2.32b 0.61 2.34b 0.62 2.32b 
heptad heptad he tad 
Table 3·10 continued 
1 . 13 
0.73 
0.57 
0.66 
0.60 
0.55 
0.65 
0.70 
TypeN Type V 
Period 
10.89 fJh1) . 
9.94 ) a 0.96 
7.42 (P/4) 0.67 
3.35 (P19) 0.77 
2.69 (Pil l ) 0�77 
2.46 (PIl2) 0.48 
2.33b 0.45 
P = 29.8 res 
(P'= 10.9 res) 
(P = 29.6 res) 
. 
2.34b 
he tad 
. . 
0.53 2.32b 0.41 
he tad 
CHAPrER 3: Sequence Homology 
3 . 7 Consensus Sequences for IF Chain Types 
page 78 
Consensus sequences for the rod domain and for the HI and H2 segments of the IF 
types Ib, ITa, llb, Ill, IV-L, IV-M, IV-H and V are shown in Figure 3-4, as is  the 
single type la sequence available. An overall consensus IF sequence is also shown. In 
those situations where several residues occur with approximately equal probability 
then each residue is listed. In some case, merely the residue type is listed, ie. apolar -
Ap; acidic - Ac; basic - Ba. Residues that are conserved within a chain type but differ 
between the types may be considered as characteristic markers of these chains types. 
In addition, they clearly indicate probable and potentially important variations in 
molecular function and local structure among the types. 
The homology study permits an analysis of the most conserved positions within the 
heptad substructure or residue types within each position. The results of this analysis 
are listed in Table 3- 1 1  and confmn an earlier report by Parry and Fraser ( 1985) that 
the 'inner' positions a and d of the heptad were those most highly conserved, that the 
next most conserved positions were the k and � residues (ie, those immediately 
adjacent to a and d in axial projection) and finally those least conserved lay in the 
'outer' positions of the coiled-coil, ie. Q, k and f. This emphasizes that residues in 
position a and Q., which are predominantly apolar, are crucial to the integrity of the 
coiled-coil structure and that residues in the k and � positions, which are commonly 
charged, have considerable importance in providing further stability to the structure 
and also in specifying the relative polarity of the constituent chains through the 
maximization of ionic interactions. Residues in positions 12, £ and f have particular 
importance in the process of molecular aggregation but are more easily able to tolerate 
sequence variations without losing their ability to specify precise modes of linear and 
lateral assembly. This is especially true for the outermost position f. 
Heptad Positions 
a 12 c d e f g 
Mixed homology tabl& 74.9 59.5 56.6 78 .0 7 1 .4 52.7 66.3 
Identity homology tableb 60. 1 5 1 .3 48. 1 67.4 60. 1 4 1 . 3  57.4 
Acidic homology tableC 1 .0 28.5 29.4 7 .3  25 .3  22 .0 30.2 
Basic homology tableC 12.2 1 5. 1  1 1 .9 2.2 29.7 1 7 .6 19 .7 
Apolar homology tablec 7 1 .9 7 . 8  9.8 72.4 1 6.4 13 .5  1 5. 1  
Table 3-1 1 Mean residue homology scores (hr) calculated for each position in the heptad 
substructure using a selection of homology criteria. a See Table 3 - 1 .  b As for Table 3- 1 but with 
zeros everywhere except along the diagonal. c See Table 3-2. 
CHAPTER 3: Sequence Homology page 79 
la 
Ib 
Ba 
Db 
m 
IV·L 
IV·M 
IV·H 
v 
ComeD. 
la 
Ib 
Ba 
Db 
m 
la 
Ib 
Ba 
lib 
Sep!lODlHl 
. S S P . ; - P � � S T S Y KR R Y V B T P R V H I S S V R S OY S T AR S AY S S Y S A P V S  
. S YT L D S L O NP S S AYR R P V T B T R S S P S R V S OS P S S O P R SQ S S WS R O S P S T V S S S Y KR S � � 
. M S P O : A DA L L OA P P A P L HO OO S L H Y : L : R K : O : OO TR S A AO S S S O P HSW : RT S 
. O L L · O  
. P CS I T T V S VN BS L L T P L NL B I DP N & QC VKQ B 
. P VC :  P : O I  Q B V T  .. N QS L L T P  L NL - I Dp ·  I Q " V" . B 
• V D  P S  L A I A .. NT , P K ·  T R T r .. 1:-"'ft1!'''''' 
S L S V R R S Y S S S S OS O S L � P S L B NL D � S Q VA A I S ND L K S I R � QBI��).�t 
P R L t Y S S AML S S AB S S L D P S Q S S S L L I O OS O P OO D YK L S R S N  
S S V S A S P S R P R OAOA AS S T D S L DT L S NO P B O C : VA : : : : R S ·  
Figure 3-4 The consensus sequences for each of the IF chain types and sub types are l isted and 
aligned to aid comparison. The segment boundaries are also marked and extend from segment HI  
through the rod domain to segment H2 (where present). Those residues that are conservatively 
substituted are shown boxed and those that are absolutely maintained in a chain type or subtype are 
shown shaded. Residues which are identical across all chain types are shaded and boxed. The overall 
consensus sequence (Consen.) is also listed. The symbols used in this figure have the following 
meanings: ., inserted blank; ?, incomplete sequence; ., chain termination; «, continuation towards the 
N-tenninus; », continuation towards the C-terminus; -, position of low homology where a consensus 
cannot be assigned; 0, the site of a 70 residue replacement for residues 99- 126 (ie. a net 42 residue 
insertion); Ac, acidic residues aspartic acid and glutamic acid; Ba, basic residues arginine and lysine; 
Ap, apolar residues alanine, isoleucine, leucine, methionine, phenylalanine, tyrosine and valine. 
CHAPrER 3: Sequence Homology page SO 
la 
Ib 
IIa 
Db 
m 
la 
Ib 
IIa 
Db 
m 
la 
Ib 
IIa 
Db 
m 
IV-L 
IV-M 
IV-H 
v 
CoaoaL 
• � OV O : V NV .  
O E O V O - V N � S V S - S � S S S .  
.. p .. - , T  P - S - L N .. R E T  S � .. - - - .. S .. - H ,  K R  T � .. I KT  V E T R D O B  V I - B t 
P T S V OS � T � OY S Q S S QV P OR S AY S OL Q ; S S Y LM S : R : P P : Y YT S H V Q  
i;iii;:�;.�.,I' .. T P t O S I T  G P  L Y T HR , P S � T I S S K I  Q K T  K V E A P K L K V Q H K P V E B I I B B  
P OP : P P S L : E O L P K I P S � S T H I K V KS E E : I K V V B KS E K E T V I V E : QT 
: Q : : . D V � . 
a Q - B V B B T I B A - K A B B AK DB P P S B O B A B B B : K " K . B . � V B D E K S EMBB t L T A I . E E L AA S A E KE E K I E - A E B KE E E . � l I V T B � V T B E E .. K E A ' �  ; E O I B t B � O E B I E ' E � O t t : T .  
Figure 3-4 continued 
CHAPfER 3: Sequence Homology page 8 1  
Type I TypeII Type ill TypeN Type V 
Trp 65% Phe 76% Cys 90% Trp 65% Tyr 44% 
Tyr 44% Glu 54% Tyr 42% Glu 53% Phe 33% 
GIu 41% Lys 53% GIy 39% Leu 47% Leu 30% 
Pro 41% Tyr 45% His 31% Arg 36% Glu 26% 
Leu 38% His 42% Leo 30% De 35% De 21%  
GIn 36% De 42% Arg 26% Tyr 26% 
Arg 33% Arg 41% Glu 24% GIn 24% 
Ala 40% 
Leu 38% 
Gly 38% 
Table 3-12 Percentage occunence of highly conserved residues in the rod domain (�) 
Specific amino acids that are highly conserved in some position in the rod domain 
sequence (h�%) are listed in Table 3-12. Note the generally low values for type V. 
This is due to the relatively low homology between the lamins and the Helix pomotia B 
protein. Generally, the observations are consistent with the pattern of residues 
conserved in other families of related proteins. For example, tryptophan, tyrosine and 
phenylalanine are generally conserved at a high level, possibly as a consequence of 
their known role in providing regions of structure with a high stability (Ruber, 1 979; 
Burley and Petsko, 1985). Features more specific to individual IF chain types include 
the high percentage conservation of glutamic acid and proline residues in ' acidic' type I 
chains, of glutamic acid, lysine, arginine and histidine in 'neutral-basic' type IT chains, 
and of cysteine and glycine in type ill chains. 
In conclusion the data again illustrate the probability that IF containing different chain 
types will exhibit polymorphism. 
3 . 8  Summary 
The homology study has revealed several novel aspects of IF structure as well as 
confirming other, established ideas. The general classification of IF proteins into 
several types is reflected in the difference in homology scores between sequences 
belonging to the same type and sequences from different types. In partic ular, 
neurofllament IF chains are confirmed to be distinct from the type III chains and have 
been classified as type IV. On the basis of homology, the neurofilaments have also 
been shown to comprise at least three sub-types, ' light' (NF-L), 'medium' (NF-M) 
and 'heavy' (NF-H). The delineation of keratins into two sub-types ( 'hard' and 'soft') 
on the basis of cell source and chemical character is also apparent in the segment 
CHAPTER 3 :  Sequence Homology page 82 
homology scores (hs) for the rod domain segments (Section 3.3. 1).  Also, the 
relatively low homology between the type V lamin chains and the Helix pomotia B 
chain suggests that further subdivision of this class may be necessary. 
An important conclusion that can be made from the differing degrees of homology 
exhibited by the various chain types is that IF composed of type I/type II chains, type 
III chains, type N chains or type V chains may have slightly different structures (ie, 
exhibit polymorphism). This may be manifested in minor variations of molecular 
packing or in the precise geometry of the surface lattice of the IF on which the 
tetrameric structural unit is known to be positioned (see Chapter 1) .  Type V IF 
molecules have already been shown to form a variety of related polymorphic forms 
(Zackroff et ai, 1984; Aebi et ai, 1986; Parry et ai, 1987a; also see Section 4.3). 
There are highly significant peaks in the Fourier transforms of the arrays of the 
conserved acidic, the conserved basic and the conserved large apolar residues within 
the rod domains of a particular chain type. These periods are related to one another and 
indicate a long-range period comparable (although not related) to the length of the 
segment from which the array originated. Primarily, however, the most significant and 
common peaks (-9.55 and -9.85 residues in segments IB and 2 respectively) parallel 
exactly those previously discovered in the linear distribution of all acidic, basic and 
apolar residues (considered separately). The evidence is now overwhelming that the 
conservation of these residues is crucial to specifying the structure of the coiled-coil 
and the modes of aggregation of IF molecules through ionic interactions. 
Subdomains within the amino- and carboxy-terminal domains have been more fully 
delineated than has been possible previously. The original scheme, which was 
developed for the keratins, is not as relevant for the other IF chains although the 
homologous regions, HI and H2 can be assigned. New evidence has been presented 
for the existence of a small HI subdomain in type Ib mammalian chains. In addition, 
both HI and H2 subdomains are present in the type N chains and in the type V 
lamins. 
The keratin chains from amphibia are readily classified according to the type IIII 
nomenclature derived initially for mammalian keratins. However, the type I Xenopus 
chains show less homology to their mammalian counterparts than do the type 11 
chains. This is also true of the hard mammalian keratins when compared to the 
epidermal keratins and may be an indication of a greater conservation in the primary 
structure of the rod domains in the type II chains compared to that in the type I chains. 
Consensus sequences for the various IF sUbtypes have been presented. Highly 
CHAPrER 3: Sequence Homology page 83 
conserved residues imply positions crucial either to the snucture or function of IF in 
general; other residues that are less well conserved may nonetheless represent 
important elements of specialized structure or function. Residues within the .il and d 
positions are shown to be conserved at a higher level than those in the other positions: 
this reflects their importance in specifying the coiled-coil structure. The k and � 
positions, important for the alignment of chains in the molecule, are the next best 
conserved followed by the 'outer' positions, h. k and f. It is apparent that functional 
and structural diversity can be built into the rod domain without upsetting the 
fundamental organization of the IF molecule. 
More data, especially the sequences of hard a-keratins, neurof'llaments and type V 
proteins, are required to verify some of the conclusions presented in this Chapter and 
to extend others. 
CHAPrER 4: Secondary and Tertiary Structure 
4.  SECONDARY AND TERTIARY 
STRUCTURE 
page 84 
Ionic interactions have been shown to play an important but somewhat different role 
from apolar and hydrogen-bond interactions in the aggregation of fibrous biopolymers 
(see, for example, Hulmes et aI, 1973; Parry, 1975, 198 1 ;  McLachlan and Stewart, 
1976). The range of influence for ionic interactions is much greater than that for apolar 
interactions - the electrostatic force which comprises the ionic interaction is inversely 
related to the separation of the participating charged atoms (ignoring the effects of the 
intervening medium) whereas the van der Waals attraction that mediates apolar 
interactions attenuates by the sixth power of the separation. Also, charged residues 
generally have longer and more flexible sidechains than the apolar residues and these 
sidechains are often situated on the periphery of the molecule. In contrast, apolar 
residues tend to be 'internalized' to shield them from the polar, aqueous environment 
of a biological system. Hence, the ionic interactions cause adjacent molecules to come 
to a close and suitable conjunction prior to the final 'docking'  provided by apolar and 
hydrogen-bond interactions. It follows then that the ionic interactions are often crucial 
in specifying the mode of chain and molecular assembly and that the apolar residues 
serve chiefly to stabilize the molecules and maintain them in a suitable configuration 
once they have been assembled. For this reason, theoretical studies on the hierarchical 
structure of IF and other fibrous proteins have paid particular attention to the positions 
of the charged residues in the amino acid sequences in order to determine the most 
likely mode of assembly based on mutual attraction between oppositely charged 
residues. 
Interchain ionic interactions are calculated for rat peripherin, the lamin proteins and the 
Helix pomotia B protein in order to determine the likely modes of chain aggregation. 
These are compared with previous studies on other IF chains. One-dimensional ionic 
interactions between molecules are also studied for these proteins. In particular, the 
assembly of chains into the molecule and the determination of the likely mode of 
molecular aggregation for the nuclear larnin proteins is examined. Maxima in the ionic 
interaction curves for larnin molecules are used to find likely axial staggers between 
the molecules in vivo and in vitro, and which in turn lead to a model that is consistent 
with the electron microscope data obtained from negatively-stained lamin paracrystals. 
The models are also compared with data from filamentous structures of lamin 
molecules and the in vivo lamin mesh work. Suitable physical data, such as the axial 
OIAPI'ER 4: Secondary and Tertiary SbUCture page 85 
periodicities that may be derived from paracrystalline structures, are currently very 
scarce for IF chains and similar studies have only been undertaken on a GF AP protein 
fragment (Stewart et ai, 1989b). 
Finally, an attempt is made to extend further the one-dimensional techniques for 
calculating inter-molecular ionic interactions to a more general, three-dimensional 
method. Preliminary results of this study are reported and show that appreciable 
progress has been made towards refining the technique into a useful tool for predicting 
molecular structure based on amino-acid sequence information. 
4 . 1  ID Ionic Interactions Between Chains 
Interchain ionic interactions have been studied previously for a selection of IF proteins 
(for example Parry et ai, 1986). In all cases a parallel in-register arrangement  of 
chains has been preferred, a result which is in agreement with the physico-chemical 
data described in Chapter 1 .  In this section, the interchain ionic interactions which 
stabilize homodimeric molecules are examined for the recently sequenced IF proteins, 
peripherin, human lamins A and C, Xenopus lamins A and B and the Helix pomotia B 
protein. 
Residues in heptad positions a and � on one chain may interact with residues on the 
other chain in the axially adjacent ,  position, and similarly for the Q. and , positions 
with the axially adjacent k position. Only segments of unbroken heptad substructure 
are examined in this study and hence the rod domain of the IF chains is  analysed 
piecemeal and the results for each coiled-coil segment are summed. Segment 2B, for 
example, has been subdivided into two sections of continuous coiled-coil since a 
discontinuity in heptad phasing occurs near the centre of the segment: the amino­
terminal piece, 2B l ,  is 58 residues in length and the carboxy-terminal piece, 2B2, is 
62 residues. The linker Ll of the lamin and Helix B proteins is reduced to a 
discontinuity in the coiled-coil structure of segments l A  and 1B  and so the lengths of 
these segments are 4 1  and 147 residues respectively. In types I-IV IF chains segment 
lA  is 35 residues in length and segment IB is 101 residues. 
Chain segments are staggered axially in integral heptad steps so as to maintain the 
internal apolar core of the coiled-coil dimers. Ionic interactions between residues in the 
appropriate heptad positions of adjacent chains are scored according to their potential 
to attract (+ 1 )  or repel (- 1 )  each other, or are scored .zero if they are considered non­
interacting. Scores are made for each stagger; in addition a total score is calculated. 
The results are given in Table 4- 1 .  In all cases the highest positive score is found for 
parallel in-register chains. This is consistent with the generalized model for the IF 
CHAPrER 4: Secondary and Tertiary Structure page 86 
Stagger P,ri�h,rin: Prurnll,lChrunS 
(heptad) lA 1B  2A 2Bl 2B2 Total 
0 2 10 0 4 2 1 8  
± 1  2 4 0 0 0 6 
±2 -4 3 - 1  0 0 -2 
Stagger feri12herin; Anti12ara11,1 Chains 
(heptad) lA IB  2A 2Bl 2B2 Total 
-2 4 3 0 0 -2 5 
- 1  - 1  -4 -1 0 -5 - 1 1  
0 -2 -2 0 -2 -2 -6 
1 -4 -7 0 -5 -5 -21 
2 4 -6 0 2 -2 -2 
Table 4-1a Interchain ionic interactions for rat peripherin. The chains are staggered by integral 
multiples of the heptad (ie, seven residues) in order to maintain the integrity of the coiled-coil 
structure. Segment 2B has been analysed in the two pieces of continuous heptad structure either side 
of the discontinuity near the centre of 2B. The length of 2B I and 2B2 is 58 and 62 residues 
respectively. 
Stagger 
(heptad) 
0 
± 1  
±2 
Stagger 
(heptad) 
-2 
- 1  
0 
1 
2 
Hwnan Lamins A a.nd C; faralld CbaiDS 
lA IB  2A 2B1 2B2 
6 1 8  0 4 8 
-2 1 0 1 -4 
- 1  4 0 1 -4 
Human Lamins A and C; Anti12arall,l Chains 
lA 1B 2A 2B1 2B2 
- 1  -2 0 0 6 
-2 1 - 1  -2 -9 
2 -6 0 2 8 
2 6 0 - 1  -5 
-6 0 0 4 0 
Total 
36 
-4 
8 
Total 
-3 
- 1 3  
6 
2 
-2 
Table 4-1b Interchain ionic interactions for human lamins A and C which have a common rod 
domain sequence. See the caption to Table 4-la for details. Also, due to the continuation of heptad 
structure into the Link region Ll the lengths of segments lA and IB  are 4 1  and 147 residues 
respectively. 
CHAPTER 4: Secondary and Tertiary Structure page 87 
Stagger X.e.lJQ1lwlamin A; farall,l Chains 
(heptad) lA 1B 2A 2B1 2B2 Total 
0 6 18  0 4 8 36 
±1 -2 1 0 1 -4 -4 
±2 -1  4 0 1 -4 8 
Stagger X.'CW,llW larnin A; Annl2mll�l ChainS 
(heptad) lA 1B 2A 2B1 2B2 Total 
-2 -1 -2 0 0 6 -3 
- 1  -2 1 - 1  -2 -9 - 1 3  
0 2 -6 0 2 8 6 
1 2 6 0 - 1  -5 2 
2 -6 0 0 4 0 -2 
Table 4-1c Interchain ionic interactions for Xenopus lamin A. See the caption to Table 4-1b. 
Stagger 
(heptad) lA 
0 2 
±1  0 
±2 0 
Stagger 
(heptad) lA 
-2 0 
- 1  0 
0 1 
1 -2 
2 -3 
Xe.l1Q/J.wlamin B: Parallel Chain� 
1B 2A 2B1 2B2 
10 0 2 
2 0 - 1  
7 0 2 
Xeno12wlamio B; Aotil2w:all,1 Chains 
1B 2A 2B1 
-6 0 -2 
0 - 1  0 
0 0 2 
4 0 - 1  
2 0 0 
8 
-3 
3 
2B2 
4 
-8 
6 
-5 
-2 
Total 
22 
-2 
8 
Total 
-4 
-9 
9 
-4 
-3 
Table 4-1d Interchain ionic interactions for Xenopus lamin B. See the caption to Table 4- 1b. 
CHAYrER 4: Secondary and Tertiary Sttucture 
Stagger 
(heptad) lA 
0 2 
±1  2 
±2 -2 
Stagger 
(heptad) lA 
-2 3 
- 1  -3 
0 -7 
1 2 
2 1 
Helix l1fltllQria B; Paral1�l Chain� 
IB 2A 2B1 
12  0 4 
-8 0 - 1  
7 0 1 
H e.lix Ilomoria B; Anti12arall�l ChainS 
IB  2A 2B1 
-6 0 0 
7 - 1  -2 
-4 0 2 
0 0 -2 
2 0 2 
page 88 
2B2 Total 
0 1 8  
2 -5 
-3 3 
2B2 Total 
-4 -7 
5 -6 
-6 -15 
0 0 
-4 1 
Table 4- 1e Interchain ionic interactions for Helix pomotia B. See the caption to Table 4- 1b. 
molecule and provides further support for it. Furthermore, the interchain ionic 
interactions for the human and Xenopus lamin A proteins are identical. Examination of 
the charged residues in the � d.. k and � positions of the heptad-containing regions for 
both sequences reveals that only six substitutions are made and in all cases the charge 
character is conserved (ie, the basic residues arginine and lysine may be interchanged 
as can aspartic acid and glutamic acid; see Figure 2-2). 
The total scores for parallel chains are summarized in Table 4-2 along with those for 
several other IF proteins. Peripherin (Table 4-2a) is very similar to other type In IF. 
The lamin A chains (Table 4-2b) reveal particularly high scores for the in-register 
arrangement: they are at least twice as great as for the type I to IV chains. This is a 
reflection in part of the greater number of charged residues in the rod domain of Iamin 
A (for example, human lamin A has 1 1 8  charged residues in segments I B  and 2 
combined, compared to an average of 90 for types I-IV IF: Parry et ai, 1986) but also 
indicates that a greater proportion of these charged residues are in a position to make 
interchain ionic bonds than for other IF. Table 4-3 lists the total ionic interactions per 
heptad pair for the coiled-coil domains of segment 1 ,  segment 2 and the rod. This may 
be viewed as an "interaction density". Only the scores for the parallel in-register 
arrangement of chains are included. Segment 1 generally has a much greater density of 
interchain ionic interactions than segment 2 although in the case of Xenopus Iamin B 
the scores are similar. The lamin A proteins show the greatest interaction density (0.89 
in segment 1 )  and the Helix pomotia B protein has the lowest (0.20 in segment 2). 
One conclusion is that lamin A molecules inherently have much more stability than 
those of any type I-IV IF. 
CHAPTER 4: Secondary and Tertiary Structure page 89 
Stagger H50K H56K COD RELV MGFAP PNF-L Peripherin 
(heptad) 
0 8 12 14 16 18 12 1 8  
± 1  5 6 2 5 1 - 1  6 
±2 4 3 -4 0 -8 1 -2 
Table 4-2a Interchain ionic interactions for parallel chains from a selection of type I-IV IF 
proteins (Parry et ai, 1986) including peripherin. H50K and H56K are type I and type 11 IF chains 
respectively from human epidermal keratin; CGD - chicken gizzard desmin, HEL V - hamster eye lens 
vimentin, MGF AP - mouse glial fibrillary acidic protein and peripherin are type III IF chains; and 
PNF-L -porcine neurofilament light chain is a type IV IF chain. Sources are listed in Table 2-1 .  
Stagger Human lamins Xenopus lamin Xenopus lamin Helix pomotia 
(heptad) Aand C  A B B 
0 36 36 22 1 8  
± 1  -4 -4 -2 -5 
±2 8 8 8 3 
Table 4-2b Interchain ionic interactions for parallel chains from V IF. Human lam ins A and C 
share a common rod domain sequence and hence have identical interaction scores. The rod domain 
sequence for Xenopus lamin A is also very similar to the human lamins A and C. 
Protein lA+IB 2A+2Bl+2B2 Total 
Peripherin 12 in 193/7 = 0.62 6 in 196/7 = 0.30 1 8  in 392/7 = 0.46 
Human Lamins A&C 24 in 266/7 = 0.89 12 in 196/7 = 0.60 36 in 465/7 = 0.77 
Xenopus Lamin A 24 in 266/7 = 0.89 12 in 196/7 = 0.60 36 in 465/7 = 0.77 
Xenopus Lamin B 12 in 266/7 = 0.45 10 in 196/7 = 0.50 22 in 465/7 = 0.47 
Helix pomotia B 14 in 266/7 = 0.52 4 in 196/7 = 0.20 1 8  in 465/7 = 0.39 
Table 4-3 Interchain ionic interactions per heptad pair for the proteins listed in Table 4- 1 .  Only 
interactions for parallel in-register chains are included. The heptad regions in segments 1 and 2 have 
been combined separately (lA+IB and 2A+2B l+2B2 respectively) and together (Total). The count of 
heptads for each combined group is given in fractions of seven for conciseness. 
CHAPrER 4: Secondary and Tertiary Structure 
4 . 2  iD Ionic Interactions Between Molecules 
page 90 
A very simple model of fibrous protein structure has been used successfully to relate 
the amino acid sequence of collagen to the relative axial positions of type I collagen 
molecules in the fibrils found in vivo (Hulmes et ai, 1973). The mechanism on which 
the preferred mode of assembly was based was found to be attributable largely to ionic 
interactions with a smaller contribution from apolar interactions; hydrogen bond 
interactions between hydrophilic residues were not studied. In the model of Hulmes 
et al ( 1973), the amino acid residues in an a-chain were spaced evenly in a linear, 
one-dimensional array. Two such arrays, each representing a single chain in adjacent 
molecules, were aligned either mutually parallel or antiparallel. Starting from a 
position of maximum relative stagger (ie, minimum overlap) the arrays were moved 
past each other and a score of possible apolar and ionic interactions between residues 
on the different chains was generated for each position of relative axial displacement 
(or stagger). Charged residues lying within three linear residue translations of each 
other were assumed to be able to interact; for apolar residues the interaction range was 
reduced to two. The scores revealed a general trend toward greater potential for 
interactions as the molecules become increasingly overlapped, as was expected. 
However, overlaid on this background was a regular pattern of peaks with notable 
maxima occurring at intervals of 234 residues (ie, -67 nm), a value previously 
determined from X-ray diffraction and electron microscope data and which 
corresponds to the axial stagger between molecules in the fibril (Bear, 1944; Hodge 
and Petruska, 1963). 
The method of Hulmes et al (1973) for calculating the potential number of ionic 
interactions between charged residues from different molecules of collagen has also 
been applied to several of the a-fibrous proteins such as tropomyosin (Parry, 1 975; 
McLachlan and Stewart, 1976), myosin (Parry, 198 1 ;  McLachlan and Karn, 1983), 
paramyosin (Cohen et ai, 1987) and IF (Crewther et ai, 1 983; Fraser et ai, 1985, 
1986; Parry et ai, 1987 a). Fraser et al (1985) studied all possible combinations of the 
IF rod domain segments IB and 2: IBU- IBU, IBU- IBD, 2U-2U, 2U-2D, IBU-2U, 
IBU-2D where IB and 2 are the major rod domain segments and U (up) and D 
(down) designate the chain directions with regard to the progression from the N- to the 
C-terminal end of the segment. Note that U-U and D-D are equivalent as are U-D and 
D-U. Due to the large number of maxima in the resulting interaction curves, a 
technique was employed for estimating the quantiles of scores from 1000 randomized 
sequences at each stagger of the chains. Only those maxima above some specified 
cutoff were deemed to be significant. The conclusion of this study was that significant 
CHAFfER 4: Secondary and Tertiary Structure page 91  
maxima in the interaction curves were indeed consistent with the dimensions of the 
surface lattice of a-keratin as determined by X-ray diffraction analysis, and favoured 
models were proposed as a basis for further studies. 
The one-dimensional ionic interaction study on IF molecules was extended by Parry 
et al ( 1987a) to include the recently sequenced human lamin proteins (McKeon et aI, 
1986; Fisher et ai, 1986). Lamin presented a simpler case than other IF in that the 
entire rod domain was predicted to be a-helical and hence the spacing between 
residues in the entire rod domain could be assumed constant. This meant that the 
analysis need only be performed for two cases - the rod domains of parallel (U-U) and 
antiparallel (U-D) molecules - rather than for combinations of smaller segments of the 
rod domains of the molecules as has proved necessary for other IF types. Also, the 
lamin molecules were assumed to be homodimeric in the rod domain as the sequences 
of human lamins A and C are identical from the N-terminus right through to a point in 
the non-helical C-terminal domain. The ionic interaction curves for lamin molecules 
are shown in Figure 4- 1 .  
The general trend for the two interaction curves is an approximately linear increase in 
interaction score with increasing overlap - an indication of the relative evenness in the 
distribution of charged residues within the rod domain. The interaction curve for 
parallel homodimeric molecules (for example, Figure 4-1 ,  U-U) shows a substantial 
dip in the score at zero stagger. This is due to the exact alignment of charged residues 
of the same sign from both homodimers which effectively reduces the chances of 
finding an oppositely charged residue within the interaction range - in this case the 
interaction range is ±1 residue translation and hence the reduction is by a factor of 
approximately one in three. 
The relationship between ionic interaction score and overlap is not sufficiently regular 
to allow significance to be assigned to peaks in the curve on an analytical basis. This is 
especially apparent in the top graph of Figure 4- 1 at zero stagger (as described above) 
and in the lower graph of Figure 4- 1 where there are several wobbles in the general 
trend for positive staggers. The empirical strategy employed by Fraser et al ( 1985) 
was also used by Parry et al ( 1987a) in analysing the ionic interaction curves of lamin 
although it was revised to take into account one of the grosser non-uniformities in the 
distribution of residues in the heptad substructure (described below). Also, unlike 
keratin, the lamin molecules were assumed to be homodimeric and consequently, 
when calculating significance levels, the same randomized sequence was used for both 
chains of both molecules. 
� o u 
c.n 
I:::::: o .-.... 
J 
u 
·S o -
150 
125 
100 
75 
50 
25 
Human Lamin A: U-U 
O+-----�r_----_.--------�----�r_----�--------�----�------�----�--�----�----�--�------�----��----� 
150 
125 
100 
75 
50 
25 
O+---�----�--�----�--�----�--�----�--�----�--�----�--�----���--� 
-400 -300 -200 -100 o 100 200 300 400 
Residue Translation 
Figure 4·1 Ionic interaction curves for parallel (U-U) and antiparallel (U-D) rod domains of human larnin A. The curves for human larnin C 
are the same as these because the rod domain sequences are identical. Figure adapted from Parry et al (1986). 
CHAFfER 4: Secondary and Tertiary Structure page 93 
The original method of Trajstman and Lucas for identifying interaction scores that 
were above some predetermined limit of significance was described by Fraser et al 
( 1985). In this procedure the residues of each of the four chains were randomly 
reassigned to positions within the sequence and a new interaction curve was calculated 
in the same manner as for the originals. This randomization and recalculation was 
carried out 1000 times and an empirical distribution of scores was generated for each 
relative stagger of the molecules .. From each distribution a quantile for the particular 
stagger was determined (So) and then compared to the actual ionic interaction score of 
the rod domain sequences (S) at the same stagger in order to determine if the actual 
score was significant or not. The cutoff level used by Fraser et al ( 1985) was 0.975. 
A modified version of this method was developed by Parry et al ( 1987a) in order to 
take into account the high occupancy rate (-75%) of the � and d positions in the heptad 
by apolar residues (an essential feature of the coiled-coil structure, Parry and Fraser, 
1985). The new procedure involved grouping the residues in the i! and d positions 
separately from those in the other positions. Reassignment on a random basis was 
made independently within the two groups so that the i! and Q. positions remained 
-75% apolar. An additional refinement to the method was to generate only one 
randomized chain for each of the 1000 iterations and to infer the other three chains 
from the first rather than to generate four separately randomized chains as per Fraser 
et al ( 1985). This is appropriate for the type ill-V chains where two homodimeric 
molecules (ie, four identical chains) are being examined. As before, a level of 0.975 
was used as a test of significance. 
Other refinements along similar structural lines are possible for this analysis. 
However, the gross simplification of the chain structure from a supercoiling of 
a-helices to a pair of linear arrays probably means that such efforts would not 
significantly improve the accuracy of predicting the packing modes available to the 
molecules in vivo. Another approach which could improve the accuracy of the method 
is to weight differently the four combinations of ionic interactions: Asp-Lys, Asp-Arg, 
Glu-Lys, Glu-Arg. This could be done according to the frequency of such 
combinations in a database of proteins with known structure. However, insufficient 
data are available at present to warrant this degree of sophistication. 
The modelling of a three-dimensional coiled-coil as a simple linear array of residues 
does, of course, ignore the radial and azimuthal information about the residues and 
only utilizes their axial positions. This simplification of the chain structure has been 
made primarily to render the analysis computationally tractable. However, the 
assumption that the axial coordinate of the residues is of greater relevance for this 
CHAPrER 4: Secondary and Tertiary Structure page 94 
analysis than the other coordinates does have some justification especially with regard 
to the range of influence of the relatively long and flexible sidechains of the charged 
residues. These particular sidechains may extend for a distance comparable to the 
diameter of the coiled-coil (and the equivalent structure in collagen) and it is 
conceivable that such arm-like sidechains on the outside of the cylinder-like structure 
of the coiled-coil will have more freedom of movement in planes transverse to the 
cylinder axis than in a direction parallel to the axis. Good evidence to support this 
view has been obtained from nuclear magnetic resonance on the peptide backbone of 
collagen (Torchia and Van der Hart, 1976). A restriction on the axial motion of 
sidechains would imply that precision in specifying the axial coordinate of a residue is 
of greater importance than for the other coordinates. This is especially true for 
determining the axial range of influence of the sidechains. For this reason, the one­
dimensional simplification has justification beyond that of computational tractability. 
Maxima from the interaction curves for the rod domains of human lamin A molecules 
(Figure 4- 1 )  that are above the 0.975 quantile are listed in Table 4-4. One of the 
highest peaks occurs for a stagger of L\z(U -U) -equal to ±4 residues. Although this is 
unimportant for the purposes of building a model to describe the axial aggregation of 
molecules it provides an important mechanism for the lateral accumulation of 
molecules to generate width for the final structure. The rates of lateral and axial 
accretion of molecules are not known, and neither is the diameter-limiting mechanism 
understood. It is nonetheless worthwhile to point out that a mechanism does exist for 
lateral assembly and that this mechanism, based as it is on ionic interactions, is similar 
in principle to that proposed for the axial aggregation of the lamin molecules. 
Ionic interaction curves have also been calculated for the rod domains of rat peripherin 
molecules and significant scores are listed in Table 4-5. This type III IF protein is 
predicted to have the "classical" IF rod domain structure which includes the usual 
interruptions in the coiled-coil structure and the ionic regularities, namely the link 
regions Ll and L12. Hence the sequence has been subdivided into the two major 
coiled-coil domains, segments IB  and 2, and all combinations of segments and 
relative orientations have been analysed. The scores may be compared to those for the 
other type ill IF proteins, chicken gizzard desmin and hamster eye lens vimentin, 
which have been studied in a similar manner by Fraser et at ( 1985). 
The scores (S) and quantile scores (So) are generally higher for peripherin than the 
other type ill chains in interactions involving segment IB but are similar for segment 
2-segment 2 arrangements. The differences in the numbers of charged residues 
between these sequences appears to be sufficient to explain this discrepancy: 
CHAPrER 4: Secondary and Tertiary Structure page 95 
Parallel molecules Antiparallel molecules 
�z(U-U) S So L\z(U-D) S So 
±345 8 8 -264 56 52 
±31 1  24 24 - 184 84 84 
±297 32 30 - 164 96 92 
±294 32 32 - 125 1 16 1 12 
±284 40 36 - 1 17 124 1 12 
±28 1 40 38 - 1 14 1 12 1 12 
±267 44 42 -105 1 16 1 16 
±262 44 44 15 164 148 
±229 58 58 24 148 144 
±193 70 70 60 140 132 
±164 84 84 94 124 124 
±136 100 96 145 100 100 
±123 106 102 160 96 96 
±1 14 106 104 282 40 40 
±4 158 146 288 36 36 
340 12 12 
Table 4-4 Significant ionic interactions between the rod domains of human lamin A molecules 
(or lamin C molecules) as a function of relative axial stagger (see Figure 4-1). l1z(U-U) and l1z(U-D) 
signify the relative staggers between parallel and antiparallel chains respectively. Only those scores 
(S) greater than or equal to the 0.975 quantile (So) are listed. 
peripherin has 50 charged residues in segment IB and 52 in segment 2 ( IB :  30 acidic 
- 20 basic, 2: 30 acidic - 22 basic); chicken gizzard desmin has 42 and 52 charged 
residues in segments IB and 2 respectively ( lB: 25- 17, 2: 30-22); and hamster eye 
lens vimentin has 45 and 52 charged residues in segments IB and 2 respectively ( lB :  
29- 16, 2: 3 1-21). The total counts for the segment 2 sequences are identical amongst 
the three proteins and hence the similarity in scores. However, segment I B  of 
peripherin contains more charged residues and the chances of interactions being scored 
are greater. 
Ionic interaction curves have also been calculated for Xenopus lamins A and B and the 
Helix pomotia B proteins. These are shown in Appendix D along with tables of the 
significant peaks. Until such time as new structural data have been obtained for these 
proteins no further progress can be made in establishing a model for their packing 
(such as that described in the next section for the human lamin proteins). However, in 
the case of Xenopus lamin A, which is highly homologous to the human lamins A and 
C in sequence, the following study may be expected to apply equally well to its 
CHAPTER 4: Secondary and Tertiary Structure page 96 
Parallel molecules Antiparallel molecules 
Az(U-U) S So Az(U-D) S So 
IBU - IBU  ±54 32 30 IBU- IBD 3*  68 60 
±4* 58 58 
2U -2U  ±138 6 6 2U-2D -65 44 36 
±1 l 3  20 18  -34* 48 48 
±103 24 20 4* 60 56 
±93 24 24 1 3* 64 52 
±72* 32 30 34* 44 44 
±63* 34 42 43* 56 40 
±34* 44 42 53* 40 40 
±26* 46 44 64* 36 36 
±5* 64 50 74* 32 32 
100 28 24 
1 10 24 20 
12 1  16  16  
1 32 12  12  
135 8 8 
143 4 4 
IBU - 2U  - 1 33 10 10 IBU-2D -85* 32 32 
-94* 34 28 -34* 50 48 
-82* 32 32 -25* 46 46 
-7 1 * 38 38 4* 48 46 
-55* 46 44 33* 36 34 
-43* 48 48 44* 32 30 
- 14  48 48 53* 26 26 
Table 4-5 Significant ionic interactions between rod domain segments of peripherin molecules as 
a function of relative axial stagger (see Appendix D, Figure D-I). ru:(U-U) and ru:(U-D) signify the 
relative staggers between parallel and antiparallel chains respectively. Only those scores (S) greater 
than or equal to the 0.975 quantile (SO) are listed. Those staggers that correspond to maxima in the 
ionic interaction curves for the type ill proteins studied by Fraser et al (1985) are marked with an 
asterisk. 
molecular aggregation as to that of the human lamins. 
4 . 3  Modell ing Lamin 
A number of studies on in vivo and in vitro assemblies of lamin IF have been 
reported, but data that prove of particular relevance to this study are as follows: 
1 )  Lamins prepared under conditions of low ionic strength from isolated nuclei or 
CHAPfER 4: Secondary and Tertiary Strucb.U'e page 97 
from cytoskeletal preparations form paracrystals with a well defmed banding 
pattern. The period originally reported for this pattern was 37-40 nm (Zackroff 
et aI, 1984; Goldman et ai, 1986). However, Parry et al ( 1987a) repeated these 
measurements on a number of negatively-stained paracrystals and showed that the 
true axial periodicity was 48±2 nm - a value somewhat greater than that 
previously stated but which nonetheless appeared to be consistent with the 
original figures of Zackroff et al ( 1984). 
2) Aebi et al (1986) studied the nuclear lamina from isolated nuclear envelopes of 
Xenopus oocytes and revealed the presence of an orthogonal lattice of side 
52 nm. In addition, a 25 nm beading was observed in Triton-treated preparations 
of the Xenopus lamin mesh work which had been washed thoroughly in low ionic 
strength buffer prior to freeze-drying and metal shadowing. This 25 nm beading 
was also observed in filamentous assemblies formed in vitro from 1 :  1 mixtures of 
purified rat liver lamins A and C dialysed against low ionic strength buffer. The 
25 nm period of the beading is almost exactly half the period of the lamina 
meshwork (52 nm) and the negatively-stained paracrystals (48 nm). 
Parry et al ( l987a) suggested that the packing of the molecules in the paracrystalline 
structures, the in vivo lattice and the in vitro filamentous structures were related. In 
particular they suggested that the in vitro filamentous assemblies were laterally 
compressed versions of the in vivo lattice. In order to investigate this proposal, model 
building studies were performed which incorporated the observed banding structure 
data and the likely relative staggers of lamin molecules deduced from the ionic 
interaction studies (Table 4-4). The observed electron microscope data from 
measurements made on the negatively stained paracrystals (Parry et aI, 1987a) were: 
I ) The axial period of the alternating light-staining and dark-staining bands was 
found to be 48±2 nm. This value is a little lower than the lattice spacing of 52 run 
observed by Aebi et al ( 1986, see above) in the nuclear lamina of Xenop us 
oocytes. The difference is probably attributable to shrinkage effects from sample 
preparation methods used in electron microscopy and also to periodic variations in 
the structural regularity and stability of the assembly rather than to alternate modes 
of packing. 
2) The relative widths of the light- and dark-staining bands were 0.49±O.04: 
0.5 I±O.04 respectively. 
In addition, four other observations of relevance were noted (see Figure 4-2 and also 
Figures 1 and 2 of Parry et ai, 1987a): 
3) The dark-staining band corresponded to a region where the molecular packing 
was of relatively low density. This was particularly apparent in electron micro­
graphs of small, newly-forming lamin paracrystals where individual rod-like 
CHAPrER 4: Secondary and Tertiary Structure page 98 
Figure 4-2 Electron micrographs of negatively-stained human lamin A paracrystals. Globular 
domains are apparent in the paracrystal on the left towards the end of the structure. These are assumed 
to correspond to the large C-tenninal domains of the molecule. A smaller, newly fonning paracrystal 
is shown on the right (not to scale) and reveals rod-like threads traversing the dark bands. Figures 
courtesy of Drs A.E. and R.D. Goldman. 
threads could be seen. These threads were interpreted as being the rod domains of 
the lamin molecules. A greater density of proteinaceous material was evident in 
the light-staining bands of the paracrystals and this was probably due in pan to 
the large C-terminal domains of the molecules. 
4) The small, newly forming paracrystals revealed evidence of dihedral symmetry. 
This was manifested by the symmetry of the banding pattern and by the light­
staining bands terminating both ends of the structures. This latter point also 
indicated that nuclear lamin molecules must terminate at, or close to, the 
boundaries of these bands. 
5) Arrays of molecules in larger paracrystals also showed evidence of dihedral 
symmetry through the appearance of globular attachments at both ends of the 
structures. The globules were believed to be the large C-temrinal domains of the 
lamin molecules and were confmed to the light-staining bands (see, for example, 
Figure 3 of Zackroff et al, 1984). 
6) No axial substructure is apparent in either the light- or dark-staining bands and it 
is reasonable to assume that the packing within both bands is uniform at the low 
CHAPrER 4: Secondary and Tertiary Structure page 99 
resolution available. 
The evidence above suggests that the molecules in the paracrystals are arranged in an 
antiparallel manner and that they contain equal numbers of oppositely directed 
molecules. These data were used in conjunction with data from the intennolecular 
ionic interaction studies to generate a model for the molecular aggregation of human 
lamin A. 
Several parameters used in the model-building process must be defined. Distances 
along the axis of the paracrystal are measured in terms of 'residue translations' (rather 
than nanometres) where a residue translation is defmed as the average axial rise per 
residue in an a-helical coiled-coil conformation. The value used here is 0. 1485 nm 
(see, for example, Fraser et al, 1972 and Fraser and MacRae, 1973b). Other 
parameters are defmed as follows (see Figures 4-3 and 4-4) : 
I = length of the rod domain of the lamin molecules. There are 356 residues in 
the rod domains of human lamins A and C (McKeon et ai, 1986; Fisher 
et ai, 1986; Parry et ai, 1986). 
p = periodicity of the paracrystals. In this case the period is taken as 48 nm, or in 
terms of residue translations: 
48 nm 
0. 1485 nm res- l 
= 323 residues 
TIris value is also considered equivalent to the lattice spacing observed in 
Xenopus oocytes. 
s = smallest relative axial stagger between a dimer of antiparallel molecules that 
generates the putative four chain asymmetric unit 
dn = effective axial length of the N-terminal non-a-helical domain of the larnin 
molecule in the negatively-stained paracrystal. 
de = effective axial length of the C-terminal non-a-helical domain of the lamin 
molecule in the negatively-stained paracrystal. 
Several intermolecular interactions are made by the asymmetric unit and these may be 
associated with molecules having the following relative staggers (see Figures 4-3 and 
4-4): 
(i) �z(U-U) = P 
(ii) �(U-D) = s 
(ill) �*(U-D) = s 
-
P 
(iv) �t(U-D) = s + p 
(twice) 
provided s + p < I 
�z refers to the difference in the axial coordinate between rod domains of different 
molecules and is measured in residue translations (Fraser et al, 1985). U and D refer 
to the 'Up' and 'Down' orientations of the molecules as defined in Section 4.2. The 
CHAPrER 4: Secondary and Tertiary Structure 
de 
I 
do 
(a) 
L\z(U-U) = p 
(twice) 
(b) 
page 100 
Figure 4-3 Molecular structure for Model A of lamin paracrystals. (a) Structure of the asymmetric 
unit which comprises a dimer of molecules staggered by an axial distance s. This distance is specified 
by three possible sets of ionic interactions, one of which is shown as a shaded rectangle between the 
molecules. (b) The two other types of ionic interactions possible for Model A are shown shaded 
between dimers of molecules. These specify the larger scale 'Structure of the paracrystal. The light­
staining bands observed in electron micrographs of the negatively-stained paracrystals correspond to 
regions containing the C-terminal domains. 
CHAPTER 4: Secondary and Tertiary Structure 
(a) 
Az(U-U) = p  
(twice) 
page 101 
(b) 
Figure 4-4 Molecular structure for Model B of lamin paracrystals. (a) Structure of the asymmetric 
unit which comprises a dimer of molecules staggered by a distance s. This distance is specified by 
four possible sets of ionic interactions. one of which is shown as a shaded rectangle between the 
molecules. (b) The three other types of ionic interactions possible for Model B are shown shaded 
between dimers of molecules. These specify the larger scale structure of the paracrystal. The light­
staining bands observed in electron micrographs of the negatively-stained paracrystals correspond to 
regions containing the C-terminal domains. 
CHAPTER 4: Secondary and Tertiary Structure page 102 
asterisk (*) and dagger (t) symbols indicate alternative values for L\z(U-D). 
The assumption of uniform packing within each of the light- and dark-staining bands 
of the paracrystal restricts model-building to two general classes of structure (Figures 
4-3 and 4-4). In Model A, the antiparallel molecules in the dimer are approximately 
half-staggered and in Model B the molecules in the dimer are largely overlapped. Thus 
in Model A, the distal ends of the C-tenninal domains are in near axial proximity, 
whereas in Model B it is the proximal ends that are in near axial register. Note that the 
large C-terminal domains are likely to have a fairly extended conformation so as to 
span half the width of a light-staining band and maintain a near uniform density of 
packing. This will be discussed in greater detail later but it is worth pointing out that 
uniform packing could also be achieved in Model A by further extending the C­
terminal domains so that they span the full width of a light-staining band. 
The ionic interaction data from Section 4.2 can now be used to find suitable values for 
p, s, do and de. Consider Model A (Figure 4-3). A suitable maxima for p from the 
interaction curve for parallel molecules is 324 residues (48 . 1  nm) which has a score 
o( 14 ionic interactions. The length of the region containing the terminal domains (ie, 
the light-staining band in the paracrystals) is: 
(0.49±O.04)p = 159±12 residues 
and this may also be described by the expression: 
and hence 
(0.49±O.04)p = I + s - p + 2dn where I = 356 and p = 324 
2dn + s = 127±12 residues 
The interactions between antiparallel molecules (U-D) occur at relative staggers of s 
and s - p and it follows that the appropriate interaction curve should be examined for 
pairs of interactions with a relative separation of p (324 residues). One such pair 
would occur when 
s = 160 residues (96 interactions) 
and 
s - p = - 164 residues (96 interactions) 
Both of these staggers correspond to interactions above the 0.975 quantile level. By 
substitution: 
dn = - 1 6±6 residues ::::: -2 nm 
This implies that the N-terminal domains (which contain only 30 residues) are so small 
that they are not visualized in the electron micrographs of the paracrystals. 
Taking the length of the C-terminal domains as approximately half the width of the 
light-staining bands as discussed previously, the likely value of de is given by the 
CHAPrER 4: Secondary and Tertiary Structure 
expression: 
d p + s - I c = 2 
= 64 residues (9.5 run) 
page 103 
This length implies a highly elongated C-terminal domain and suggests that there must 
be important structural and functional reasons for such a conformation. As mentioned 
above, it is also possible that the C-terminal domains extend across the width of the 
light-staining bands in which case the expression for de becomes:  
de = p + s - I 
In other words, de could be twice as long as before. 
The apparent globular form of the C-terminal domains observed for isolated nuclear 
lamin molecules from rat by Aebi et al (1986) does not discount the possibility of an 
elongated structure. Myosin, for example, was originally believed to have spherical 
heads but under better preparative procedures was later shown to have elongated heads 
with a high aspect ratio (Elliott and Offer, 1978). Certainly the volume of the pair of 
C-terminal domains in the lamin molecule is large (fable 4-6). These domains are also 
clearly apparent in the micrographs of Zackroff et al (1984). The volume of a single 
C-terminal domain of human lamin A calculated from the amino acid sequence is about 
37  nm3 (Table 4-6). As this could lead to a structure of length 9.5  nm and 
approximate diameter 2.2 run, the volume data are clearly not inconsistent with a 
structure extending over half the width of a light-staining band of the lamin 
paracrystal. 
Analysis of the amino acid sequence of the C-terminal domains of human lamins A 
and C has revealed extensive but localized runs of charged residues. Furthermore, the 
relatively high serine content of regions of this domain suggests that they may have 
importance as potential sites for hyperphosphorylation (Fisher et ai, 1986). Both of 
these observations indicate a high charge count and hence an elongated structure for 
the C-terminal domain is again predicted. It is reasonable to assume that lamins from 
other sources will have similar structures (cf. Types I-IV IF) although additional 
physico-chemical data will be necessary to provide proof of this. The Xenopus lamin 
A sequence is highly homologous to human lamin A, suggesting that it will share a 
common structure with its human analogue. 
In Model B (Figure 4-4), the smallest but highest scoring value of �z(U-D) occurs 
when s is 15 residues ( 164 interactions). Using the same value of p as for Model A 
(p = 324 residues), the value for �z*(U-D) will be -309 residues and corresponds to 
16  interactions. For Model B, 
s + P = 339 residues < I 
CHAPrER 4: Secondary and Tertiary Structure page 104 
Human Lamin A Human Lamin C 
Residue Numbers Total Numbers Total 
Volume of Residues Volume of Residues Volume 
(nm3) (nm3) (nm3) 
Ala 0.089 17 1 .5 1 3  1 1  0.979  
Arg 0. 1 82 21 3.822 17 3.094 
Asn 0. 123 1 1  1 .353 7 0 .861  
Asp 0. 1 17 14 1 .638 1 1  1 .287 
Cys 0. 107 5 0.535 1 0. 107 
GIn 0. 149 13 1 .937 9 1 .34 1 
GIu 0. 145 13 1 .885  12  1 .740 
Gly 0.063 31  1 .953 16 1 .008 
His 0. 153 8 1 .224 7 1 .07 1 
ne 0. 168 5 0.840 3 0.504 
l.eu 0. 168 17 2.856 12 2 .0 1 6 
Lys 0. 174 1 1  1 .914 1 0  1 .740 
Met 0. 168 3 0.504 2 0 .336 
Phe 0. 197 5 0.985 4 0.788  
Pro 0. 130 12 1 .560 7 0.9 1 0  
Ser 0.094 44 4. 136 23 2. 1 62 
Thr 0. 199 20 3.980 14  2 .786 
Trp 0.235 4 0.940 4 0.940 
Tyr 0. 199 4 0.796 1 0. 1 99 
Val 0. 141 19 2.679 14 1 .974 
Totals 277 37.050 185 25.843 
Table 4-6 Volume calculation for the C-terminal domains of human lamins A and C based on 
the residue volume data collected by Parry and Baker (1984) and the lam in sequence data of Fisher 
et ai (1986). The sequences of human lamins A and C are identical from the N-terminal domain to the 
179th residue of the C-terminal domain. Thereafter, lamin A continues for 98 residues and lamin C for 
6.  
and so interaction (iv) above will be viable. The nearest maxima is at 
hzt(U-D) = 340 residues (12 interactions) 
The width of the light-staining bands for Model B is (as before) 
(0.49±O.04)p = 159±12 
and this may also be described by the expression: 
Hence 
(0.49±O.04)p = 2de - p - s + I where I = 356 and p = 324 
de = 71±6 residues (1 1 nm) 
CHAPrER. 4: Secondary and Tertiary Structure page 105 
Model A Model B 
Interaction Stagger Score Stagger Score 
I(residue translations) (residue translations) 
Az(U-U) = P (twice) 324 14+14 324 1 4+14 
Az(U-D) = 8  160 96 15 164 
Az*(U-D) = 8 - p  -164 96 -309 16  
Azt(U-D) = 8 + P (8 + �_ < I 340 12 
Total 220 Total 220 
Table 4-7 The sets of ionic interactions used in Models A and B (Figures 4-3 and 4-4). 
A summary of the ionic interactions in the asymmetric unit for the two models is listed 
in Table 4-7. The total interactions for Models A and B (220) are identical. Important 
differences, however, do exist between the two models. For example, in Model A it is  
the rod-rod ionic interactions that specify both the structure of the dimer and the large 
scale axial coherence of the molecular assembly whereas in Model B the rod-rod 
interactions specify the dimer and it is predominantly the C-terminal domain to rod 
interactions that specify the axial coherence of the structure. An interesting observation 
is that the link L2 and the ' stutter' in segment 2B are in approximate ax,ial alignment in 
Model A but are not so related in Model B. The significance (or otherwise) of thi s  
point has yet to be ascertained. 
Both models use the same stagger between the dimer of molecules (.1z(U-U) = p = 
324 residues) which results in the axial alignment of the fIrst and last 32 residues of 
the rod domain. Although this stagger does not represent a major maximum in the 
ionic interaction curve, it does correspond closely to the observed period of the 
banding pattern in the negatively-stained paracrystal and in addition there are several 
points of note about the 32 residue sub-sequences (Figure 4-5). For example, a 
number of identical residues lie in exact axial register. There is also a small region 
which is highly hydrophobic in character within each sub-sequence and these regions 
are axially adjacent at this stagger. These regions exhibit the highest concentration of 
hydrophobicity within the entire rod domain of the lamin sequence and appear to 
indicate that localized apolar interactions may also be important in specifying the axial 
assembly of lamin molecules. No apolar interaction curves have been calculated, 
however, as the distribution of apolar residues outside of the .il and si positions of the 
heptad has been found to be random (Parry et ai, 1977; McLachlan and Stewart, 
1982; Dowling et ai, 1983). 
Another feature of the 32-residue sub-sequences is that each includes one of the two 
CHAPI'ER 4: Secondary and Tertiary SbUcture 
31 - 46 
355 - 370 
47 - 62 
371 - 386 
K E D  
L D E 
R V R S L E T E N A !9 L R L IRl I 
E I H A Y  R K L L E � E E E � L - -
page 106 
Figure 4-5 The sub-sequences that comprise the interacting portions of the rod domains of the 
lamin sequences staggered by Az(U-U) = P where P is 324 residues fm-both Models A and B. Residues 
are numbered from the start of the N-tenninal domain which is 30 residues in length. The top segment 
begins at the N-terminus of the rod domain and the lower segment finishes at the C-tenninus of the 
rod domain. Identical residues in the two sequences are boxed and conservative substitutions are 
underlined. The shaded box shows two regions which have the highest hydrophobicity in the rod 
domain of lamin molecules. 
highly conserved sequences found in all IF chains in segment lA  and towards the end 
of segment 2B (see, for example, Parry and Fraser, 1985). The features observed 
within these two interacting regions of the rod domain suggest that one of their major 
roles is to specify the axial aggregation of parallel lamin molecules and, by 
implication,  the axial assembly of other IF molecules since these chains also contain 
similar highly conserved sequences. 
Two perpendicular diads are seen in the negatively stained paracrystals and these occur 
at the centre of both the light- and dark-staining bands. Using the parameters defined 
before, the positions may be expressed as: 
Dark-staining band: I + s 2
- P = 96.0 residues (Model A) 
or 23.5 residues (Model B) 
Light-staining band: I +2 
s 
= 258.0 residues (Model A) 
or 1 85.5 residues (Model B) 
The possible diad position at residue 96 lies close to the position of the start of the 42 
residue insert in segment 1B of type V IF molecules which is not present in other IF. 
Also, the possible diad at residue 23.5 lies within the highly conserved sequence in 
segment lA. The relevance of these observations is unclear at this stage. 
No preference has yet been stated for one of the proposed Models over the other. 
Some evidence is available, however, to suggest that Model A is more likely to be 
valid than Model B. As detailed in the start of this Section, a 'beading'  of period 
24-25 run has been observed in filamentous structures prepared under conditions of 
low ionic strength (Aebi et ai, 1986). Assuming that the molecular packing in the 
CHAPrER 4: Secondary and Tertiary Structure page 107 
paracrystals is identical to that in the filamentous structures. examination of the Models 
shows that a 25 run period can be generated from Model A by realigning the large 
C-terminal domains perpendicular to that axis of the molecule (Figure 4-6). The 
proximal ends of the C-terminal domains would occur on average at alternate intervals 
of p - s and s ( 164 and 160 residues respectively. or 24.4 and 23.7 nm). The 
difference in orientations of the terminal domains using different staining and 
visualization procedures may be related to ionic conditions since beading becomes 
more apparent as ionic strength is lowered (Aebi et al. 1986). In Model B the 
C-terminal domains are spaced by the period of the paracrystal (48 nm) and so could 
not generate a 25 run period in such a simple way. 
Aebi et al ( 1986) also noted that washing the Triton-extracted nuclear lamin 
meshwork from Xenopus oocytes with low ionic strength buffer resulted in a 24-
25 nm beading in the orthogonal lattice. This suggests that the packing of lamin 
molecules in the lattice is similar to that in the filamentous structures. As before. 
assuming that the molecular packing in the lattice is the same as that suggested in 
Model A for the paracrystal. then the beading can be explained in the same manner as 
for the filamentous structures. Model A can also explain in a straightforward way how 
the lattice structure may be disassembled by hyperphosphorylation.  For reasons 
already discussed it is likely that many (but not necessarily all) of the sites of 
phosphorylation in lamin molecules will be located in the C-terminal domain (see also 
Fisher et al. 1986). The proximity of the rod and C-terminal domains means that the 
marked change of the charge profile of the C-terminal domain on phosphorylation 
would be expected to destabilize the ionic interactions between them. It could possibly 
also have a similar effect on adjacent rod-rod interactions and hence cause 
dissociation. Conversely. dephosphorylation would pennit self-assembly of the lattice 
structure. 
The lattice nodes would be expected to correspond to one of the two perpendicular 
diads observed in the negatively-stained paracrystals by consideration of symmetry. In 
either case, the large C-terminal domains would appear to make only a marginal 
contribution to the stability of the cross-over regions. The nodes would therefore 
consist essentially of intersecting rod domains. If the lattice is collapsed by rotating the 
intersecting rod domains relative to each other a paracrystalline structure such as 
described by Model A could be produced (Figure 4-7). Another a-fibrous protein, 
tropomyosin, reveals two such packing modes coexisting in the same structure (Figure 
4-8) which provides support for a similar duality in lamin structure. 
CHAPrER 4: Secondary and Tertiary Structure page 108 
(a) (b) 
Figure 4-6 Schematic representation of the two modes of molecular assembly capable of 
explaining the stain distribution in the electron micrographs of negatively-stained lamin paracrystals. 
(a) In Model A the molecules comprising the dimer are antiparallel and approximately half- staggered 
whereas in (b) the molecules are antiparallel and almost completely overlapped. In the lower parts of 
both diagrams the large C-terminal domains have been oriented perpendicular to the axis of the 
paracrystal. In (a) this generates a quasi-repeat of ... 25 nm but in (b) the repeat is twice that of (a). 
CHAFfER 4: Secondary and Tertiary Structure page 109 
I 
Figure  4-7  Possible transformation between the lattice and paracrystalline forms of larnin 
structure. The large C-terminal domains are drawn as filled circles and, in the collapsed lattice (to the 
right), indicate the approximate positions of the light-staining bands in the paracrystal. See also 
Figure 4-8. 
CHAPI'ER4: Secondary and Tertiary Structure page 1 10 
Figure 4-8 Electron micrograph of tropomyosin revealing two packing modes coexisting in the 
same structure. The crystalline lattice form at the lower left merges with the banded tactoid at the 
upper right. This lattice-filament dual ity may be similar to that proposed for lamin (Figure 4-7). 
Figure courtesy of Or C. Cohen. 
4 . 4  3D Ionic Interactions Between Molecules 
Difficulties do arise with the linear array model of the coiled-coil used in the ID  
method of calculating interactions because of the loss of azimuthal and radial 
information about each residue. The linear model would allow the possibility of 
interactions between residues in different molecules which the constraints of distance 
and orientation (ie, steric hindrance) would normally disallow. In particular, residues 
internal to the dimer of coiled-coils (ie, those in the � and .d positions of the heptad) 
would rarely be in a position to allow interactions to occur with those in other 
molecules. Also, residues with identical axial coordinates but on opposite sides of a 
two-molecule assembly (for example) would be physically unable to interact with each 
other because of their separation. An advantage of a three-dimensional study is that 
several criteria previously not included can be built into the model and other criteria 
can be defined more realistically. For example, the distance limit previously imposed 
CHAYfER 4: Secondary and Tertiary Structure page 1 1 1  
No NI M P (nm) N 
1 24 84 12 .12 3.6522 
1 26 91  1 3.21 3.6400 
1 28 98 14.29 3.6296 
1 30 105 15.37 3.6207 
1 32 1 12 1 6.44 3.6129 
1 36 126 18.58* 3.6000 
Table 4-8 Values used to derive a set of five pitch lengths for the 3D intmlction study. NO is the 
number of turns of major helix; NI is the number of turns of minor helix; P is the pitch length of the 
coiled-coil; M is the number of turns in the repeat distance, c, where c=NoP. Also included are the 
values that correspond closely to the pitch length (*) used by Parry and Suzuki (1969) in their 3D 
study of intrachain potential energy of a-helices and coiled-coil ropes. 
between interacting residues was defined in terms of a range of linear residue 
translations. However, the cummulative azimuthal displacements of residues on an 
a-helix are approximately 103°, 206°, 309° (= -5 1°), 412° (= 52°) and so on. Hence, it 
is possible that while one residue may be in a position to interact with a particular site 
on an adjacent molecule, its nearest neighbour on the a-helix may be rotated too far 
towards the inside of the coiled-coil to do so. However, the third or fourth neighbour 
along (at azimuths of -51 ° and 52° and axial displacements of 0.45 nm and 0.59 nm 
respectively) may be in a better position to participate in interactions at the site and may 
even compete with the original residue in making a bond between the molecules. 
The pitch length of the IF coiled-coil must also be included as a variable in the three­
dimensional analysis since it has not yet been accurately determined. Five values were 
chosen that cover a range of 'reasonable' pitch lengths suggested by physical data 
(Cohen and Holmes, 1963; Elliott et ai, 1968; Fraser et al, 197 1 ;  Parry, 1 975;  
Phillips et al, 1986). These values are listed in Table 4-8 along with other parameters 
for determining the pitch length of the coiled-coil. 
An assumption made to simplify the three-dimensional study is that the two molecules 
are related either by a parallel diad (parallel molecules) or by a perpendicular diad 
(antiparallel molecules). This has the effect of matching the azimuthal coordinates for 
the molecules (see Figure 4-9). The independent variables are thus reduced from four 
(pitch length of the coiled-coil, relative axial stagger between molecules and two 
azimuthal coordinates, one for each molecule) to three and makes the study 
computationally manageable. Note that the sense of rotation of the molecules is 
different for the parallel and antiparallel cases. 
CHAPTER 4: Secondary and Tertiary Structure page 1 12 
�- -
\ 
(a) (b) 
Figure 4-9 Relative directions of rotation for the three-dimensional modelling of molecules. (a) 
Parallel (V-V) and (b) antiparallel (V-D) molecules. The dashed lines indicate the axes of symmetry 
used 10 derive one molecule from the other. 
4 . 4 . 1 Generation of Coordinates for a Coiled-Coil Molecule 
A coordinate system is used whereby the z-axis is aligned along the axis of the coiled­
coil and increases with the residue sequence from the amino- to carboxy-terminus of 
the protein chain. The x-axis is arbitrarily aligned so that the axes of the two a-helices 
at their starting positions lie on it: the two a-helices are equidistant from the origin but 
are displaced in opposite directions along the x-axis. A starting set of coordinates for 
the atoms comprising a single residue on an a-helix with 3.6 residues per turn is given 
in Table 4-9. Using these data a coiled-coil with radius ro and pitch length P may be 
generated using the methods described by Crick (1953) and Fraser et al ( 1964c). The 
coordinates of residues in a second, parallel strand coiled-coil are determined by the 
application of a parallel diad. Coiled-coils consisting of anti parallel chains have not 
been studied as there is no experimental evidence to suggest that such an arrangement 
exists in any a-fibrous protein. 
The coiled-coil pitch length is specified in part by the number of residues per turn in 
the undistorted a-helix.  Hence it would appear necessary to use a different starting set 
of coordinate data for coiled-coils of different pitch lengths. However, the analysis of 
CHAPTER 4: Secondary and Tertiary Structure page 1 13 
Atom 1 x�O (nm) YiO (nm) Zio (run) riO (run) 0iO e) 
N 1 0. 105 1  -0. 1 1 39 -0.0043 0. 1550 -47.3  
H 2 0. 1227 -0. 1005 -0. 1017  0. 1586 -39 .3  
Ca 3 0.2159 -0.073 1 0.0830 0.2279 - 1 8.7 
Ha 4 0.2608 -0. 1617  0. 1297 0.3069 -3 1 . 8  
C' 5 0. 1632 0.0206 0. 1919 0. 1 645 7 .2  
0 6 0. 1 858 -0.0016  0.3 1 1 8  0. 1 858 -0.5 
Cp 7 0.32 18  0.0000 0.0000 0.3218  0 .0  
Table 4-9 Starting set coordinates of  a residue in an undislOrted a-helix (Parry and Suzuki, 
1969). The a-helix is right-handed with a unit rise of 0.15  nm. There are 18 residues in 5 turns of 
a-helix and hence the unit twist is +100°. Coordinates of an extended section of a-helix can be 
generated by application of the following expressions: 
run = riO 
0im = 0iO + lOOm 
zim = ZiO + O.15m 
The symbols i and m refer to the number of the atom in the reference set and to the residue number 
respectively. The standard set of coordinates in the Table above refer to the zeroth residue, ie m=O. 
the packing of the coiled-coil molecules described in this section is of low resolution 
and is not sensitive to small changes in coordinate data. The coordinates for the coiled­
coils comprising a pair of 1B segments with a pitch length of 12. 1  nm are given in 
Table 4- 10 and plotted in Figure 4- 10. It is imponant to point out that by arranging for 
the �-carbon atom of the starting coordinate data set to lie on the x-axis, the sidechains 
of the apolar residues in the il and d positions of the two strands are approximately 
enmeshed to generate close to optimal knob-hole packing. 
Having generated the coordinates of the two strands of a single coiled-coil molecule, 
the molecule is then translated a distance of 1 nm along the x -axis from the origin. The 
coordinates of residues in a second coiled-coil rope are calculated by the application of 
either a perpendicular diad, ie a two-fold axis parallel to the y-axis (antiparallel 
molecules) or a parallel diad, ie a two-fold axis parallel to the z-axis (parallel 
molecules) (see Figure 4-9). The exact separation of the molecules has been chosen in 
a somewhat arbitrary fashion and indeed some of the outermost �-carbon atoms from 
the two molecules appear to overlap. Although this arrangement is not feasible from an 
energetic viewpoint, the relatively large range allowed for interactions avoids any 
difficulties that might arise from the imprecise location of the molecules. 
4 . 4 . 2  Determination of Interactions 
The sidechains of the amino acids originate at the �-carbon atoms, the coordinates of 
which are readily calculable for a regular coiled-coil structure. However, limited 
freedom of rotation about each of the single bonds in the sidechains prevents a 
CHAPfER 4: Secondary and Tertiary SlIUcture page 1 14 
Mol�ule 1 MQlecule 2 
Coiled-Coil 1 Coiled-Coil 2 Coiled-Coil 1 Coiled-Coil 2 
Residue x y z x y z x y z x y z 
144-b E 1.93 -0. 1 1  0.00 0.07 0. 1 1  0.00 -1 .93 -0. 1 1  14.49 -0.07 0.1 1 14.49 
145-c M 1 .76 0.56 0.36 0.24 -0.56 0.36 - 1 .76 0.56 14. 13 -0.24 -0.56 14. 1 3  
147� E 1.47 -0.43 0.4 1 0.53 0.43 0.41 - 1 .47 -0.43 14.09 -0.53 0.43 14.09 
148-f E 2.05 0.00 0.68 -0.05 0.00 0.68 -2.05 0.00 13.81 0.05 0.00 13 .8 1  
149-g E 1.47 0.43 0.96 0.53 -0.43 0.96 - 1 .47 0.43 13 .53 -0.53 -0.43 13 .53 
151-b R 1 .76 -0.56 1 .01 0.24 0.56 1.01 -1 .76 -0.56 13 .48 -0.24 0.56 13 .48 
152-c E 1 .93 0.1 1  1 .37 0.07 -0.1 1 1 .37 - 1 .93 0. 1 1  13 .12  -0.07 -0.1 1 13 . 12  
1� R 1 . 19 -0.61 1 .42 0.8 1 0.61 1.42 - 1 .19 -0.61 13 .08 -0.8 1 0.61 13 .08 
155-f R 1.91 -0.53 1 .69 0.09 0.53 1 .69 - 1 .9 1  -0.53 12.80 -0.09 0.53 1 2.80 
156-g Q 1 .63 0.14 1 .97 0.37 -0.14 1 .97 - 1 .63 0.14 12.52 -0.37 -0.14 1 2.52 
158-b D 1.38 -0.86 2.02 0.62 0.86 2.02 - 1 .38 -0.86 12.47 -0.62 0.86 1 2.47 
159-c A 1 .86 -0.38 2.38 0.14 0.38 2.38 - 1 .86 -0.38 12. 1 1  -0.14 0.38 1 2. 1 1  
161� T 0.86 -0.63 2.43 1 . 14 0.63 2.43 -0.86 -0.63 12.07 - 1 .14 0.63 1 2.07 
162-f G 1 .53 -0.91 2.70 0.48 0.91 2.70 -1 .53 -0.91 1 1 .79 -0.48 0.9 1 1 1 .79 
163-g Q 1 .61  -0. 19 2.98 0.39 0.19 2.98 - 1 .61 -0. 19 1 1 .5 1  -0.39 0.19 1 1 .5 1 
165-b A 0.89 -0.93 3.03 1 . 1 1 0.93 3.03 -0.89 -0.93 1 1 .46 - 1 . 1 1  0.93 1 1 .46 
166-c R 1.56 -0.76 3.39 0.44 0.76 3.39 - 1 .56 -0.76 1 1 . 1 1  -0.44 0.76 1 1 . 1 1 
168� E 0.57 -0.47 3.43 1 .43 0.47 3.43 -0.57 -0.47 1 1 .06 -1 .43 0.47 1 1 .06 
169-f V 1 .00 -1 .05 3.7 1  1 .00 1 .05 3.71 -1 .00 -1 .05 10.78 - 1 .00 1 .05 10.78 
170-g E 1 .43 -0.47 3.99 0.57 0.47 3.99 -1 .43 -0.47 10.50 -0.57 0.47 10.50 
172-b D 0.44 -0.76 4.04 1 .56 0.76 4.04 -0.44 -0.76 10.45 - 1 .56 0.76 10.45 
173-c N 1 . 1 1  -0.93 4.40 0.89 0.93 4.40 - 1 . 1 1 -0.93 10.10 -0.89 0.93 10.10 
175� L 0.39 -0. 19 4.44 1.61 0. 19 4.44 -0.39 -0. 19  10.05 - 1 .61 0. 19 10.05 
176-f D 0.48 -0.91 4.72 1 .53 0.91 4.72 -0.48 -0.91 9.77 - 1 .53 0.91 9.77 
177-g N 1 . 14 -0.63 5.00 0.86 0.63 5.00 -1 .14 -0.63 9.49 -0.86 0.63 9.49 
179-b Q 0.14 -0.38 5.05 1 .86 0.38 5.05 -0.14 -0.38 9.44 - 1 .86 0.38 9.44 
l8O-c K 0.62 -0.86 5.4 1  1 .38 0.86 5.41 -0.62 -0.86 9.09 -1 .38 0.86 9.09 
182� K 0.37 0. 14 5.45 1 .63 -0.14 5.45 -0.37 0.14 9.04 - 1 .63 -0.14 9.04 
183-f Q 0.09 -0.53 5.73 1 .91  0.53 5.73 -0.09 -0.53 8.76 - 1 .91 0.53 8.76 
184-g K 0.81 -0.61 6.01 1 . 19  0.61 6.01 -0.8 1 -0.61 8.48 - 1 . 19 0.61 8.48 
186-b Q 0.07 0.1 1  6.06 1 .93 -0.1 1 6.06 -0.07 0. 1 1  8.44 - 1 .93 -0.1 1 8.44 
187-c E 0.24 -0.56 6.42 1.76 0.56 6.42 -0.24 -0.56 8.08 -1 .76 0.56 8.08 
189� I 0.53 0.43 6.46 1 .47 -0.43 6.46 -0.53 0.43 8.03 - 1 .47 -0.43 8.03 
19O-f Q -0.05 0.00 6.74 2.05 0.00 6.74 0.05 0.00 7.75 -2.05 0.00 7.75 
191-g L 0.53 -0.43 7.02 1.47 0.43 7.02 -0.53 -0.43 7.47 - 1 .47 0.43 7.47 
193-b Q 0.24 0.56 7.07 1 .76 -0.56 7.07 -0.24 0.56 7.43 - 1 .76 -0.56 7.43 
194-c E 0.07 -0. 1 1  7.43 1 .93 0. 1 1  7.43 -0.07 -0. 1 1  7.07 - 1 .93 0. 1 1  7.07 
196-e E 0.8 1 0.61 7.47 1 . 19  -0.6 1 7.47 -0.8 1 0.61 7.02 - 1 . 19 -0.61 7.02 
197-f N 0.09 0.53 7.75 1 .91 -0.53 7.75 -0.09 0.53 6.74 - 1 .9 1  -0.53 6.74 
198-g N 0.37 -0. 14 8.03 1 .63 0.14 8.03 -0.37 -0.14 6.46 - 1 .63 0.14 6.46 
200-b A 0.62 0.86 8.08 1 .38 -0.86 8.08 -0.62 0.86 6.42 -1 .38 -0.86 6.42 
201-c A 0. 14 0.38 8.44 1.86 -0.38 8.44 -0. 14 0.38 6.06 - 1 .86 -0.38 6.06 
203� R 1 . 14 0.63 8.48 0.86 -0.63 8.48 -1 . 14 0.63 6.0 1 -0.86 -0.63 6.01 
204-f A 0.48 0.9 1  8.76 1 .53 -0.9 1 8.76 -0.48 0.9 1 5.73 -1 .53 -0.9 1 5.73 
205-g D 0.39 0. 19 9.04 1.6 1  -0.19 9.04 -0.39 0. 19 5 .45 - 1 .61 -0.19  5 .45  
207-b D 1 . 1 1  0.93 9.09 0.89 -0.93 9.09 -1 .1 1 0.93 5 .4 1  -0.89 -0.93 5.4 1 
208-c A 0.44 0.76 9.44 1 .56 -0.76 9.44 -0.44 0.76 5 .05 -1 .56 -0.76 5.05 
Table 4-10  Coordinates of a pair of antiparallel I B  segments for chicken gizzard desmin. The 
pitch length of the coiled-coils is 12.1 nm. Residue numbers listed are relative to the amino-terminus 
of the chain and are followed by the heptad positions of the residues and their single letter code. 
Residues in the innermost positions of the heptad, it and d, have not been included as they are assumed 
to be incapable of forming intermolecular interactions. The x,y,z coordinates are measured in  
nanometres relative to an origin midway between the starting residues of the molecules: the z-axis is 
directed parallel to the axes of the molecules and the x-axis is colinear with the frrst residues from the 
amino-terminal end of the segments occupying the outermost ill position of the heptad. These 
coordinates are plotted in Figure 4-10. 
CHAPTER 4: Secondary and Tertiary Structure page 1 15 
Mol�ul� 1 MQlecul�2 
Coiled-Coil 1 Coiled-Coil 2 Coiled-Coil 1 Coiled-Coil 2 
Residue x y z x y z x y z x y z 
21O-e T 1.43 0.47 9.49 0.57 -0.47 9.49 -1 .43 0.47 5.00 -0.57 -0.47 5.00 
21 1-f L 1 .00 1 .05 9.77 1.00 -1.05 9.77 -l .OO 1.05 4.72 - 1 .00 -1 .05 4.72 
212-g A 0.57 0.47 10.05 1.43 -0.47 10.05 -0.57 0.47 4.44 - 1 .43 -0.47 4.44 
214-b I 1 .56 0.76 10.10 0.44 -0.76 10. 10 - 1 .56 0.76 4.40 -0.44 -0.76 4.40 
215-c D 0.89 0.93 10.45 1. 1 1  -0.93 10.45 -0.89 0.93 4.04 -1 . 1 1 -0.93 4.04 
217-e E 1 .61 0.19 10.50 0.39 -0.19 10.50 - 1 .61  0.19 3.99 -0.39 -0.19 3.99 
218-f R 1 .53 0.91 10.78 0.48 -0.91 10.78 -1 .53 0.91 3.7 1 -0.48 -0.91 3.7 1 
219-g R 0.86 0.63 1 1 .06 1 .14 -0.63 1 1 .06 -0.86 0.63 3.43 -1 . 14 -0.63 3.43 
221-b E 1 .86 0.38 1 1 . 1 1  0. 14 -0.38 1 1 . 1 1  -1 .86 0.38 3.39 -0. 14 -0.38 3.39 
222-c S 1 .38 0.86 1 1 .46 0.62 -0.86 1 1 .46 -1 .38 0.86 3.03 -0.62 -0.86 3.03 
224-e Q 1 .63 -0. 14 1 1 .5 1 0.37 0.14 1 1 .51 -1 .63 -0. 14 2.98 -0.37 0.14 2.98 
225-f E 1 .91  0.53 1 1 .79 0.09 -0.53 1 1.79 -1 .9 1  0.53 2.70 -0.09 -0.53 2.70 
226-g E 1 . 19 0.61 12.07 0.81 -0.61 12.07 -1 .19 0.61 2.43 -0.81 -0.61 2.43 
228-b A 1 .93 -0. 1 1  12. 1 1 0.07 0. 1 1  12. 1 1  - 1 .93 -0. 1 1  2.38 -0.07 0. 1 1  2.38 
229-c F 1 .76 0.56 12.47 0.24 -0.56 12.47 -1 .76 0.56 2.02 -0.24 -0.56 2.02 
231-e K 1.47 -0.43 12.52 0.53 0.43 12.52 -1 .47 -0.43 1.97 -0.53 0.43 1.97 
232-f K 2.05 0.00 12.80 -0.05 0.00 12.80 -2.05 0.00 1 .69 0.05 0.00 1 .69 
233-g V 1 .47 0.43 13.08 0.53 -0.43 13.08 -1 .47 0.43 1.42 -0.53 -0.43 1 .42 
235-b E 1 .76 -0.56 13.12 0.24 0.56 13. 12 -1 .76 -0.56 1.37 -0.24 0.56 1 .37 
236-c E 1 .93 0. 1 1  13.48 0.07 -0. 1 1  13.48 -1 .93 0. 1 1  1 .01  -0.07 -0.1 1 1 .01  
238-e I 1 . 19 -0.61 13.53 0.81 0.61 13.53 -1 . 19 -0.61 0.96 -0.81 0.61 0.96 
239-f R 1 .9 1  -0.53 13.81 0.09 0.53 13.81 - 1 .9 1  -0.53 0.68 �.09 0.53 0.68 
240-g E 1 .63 0.14 14.09 0.37 -0.14 14.09 -1 .63 0. 14 0.4 1 -0.37 -0.14 0.4 1 
242-b Q 1 .38 -0.86 14.13 0.62 0.86 14. 13 -1 .38 -0.86 0.36 -0.62 0.86 0.36 
243-c A 1 .86 -0.38 14.49 0.14 0.38 14.49 -1 .86 -0.38 0.00 -0.14 0.38 0.00 
Table 4-10 continued 
meaningful prediction of their atomic coordinates. In order to reflect the mobility of the 
sidechains, a "volume of influence" is calculated: if the �-carbon atoms of a pair of 
residues from adjacent molecules lie within the volume, then the residues have the 
potential to interact The volume chosen is an ellipsoid where 
(�X)2 (�y)2 (�z)2 1 
0.82 + 0.82 + 0.32 < 
and �x, �y and �z are the differences in the x, y and z coordinates respectively of the 
�-carbon atoms of a pair of residues. This volume implies that interacting residues may 
be separated by up to 0.8 nm azimuthally (ie, in the x-y plane) but by only 0.3 nm 
(-2 residues) axially. Justification for this restriction on the axial freedom of the 
sidechains has already been discussed (Section 4.2). 
Interactions studied include those between charged residues (lysine or arginine with 
aspartic acid or glutamic acid), hydrophilic residues (serine, threonine, glutamine, 
asparagine, arginine and tyrosine) and apolar residues (leucine, valine, isoleucine, 
methionine, phenylalanine, tyrosine and alanine). Some residues may exhibit multiple 
characteristics: for example, tyrosine has both apolar and hydrophilic propenies  and 
arginine has both charged and hydrophilic propenies. In addition to the three types of 
CHAPrER 4: Secondary and Tertiary StruCb.U'e page 1 16 
Figure 4-10 Schematic view of two IB segments, each a dimes of coiled-coil chains. TIle segments 
have been rotated slightly out of the plane and some perspective has been added 10 enhance the view. 
The line segments shown connect the positions calculated for the Cp atoms in Table 4-10, although 
the residues int.emallO the coiled-ooils have been included here. 
interaction calculated directly, a sum of the three is also made. 
At this relatively early stage in the development of the three-dimensional technique, all 
the interactions have been weighted similarly with the exception of those involving 
alanine. Few physical data are available to allow useful modification of the weighting 
scheme and it is worth considering the success of the one-dimensional study which is 
based on a very simple model of protein structure. Down-weighting the interactions 
involving alanine, however, is based on its relative smallness compared to the other 
hydrophobic residues in its group. The interaction of alanine with a large hydrophobic 
residue is weighted by 0.25 and alanine with itself is weighted by 0.252. In addition, 
the criterion used for interaction distance is simple: no allowance is made for the 
different flexibilities or lengths of the sidechains or for the different rates at which the 
interactions attenuate over distance. 
With three independent variables (pitch length, relative axial stagger and a common 
azimuth for the molecules) and four dependent variables (ionic, hydrogen-bond, 
apolar and total interaction scores) the study generates a large amount of data. For 
example, analysis of segment IB involves varying the axial stagger between ±15  nm 
in 0.2 nm steps and the relative rotation from 0° to 180° in 2° steps (because the 
molecules are homodimeric and their rotations are equal, rotations from 1 80° to 360° 
are equivalent to those from 0° to 180°) and results in 27000 scores being generated for 
each pitch length and interaction type. These data can be represented compactly by an 
interaction map such as that shown in Figure 4- 1 1  where rotation is represented along 
the horizontal axis, stagger is along the vertical axis and the interaction score is 
represented by a grey level on the graph - darker areas indicate higher interaction 
scores. The four different types of interaction are shown together as a mosaic in 
Figure 4-1 1  and a separate mosaic is calculated for each pitch length. 
CHAPI'ER 4: Secondary and Tertiary Sttucture 
+t 
o 
o 
1800 
Azimuth 
.. :.:.:.:.:.:.:.:.:.:.:-:.:.:.: .
.
.
. 
1 800 
Azimuth 
page 1 17 
Figure 4- 1 1  Interaction map for parallel 1B  segments of chicken gizzard desmin. The pitch length 
of the coiled-coil is 12.1 nm. Higher scores are represented by darker shading. Stagger is displayed in 
units of the segment length. In this case, segment 1B  is 101  residues long corresponding to 1 5  nm 
(assuming a constant axial rise per residue of 0. 1485 nm). 
4 . 4 . 3  Analysis of the Interaction Maps 
A selection of the highest scoring total interactions for segment IB  of chicken gizzard 
desmin are listed in Table 4- 1 1  (antiparallel segments) and 4- 1 2  (parallel segments). 
No attempt at generating statistical levels of significance have been made. Such 
calculations are straightforward in principle and would follow the method detailed in 
Section 4.2. However, the enonnous computational effort involved disallows any 
attempt at this at the present time. 
The scores recorded in Tables 4- 1 1  and 4- 12  are typically about 25% of the maximum 
scores calculated by Fraser et al ( 1985) in their one-dimensional analysis of ionic 
residues from the same protein. This difference arises primarily from the distance 
constraints placed on the interaction parameters in the three-dimensional study. A 
CHAPTER 4: Secondary and Tertiary Sttucture page 1 18 
Pitch (nm) Azimuth (0) Stagger (residues) Total Score 
1 6.4 132 - 12.00 16.00 
1 2. 1  6 -9.33 1 3.00 
1 4.2 108 -9.33 14.50 
1 5.3 108 -9.33 14.50 
1 3.2 14 -4.00 12.00 
14.2 150 -4.00 14.00 
1 5.3 178 -4.00 14.25 
1 6.4 174 -4.00 15.25 
1 2. 1  146 0.00 15.63 
1 3.2 134 4.00 13.50 
1 4.2 82 33.33 13 .00 
1 5.3 100 33.33 1 3.00 
Table 4- 1 1  A selection of the highest three-dimensional interaction scores for antiparaIIel 1B  
segments of  molecules of chicken gizzard desmin. The total score is the sum of  the ionic, hydrogen 
bonding and apolar interaction scores. Due to the symmetry of the homodirneric molecules, the same 
scores are found for azimuths offset by 1800 from those listed. The stagger is recorded in units of 
residue translation in axial projection (=0.1485 om). 
. 
simple example illustrates this point. In the one-dimensional study, ionic interactions 
between two 1B segments are scored in pairs. This occurs because the molecule is 
assumed to consist of a pair of parallel, in-register chains and when modelled as a 
single linear sequence there is effectively two identical residues in each position. These 
pairs of residues are assumed to interact in unison. However, in the three-dimensional 
model, the previously 'colinear' residues are now radially opposed about the axis of 
the molecule and are unlikely to contribute to the interaction score simultaneously. 
This implication alone of the three-dimensional structure of the molecules inherently 
reduces the scores by a factor of about two. 
Higher interaction scores are revealed between antiparallel rather than parallel 
segments although the difference is not great. Also, higher scores are apparent at the 
extremes of pitch length studied (12.1 and 16.4 nm): however this may be an artifact 
of the geometry of the structures studied and the fixed relationship between the 
azimuths of each molecule. The axial length of segment IB is  approximately 1 5  nm 
( 101 residues each with an axial rise of -0. 1485 nm), a value which is of the same 
order as the pitch lengths studied. The 16.4 nm pitch length arranges the residues of 
segment 1B  on less than one pitch of coiled-coil. As the pitch length is reduced, the 
angle of the pitch becomes shallower (ie, closer to a plane transverse to the axis of the 
molecule) and the number of residues within interaction range of each other decreases 
CHAPrER 4: Secondary and Tertiary Structure page 1 19 
Pitch (nm) Azimuth (0) Stagger (residues) Total Score 
1 2. 1  -150 (= -30°) -±2.00 14. 19  
1 3.2 0 ±1 .33 14.06 
1 4.2 -103 0.00 1 2.06 
1 5.3 124 0.00 13.00 
1 6.4 140 ±2.67 14.00 
1 3.2 -91 ±21 .33 1 2.00 
14.2 -103 ±21 .33 13.00 
1 5.3 -1 15 ±21 .33 1 2.00 
Table 4-12 A selection of the highest three-dimensional interaction scores for parallel IB  
segments of molecules of chicken gizzard desmin. The total score is the sum of the ionic, hydrogen 
bonding and apolar interaction scores. Due to the symmetry of the homodimeric molecules, the same 
scores are found for azimuths offset by 1800 from those listed. The stagger is recorded in units of 
residue translation in axial projection (=0.1485 nm). The '-' symbol indicates that the score was 
common to a range of azimuths or staggers of which only the central one is listed here. 
slightly. Further reduction of the pitch length to below the axial length of segment IB 
causes the residues to be arranged on more than one pitch and hence the numbers of 
residues within interaction range will increase. 
A feature of Table 4-12 (parallel segments) is that the azimuths of the high scoring 
orientations appear to increase with increasing pitch length. Such a trend is not 
apparent in the antiparallel case. This result can also be attributed in part to the 
geometry of the structure. In the case of parallel segments, the azimuth increases with 
the pitch length in order to "maintain" optimal interactions whereas for antiparallel 
segments no adjusttnent in relative azimuth is possible with the implementation of the 
perpendicular diad axis assumed. 
The results presented here do not reveal any clear preference for either the pitch length 
of the coiled-coil or for the relative orientations of the molecules. Rather than a direct 
failing of the model, the problem probably lies in the lack of refinement of some 
parameters. For example, some weighting scheme for the different kinds of interaction 
as well as separate criteria for their ranges would be preferable to the arbitrary methods 
used here. However, until some physical evidence is pr�sented no detailed 
specifications of these parameters can be made realistically. The excessive 
computational load also prevents a systematic examination of different combinations of 
feasible parameters. 
A limitation of this molecular model is that the coiled-coil is assumed to be a regular 
structure. However, the helix geometry suffers local distortions which are likely to 
affect both the positions and orientations of the amino acid sidechains as well as the 
CHAYrER 4: Secondary and Tertiary Structure page 120 
pitch length. In addition to being irregular, the molecular structure is also dynamic and 
there is evidence that the non-covalent bonds between residues are frequently made 
and remade. Nuclear magnetic resonance studies on IF proteins by Mack et al ( 1988) 
suggest that the peptide backbone in the rod domain is relatively stiff but that the 
sidechains are very mobile: leucine sidechains revealed two states that were related by 
rotations of -1200 about both the Ca-Cp and CP-Cy bonds. Also, the flexibility of the 
protein backbone (which was examined in Chapter 2) is further evidence that IF 
proteins do not have a 'fIxed' conformation but instead experience constant motion of 
the residues relative to each other. Both of these points are currently incorporated in 
the relatively loose defInition of the interaction distance and it is possible that a more 
exact specifIcation may not be feasible. 
4 . 5  Summary 
Interchain ionic interactions have been calculated for the lamin proteins, peripherin and 
the Helix pomotia B protein. In each case, a parallel in-register arrangement of chains 
is clearly preferred in accordance with previous studies on other IF and a-fibrous 
protein chains. Generally, segment 1 contributes a greater proponion of ionic 
interactions to the total than segment 2. Of special interest is the unusual intensity of 
interactions predicted for the human and Xenopus lamin A chains: the scores of these 
proteins (36) are fully twice that of peripherin (18) for example. These results indicate 
1 ) that the proteins studied follow the general plan of IF structure in so far as the 
molecules are comprised of parallel, in-register chains, 
2) that within the molecule, segment 1 is relatively more stable than segment 2,  
and 
3) that the lamin A molecules are particularly stable compared to the other IF 
molecules studied. 
This last point may indicate that lamin molecules require greater structural integrity 
than the cytoplasmic IF possibly to meet the special demands made on the lamin 
meshwork in supporting the nuclear envelope. 
Intermolecular ionic interactions have also been studied for this same group of 
proteins. A brief discussion of the original method for calculating one-dimensional 
ionic interactions for collagen and a-fIbrous proteins is presented in Section 4.2. This 
method has been used as the starting point for the development of a more sophisticated 
approach which takes into account the uneven distribution of apolar residues in coiled­
coil segments. The positions of maxima in the interaction profiles for peripherin are 
found to compare well with those from a previous study on other type m proteins and 
provide further confmnation that peripherin is indeed a type ITI IF protein. The results 
CHAPTER 4: Secondary and Tertiary Sb'Ucture page 121 
for human lamins A and C are also of significance in that they can be combined with 
physical data on lamin paracrystal and lattice structures and used to generate a 
plausible model for the structures. The interactions for the remaining proteins are 
included for completeness but no further conclusions are possible without additional 
physical data. 
The model building study on human lamins A and C attempts to rationalize a number 
of observations made on the various lamin polymorphs by explaining the disparate 
data with a common model of molecular packing. The one-dimensional ionic 
interaction data are used to supply constraints on the modes of aggregation of. the 
molecules and a single preferred model is presented. The model is simple in form, 
comprising a half-staggered array of antiparallel lamin molecules with each molecule 
consisting of a pair of parallel, in-register chains. It also predicts that the large 
C-terminal domains are highly extended. The axial aggregation of lamin molecules is 
proposed to be specified largely by ionic interactions. Apolar interactions between 
similarly directed segments corresponding to the highly conserved sequences found in 
all IF proteins in segment lA and at the C-tenninal of segment 2B are also considered 
to be of importance. 
An additional feature of the proposed model for the packing of lamin molecules i s  the 
possibility of dual lattice-paracrystal structures. This is illustrated diagrammatically 
with the large C-tenninal domains giving rise to the banding pattern observed in  the 
paracrystal on the collapse of the lattice form. Evidence for such a duality is given in 
an electron micrograph for another a-fibrous protein where two different forms of 
packing (crystalline lattice and paracrystal) are apparent in the same structure: one 
packing mode merges smoothly with the other. 
The mode of packing proposed here for lamin molecules is closely analogous to that 
suggested previously for other IF molecules (Geisler et ai, 1985b; Parry and Fraser, 
1985; Steinert and Parry, 1985). However lamin molecules are capable of adopting a 
variety of structural forms while IF molecules generally form only filaments of 
'native' or 'near-native' dimensions ("" 10 nm diameter) although paracrystals formed 
from fragments of a type ill IF protein have been reported recently (Quinlan et  ai, 
1989; Stewart et ai, 1989a,b). The polymorphism of lamin may be attributable to the 
fundamental differences in lamin structure compared to the other IF molecules. One 
such difference is the relative simplicity of the structure for lamin which is predicted to 
be a-helical throughout the rod domain and which has fewer breaks in the heptad and 
ionic substructures than other IF molecules. Consequently, lamin molecules are more 
akin to the a-fibrous proteins in muscle such as tropomyosin,  paramyosin and 
CHAPTER 4: Secondary and Tertiary Structure page 122 
myosin, all of which exhibit considerable polymorphism in vitro. 
The role of phosphorylation as a control mechanism for the assembly of the nuclear 
lamina is well established. The nuclear lamina is known to disassemble concomitant 
with phosphorylation. Serine and threonine residues are generally considered to be 
potential sites for phosphorylation and this has been observed in the lamin proteins by 
Ottaviano and Gerace (1985). The model for aggregates of lamin molecules outlined 
above also provides a basis for understanding the dissociation of the lamin network. 
The phosphorylation sites within the terminal domains would lie in close proximity to 
the rod domains. The axial cohesion of the lamin structure could then be destabilized 
by the addition of extra charged groups. Presumably the intermolecular ionic 
interactions will allow reassembly of the structure once the charged moieties have been 
removed. 
An interesting observation by Ottaviano and Gerace (1985) is that low concentrations 
of phosphate were associated with assembled lamins in growing interphase cells. 
Gerace (1986) suggested that this might allow local weakening of the lamin mesh work 
so that newly synthesized lamins could be incorporated. This idea is also consistent 
with the model above although the nature of the insertion of new material into the 
orthogonal lattice is not clear. In particular, the method of extending the lattice while 
maintaining a constant cross-over spacing is difficult to visualize. It is possible that 
some temporary distortion to the orthogonal lattice is required: the lattice spacing might 
expand between two parallel filaments to allow a third, preassembled filament to be 
inserted. 
A refinement to the interaction method is offered in Section 4.4. The one-dimensional 
approximation to the molecular structure is modified to a three-dimensional model. 
This allows a more realistic consideration of the potential for interactions to be made 
between residues on adjacent molecules although at this stage there are insufficient 
data for proper specification of the limits and relative weights associated with different 
types of interaction. Preliminary results are presented for a type ill IF protein, chicken 
gizzard desmin, but unfortunately no clear conclusions can be made on the likely 
modes of molecular packing. 
CHAYrER 5: Summary page 123 
5. SUMMARY 
A number of aspects of IF chain and molecular structure, as well a s  molecular 
aggregation, have been examined in this thesis. These include the extent of homology 
among the IF proteins, the delineation of periodicities in the sequences of structural 
domains of IF proteins, the distribution of amino acid residues within the heptad 
substructure, the flexibility of the peptide backbone, the packing of chains in the 
dimeric molecule, and the axial packing of molecules in the IF. To a large degree the 
studies have been based on the amino acid sequences of the IF proteins. This approach 
has proved remarkably successful in that it has allowed much of the structural 
hierarchy to be specified from the primary structure right up to the level of molecular 
organization in the IF. 
Several new conclusions can now be made about IF chain classification. In addition, 
other established ideas can be confinned. 
1 )  Subtyping of the type I and IT keratin IF chains has been proposed on the basis 
of sequence homology within the rod domain. The 'hard' keratins  are termed 
types la and lla and the 'soft' or epidennal keratins are tenned Ib and llb. 
2) The peripherin chain is confirmed as a type ill IF chain in spite of its location in 
neuronal tissue and its coexpression with the type IV neurofilament chains. 
Examination of the gene structure by others had already indicated that peripherin 
shared a greater homology with type III IF proteins than type IV but the 
Fourier, homology and ionic interaction studies presented here all provide 
confirmatory evidence for this classification. 
3) The classification of neurofilament chains as a type IV group separate from the 
type III IF proteins is also confinned by the homology studies. In addition, the 
NF-L, NF-M and NF-H subtyping, which were assigned initially on the basis of 
the molecular weights of the proteins, is shown to be reflected in the homology 
within the rod domains and indeed throughout the entire amino acid sequences. 
4) The type V chains may require sUbtyping based on the evidence of the sequence 
disparities between the karyoskeletal and cytoskeletal members of the group. 
Although simple nomenclature is clearly desirable, the more detailed specification of IF 
groupings described above is a response to the increasing structural diversity apparent 
from the large body of IF protein sequence information now available. A greater 
understanding of the protein structure is reflected in the new subclasses. In some cases 
the finer divisions are revealed on a macroscopic scale as, for example, between the 
CHAPTER 5: Summary page 124 
'hard' and ' soft' keratins. 
Some of the results obtained in this work are in good agreement with previously 
established features of IF protein structure. The Fourier transfonn studies on the 
newly sequenced type V IF proteins and the type ITI peripherin protein reveal 
periodicities common to all IF. In particular, a 9- 10  residue period in the linear 
distribution of acidic and basic residues is present in the transforms. This period is 
also evident in Fourier transforms of the homology curves for acidic and basic 
residues in each of the IF types, indicating that it is conserved among all IF proteins. 
The ubiquity of this period provides overwhelming evidence of its importance in 
specifying the modes of aggregation of molecules in the IF. 
An interesting feature of some of the transforms for type V IF chains is the -9.23 
residue period which has been observed previously only in the type I keratins. This 
period raises some new possibilities for the interaction of coexpressed IF networks in 
the cytoplasm (ie, keratin-containing IF and the Helix pomotia B-containing IF) 
although it may only reflect a common structural organization adopted by the separate 
networks. In addition, the role of lamin B as a possible nuclear anchorage site for the 
cytoplasmic IF has been raised (Georgatos and Blobel, 1987) and again the -9.23 
residue period common in lamin B and the type I keratins indicates that a means for 
direct interaction between the IF networks does exist. The problem of how the 
cytoskeletal and karyoskeletal IF interact across the nuclear membrane has yet to be 
resolved. Some evidence does exist, however, that parts of both networks are 
localized around the nuclear pores and hence may maintain connectivity through the 
pore structure. 
Another structural feature of the coiled-coil regions that is confirmed here is the 
dominance of apolar residues in the internal i!. and !! positions of the heptad 
substructure. The examination of residue occupancy of the heptad positions presented 
in Chapter 2 is based on a greatly expanded protein sequence database from that of 
Parry and Fraser ( 1985). Nonetheless, it confmns their measurement of an occupancy 
rate of -75% for apolar residues in the internal positions of the heptad. The average 
homology scores of residues in these positions is also higher than for residues in other 
positions of the heptad. However the different numbers of individual apolar residues 
in the structurally similar i!. and !! positions emphasize that they are not 
stereochemically identical. Residues in the adjacent � and g positions reveal the second 
highest mean homology scores and in the remaining positions,  the homology is 
significantly lower. It is apparent that functional and structural diversity can be built 
into the 'outer' heptad positions, 12, £ and i, domain without upsetting the molecular 
CHAFfER 5: Summary page 125 
structure. 
The conclusion that the lamin A molecule is stabilized to a greater degree by interchain 
ionic interactions than other IF molecules is interesting. The nuclear lamina provides 
much of the structural integrity of the nuclear envelope and. in addition. provides a 
scaffold for the nuclear pore structures and an anchor for interphase chromatin. 
Presumably these structural demands on the lamina require a greater stability for the 
lamin ftlaments than is needed for the cytoplasmic IF. 
An area of interest that has not been addressed in this thesis is the characterization of 
the portions of protein sequences that specify either homodimeric or heterodimeric 
molecules. Reconstituted keratin ftlaments are formed from mixtures of type I and IT 
proteins whereas the other IF readily re-form monocomponent filaments (Steinen. 
198 1 a) although the various type ITI chains are capable of copolymerization in vitro 
(Steinen et al. 198 1 .  1982). The reason for this dichotomy of dimer structure is 
currently unknown: neither are the sequence-specific features that give rise to 
heterodimeric molecules on one hand (types I and IT) and homodimers on the other 
(types m. IV and V). Some progress has been made. however. in identifying the 
broad regions of keratin rod domains that recognise the complementary polypeptide in 
the heterodimer although no explanation of the mechanism is yet possible (Magin 
et al. 1987). The reason for this extra complexity in the structure of keratin molecules 
is also obscure. One possibility is that the rod domain sequences largely dictate the 
unique pairing of particular type I chains with their type IT panners (in addition to 
specifying the axial coherence of the filament) and that the terminal domains have some 
special functional role that could not be achieved by a homodimeric structure. 
A further elaboration on this theme is possible. Cohen and Parry ( 1986) have made an 
interesting observation that intracellular a-helical coiled-coil proteins are two-stranded 
(for example. IF proteins. tropomyosin. paramyosin and myosin) whereas all tbree­
strand coiled-coil proteins are extracellular and include fibrinogen (Doolittle et ai, 
1978;  Parry. 1978). laminin and, more recently. the gp 17 tail fibre of the 17 
bacteriophage (Steven et al. 1988). In addition they noted that all two-stranded coiled­
coil molecules have the same parallel-chain, in-register structure (Cohen and Parry, 
1989). These protein sequences also contain the information which specify whether 
the molecule is a two or three stranded rope and again it is of interest to determine how 
and where this is specified by the amino acid sequences. 
The rod domain of the IF chain is predicted to be less flexible on average than the 
terminal domains. This is in agreement with the perceived role of the rod as specifying 
the axial integrity of the filament (ie, the 'core')  and with the terminal domains being 
CHAPfER 5: Summary page 126 
located at least partly on the fllament periphery. Some portions of the rod of highest 
flexibility are located near the heptad stutters and may facilitate distortion of the coiled­
coils to accommodate the break. The structural implications of the disruptions to the 
structure of the coiled-coil, however, remain obscure. 
Several of the studies presented here are hindered somewhat by insufficient sequence 
data. Additional information from the 'hard' keratins is required to allow homology 
comparisons to be made and to more fully delineate the boundaries between the 
terminal sub-domains. Few type V sequences are available: in particular, sequences 
from lamin B proteins and cytoplasmic type V IF (analogous to the Helix pomotia B 
protein) are required. Both of these proteins are of special interest: the lamin B because 
of its suggested role in connecting the karyo- and cyto-skeletal IF networks and the 
Helix protein which appears to represent a new class or sub-class of IF chain. This 
latter protein shares an extended 1 B segment with the lamin proteins but otherwise the 
homology between it and the lamins is not as great as that amongst the lamins alone. 
The location of the Helix protein within the cell is also different from that of the lamins 
and hence it is not unlikely that their functions are dissimilar. The potential for 
interaction between keratins and the Helix protein has been discussed but evidence to 
substantiate the occurrence of such aggregation is required. 
Differences between the amphibian and mammalian keratins and also between the hard 
and soft keratins indicate that the rod domains of type IT chains have retained a greater 
degree of homology overall than have the type I chains. This conclusion raises the 
question of why one partner of the obligate heterodimeric molecule should change at a 
greater rate than the other. It seems possible that the structural or functional roles of the 
rod domains are slightly different for the type I and IT chains. 
The method for calculating intermolecular ionic interactions based on a one­
dimensional model of molecular structure remains a useful tool for predicting likely 
modes of aggregation. Some improvement on the technique is still possible, especially 
with regard to establishing a weighting scheme for the four possible types of 
interaction (Asp-Lys, Asp-Arg, Glu-Lys, Glu-Arg). Some physical basis for 
determining the appropriate scale factors is required, however. An empirical basis is 
also possible whereby potential ionic interactions measured for some protein of known 
structure (preferably similar to the coiled-coil of IF) provide the relative probabilities in 
much the same manner as the temperature factor data of a group of proteins allow 
chain flexibility to be predicted (Section 2.3). 
The three-dimensional method also shows promise although greater attention to the 
detail of the structure and the parameters of the interactions is required. A problem 
CHAPTER 5: Summary page 127 
inherent in the method is determining the positions of the amino acid sidechains. The 
model used for the rod domain of the IF molecule assumes a regular coiled-coil 
structure but it is apparent that in vivo the coiled-coils will suffer local distortions 
(Phillips et ai, 1986) in much the same way as a-helices in crystalline globular 
proteins have geometries noticeably different from the classical a-helix. This problem 
is compounded further by the mobility of the peptide backbone and the sidechain s  as 
indicated by NMR studies (Torchia and van der Hart, 1976; Mack et ai, 1988). The 
uncertainty in sidechain position is currently overcome by specifying a volume within 
which the J3-carbons of the two structures must exist if an interaction is to take place. 
However, the results presented here imply that the method is not yet selective enough 
to yield unequivocal results. Unfortunately it is not clear how the selection criteria may 
be improved at the present time. 
Polymorphism has been observed in aggregates of the nuclear lamin proteins and has 
provided extra sources of structural infonnation about molecular packing. These data 
have been used in combination with intennolecular ionic interaction studies to build a 
model for the axial packing of lamin molecules. This model can explain in a simple 
way the effect of phosphorylation in dismantling the lamin mesh work during mitosis 
and the subsequent reassembly on dephosphorylation. In addition, the transition 
between the observed lattice and paracrystalline forms is outlined. Predictions in this 
work and in Conway and Parry (1988) of polymorphism in the type I-IV IF have been 
confmned by the mouse GFAP paracrystals described recently (Quinlan et ai, 1 989; 
Stewart et ai, 1989a,b). Variations in structure among IF from different chain types 
are also predicted based on the differing homology amongst the rod domains of the 
type I-IV chains. 
Stewart et al ( 1989b) have developed a model for the packing of rod domain 
fragments of mouse GF AP in paracrystalline structures. This model bears some 
similarities to that proposed for lamin paracrystals in Chapter 4 although the relative 
axial staggers of molecules are somewhat different More axial substructure is apparent 
in the GF AP paracrystals than in lamin and this may result from the absence of the 
terminal domains in the GF AP fragments. [As an aside, it may be posited that if the 
terminal domains were cleaved from lamin molecules, the rod domains would 
aggregate in very well-ordered paracrystals]. However, an identical packing mode for 
both GFAP and lamin molecules is unlikely due to the differences in the lengths of the 
1B  segments in the chains. 
The terminal domains are probably the least well understood structural elements of IF 
proteins. In many instances, phosphorylation sites have been identified within these 
CHAFfER 5: Summary page 128 
domains but little progress on their tertiary conformation or their functions has been 
made. Certainly the terminal domains do not appear to be so crucial as the roo domain 
for maintaining the axial structure of the filament and it is probable that the functions of 
different IF are specified largely by the terminal domains. In some cases, the length of 
the terminal domain far outruns that of the rod domain (the NF-H chains are an extreme 
example): in effect this implies a larger potential for function than for those IF chains 
where the terminal domains are short. The location of at least parts of these domains 
on the exterior of the filament suggests that their role is one of interacting with other 
proteins in the cell. The structure of the terminal domains, however, does not lend 
them to X-ray diffraction study in the same manner as the rod domains and hence 
future work on this aspect of the IF protein structure must rely on electron 
microscopy, chemical studies and predictive structural techniques. Alternatively, it  
may be possible to crystallize the terminal domains individually. 
The special features of the recently characterized type V IF group raise many questions 
about IF proteins in general. For instance, is the extended IB segment a remnant of the 
postulated ancestor to all IF proteins or a development from it? Does the molecular 
structure containing the extended IB segment differ from that for the type I-IV IF 
molecule? Where these type V IF coexist with other IF, how do the different networks 
- interact? The protein sequence studies here do not answer these questions but 
hopefully add to the ever-increasing body of knowledge about the IF proteins that will 
someday allow a fuller explanation of their structure and function. 
In addition to IF structure, the role of IF in the cell has not yet been fully determined. 
The proposed connections between the cell nucleus via the cytoplasmic IF to the cell 
membrane suggests several possibilities: positioning of the nucleus, maintenance of 
cell shape and signal transmission or sensing between the nucleus and the cell surface. 
The further connection between IF of neighbouring cells through the desmosomes 
carries these ideas onto a larger scale and allows the possibility of physical cell-to-cell 
communication via the IF networks. 
An area of direct application for this kind of work is in the investigation of cell 
dysfunction in so far as the IF are involved. For instance, Osborn and Weber ( 1982, 
1983, 1986) have outlined the advantages for tumour diagnosis by IF typing. Certain 
types of tumour express particular IF types (eg, carcinomas are characterized by 
specific cytokeratins) and in addition the tumour will generally continue to express the 
same IF as the original cells. The use of IF antibodies may aid in the classification of 
tumours as an alternative to the usual histological procedures. 
Skin disorders related to keratin IF have been under investigation by several groups. 
CHAPrER 5: Summary page 129 
Steinert et al (198Oc) have detennined that some skin disorders can be associated with 
abnormalities in the keratin chain, specifically in the non-a-helical tenninal domains. 
Tyner and Fuchs ( 1986) describe keratin chains from human epidermis that are 
expressed during periods of rapid growth such as wound-healing. Some chronic skin 
diseases are related to uncontrolled proliferation of these chains and hence the 
regulatory mechanisms involved are a subject of interest as well as the structural 
differences between the keratins expressed during normal tissue growth and during 
rapid regrowth. Trachtenberg et al (1986) and Trachtenberg ( 1987) have investigated 
the structure of keratin IF in a specific chronic skin condition - psoriasis. The keratin 
IF contained in psoriatic cells is reduced in amount and in addition appears to be 
proteolytic ally degraded. The importance of defining the structural hierarchy and 
function of IF is clear: abnormalities on a microscopic level can and do result in 
tangible and serious defects occurring. 
The advances made in recent years in determining the structure and spatial arrangement 
of IF proteins have been great. Acknowledgement of the homology between the 
keratins and the non-epithelial IF has allowed a coalescence of these previously 
separate fields of study and has resulted in many valuable insights into the structure 
and function of IF. The recent inclusion of the type V proteins into the IF family has 
revealed further variations on the common structural framework of the IF molecule and 
suggested that IF form part not only of the cytoplasmic skeleton but also of the 
karyoskeleton. Other work on potential IF-binding protein components of the 
desmosomes, such as the desmoplakins, indicates the possibility of an intercellular 
connection between the IF networks. It is clear that IF is  an element of the cell 
structure that has come of age and can no longer be regarded as merely the cell debris 
in centrifuge experiments. 
High quality X-ray diffraction and electron microscope data will be needed if the 
structure of IF is to be determined in greater detail than at present It is to be hoped that 
fragments of IF or of IF molecules may soon be available for crystallization in a form 
suitable for X-ray analysis. In particular, the pitch length of the of the supercoil and 
the nature of the defonnations in the supercoil structure at the heptad discontinuities are 
aspects of the structure that could be revealed by such studies. 
In addition to the established techniques of X-ray diffraction and electron microscopy, 
new methods are now being used to investigate IF structure, and indeed protein 
structure in general. Site-directed mutagenesis is beginning to reveal new aspects of 
the structure of IF chains such as the nuclear localizational signal for the nuclear lamin 
proteins (Heald et al, 1988; Loewinger and McKeon, 1988). This technique has great 
CHAPTER 5: Summary page 130 
potential for probing specific features of IF molecules and their modes of assembly. 
Nuclear magnetic resonance studies are being used to determine information about the 
dynamics of the IF molecule and have shown that the positions of the amino acid 
residues of these proteins and their sidechains are not as flXed as was once thought. 
Advances in preparative techniques for electron microscopy are also likely to yield 
finer details of the molecular and IF structure by allowing samples to be preserved at a 
higher level of fidelity than is currently possible. Image enhancement and analysis 
techniques used in conjunction with electron microscopy will surely result in improved 
resolution of structural detail. Scanning transmission electron microscopy has been 
used for determining mass-per-unit-Iength data on IF and shows potential for 
investigating the early stages of IF assembly. 
The rate of progress in advancing our knowledge about the IF proteins has been 
remarkable in the past decade. However, the emphasis in the field is now moving 
away from IF structure per se and towards the IF-associated proteins such as 
desmoplakin, fibronectin, and fIlaggrin. Only when the complex interplay between the 
IF and its associated proteins is understood will the full function of the cytoskeleton 
become apparent We may expect significant developments in this exiting area of cell 
biology will occur during the 1990's. 
APPENDIX A: Zero-Filling prior to Fourier Transformation page 131 
ApPENDICES 
Appendix A : Zero-Filling prior to Fourier Transformation 
The operation of zero-filling prior to Fourier transfonnation has the effect of 
improving the resolution in Fourier space. Generally, zero tilling is perfonned by 
extending the data record by the addition of data points to the end of the record. For 
example, the Fourier transfonns described in Chapter 2 involve sequence arrays of 
101  to 356 data points which are zero-filled to a total length of 2048 points. The 
improved resolution in the Fourler space is achieved at the cost of introducing small 
side lobes for each peak. The consequences of the zero-filling operation will be 
described here. 
Zero-filling may be described as the truncation of a long, hypothetical data record, 
d(x), by multiplication with a rectangle function, TI(x). This rectangle function is 
defined as 
II(x) = { � 
o�<w 
W�<P 
where W is the width of the non-zero data segment and P is the period, or full width 
of the filled record. For the purposes of zero-filling, A is typically set equal to unity. 
The discrete Fourier transform of II(x) is defmed as 
F{II(x)} = � JII(x).expc2;iXf)dx 
IF {II(x)} 1  
W 
- - exp --1 JA (
-21tixf)dx 
- P P 
_ -A [ (-21tixf)]W - 21tif exp p 0 
-A [ -21tiWf ] = 21tif exp( p -) - I 
A [ (1tiWf) ( -1tiWf)] (-1tiWf) = 21tif exp -p- - exp -p- exp -p-
A . (1tWf) (-7tiWf) = -sm - exp --1tf P p 
W sin(1t;f) = A-P 'W& 1tp 
= A � sinc(1t � f) 
= A � sinc(a) 
. sin(x) where smc(x) = -x-
W 
where a = 1t p f 
APPENDIX A: Zero-Filling prior to Fourier Transformation page 132 
The intensity function is simply the square of the Fourier transform, 
IF{IT(x)} 12 = [A � ] 2 sinc2(a) 
See Figure A-I for a graphical representation of these functions. Minima occur for 
sin(a) = 0 a � 0 
or, sin(1t � f) = 0 
which implies that '; f = n where n = I ,  2, 3 ,  . . .  
p or, f = n W 
Maxima occur when the slope of the sinc function is zero, ie 
:a sinc(a) = 0 
cos(a) 
_ 
sin(
2
a) = 0 a a 
� [acos(a) - sin(a)] = 0 
a 
acos(a) - sin (a) = 0 
sin(a) a = cos(a) 
a = tan(a) 
ie, a "" 0, ±1 .43031t, ±2.45901t, ±3.47071t, . . .  
P P P or f "" 0, ±1 .4303 W' ±2.4590 W' ±3 .4707 W' 
The half width limits of the main intensity peak are given by: 
[sin�a)] 2 = � 
The half width is then 
a = ±..fi sin(a) 
ie, a "" ± 0.44291t 
P or, f :::: ± 0.4429 W 
.1a :::: 0.88587t 
p 
or, .1f :::: 0.8858 W 
The ratios of the main intensity peak to the fIrst few side lobes are: 
at a = 0, sinc2(a) = I 
at a = 1 .43037t, sinc\a) :::: 0.047 19 
at a = 2.45907t, sinc2(a) ::::: 
at a = 3.47077t, sinc2(a) ::::: 
0.01648 
0.008340 
1 ::::: 2 1 . 19 
I ::::: 60.68 
1 ::::: 1 19.90 
APPENDIX A: Zero-Filling prior to Fourier Transformation page 1 33 
A 
o 
1 
Rectangle Function TI(x) = {A: 05.x<W 
W 
I 
-2.4590 P/W 
-3 .4707 P/W -1 .4303 P/w 0 
0: WSx<P 
P 
- sinc(1tfW/P) 
- sinc2(1tfW/P) 
In Width = 0.8858 P/W 
I 
2.4590 P/W 
1 .4303 P/w 3.4707 P/w 
-
x 
Figure A-I The rectangle function IT and its Fourier transform pair, the sine function. The 
intensity of the sinc function is also shown along with the positions of the first three side lobes and 
the half width of the central peak in the intensity distribution. 
Using the length of segment IB as an example and zero-filling to 2048 points, 
W = 101 
P = 2048 
A = 1 
then the rectangle function is defined as 
TI(x) = { � 
OSx<lOl 
101Sx<2048 
APPENDIX A: Zero-Filling prior to Fourier Transformation 
and the intensity of its Fourier transform is 
F{II(x)}2 = [A � ] 2 sinc2(a) 
[ 101  ] 2 . 2 101 = 2048 SInC (1t 2048 f) 
The half width of the main intensity peak is 
p 
f "'" 0.8858 W 
"'" 17.96 
and the first sidelobes occur at frequencies of 
p f "'" ± 1 .4303 W 
== ± 29.00 
page 134 
W where a = 1t P f 
The effect of multiplying the data record by a second function (ie, the II function) is to 
convolute the Fourier transform of the data by the Fourier transform of the second 
function. Convolution by the sinc function causes peaks in the data transform to be 
broadened by the width of the central peak of the sine function as well as allowing 
"leakage" of peaks into the areas defined by the sidelobes of the sine function. The 
relative strengths of the side lobes to the main peak are not large (less than 5%) and 
generally their effect on the Fourier plane can be ignored. However in the case of large 
peaks, such as the zero-frequency spike produced when the baseline has not been 
corrected for, the side lobes become significant relative to the rest of the Fourier data. 
The zero-frequency peak may be eliminated by performing a baseline correction 
operation prior to the Fourier transform. 
APPENDIX B: Fourier Transfonns page 135 
Appendix B : Fourier Transforms 
The Fourier transfonns of charged residues in the rod domains of peripherin, human 
lamins A and C, Xenopus lamins A and B and the Helix pomotia B protein are 
presented in this appendix. These curves are discussed in Chapter 2. The relationship 
between the frequency axis and periods in the sequence is: 
period = freq��ncy where N f = size of the data set. 
The value of Ne is 2048 for all of the transforms included here. Some periods of 
structural significance with their corresponding frequencies are: 
1 )  the 9 to 10 residue period (lE frequencies from 228 to 205), 
2) the heptad period ( 1 st order =< 293; 2nd order s 585; and 3rd order = 878). 
Although the heptad is defined mainly by apolar residues, nonetheless the charged 
residues can reveal some regularities of this order due to their rarity in the il and d 
positions and their conservation in the � and � positions. 
Peripherin 
Period Segment 1B Segment 2 
(residues) acidics basics acidics basics 
1 3.045 4.7 12  
10.343 5.576 
9.894 7.359 
9.752 7.39 1 
9.6 15  5 .639 
6.9 19 6.099 
4.008 5.292 
3 .7 10 4.530 
3.47 1 3.078 
2.852 5.022 
2.833 4.353 
2.330 5.580 
2.248 4.873 
2.24 1  4.268 
Table B-1 A selection of peaks in the Fourier transforms of acidic and basic residues in the major 
coiled-coil segments of peripherin (Figure B-t). Their scaled intensities are shown together with the 
period to which the peaks correspond All peaks above an arbitrarily chosen cutoff intensity of 4.0 are 
listed although lower peaks are included where they coincide with periods that may be related to 
known structural features. 
APPENDIX B: Fomier Transfonns 
8 
6 
4 
8 
a 6 .
� 
.s 4 
8 
a 6 
.� 
..s 4 
8 
a 6 
.� 
..s 4 
2 
Peripherin. Segment IB: Fr of acidic residues 
Peripherin. Segment IB: Fr of basic residues 
Peripherin. Segment 2: Fr of acidic residues 
Peripherin. Segment 2: Fr of basic residues 
page 136 
o�����--��������-&�--��------��--���� 
o 200 400 600 800 1000 
Frequency 
Figure B-t Fourier transfonns of the dispositions of acidic residues (top graphs) and basic residues 
(lower graphs) in rod domain segments of rat peripherin. A selection of peaks are listed in Table B- 1 .  
Period is related to frequency by the expression: 
.od 2048 pen = frequeocy 
APPENDIX B: Fourier Transfonns page 1 37 
Hymm LaminS A and (; 
Period Segment 1B Segment 2 Rod domain 
(residues) acidics basics acidics basics acidics basics 
12.962 4 .504 
1 1 .838 4.018  
1 1 .636 6. 1 36 
9.942 6.232 3.476 4.4 17  
9.894 6.272 
9.309 6.245 3.654 
8.569 4.235 5 .874 
6.606 4.526 6.053 
5 .069 7.352 
5.044 6. 1 23 
3.568 4.045 
3.374 4.420 
3.098 5.248 4 . 1 1 3  
3.025 5.52 1 
3.02 1 8.346 
2.825 4.237 
2.768 4.805 
2.760 4. 1 53 
2.528 6.067 
2.465 5.62 1 
2.343 4.869 
2.338 6.979 
2.3 1 7  4.298 
2.278 4.733 
2. 122 6.07 1 5 . 1 29 
2.048 4.204 
2.036 5 .349 
Table B-2 A selection of peaks in the Fourier transfonns of acidic and basic residues in the major 
coiled-coil segments of human lamins A and C (as for Table B-1). See Figure B-2. 
APPENDIX B: Fourier Transfonns page 138 
Human Lamin A, Segment 1: FT of acidic residues 
8 
2;- 6 ... 
� 
£ 
4 c .... 
2 
0 
Human Lamin A, Segment 2: FT of acidic residues 
8 
2;- 6 
.� c 4 -
2 
0 
Human Lamin A, Rod Domain: FT of acidic residues 
8 
2;- 6 'r.;; c 
S c 4 -
2 
0 
0 200 400 600 800 1000 
Frequency 
Figure B-2a Fourier transforms of the dispositions of acidic residues in rod domain segments of the 
human lamin A protein. These transforms are also equivalent to those for the human lamin C protein 
which has a primary rod domain sequence identical to that of lamin A. A selection of peaks are listed 
in Table B-2. Period is related to frequency by the expression: 
'od 2048 pen = frequeocy 
APPENDIX B: Fourier Transforms page 139 
Human Lamin A, Segment 1 :  Pr of basic residues 
8 
� 6 'fil 
� 
.s 4 
2 
0 
Human Lamin A, Segment 2: Pr of basic residues 
8 
� 6 
'fil c 
£ c 4 
..... 
2 
0 
Human Lamin A, Rod Domain: Pr of basic residues 
8 
� 6 
'fil c £ 
.5 4 
2 
0 
0 200 400 600 800 1000 
Frequency 
Figure B-2b Fourier transforms of the dispositions of basic residues in rod domain segments of 
the human lam in A protein, These transforms are also equivalent to those for the human lamin C 
protein which has a primary rod domain sequence identical to that of lamin A. A selection of peaks are 
listed in Table B-2, Period is related to frequency by the expression: 
'od 2048 pen = r-n u"'tuency 
APPENDIX B:  Fourier Transfonns 
Period Segment IB  
(residues) acidics basics 
l 3 .653 
1 1 .636 
9.942 6.089 3 . 1 17 
9.894 
9.309 6.903 
8.569 
6.966 3.066 
6.606 
6.585 5. 369 
6.46 1 
5 .069 
5.057 
3 .374 
3 .098 4.560 
3.025 6.924 
2.768 
2.528 
2.507 
2.465 
2.335 6. 8 12 
2.278 4.384 
2. 1 27 
2. 1 24 
2.086 
2.079 4. 145 
2.038 
Xe.no/lW Lamin A 
Segment 2 
acidics basics 
4.708 
7.368 
6.956 
5.731  
page 140 
Rod domain 
acidics basics 
4.38 3  
3.979 
5.27 1 
4.638 
5.449 
5.259 
4.0 1 8  
6.254 
4. 1 23 
5. 1 63 
5.056 
4.020 
6.620 
4.968 
4.002 
5.544 
Table B·3 A selection of peaks in the Fourier transforms of acidic and basic residues in the major 
coiled-coil segments of Xenopus lamin A (as for Table B-1). See Figure B-3. 
APPENDIX B: Fourier Trnnsfonns page 141 
Xenopus Lamin A, Segment 1:  Fr of acidic residues 
8 
� 6 
.;n c 
£ 
4 .El 
2 
0 
Xenopus Lam in A, Segment 2: Fr of acidic residues 
8 
.� 6 .., 
Y 
.El 4 
2 
0 
Xenopus Lamin A, Rod domain: Fr of acidic residues 
8 
� 6 .;n 
Y c 4 -
2 
0 
0 200 400 600 800 1000 
Frequency 
Figure B·3a Fourier transfonns of the dispositions of acidic residues in rod domain segments of the 
Xenopus lamin A protein. A selection of peaks are listed in Table B-3. Period is related to frequency 
by the expression: 
.od 2048 pen = frequeocy 
APPENDIX B: Fowier Transfonns page 142 
Xenopus Lamin A, Segment 1: FT of basic residues 
8 
� 6 
.� 
..s 4 
2 
0 
Xenopus Lamin A, Segment 2: FT of basic residues 
8 
l;> 6 
.� 
9 
4 ..s 
2 
0 
Xenopus Lamin A, Rod domain: FT of basic residues 
8 
l;> 6 
.� 
9 4 s:: 
-
2 
0 
0 200 400 600 800 1000 
Frequency 
Figure B-3b Fowier transforms of the dispositions of basic residues in rod domain segments of 
the Xenopus lamin A protein. A selection of peaks are listed in Table B-3. Period is related to 
frequency by the expression: 
'od 2048 
pen = frequeocy 
APPENDIX B: Fourier Transfonns 
Period Segment lB 
(residues) acidics basics 
14.629 4:507 
12. 190 
1 1 .703 4.44 1 
1 1 .57 1 
10.089 
9.942 
9.846 7.405 
9.225 4.603 
8.29 1 4.505 
7.341 
4.267 
3.568 
3.525 4.004 
3.5 1 3  
3.030 6.207 
3.025 
2.8 17  4.227 
2.775 
2.768 
2.333 6. 1 3 1  
2.243 4.246 
2.238 
2.038 
2.022 
XeCJ.QIl.W Lamin B 
Segment 2 
acidics basics 
4.403 
3. 1 39 
4.025 
4.052 
page 143 
Rcxidomain 
acidics basics 
4.509 
5.504 
3 . 1 83 
7 .964 
4.928 
5 . 1 4 1  
4.98 1 
4.490 
5.903 
3 .585 
4 .889 
4.863 
Table 8·4 Peaks in the Fourier transfonns of acidic and basic residues in the major coiled-coil 
segments of Xenopus lamin B (as for Table B-1). See Figure B-4. 
APPENDIX B:  Fourier Transfonns page 144 
Xenopus Lamin B. Segment 1 :  FT of acidic residues 
8 
>. 6 
... 
. � £ 4 c 
-
2 
0 
Xenopus Lamin B. Segment 2: FT of acidic residues 
8 
?;- 6 
'/iI 
� c 4 
-
2 
0 
Xenopus Lamin B. Rod Domain: FT of acidic residues 
8 
?;- 6 
'1i5 
§ c 4 -
2 
0 
0 200 400 600 800 1000 
Frequency 
Figure 8·4a Fourier transfonns of the dispositions of acidic residues in rod domain segments of the 
XeMpus lam in B protein. A selection of peaks are listed in Table B-4. Period is related to frequency 
by the expression: 
'od 2048 pen = frequeocy 
APPENDIX B: Fourier Transfonns page 145 
Xenopus Lamin B, Segment 1:  FT of basic residues 
8 
?;> 6 
'<i) c 
£ 
.s 4 
2 
0 
XenopusLamin B, Segment 2: FT of basic residues 
8 
>- 6 
-
'� 
£ c 4 
-
2 
0 
Xenopus Lam in B, Rod Domain: FT of basic residues 
8 
?;> 6 
'� 
£ c 4 -
2 
0 
0 200 400 600 800 1000 
Frequency 
Figure B-4b Fourier transfonns of the dispositions of basic residues in rod domain segments of 
the Xenopus lamin A protein, A selection of peaks are listed in Table B-4. Period is related to 
frequency by the expression: 
'od 2048 pen = frequency 
APPENDIX B: Fourier Transfonns 
Period Segment 1B 
(residues) acidics basics 
12. 190 5. 146 
10.894 4.345 
10.039 4.310 
9.846 
9.526 
9.352 
9.309 5.673 
9.267 
9.225 
9.062 
8.225 4.057 
6.502 
5.347 
4. 1 12 
3.5 1 3  4.614 
2.95 1 4.339 
2.926 
2.840 
2.447 
2.263 
2. 167 
2. 158 
2.077 4. 1 66 
He.lix 12omotia B 
Segment 2 
acidics basics 
6.3 10 
5.574 
4.739 
4.509 
5.088 
page 146 
Rod domain 
acidics basics 
7.362 
4.273 
3 .028 
3.469 
5 .540 
9.860 
4.530 
5. 108 
4. 1 20 
4.325 
4.279 
4. 177 
Table 8-S Peaks in the Fourier transfonns of acidic and basic residues in the major coiled-coil 
segments of the Helix pomotia B protein (as for Table B-1). See Figure B-S. 
APPENDIX B: Fourier Transfonns page 147 
Helix B, Segment 1:  FT of acidic residues 
8 
Helix B, Segment 2: FT of acidic residues 
8 
6 
4 
Helix B, Rod Segment: FT of acidic residues 
8 
� 6 
'� 
.s 4 
o 200 400 600 800 1000 
Frequency 
Figure B-Sa Fourier transfonns of the dispositions of acidic residues in rod domain segments of the 
Helix pomotia B protein, A selection of peaks are listed in Table B-S. Period is related to frequency by 
the expression: 
'od 2048 pen = freq ueocy 
APPENDIX B: Fourier Transforms page 148 
HelixB, Segment 1: Ff of basic residues 
8 
� 6 o� 
c 4 -
2 
0 
Helix B, Segment 2: Ff of basic residues 
8 
� 6 0<i! 
c £ 
.s 4 
2 
0 
Helix B, Rod Segment Ff of basic residues 
8 
� 6 0<i! 
c £ 
.s 4 
2 
0 
0 200 400 600 800 1000 
Frequency 
Figure B-Sb Fourier transforms of the dispositions of basic residues in rod domain segments of 
the Helix pomotia B protein ° A selection of peaks are listed in Table B-50 Period is related to 
frequency by the expression: 
00d 2048 pen = frequeocy 
APPENDIX C: Curve Smoothing page 149 
Appendix C : Curve Smoothing 
Curve smoothing is performed by convolution of the raw data with a Gaussian 
function of the form : ( 4 In  2 x2 ) g(x) = exp -
62 
where -oo<x<oo and 6 is the half width of the Gaussian function. The Fourier 
transform G(X) of the Gaussian function g(x) is given by 
( 1t2 62 X2 ) G(X) = F{g(x)} = exp - 4 In 2 
where X is the Fourier (or reciprocal) space coordinate. For ease of computation, the 
smoothing operation is performed in Fourier space in the following manner : 
FI' invFf 
h(x) * g(x) ::) H(X) · G(X) = S(X) ::) s(x) 
where H(X) and S(X) are the Fourier transforms of the unsmoothed h(x) and the 
smoothed s(x) respectively. The * is the symbol for convolution and the • symbolizes 
multiplication. Although this series of operations looks more complicated than directly 
convoluting h(x) and g(x), it is in fact computationally faster because it involves fewer 
basic arithmetic operations. 
APPENDIX D: Intennolecular Ionic Interactions page 150 
Appendix D : Intermolecular Ionic Interactions 
Ionic interactions for the sequences studied in Section 4.2 are listed here. The scores 
for human lamins A and C and for peripherin, which are also given in Section 4.2, are 
reproduced here for completeness. In addition, the interaction curves corresponding to 
the tables are shown in this Appendix. 
Parallel molecules Antiparallel molecules 
.1z(U-U) S So .1z(U-D) S So 
IBU- I BU  ±54 32 30 IBU-IBD  3 68 60 
±4 58 58  
2U -2U  ±138 6 6 2U-2D -65 44 36 
±1 13  20 1 8  -34 48 48 
±103 24 20 4 60 56 
±93 24 24 1 3  64 52 
±72 32 30 34 44 44 
±63 34 42 43 56 40 
±34 44 42 53 40 40 
±26 46 44 64 36 36 
±5 64 50 74 32 32 
100 28 24 
1 10 24 20 
1 2 1  16 1 6  
1 32 12  12  
1 35 8 8 
143 4 4 
IBU - 2U  - 133 10 10 IBU-2D -85 32 32 
-94 34 28 -34 50 48 
-82 32 32 -25 46 46 
-7 1 38 38 4 48 46 
-55 46 44 33 36 34 
-43 48 48 44 32 30 
- 14 48 48 53 26 26 
Table D-l Significant ionic interactions between rod domain segments of peripherin molecules as 
a function of relative axial stagger (see Figure D- l). l\z(U-U) and l\z(U-D) signify the relative staggers 
between parallel and antiparallel molecules respectively. Only those scores (S) greater than or equal to 
the 0.975 quantile (SO) are listed. 
APPENDIX D: Intennolecular Ionic Interactions 
� 60 u C"/.l 
c: o .::1 40 
� 
� 20 
·S  
.s 
� 60 
C"/.l 
c: o .::1 40 
� 
£ 
� 20 
·S 
.s 
Peripberin: 1BU-IBU 
Peripherin: 2U-2U 
page 151 
O��
�-------------------+------------------��� 
� 60 u Cl) 
c: o .::1 40 
� 
� 20 
·S o -
Peripherin: 1BU-2U 
O�-
-----�--------�--�--�--�--��----��------� 
-ISO -100 -50 o 50 
Residue Translation 
Figure D·la Ionic interaction curves for parallel segments of rat peripherin. 
100 150 
APPENDIX 0: Intermolecular Ionic Interactions 
� 60 (,) Cl) 
c: o 
·d 40 
j (,) 20 
·S o -
Petipherin: IBU-IBO 
page 152 
O+-------��------�------�------��--���------� 
� 60 (,) Cl) 
c: o ·0 40 
� 
� 20 
·s o -
Peripherin: 2U-20 
o��----------------------------------------��� 
� 60 (,) Cl) 
c: o ·0 40 
e 
£ 
� 20 
] 
Perioherin: IBU-20 
o�------��----�------��----�------�------� 
-150 -100 -50 o 50 
Residue Translation 
Figure D -1 b Ionic interaction curves for antiparallel segments of rat peripherin. 
100 150 
APPENDIX D: Intennolecular Ionic Interactions page 153 
Parallel molecules Antiparallel molecules 
dZ(U-U) S So dz(U-D) S So 
±345 8 8 -264 56 52 
±3 1 1  24 24 - 1 84 84 84 
±297 32 30 - 164 96 92 
±294 32 32 - 125 1 16 1 12 
±284 40 36 - 1 17 124 1 12 
±28 1  40 38 - 1 14 1 12 1 12 
±267 44 42 - 105 1 16 1 16 
±262 44 44 15  164 148 
±229 58 58 24 148 144 
±193 70 70 60 140 1 32 
±164 84 84 94 124 1 24 
±136 100 96 145 100 100 
±123 1 06 102 160 96 96 
±1 14 106 104 282 40 40 
±4 158 146 288 36 36 
340 12  1 2  
Table D-2 Significant ionic interactions between the rod domains of human lam in A molecules 
(or lamin C molecules) as a function of relative axial stagger (see Figure D-2). �(U-U) and �(U-D) 
signify the relative staggers between parallel and antiparaUel molecules respectively. Only those 
scores (S) greater than or equal to the 0.975 quantile (SO) are listed. 
150 
125 
100 
75 
50 
25 
Human Lamin A: U-U 
O �--��--�----r---�----T---�----__ --�----�--��--�----�---r----�---T�� 
150 
125 
100 
75 
u 50 
'S 
� 25 
Human Lamin A: U-D 
O �--��--�----r---�----T---�----__ --�----�--��--�----�---r----����� 
-400 -300 -200 -100 o 100 200 300 
Residue Translation 
Figure D·2 Ionic interaction curves for parallel (V-V) and antiparallel (U-D) rod domains of human lamin A. The curves for human lamin C 
are the same as these because the rod domain sequences are identical. Figure adapted from Parry et al (1986). 
400 
APPENDIX D: Intermolecular Ionic Interactions page 1 55 
Parallel molecules Antiparallel molecules 
�z(U-U) S So Az(U-D) S So 
±345 8 8 -295 36 36 
-1:297 30 30 -264 56 52 
-1:282 40 36 - 184 92 84 
-1:267 46 42 - 164 96 88 
-1:263 46 44 - 144 100 100 
-1:229 56 56 - 125 1 16 108 
± 136 98 94 - 1 17 1 16 1 12 
±123 102 100 -69 1 28 1 28 
±1 15  104 102 -36 140 140 
-1:21 140 138 -24 144 144 
±4 154 146 0 152 1 52 
15 164  148 
24 156 144 
60 140 1 32 
69 128 128 
94 120 120 
146 100 100 
283 40 40 
288 36 36 
340 1 2  12  
Table D·3 Significant ionic interactions between the rod domains o f  Xenopus lamin A molecules 
as a function of relative axial stagger (see Figure D-3). �(U-U) and �z(U-D) signify the relative 
staggers between parallel and antiparallel molecules respectively. Only those scores (S) greater than or 
equal to the 0.975 quantile (SO) are listed. 
� o u 
CZl 
§ 
·0 u 
e 
£ 
.s 
u ..... 
§ -
150 
125 
100 
75 
50 
25 
Xenopus Lamin A: U -U 
0+---��--�----�--�----�--�----�--�----�--�----�--��--�----�--��--4 
150 
125 
100 
75 
u 50 
·2 
� 25 
Xenopus Lamin A: U-D 
O+---��---r----�--�----�----�--�----�----�--�----�--��--�----�--�----� 
-400 -300 -200 -100 o 100 200 
Residue Translation 
Figure 0-3 Ionic interaction curves for parallel (V-U) and antiparallel (V-D) rod domains of Xenopuslarnin A. 
300 400 
APPENDIX 0: Intermolecular Ionic Interactions page 157 
Parallel molecules Antiparallel molecules 
�z(U-U) S So �-D) S So 
±352 4 4 -353 4 4 
±339 10 10 -313 24 24 
±294 32 28 -283 36 36 
±28 1  38 32 -281 36 36 
±262 46 38 -264 44- 44 
±259 40 38 - 158 84 80 
±251  50 42 -36 1 32 1 20 
±242 46 44 -6 1 28 128 
±151  80 76 4 1 28 128 
±1 14 90 88 53 1 12 1 12 
±4 1 34 124 86 104 104 
130 92 92 
160 92 84 
17 1 84 76 
221 64 60 
239 56 52 
260 48 44 
28 1 40 36 
330 16  16  
340 12  1 2  
352 4 4 
Table D-4 Significant ionic interactions between the rod domains of Xenopus lamin B molecules 
as a function of relative axial stagger (see Figure 04). �(U-U) and �(U-O) signify the relative 
staggers between parallel and antiparallel molecules respectively. Only those scores (S) greater than or 
equal to the 0.975 quantile (SO) are listed. 
� 
o u 
Cl) 
s:: o 
.� u 
� 
� 
..s 
u ..... 
s:: o -
� o u 
Cl) 
s:: o 
·c 
� 
..s 
u 
·2 o -
150 
125 
100 
75 
50 
25 
Xenopus Lamin B:  U-U 
O+-----��-----r--------�----_,--------�--------�----_,--------�--------r_----_T--------�----��----_r--------�----��----� 
150 
125 
100 
75 
50 
25 
Xenopus Lamin B: U-D 
O +-----�----��---- --�--�---- --r_--�----�--�----__ --�----�--��--�----���--� 
-400 -300 -200 -100 o 100 200 300 400 
Residue Translation 
Figure D·4 Ionic interaction curves for parallel (V-V) and antiparallel (V-D) rod domains of Xenopuslarnin B. .... U\ 00 
APPENDIX 0: Intennolecular Ionic Interactions page 159 
Parallel molecules Antiparallel molecules 
L\z(U-U) S So &(U-D) S So 
±348 6 4 -349 4 4 
±335 10 10 -341 8 8 
±3 10 20 20 -334 12  12  
±300 22 22 -322 16 16 
±198 56 52 -264 52 40 
±180 58 58 -214 56 56 
±160 68 66 - 199 64 60 
±153 72 68 - 164 72 72 
±125 80 76 - 155 76 76 
±1 14  82 80 - 142 84 80 
±105 84 82 -125 92 84 
±77 94 92 - 105 88 88 
±6 124 1 12 -96 92 92 
-79 104 · 100 
-40 108 108 
-34 120 1 12 
133 80 80 
326 16 16 
336 12  12 
348 4 4 
Table D-S Significant ionic interactions between the rod domains of Helix pomotia B molecules 
as a function of relative axial stagger (see Figure 0-5) . .1z(U-U) and !\Z(U-D) signify the relative 
staggers between parallel and antiparallel molecules respectively. Only those scores (S) greater than or 
equal to the 0.975 quantile (SO) are listed. 
� o u 
Cl) 
r:: o 
'p 
� .... 
� 
r:: -
u ..... 
r:: o -
150 
125 
100 
75 
50 
25 
Helix pomotia B: U-U 
O +---�----�---------r--------�--------�--------�--------��--__ ----r-��----; 
� o u 
Cl) 
r:: 
o 
'p 
� .... 
B 
r:: -
1 50 
125 
100 
75 
u 50 
'8 
� 25 
O+---��--�----�---r----�--�----�--�----�--�----�--��--�----���--� 
-400 -300 -200 -100 o 100 200 
Residue Translation 
Figure D-S Ionic interaction curves for parallel (U-U) and antiparallel (U-D) rod domains of the Helix pomotia B protein. 
300 400 
BmUOORAPHY page 161 
I BIBLIOGRAPHY 
Aebi, U., Fowler, W.E., Rew, P. and Sun, T.-T. ( 1983) I. Cell Biol., 97, 
1 131-1 143 
The fibrillar substructure of keratin filaments unravelled 
Aebi, U., Cohn, I., Buhle, L. and Gerace, L. ( 1986) Nature (London), 323, 
560-564 
The nuclear lamina is a meshwork of intermediate-type filaments 
Ahmadi, B.  and Speakman, P.T. (1978) FEBS Lett., 94, 365-367 
I 
Suberimidate crosslinking shows that a rod-shaped, low cystine, high helix protein 
prepared by limited proteolysis of reduced wool has four protein chains 
Ahmadi, B. ,  Boston, N.M., Dobb, M.G. and Speakman, P.T. ( 1980) in "Fibrous 
Proteins: Scientific, Industrial and Medical Aspects", eds D.A.D. Parry and 
L.K.Creamer, Academic Press, London, 2, 161- 166 
Possible four-chain repeating unit in the microfibril of wool 
Aletta, I.M., Shelanski, M.L. and Greene, L.A. ( 1989) J. BioI. Chem., 264, 
4619-4627 
Phosphorylation of the peripherin 58-kDa neuronal intermediate filament protein: 
Regulation by nerve growthfactor and other agents 
Ambrose, E.1., Elliott, A. and Temple, R.B. (1949) Nature (London), 163, 859-862 
New evidence on the structure of some proteins from measurements with polarized 
infrared radiation 
Astbury, W.T. and Bell, F.O. ( 1941)  Nature (London), 147, 696-699 
Nature of the intramolecular fold in a-keratin and a-myosin 
Astbury, W.T. and Marwick, T.e. (1932) Nature (London), 130, 309-3 10 
X -ray interpretation of the molecular structure of feather keratin 
Astbury, W.T. and Street, A. (1931 )  Phil. Trans. Roy. Soc. London, 230A , 75- 10 1  
X-ray studies on the structure of hair, wool and related fibres I: General 
BmuOGRAPHY page 162 
Astbury, W.T. and Woods, H.J. (1933) Phil. Trans. Roy. Soc. London, 232A, 
333-394 
X-ray studies on the structure of hair, wool and relatedfibres //: The molecular 
structure and elastic properties of hair keratin 
Astbury, W.T., Reed, R. and Spark, L.C. (1948) Biochem. l., 43, 282-287 
An X -ray and electron microscope study of tropomyosin 
Bader, B.L., Magin, T.M., Hatzfeld, M. and Franke, W.W. ( 1986) EMBO J. , 5,  
1865- 1875 
Amino acid sequence and gene organization of cytokeratin no. 19, an exceptional 
tail-less intermediate filament protein 
Bamford, C.H., Brown, L., Elliott, A., Hanby, W.E. and Trotter, I.F. ( 1952) Nature 
(London), 169, 357-358 
Structw-e of synthetic polypeptides 
Bear, R.S . ( 1944) J. Amer. Chem. Soc., 66, 1297-1305 
X-ray diffraction studies onprotein/ibres: I. The large fibre-axis period of 
collagen 
Birbeck, M.S.C. and Mercer, E.H. (1957) J. Biophys. Biochem. Cytol., 3, 203-2 14 
The electron microscopy of the human hair follicle: I. Introduction and the hair 
cortex 
Bragg, W.L., Kendrew, J.C. and Perutz, M.F. ( 1950) Proc. Roy. Soc. London,  
203A, 321-357 
Polypeptide chain configurations in crystalline proteins 
Brown, L. and Trotter, LF. (1956) Trans. Faraday Soc., 52, 537-548 
X -ray studies of PolY-L-alanine 
Burke, B. and Gerace, L. (1986) Cell, 44, 639-652 
A cell free system to study reassembly of the nuclear envelope at the end of mitosis 
Burley, S .K. and Petsko, G.A. ( 1985) Science, 229, 23-28 
Aromatic-aromatic interaction: A mechanism of protein structure stabilization 
Chou, P.Y. and Fasman, G.D. (1974) Biochemistry, 13, 222-245 
Prediction of protein conformation 
BmuOGRAPHY page 163 
Chou, Y.-H., Rosevear, E and Goldman, R.D. (1989) Proc. Nat!. Acad. Sci. USA, 
86, 1 885- 1889 
Phosphorylation and disassembly of intermediate filaments in mitotic cells 
Cochran, W., Crick, F.H.C. and Vand, V. ( 1952) Acta Crystallogr., 5, 581 -586 
The structure of synthetic polypeptides 1. The transforms of atoms on a helix 
Cohen, C. and Holmes, K.C. ( 1963) J. Mol. BioI., 6, 423-432 
X-ray diffraction evidence of a-helical coiled-coils in native muscle 
Cohen, C. and Parry, D.A.D. ( 1986) Trends Biochem. Sci. , 11 ,  245-248. 
a-Helical coiled-coils - a widespread motif in proteins 
Cohen and Parry, D.A.D. ( 1989) Proteins :  Structure, Function and Genetics, 6, in 
press. 
a-Helical coiled coils and bundles: How to design an a-helical protein 
Cohen, C., Lanar, D.E. and Parry, D.A.D. ( 1987) Biosci. Rep., 7, 1 1 - 16 
Amino acid sequence and structural repeats in schistosome paramyosin match 
those of myosin 
Conway, J.F. and Parry, D.A.D. ( 1988) Int. J. BioI. Macromol., 10, 79-98 
Intermediate filament structure: 3. Analysis of sequence homologies 
Conway, J.F. and Parry, D.A.D. (1989) in "Cytoskeletal and Extracellular Proteins :  
Structure, Interactions and Assembly", eds U. Aebi and J .  Engel, Springer Series 
in Biophysics, Springer-Verlag, Berlin, 3, 140- 149 
Structural and spatial organization of intermediate filament and nuclear lamin 
proteins 
Conway, J.F., Fraser, R.D.B., MacRae, T.P. and Parry, D.A.D. ( 1989) in "The 
Biology of Wool and Hair", eds G.E. Rogers, P.J. Reis, K.A. Ward and R.e. 
Marshall, Chapman and Hall, London and New York, 127- 144 
Protein chains in wool and epidermal keratin IF: Structural features and spatial 
arrangement 
Crewther, W.G. (1976) Proc. 5th Int. Wool Text. Res. Conf. Aachen, 1, 1 - 10 1 
Primary structure and chemical properties of wool 
Crewther, W.G. and Harrap, B.S.  (1967) J. BioI. Chem., 242, 43 10-43 19  
The preparation and properties of a helix-rich fraction obtained by partial 
proteolysis of low sulphur S-carboxymethylkerateine from wool 
BmUOORAPHY page 164 
Crewther, W.G. and Dowling, L.M. (1971)  Appl. Polym. Symp., 18, 1 -20 
Preparation and properties of large pep tides from the helical regions of the low­
sulphur proteins of wool 
Crewther, W.G., Fraser, R.D.B. ,  Lennox, F.G. and Lindley, H. ( 1965) Adv. 
Protein Chem., 20, 191-346 
The chemistry of keratins 
Crewther, W.G., Gillespie, I.M., Harrap, B.S. and Inglis, A.S.  (1966) 
Biopolymers, 4, 905-9 16 
Low-sulphur proteins from a-keratins. lnterrelations between their amino acid 
compositions, a-helix contents, and the supercontraction of the parent keratin 
Crewther, W.G., Dobb, M.G., Dowling, L.M. and Harrap, B.S.  ( 1968) in 
"Symposium on Fibrous Proteins", ed W.G. Crewther, Butterworths, 329-340 
The structW"e and aggregation of low-sulphur proteins derived from alpha keratins 
Crewther, W.G., Inglis, A.S . and McKern, N.M. ( 1978a) Biochem. I., 173, 
365-37 1 
Amino-acid sequences of a-helical segments from S-carboxymethylkerateine A :  
Complete sequence of a type-II segment 
Crewther, W.G., Gough, K.R. , Inglis, A.S. and McKern, N.M. ( 1978b) Text. Res. 
I., 48, 1 60- 162 
Sequence homologies in helical segments from a-keratin 
Crewther, W.G., Dowling, L.M., Gruen, L.C., Sparrow, L.G. and Woods, E.F. 
( 1980) Proc. 6th Int. Wool Text. Res. Conf. Pretoria, 2, 1 - 12  
Interactions between the microfibrillar proteins of wool 
Crewther, W.G., Dowling, L.M., Steinert, P.M. and Parry, D.A.D. (1983) Int. I .  
BioI. MacromoI., S, 267-274 
StructW"e of intermediate filaments 
Crewther, W.G., Dowling, L.M., Inglis, A.S., Sparrow, L.G., Strike, P.M. and 
Woods, E.F. ( 1985) Proc. 7th Int. Wool Text. Res. Conf. Tokyo, 1, 85-94 
Wool micro fibrils and intermediate filaments have common structW"al featW"es 
Crick, F.R.C. (1952) Nature (London), 170, 882-883 
Is a-keratin a coiled-coil? 
BmUOORAPHY 
Crick, F.H.C. (1953) Acta Crystallogr, 6, 689-697 
The packing of a-helices: Simple coiled-coils 
page 165 
Dautigny, A., Pham-Dinh, D., Roussel, C., Felix, J.M., Nussbaum, J.L. and Jolles, 
P. (1988) Biochem, Biophys. Res. Commun., 154, 1099- 1 106 
The large neurofilament subunit (NF-H) of the rat: cDNA cloning and in situ 
detection 
Doolittle, R.F., Goldbaum, D.M. and Doolittle, L.R. ( 1978) J. Mol. BioI., 120, 
3 1 1 -325 
Designation of sequences involved in the "coiled-coil" interdomainaI connections 
in fibrinogen: Construction of an atomic scale model 
Dowling, L.M. and Crewther, W.G. (1972) Proc. Aust. Biochem. Soc., 5, 3 
Distribution of two types of a-helical units in the low-sulphur protein chains from 
a-keratin 
Dowling, L.M., Parry, D.A.D. and Sparrow, L.G. (1983) Biosci. Rep., 3, 73-78 
Structural homology between hard a-keratin and the intermediate filament proteins 
desmin and vimentin 
Dowling, L.M., Crewther, W.G. and Inglis, A.S. ( 1986) Biochem. J., 236, 695 -703 
The primary structure of component 8c-1 , a subunit protein of intermediate 
filaments in wool keratin 
Eckert, R.L. (1988) Proc. Natl. Acad. Sci. USA, 85, 1 1 14- 1 1 1 8 
Sequence of the human 40-kDa keratin reveals an unusual structure with very high 
sequence identity to the corresponding bovine keratin 
Elleman, T.C.,  Crewther, W.G. and van der Touw, J. (1978) Biochem. J., 173, 
387-391 
Amino-acid sequence of a-helical segments from S-carboxymethylkerateine A :  
Statistical analysis 
Elliott, A. and Malcolm, B.R. (1956) Nature (London), 178, 192 
Absolute configuration and optical rotation offolded (a) polypeptides 
Elliott, A.  and Malcolm, B .R. ( 1959) Proc. Roy. Soc. (London.), 249A, 30-41 
Chain arrangement and sense of the a-helix in PolY-L-alanine fibres 
Elliott, A. and Offer, G.  (1978) J. Mol. BioI., 123, 505-5 19 
Shape and flexibility of the myosin molecule 
BmuOGRAPHY page 166 
Elliott, A., Lowy, J . ,  Parry, D.A.D. and Vibert, PJ. (1968) Nature (London), 218, 
656-659 
Puzzle of the coiled-coils in the paramyosin filament of molluscan muscles 
Fawcett, D.W. ( 1966) Amer. J. Anat., 199, 129-146 
On the occurrence of a fibrous lamina on the inner aspect of the nuclear envelope in 
certain cells of venebrates 
Filshie, B.K. and Rogers, G.E. (1961)  J. Mol. BioI., 3, 784-786 
The fine structure of a-keratin 
Fisher, D.Z., Chaudhary, N. and Blobel, G. ( 1986) Proc. Nat!. Acad. Sci. USA, 83, 
6450-6454 
cDNA sequencing of nuclear lamins A and C reveals primary and secondary 
structural homology to intennediate filament proteins 
Franke, W.W., Grund, C. and Achtstatter, T. (1986) J. Cell BioI., 103, 1933- 1943 
Co-expression of cytokeratins and neurofilament proteins in a permanent cell line: 
cultured rat PC 12 cells combine neuronal and epithelial features 
Fraser, R.D.B. and MacRae, T.P. ( 1959) Biochem. Biophys. Acta, 29, 229-240 
Structural implications of the equatorial X -ray diffraction pattern of a-keratin 
Fraser, R.D.B. and MacRae, T.P. ( 1973a) Polymer, 14, 61-67 
The structure of a-keratin 
Fraser, R.D.B. and MacRae, T.P. ( 1973b) "Confonnation in Fibrous Proteins and 
Related Synthetic Polypeptides", Academic Press, New York and London 
Fraser, R.D.B. and MacRae, T.P. ( 1983) Biosci. Rep., 3, 5 17-525 
The structure of the a-keratin microfibril 
Fraser, R.D.B. and MacRae, T.P. ( 1985) Biosci. Rep., S,  573-579 
I ntennediate filament structure 
Fraser, R.D.B. and MacRae, T.P. ( 1988) Int. J. BioI. Macromol.,  10, 178- 184 
Surface lattice in a-keratin filaments 
Fraser, R.D.B., MacRae, T.P. and Rogers, G.E. ( 1959) Nature (London), 183, 
592-594 
Structure of a-keratin 
BmUOORAPHY page 167 
Fraser, RD.B.,  MacRae, T.P. and Rogers, G.E. ( 1962) Nature (London), 193, 
1052-1055 
Molecular organization of a-keratin 
Fraser, RD.B. ,  MacRae, T.P. and Miller, A. ( 1964a) Nature (London), 203, 
1231-1233 
Molecular structure of a-keratin 
Fraser, R.D.B. ,  MacRae, T.P. and Miller, A. ( 1964b) J. Mol. Biol. ,  10, 147- 1 56 
The coiled-coil model of a-keratin structure 
Fraser, R.D.B.,  MacRae, T.P. and Miller, A. ( 1964c) Acta Crystallogr., 17, 8 13-8 16  
The F ourier transform of the coiled-coil model for a-keratin 
Fraser, R.D.B.,  MacRae, T.P. and Miller, A. ( 1965) J. Mol. BioI. ,  14, 432-442 
X -ray diffraction patterns of a-fibrous proteins 
Fraser, R.D.B. ,  MacRae, T.P. Millward, G.R, Parry, D.A.D., Suzuki, E. and 
Tulloch, P.A. ( 1971 )  Appl. Polym. Symp., 18, 65-83 
The molecular structure of keratins 
Fraser, RD.B.,  MacRae, T.P. and Rogers, G.E. ( 1972) "Keratins: Their 
composition, structure and biosynthesis", Charles C. Thomas, Springfield, USA 
Fraser, R.D.B. ,  Gillespie, J.M. and MacRae, T.P. ( 1973) Comp. Biochem. Physiol . ,  
44B, 943-947 
Tyrosine-rich proteins in keratins 
Fraser, R.D.B.,  MacRae, T.P. and Suzuki, E. ( 1976) J. Mol. BioI . ,  108, 435-452 
Structure of the a-keratin microfibril 
Fraser, R.D.B.,  Jones, L.N., MacRae, T.P., Suzuki, E. and Tulloch, P.A. ( 1980) 
Proc. 6th Int. Wool Text. Res. Conf. Pretoria, 1, 1-33 
The fine structure of the wool fibre 
Fraser, R.D.B. ,  MacRae, T.P., Suzuki, E. and Parry, D.A.D. ( 1985) Int. J. BioI. 
Macromol., 7, 258-274 
Intermediatejilament structure: 2. Molecular interactions in thefilament 
Fraser, R.D.B. ,  MacRae, T.P., Parry, D.A.D. and Suzuki, E. ( 1986) Proc. Natl. 
Acad. Sci. USA, 83, 1 179- 1 1 83 
Intermediatejilaments in a-keratins 
BmuOGRAPHY page 168 
Fraser, R.D.B., MacRae, T.P., Sparrow, L.G. and Parry, D.A.D. ( 1988) Int. J .  
BioI. Macromol., 10, 106- 1 12 
Disulphide bonding in a-keratin 
Gamier, J. ,  Osguthorpe, D.J. and Robson, B.  ( 1978) J. Mol. BioI., 120, 97- 1 20 
Analysis of the accuracy and implications of simple methods for predicting the 
secondary structure of globular proteins 
Geisler, N. and Weber, K. ( 198 1 )  Proc. Nat!. Acad. Sci. USA, 78, 4120-4 123 
Comparison of the proteins of two immunologically distinct intermediate-sized 
filaments by amino acid sequence analysis: Desmin and vimentin 
Geisler, N. and Weber, K. (1982) EMBO 1., 1, 1649- 1656 
The amino acid sequence of chicken muscle desmin provides a common structural 
model for intermediate filament proteins 
Geisler, N. and Weber, K. (1983) EMBO 1., 2, 2059-2063 
Amino acid sequence data on glial fibiillary acidic protein (GFA); Implications for 
the subdivision of intermediate filaments into epithelial and non-epithelial members 
Geisler, N. and Weber, K. (1988) EMBO J., 7, 15-20 
Phosphorylation of desmin in vitro inhibits formation of intermediate filaments; 
identification of three kinase A sites in the aminoterminal head domain 
Geisler, N., Kaufmann, E. and Weber, K. (1982a) Cell, 30, 277-286 
Proteinchemical characterization of three structurally distinct domains along the 
protojilament unit of desmin 10 nm filaments 
Geisler, N., Plessmann, U. and Weber, K. (1982b) Nature (London), 296, 448-450 
Related amino acid sequences in neurofilaments and non-neuronal intermediate 
filaments 
Geisler, N., Kaufmann, E., Fischer, S. ,  Plessmann, U. and Weber, K. ( 1983) 
EMBO J., 2, 1295-1302 
Neurofilament architecture combines structural principles of intermediate filaments 
with carboxy-terminal extensions increasing in size between triplet proteins 
Geisler, N., Fischer, S., Vandekerckhove, J., Plessmann, U. and Weber, K. ( 1 984) 
EMBO J., 3, 2701-2706 
Hybrid character of a large neurofilament protein (NF -M): Intermediate filament 
type sequence followed by a long and acidic carboxy-terminal extension 
BmuOGRAPHY page 169 
Geisler, N., Fischer, S., Vandekerckhove, J., Van Damme, J., Plessmann, U. and 
Weber, K. ( 1985a) EMBO J., 4, 57-63 
Protein-chemical characterization 0/ NF -H, the largest mammalian neurofiiamenl 
component; intermediate filament-type sequences followed by a unique carboxy­
terminal extension 
Geisler, N., Kaufmann, E . .  and Weber, K. ( 1985b) 1. Mol. BioI., 182, 173- 177 
Antiparallel orientation of the two double-stranded coiled-coils in the tetrameric 
protofilament unit of intermediate filaments 
Geisler, N., Plessmann, U. and Weber, K. (1985c) FEBS Lett. , 182, 475-478 
The complete amino acid sequence of the major mammalian neurofilament protein 
(NF-L) 
Geisler, N., Vandekerckhove, J. and Weber, K. (1987) FEBS Lett., 221, 403-407 
Location and sequence characterization of the major phosphorylation sites 0/ the 
rug h molecular mass neurofilament proteins M and H 
Georgatos, S.D. and Marchesi, V.T. ( 1985) J. Cell BioI., 100, 1955- 196 1 
The binding o/vimentin to human erythrocyte membranes: A model system/or the 
study of intermediate filament-membrane interactions 
Georgatos, S .D. and Blobel, G. ( 1987) J. Cell BioI., 105, 1 17- 125 
Lamin B constitutes an intermediate filament attachment site at the nuclear envelope 
Georgatos, S.D., Weaver, D.C. and Marchesi, V.T. (1985) J. Cell BioI., 100, 
1962- 1967 
Site specijicity in vimentin-membrane interactions: Intermediate filament subunits 
associate with the plasma membrane via their head domains 
Georgatos, S .D., Weber, K., Geisler, N. and Blobel, G. (1987) Proc. Nat!. Acad. 
Sci. USA, 84, 6780-6784 
Binding of two desmin derivatives to the plasma membrane and the nuclear 
envelope of avian erythrocytes: Evidencefor a conserved site-specijicity in 
intermediate filament-membrane interactions 
Georgatos, S .D., Maroulakou, I.  and Blobel, G. (1989) J. Cell BioI.,  108, 
2069-2082 
Lamin A, lamin B, and lamin B receptor analogues in yeast 
BmuOGRAPHY 
Gerace, L. ( 1985) Nature (London), 318, 508-509 
Traffic control and structural proteins in the eukaryotic nucleus 
Gerace, L. ( 1986) Trends Biochem. SeL, 11, 443-446 
Nuclear lamina and organization of nuclear architecture 
Gerace, L. and Blobel, G. ( 1980) Cell, 19, 277-287 
The nuclear envelope lamina is reversibly depolymerized during mitosis 
page 170 
Gerace, L. and Blobel, G. (1982) Cold Spring Harbor Symp. Quant BioI., 46, 
967-978 
Nuclear lamina and the structural organization of the nuclear envelope 
Gerace, L. and Burke, B .  (1988) Ann. Rev. Cell BioI., 4, 335-374 
Functional organization 0/ the nuclear envelope 
Gerace, L. B lum, A. and Blobel, G. ( 1978) 1. Cell BioI., 79, 546-566 
Immunocytochemical localization of the major polypeptides 0/ the nuclear pore 
complex-lamina fraction 
Glass, C., Kim, H.K. and Fuchs, E. ( 1985) J. Cell BioI., 101, 2366-2373 
Sequence and expression 0/ a human type II mesothelial keratin 
Goldman, A.E., Maul, G., Steinert, P.M., Yang, H.-Y. and Goldman, R.D. ( 1986) 
Proc. Natl. Acad. Sci. USA, 83, 3839-3843 
Keratin-like proteins that co isolate with intermediate filaments 0/ BHK-21 cells are 
nuclear lamins 
Goldman, R.D. and Dessev, G.N. ( 1989) in "The Biology of Wool and Hair", eds 
G.E. Rogers, P.J. Reis, K.A. Ward and R.C. Marshall, Chapman and Hall, 
London and New York, 87-95 
Intermediate filaments: Problems and perspectives 
Gough, K.H., Inglis, A.S. and Crewther, W.G. ( 1978) Biochem. J., 173, 373-385 
Amino-acid sequences of a-helical segments/rom S-carboxymethylkerateine A:  
Complete sequence 0/ a type-I segment 
Green, K.J., Parry, D.A.D., Steinert, P.M., Virata, M.L.A. ,  Wagner, R.M., Angst, 
B.D. and Nilles, L.A. ( 1989) J. BioI. Chem., in press 
Structure 0/ the hwnan desmoplakins: Implications for function in the desmosomal 
plaque 
BmuOGRAPHY 
Oruen, L.C. and Woods, E.F. ( 1983) Biochem. I., 209, 587-595 
Structural studies on the microfibrillar proteins of wool 
Hanukoglu, I. and Fuchs, E. ( 1982) Cell, 31, 243-252 
page 171  
The cDNA sequence of a human epidermal keratin: Divergence of sequence but 
conservation of structure among intermediate filament proteins 
Hanukoglu, I. and Fuchs, E. (1983) Cell, 33, 915-924 
The cDNA sequence of a type l/ cytoskeletal keratin reveals constant and variable 
structural dlJrnains among keratins 
Harrap, B.S.  (1963) Aust. J. BioI. ScL, 16, 231 -240 
The conformation of a soluble wool keratin derivative 
Heald, R., Loewinger, L., McKeon, F. ( 1988) J. Cell BioI., 107, 661 a  
Using mutants expressed in vitro to define dlJrnains in the nuclear lamin proteins 
required for the assembly of the nuclear envelope 
Hodge, A.I. and Petruska, J.A. ( 1963) in "Aspects of Protein Structure", ed G.N. 
Ramachandran, Academic Press, London, 289-300 
Recent studies with the electron microscope on ordered aggregates of the 
tropocollagen molecule 
Hodges, R.S., Sodek, I., Smillie, L.B. and Jurasek, L. (1972) Cold Spring Harbor 
Symp. Quant. BioI., 37, 399-4 10 
Tropomyosin: Amino acid sequence and coiled-coil structure 
Hoffmann, W. and Franz, J.K. ( 1984) EMBO J., 3, 1301- 1 306 
Amino acid sequence of the carboxyl-terminal pan of an acidic type I cytokeratin of 
molecular weight 51000 from Xenopus laevis epidermis as predictedfrom the 
cDNA sequence 
Hoffmann, W., Franz, J.K., and Franke, W.W. ( 1985) J. Mol. BioI., 184, 7 13 -724 
Amino acid sequence microheterogeneities of basic (type l/) cytokeratins of 
Xenopus laevis epidermis and evolutionary conservativity of helical and non­
helical dlJmains 
Hogg, D.McC., Dowling, L.M. and Crewther, W.O. ( 1978) Biochem. J., 173, 
353-363 
Amino-acid sequences oJ a-helical segmentsJrom S-carboxymethylkerateine A :  
Tryptic and chymotryptic pep tides from a type-II segment 
BmUOORAPHY page 172 
Hopp, T.P. and Woods, K.R. (198 1) Proc. Nat!. Acad. Sci. USA, 78, 3824-3828 
Prediction of protein antigenic determinants from amino acid sequences 
Huber, R. ( 1979) Trends Biochem. Sci., 4, 271-276 
Conformational flexibility and its functional significance in some protein molecules 
Huggins, M.L. ( 1943) Chem. Rev., 32, 195-218 
The structure of fibrous proteins 
Huggins, M.L. ( 1952) J. Amer. Chem. Soc., 74 , 3963 
Polypeptide helices in proteins 
Hulmes, D.J.S., Miller, A., Parry, D.A.D., Piez, K.A. and Woodhead-Galloway, J. 
(1973) J. Mol. Biol., 79, 1 37-148 
Analysis of the primary structure of collagenfor the origins of molecular packing 
Inagaki, M., Nishi, Y., Nishizawa, K., Matsuyama, M. and Sato, C. ( 1987) Nature 
(London), 328, 649-652 
Site-specific phosphorylation induces disassembly of vimentin filaments in vitro 
Inglis, A.S.,  Sutherland, W.J. and Woods, E.F. (1983) Proc. Aust. Biochem. Soc., 
IS, 31  
Coiled-coil particles from wool: Sequenator analysis using a stationary cup 
procedure 
lp, W., Hartzer, M.K., Pang, Y.-Y.S. and Robson, R.M. (1985) J. Mol. BioI . ,  183, 
365-375 
Assembly of vimentin in vitro and its implications concerning the structure of 
intennediate filaments 
Jonas, E., Sargent, T.D. and Dawid, LB. ( 1985) Proc. Natl. Acad. Sci. USA, 82, 
5413-5417 
Epidermal keratin gene expressed in embryos of Xenopus laevis 
Jorcano, J.L., Rieger, M., Franz, J.K., Schiller, D.L., Moll, R.  and Franke, W.W. 
(1984) J. Mol. BioI., 179, 257-28 1 
Identification of two types of keratin polypeptides within the acidic cytokeratin 
subfamily I 
Julien, J.-P., Ramachandran, K. and Grosveld, F. (1985) Biochem. Biophys. Acta, 
825, 398-404 
Cloning of a cDNA encoding the smallest neurofilament protein from the rat 
BmUOORAPHY 
Julien, I.-P., Grosveld, F., Yazdanbaksh, K., Flavell, D., Meijer, D. and 
Mushyriski, W. (1987) Biochem. Biophys. Acta, 909, 1 0-20 
page 173 
The structure of a human neurofilament gene (NF-L): A unique exon-intron 
organization in the intermediate filament gene family 
Julien, J.-P., Cote, F., Beaudet, L., Sidkey, M., Flavell, D., Grosveld, F.G. and 
Mushynski, W.E. ( 1988) Gene, 68, 307-3 14 
Sequence and structure of the mouse gene coding for the largest neurofilament 
subunit 
Karplus, P.A. and Schulz, G.E. ( 1985) Naturwissenschaften, 72, 212-213  
Prediction of chain flexibility in proteins 
Kaufmann, E., Weber, K. and Geisler, N. (1985) J. Mol. BioI., 185, 733-742 
Intermediatefilamentforming ability of desmin derivatives lacking the amino­
terminal 67 or the carboxy-terminal 27 residues 
Kendrew, J.C., Dickerson, R.E., Strandberg, B .E., Hart, R.G., Davies, D.R.,  
Phillips, D.e. and Shore, V.C. (1960) Nature (London), 185, 422-427 
Structure of myoglobin: Three-dimensional F ouner synthesis at 2 A resolution 
Knapp, B. ,  Rentrop, M., Schweizer, 1. and Winter, H. (1987) J. BioI. Chem., 262, 
938-945 
Three cDNA sequences of mouse type I keratins: Cellular localization of the 
mRNAs in normal and hyperproliferate tissues 
Krieg, T.M., Schafer, M.P., Cheng, C.K., Filpula, D., Flaheny, P., Steinen, P.M. 
and Roop, D.R. (1985) J. BioI. Chem., 260, 5867-5870 
Organization of a type I keratin gene: Evidence for evolution of intermediate 
filaments from a common ancestral gene 
Krohne, G., Wolin, S .L., McKeon, F.D. ,  Franke, W.W. and Kirschner, M.W. 
(1987) EMBO J., 6, 3801 -3808 
Nuclear lamin L1 of Xenopus laevis: cDNA cloning, amino acid sequence and 
binding specijicity of a member of the lamin B subfamily 
Kyte, J. and Doolittle, R.F. (1982) 1. Mol. BioI., 157, 105- 1 32 
A simple method for displaying the hydropathic character of a protein 
Lersch, R. and Fuchs, E. ( 1988) Mol. Cell. Biol. ,  8, 489-493 
Sequence and expression of a type Il keratin, K5, in human epidermal cells 
BmuOGRAPHY 
Lee, LD. and Baden, H.P. (1976) Nature (London), 264, 377-379 
Organisation of the polypeptide chains in mammalian keratin 
page 174 
Lees, J.F., Shneidman, P.S., Skuntz, S.F., Carden, M.J. and Lazzarini, R.A. ( 1988) 
EMBO J., 7, 1947-1955 
The structure and organization of the human heavy neurofilament subunit (NF -H) 
and the gene encoding it 
Lehner, C.F., Flirstenberger, G., Eppenberger, H.M. and Nigg, E.A. ( 1986a) Proc. 
Nat!. Acad. Sci. USA, 83, 2096-2099 
Biogenesis of the nuclear lamina: In vivo synthesis and processing of nuclear 
protein precursors. 
Lehner, C.F., Kurer, V., Eppenberger, H.M. and Nigg, E.A. ( 1986b) J. BioI. 
Chem., 261, 1 3293-1 3301 
The nuclear lamin protein family in higher vertebrates: Identification of quantitively 
minor lamin proteins by monoclonal antibodies 
Lehnert, M.E., Jorcano, J.L., Zentgraf, H., Blessing, M., Franz, J.K. and Franke, 
W.W. ( 1984) EMBO J., 3, 3279-3287 
Characterization of bovine keratin genes: Similarities of exon patterns in genes 
coding for different keratins 
Leonard, D.G.B., Gorham, J.D., Cole, P., Greene, L.A. and Ziff, E.B .  ( 1988)  J .  
Cell BioI., 106, 1 8 1- 193 
A nerve growthfactor-regulated messenger RNA encodes a new intermediate 
filament protein 
Levy, E., Liem, R.K.H., D'Eustachio, P. and Cowan, NJ. (1987) Eur. J. 
Biochem., 166, 71 -77 
Structure and evolutionary origin of the gene encoding mouse NF -M, the middle­
molecular-mass neurofilamentprotein 
Lewis, S .A. and Cowan, N.J. ( 1985) J. Cell BioI., 100, 843-850 
Genetics, evolution, and expression of the 68,O()()-mol-wt neurofilament protein: 
Isolation of a cloned cDNA probe 
Lewis, S.A. and Cowan, N.J. (1986) Mol. Cell. BioI., 6, 1529- 1 534 
Anomalous placement of introns in a member of the intermediate filament 
muitigenefamily: An evolutionary conundrum 
BIBUOORAPHY page 175 
Lewis, S.A., Balcarek, I.M., Krek, V., Shelanski, M. and Cowan, N.I. ( 1984) 
Proc. Natl. Acad. Sci. USA, 81, 2743-2746 
Sequence of a cDNA clone encoding mouse gliaIfibrillary acidic protein: Structural 
conservation of intermediate filaments 
Liem, R.K.H., Yen, S .-H., Salomon, GD. and Shelanski, M.L. ( 1978) J. Cell 
BioI., 79, 637-645 
Intermediate filaments in nervous tissue 
Loewinger, L. and McKeon, F.D. (1988) EMBO 1., 7, 2301-2309 
Mutations in the nuclear lamin proteins resulting in their aberrant assembly in the 
cytoplasm 
Lotay, S.S. and Speakman, P.T. ( 1977) Nature (London), 265, 274-276 
Three-chain merokeratinfrom wool may be afragment of the microfibril 
component macromolecule 
Lu, Y.-I, and Iohnson, P. ( 1983) Int. I. BioI. MacromoI. ,  5, 347-350 
The N -terminal domain of desmin is not involved in intermediate filament 
formation: Evidence from thrombic digestion studies 
MacArthur, I. ( 1943) Nature (London), 152, 38-41 
StrUCture of a-keratin 
Mack, I.W., Torchia, D.A. and Steinert, P.M. ( 1988) Biochem., 27, 541 8-5426 
Solid-state NMR studies of the dynamics and structure of mouse keratin 
intermediate filaments 
Magin, T.M., Hatzfeld, N. and Franke, W.W. ( 1987) EMBO I., 6, 2607-2615 
Analysis of cytokeratin domains by cloning and expression of intact and deleted 
polypeptides in Escherichia coli 
Mandelkow, E., Thomas, I. and Cohen, C. (1977) Proc. NatI. Acad. Sci. USA, 74,  
3370-3374 
Microtubule structure at low resolution by x-ray diffraction 
Marchuk, D., McCrohan, S. and Fuchs, E. ( 1984) Cell, 39, 49 1-498 
Remarkable conservation of structure among intermediate filament genes 
BmUOORAPHY page 176 
Marchuk, D., McCrohan, S. and Fuehs, E. (1985) Proc. Nat!. Acad. SeL USA, 82, 
1609-1613  
Complete sequence of a gene encoding a human type I keratin: Sequences 
homologous to enhancer elements in the regulatory region of the gene 
McKeon, F.D., Kirschner, M.W. and Caput, D. (1986) Nature, 319, 463-468 
Homologies in both primary and secondary structure between nuclear envelope 
and intermediate filament proteins 
MeLachlan, A.D ( 1978) J. Mol. BioI., 124, 297-304 
Coiled coil formation and sequence regularities in the helical regions of a-keratin 
MeLaehlan, A.D. and Stewart, M. (1975) J. Mol. BioI., 98, 293-304 
. Tropomyosin coiled-coil interactions: Evidence for an unstaggered structure 
McLachlan, A.D. and Stewart, M.  (1976) J. Mol. BioI., 103, 271-298 
The 14 {old periodicity in a-tropomyosin and the interaction with actin 
McLachlan, A.D. and Stewart, M. ( 1982) J. Mol. BioI., 162, 693-698 
Periodic charge distribution in the intermediate filament proteins desmin and 
vimentin 
MeLachlan, A.D. and Kahn, J. ( 1983) J. Mol. BioI., 164, 605-626 
Periodic features in the amino acid sequence of nematode myosin rod 
Milam, L. and Erickson, H.P. (1982) J. Cell BioI., 94, 592-596 
Visualization of a 21-nm axial periodicity in shadowed keratin filaments and 
nelUofilaments 
Myers, M.W., Lazzarini, R.A., Lee, V.M.-Y., Schlaepfer, W.W. and Nelson, D.L. 
( 1987) EMBO J., 6, 1617- 1626 
The human mid-size nelUofilament subunit: A repeated protein sequence and the 
relationship of its gene to the intermediate filament gene family 
Napolitano, E.W., Chin, S .S.M., Colman, D.R. and Liem, R.K.H. ( 1987) J. 
Neurosci., 7, 2590-2599 
Complete amino acid sequence and in vitro expression ofrar NF-M, the middle 
molecular weight neurofilament protein 
O'Donnell, (1969) Aust. 1. BioI. Sei., 22, 47 1 -488 
Studies on reduced wool: IX. The N-terminal sequence of a fragment produced by 
cleavage of component 8 with cyanogen bromide 
BmuOGRAPHY page 177 
Osbom, M. and Weber, K. ( 1982) Cell, 31, 303-306 
Intermediatejilaments: Cell-type-specijic markers in differentiation and patJwlogy 
Osbom, M. and Weber, K. (1983) Lab. Invest. , 48, 372-394 
Biology of disease. Tumour diagnosis by intermediate filament typing: A novel 
tool for surgical pathology 
Osbom, M. and Weber, K. (1986) Trends Biochem. Sci., 11, 469-472 
Intermediatejilament proteins: A multigenefamily distinguishing major cell 
lineages 
Ottaviano, Y. and Gerace, L. ( 1985) 1. BioI. Chem., 260, 624-632 
PhospJwrylation of the nuclear lamins during interphase and mitosis 
Parry, D.A.D. ( 1974) Biochem. Biophys. Res. Commun., 57, 216-224 
Structural studies on the tropomyosinJtroponin complex ofvenebrate skeletal 
muscle 
Parry, D.A.D. ( 1975) 1.Mol.Biol. ,  98, 519-535 
Analysis of the primary sequence of a-tropomyosin from rabbit skeletal muscle 
Parry, D.A.D. ( 1978) 1.Mol.Biol . ,  120, 545-55 1 
Fibrinogen: A preliminary analysis of the amino acid sequences of the portions of 
the a, f3 and "{-chain postulated to form the interdomainal link between globular 
regions of the molecule 
Parry, D.A.D. (1979) in "Fibrous Proteins: Scientific, Industrial and Medical 
Aspects", eds D.A.D. Parry and L.K.Creamer, Academic Press, London, 1,  
393-427 
Determination of structural information from the amino acid sequences of jibrous 
proteins 
Parry, D.A.D. ( 198 1) 1.Mol.Biol. ,  153, 459-464 
Structure of rabbit skeletal myosin: Analysis of the amino acid sequences of two 
fragments from the rod region 
Parry, D.A.D. (1982) Biosci. Rep., 2, 1017- 1024 
Coiled-coils in a-helix-containing proteins: Analysis of the residue types within the 
heptad repeat and the use of these data in the prediction of coiled-coils in other 
proteins 
BmUOORAPHY page 178 
Parry, D.A.D. (1989) in "Cellular and Molecular Biology of Intermediate Filaments", 
eds R.D. Goldman and P.M. Steinert, Plenum Press (in press) 
Primary and secondary structure of IF protein chains and modes of molecular 
aggregation 
Parry, D.A.D. and Suzuki, E. ( 1969) Biopolymers, 7, 199-206 
Interchain packing energies of multistranded a-helical ropes 
Parry, D.A.D. and Baker, E.N. ( 1984) Rep. Progr. Phys. ,  47, 1 133- 1232 
Biopolymers 
Parry, D.A.D. and Fraser, R.D.B. (1985) Int J. BioI. MacromoI. ,  7, 203-21 3  
Intermediatefilament structure: 1 .  Analysis of IF protein sequence data 
Parry, D.A.D., Crewther, W.G., Fraser, R.D.B. and MacRae, T.P. ( 1977) J. Mol. 
Biol., 113,  449-454 
Structure of a-keratin: Structural implication of the amino acid sequences of the 
type I and type 11 chain segments 
Parry, D.A.D., Steven, A.C. and Steinert, P.M. (1985) Biochem. Biophys. Res. 
Commun. ,  127, 1012- 1018  
The coiled-coil molecules of intermediate filaments consist of two parallel chains in 
exact axial register 
Parry, D.A.D.,  Conway, J.F. and Steinert, P.M. ( 1986) Biochem. J., 238, 305-308 
Structural studies on lamin: Similarities and differences between lamin and 
intermediate-filament proteins 
Parry, D.A.D., Conway, J.F., Goldman, A.E., Goldman, R.D. and Steinen, P.M. 
( 1987a) 1nl. J. BioI. MacromoI., 9, 1 37-145 
Nuclear lam in proteins: Common structures for paracrystalline,filamenrous and 
lattice forms 
Parry, D.A.D., Fraser, R.D.B., MacRae, T.P. and Suzuki, E. ( 1987b) in "Fibrous 
Protein Structure", eds J.M. Squire and PJ. Vibert, Academic Press, 193-21 4  
Intermediate filaments 
Parysek, L.M. and Goldman, R.D. (1987) J. Neurosci., 7, 78 1 -79 1 
Characterization of intermediate filaments in PCl2 cells 
BmuOGRAPHY page 179 
Parysek, L.M., Chisholm, R.L., Ley, C.A. and Goldman, R.D. (1988) Neuron, 1 ,  
395-401 
A type III intermediate filament gene is encoded in mature neurons 
Pauling, L. and Corey, R.B. (1950) I. Amer. Chem. Soc, 72, 5349 
Two hydrogen-bonded spiral configurations of the polypeptide chain 
Pauling, L. and Corey, R.B. (1951a) Proc. Nat!. Acad. Sci. USA, 37, 235-240 
Atomic coordinates and structure factors for two helical configurations of 
polypeptide chains 
Pauling, L. and Corey, R.B. (195 1b) Proc. Nat!. Acad. Sci. USA, 37, 729-740 
Configurations of polypeptide chains with favoured orientations around sing le 
bonds: Two new pleated sheets 
Pau1ing, L. and Corey, R.B. (1953a) Nature (London), 171, 59-61 
Compound helical configurations of polypeptide chains: Structure of proteins of 
the a-keratin type 
Pauling, L. and Corey, R.B. (1953b) Proc. Nat!. Acad. Sci. USA, 39, 253-256 
Two rippled-sheet configurations of polypeptide chains, and a note about the 
pleated sheets 
Pauling, L., Corey, R.B. and Branson, H.R. ( 1951 )  Proc. Nat!. Acad. Sci. USA, 
37, 205-21 1  
The structure of proteins: Two hydrogen-bonded helical configurations of the 
polypeptide chain 
Perutz, M.F. ( 195 1a) Nature (London), 167, 1053-1054 
New x-ray evidence on the configuration of polypeptide chains: Polypeptide chains 
in poly-r-benzyl-L-glutamate, keratin and hamoglobin 
Perutz, M.F. (1951b) Nature (London), 168, 653-654 
The 15-.4 reflection from proteins and polypeptides 
Peter, M., Kitten, G.T., Lehner, c.F., Vorburger, K., Bailer, S.M., Maridor, G. and 
Nigg, E.A. (1989) J. Mol. BioI., 208, 393-404 
Cloning and sequencing of cDNA clones encoding chicken lamins A and Bl and 
comparison of the primary structures of vertebrate A- and B-type lamins 
Phillips, G.N., Fillers, J.P. and Cohen, C. (1986) I. Mol. BioI., 192, 1 1 1 - 1 3 1  
Tropomyosin crystal structure and muscle regulation 
BmuOGRAPHY page 180 
Portier, M.-M., Brachet, P., Croizat, B. and Gros, F. ( l984a) Dev. Neurosci., 6, 
215-226 
Regulation ojperipherin in mouse neuroblastoma and rat pe12 pheochromo­
cytoma cell lines 
Portier, Mo.-M., de Nechaud, B. and Gros, F. ( 1984b) Dev. Neurosci. ,  6, 335-344 
Peripherin, a new member ojthe intermediate filament proteinjamily 
Quax, W., Egberts, W.V., Hendricks, W., Quax-Jeuken, Y. and Bloemendal, H. 
(1983) Cell, 35, 215-223 
The structure oj the vimentin gene 
Quax, W., van der Heuvel, R., Egbens, W.V., Quax-Jeuken, Y. and Bloemendal, H. 
( 1984) Proc. Nat!. Acad. Sci. USA, 81, 5970-5974 
Intermediatefilament cDNAsfromBHK-21 cells: Demonstration of distinct genes 
for desmin and vimentin in all vertebrate classes 
Quax, W., van den Broek, L., Egbens, W.V.,  Ramaekers, F. and Bloemendal, H. 
(1985) Cell, 43, 327-338 
Characterization of the hamster desmin gene: Expression andformation of desmin 
filaments in nonmuscle cells after gene transfer 
Quax-Jeuken, Y.E.F.M., Quax, W.J., and Bloemendal, H. (1983) Proc. Nat!. Acad. 
Sci. USA, 80, 3548-3552 
Primary and secondary structure of hamster vimentin predictedfrom the nucleotide 
sequence 
Quinlan, R.A. and Franke, W.W. ( 1982) Proc. Nat!. Acad. Sci. USA, 79, 
3452-3456 
Heteropolymer filaments ofvimentin and desmin in vascular smooth muscle tissue 
and cultured baby hamster kidney cells demonstrated by chemical crosslinking 
Quinlan, R.A. and Franke, W.W. ( 1983) Europ. J. Biochem., 132, 477-484 
Molecular interactions in intermediate-sizedfilaments revealed by chemical 
crosslinking: Heteropolymers of vi men tin and glial filament protein in cultured 
human glioma cells 
Quinlan, R.A., Cohlberg, I.A., Schiller, D.L., Hatzfeld, M. and Franke, W.W. 
( 1984) J. Mol. Bio!., 178, 365-388 
Heterotypic tetramer (A2D2) complexes of non-epidermal keratins isolated from 
cytoskeletons of rat hepatocytes and hepatoma cells 
BmUOORAPHY 
Quinlan, R.A., Moir, R.D. and Stewart, M. ( 1989) J. Cell Sci., 93, 7 1-83 
Expression in Escherichia coli offragments of glialfibrillary acidic protein: 
characterization, assembly properties and paracrystal formation 
page 181 
Raychaudury, A., Marchuk, D., Lindhurst, M. and Fuchs, E. ( 1986) Mol. Cell. 
BioI., 6, 539-548 
Three tightly linked genes encoding human type I keratins: Conservation of 
sequence in the 5' -unrranslated leader and the 5' -upstream regions of coexpressed 
keratin genes 
Rieger, M., Jorcano, J.L. and Ftanke, W.W. ( 1985) EMBO J., 4 ,  2261-2267 
Complete sequence of a bovine type I cytokeratin gene: conserved and variable 
intron positions in genes of poly pep rides of the same cytokerarinfamily 
Rogers, G.E. ( 1959) J. Ultrastruct. Res., 2, 309-330 
Electron microscopy of wool 
Rogers, G.E. ( 1984) Biochem. Soc. Symp., 49, 85-108 
Studies on keratin multigene families 
Rose, G.D. ( 1978) Nature (London), 272, 586-590 
Prediction of chain turns in globular proteins on a hydrophobic basis 
Sauk, J.J., Krunweide, M., Cocking-Johnson, D. and White, J.G. ( 1984) J.  Cell 
Biol., 99, 1590- 1597 
Reconstitution of cytokeratinfilaments in vitro: Further evidencefor the role of 
nonhelical peprides in filament assembly 
Scheer, U., Kartenbeck, J., Trendelenburg, M.F., Stadler, J. and Franke, W.W. 
( 1976) J. Cell BioI., 69, 1 - 1 8  
Experimental disintegration of the nuclear envelope: Evidence for pore-connecting 
fibrils 
Shelton, K.R., Higgins, L.L., Cochran, D.L., Ruffolo, 1.J. and Egle, P.M. ( 1980) 
J. BioI. Chem., 255, 10978- 10983 
Nuclear lamins of erythrocyte and liver 
Shneidman, P.S., Carden, M.J., Lees, J.P. and Lazzarini, R.A. ( 1988) Molec. Brain 
Res., 4, 217-232 
The structure of the largest murine neurofilament protein (NF -H) as revealed by 
cDNA and genomic sequences 
BmuOGRAPHY page 182 
Singer, P.A., Trevor, K. and Oshima, R.G. ( 1986) J. BioI. Chem., 261, 538-547 
Molecular cloning and characterization of the Entia B cyrokeratin expressed in 
preimplantation mouse embryos 
Skerrow, D., Matoltsy, A.G. and Matoltsy, M.N. (1973) J. BioI. Chem., 248, 
4820-4826 
Isolation and characterization of the a-helical regions of epidennaI prekeratin 
Sparrow, L.G. and Inglis, A.S. (1980) Proc. 6th Int. Wool Text. Res. Conf. 
Pretoria, 2, 237-246 
Characterization of the cyanogen bromide peptides of component 7 c, a major 
microfibrillar protein from wool 
Sparrow, L.G., Dowling, L.M., Loke, V.Y. and Strike, P.M. ( 1989) in "The 
Biology of Wool and Hair", eds G.E. Rogers, P.J. Reis, K.A. Ward and R.C. 
Marshall, Chapman and Hall, London and New York, 145- 155 
Amino acid sequences of wool keratin IF proteins 
Steinert, P.M. ( 1978) 1. �ol. BioI., 123, 49-70 
Structure of the three-chain unit of the bovine epidermal keratin filament 
Steinert, P .M. (198 1)  in "Microscopy of Proteins", ed J .R. Harris, Academic Press, 
London, 1, 125- 166 
Intermediate filaments 
Steinert, P.M. ( 1988) J. BioI. Chem., 263, 1 3333- 13339 
The dynamic phosphorylation of the human intermediate filament keratin 1 chain 
Steinert, P.M. and Parry, D.A.D. ( 1985) Ann. Rev. Cell BioI., 1, 4 1-65 
Intermediate filaments: Conformity and diversity of expression and structure 
Steinert, P.M. and Roop, D.R. ( 1988) Ann. Rev. Biochem.,  57, 593-625 
Molecular and cellular biology of intermediate filaments 
Steinert, P.M., Zimmerman, S .B., Starger, J.M. and Goldman, R.D. ( 197 8) Proc. 
Natl. Acad. Sci. USA, 75, 6098-6101 
Ten-nanometer filaments of hamster BHK-21 cells and epidermal keratin filaments 
have similar structures 
BmUOORAPHY page 183 
Steinen, P.M., Idler, W.W. and Goldman, R.D. ( 1980a) Proc. Natl. Acad. Sci. 
USA, 77, 4534-4538 
Intermediatefilaments ofbaby hamster kidney (BHK-21) cells and bovine 
epidermal keratinocytes have similar ultrastructW"es and subunit domain structures 
Steinen, P.M., Starger, I.M. and Goldman, R.D. ( 1980b) in ''Fibrous Proteins: 
Scientific, Industrial and Medical Aspects", eds D.A.D. Parry and L.K.Creamer, 
Academic Press, London, 2, 227-236 
Homologous structures of 100 A filaments and mammalian keratin filaments 
Steinen, P.M., Peck, G.L. and Idler, W.W. (198Oc) in "Biochemistry of Nonnal and 
Abnonnal Epidennal Differentiation", eds. I.A. Bernstein and M. Seiji, University 
of Tokyo Press, Tokyo, 39 1 -406 
StructW"al changes of hwnan epidermal a-keratin in disorders of keratinization 
Steinen, P.M., Idler, W.W., Aynardi-Whitrnan, M., Zackroff, R.V. and Goldman, 
R.D. ( 1 982) Cold Spring Harbor Symp. Quant. BioI., 46, 465-474 
Heterogeneity of intermediate filaments assembled in vitro 
Steinen, P.M.,  Rice, R.H., Roop, D.R., Trus, B.L. and Steven, A.C. ( 1983a) 
Nature (London), 302, 794-800 
Complete amino acid sequence of a mouse epidermal keratin subunit and 
implications for the structW"e of intermediate filaments 
Steinert, P.M.,  Steven, A.C. and Roop, D.R. (1983b) J. Invest. Dermatology, 81 ,  
1 - 14  
Strucnualfeanues of epidermal keratin filaments reassembled in vitro 
Steinert, P.M., Parry, D.A.D., Racoosin, E.L., Idler, W.W., Steven, A.C., Trus, 
B.L. and Roop, D.R. ( 1984a) Proc. Natl. Acad. Sci. USA, 81, 5709-57 1 3  
The complete cDNA and deduced amino acid sequence of a type II mouse 
epidermal keratin of 60,()()0 Da: Analysis of sequence differences between type I 
and type Il keratins 
Steinert, P.M., Jones, J.C.R. and Goldman, R.D. ( 1984b) J. Cell BioI. ,  99, 22s-27s 
Intermediate filaments 
BmuOGRAPHY page 184 
Steinert, P.M., Parry, D.A.D., Idler, W.W., lohnson, L.D., Steven, A.C. and Roop, 
D.R. ( 1985a) J. BioI. Chem., 260, 7 142-7 149 
Amino acid sequences of mouse and human epidermal type II keratins of Mr 
67,000 provide a systematic basis for the structural andfunctional diversity of the 
end domains of keratin intermediate filament subunits 
Steinert, P.M., Steven, A.C. and Roop, D.R. ( 1985b) Cell, 42, 41 1 -419 
The molecular biology of intermediate filaments 
Steinert, P.M., Idler, W.W., Zhou, X.-M., Johnson, L.D., Parry, D.A.D., Steven, 
A.C. and Roop, D.R. ( 1985c) Ann. N. Y. Acad. Sci., 455, 451-461 
Structural and functional implications of amino acid sequences of keratin 
intermediate filament subunits 
Steinert, P.M., Torchia, D.A. and Mack, lW. ( 1989) in "The Biology of Wool and 
Hair" , eds G.E. Rogers, P.J. Reis, K.A. Ward and R.C. Marshall, Chapman and 
Hall, London and New York, 1 57- 167 
Structural features of keratin intermediate filaments 
Steven, A.C. (1989) in "Cellular and Molecular Biology of Intermediate Filaments", 
eds R.D. Goldman and P.M. Steinert, Plenum Press (in press) 
Intermediatefilament structure: Diversity, polymorphism, and analogy to myosin 
Steven, A.C., Wall, J., Hainfeld, J. and Steinert, P.M. (1982) Proc. Natl. Acad. Sci. 
USA, 79., 3101 -3 105 
Structure offibroblastic intermediate filaments: Analysis by scanning transmission 
electron microscopy 
Steven, A.C., Hainfeld, J.F., Trus, B .L., Wall, J.S. and Steinert, P.M. ( 1983a) J. 
BioI. Chem., 258, 8323-8329 
The distribution of mass in heteropolymer intermediate filaments assembled 
in vitro 
Steven, A.C., Hainfeld, I.F., Trus, B.L., Wall, 1.S . and Steinert, P.M. ( 1983b) J. 
Cell BioI., 97, 1939-1944 
Epidermal keratin filaments assembled in vitro have masses-per-unit-Iength that 
scale according to average subunit mass: Structural basisfor homologous packing 
of subunits in intermediate filaments 
BmuOGRAPHY page 185 
Steven, A.C., Trus, B.L., Hainfeld, J.P. Wall, I.S. and Steinert, P.M. ( 1985) Ann. 
N.Y. Acad. ScL, 455, 371 -380 
Conformity and diversity in the structures of intermediate filaments 
Steven, A.C. Trus, B.L., Maizel, J.V., Unser, M., Parry, D.A.D., Wall, J.S ., 
Hainfeld, I.F. and Studier, P.W. (1988) J. Mol. BioI., 200, 351 -365 
Molecular substructure of a viral receptor-recognition protein. The GP 17 tail fibre 
of bacteriophage 17 
Steven, A.C., Mack, lW., Trus, B.L., Bisher, M.E. and Steinen, P.M. ( 1989) in 
"Cytoskeletal and Extracellular Proteins: Structure, Interactions and Assembly", 
eels U. Aebi and J. Engel, Springer Series in Biophysics, Springer-Verlag, Berlin, 
3, 1 5-26 
StructW'e and assembly of intermediate filaments: Multi-Jaceted, myosin-like (but 
non-motile) cytoskeletal polymers 
Stewart, M., Quinlan, R.A., Moir, R.D., Clarke, S.R. and Atkinson, SJ. ( 1989a) in 
"Cytoskeletal and Extracellular Proteins: Structure, Interactions and Assembly", 
eds U. Aebi and I. Engel, Springer Series in Biophysics, Springer-Verlag, Berlin, 
3, 1 50- 159 
The role of repeating sequence motifs in interactions between a-helical coiled-coils 
such as myosin, tropomyosin and intermediate-filament proteins 
Stewart, M., Quinlan, R.A. and Moir, R.D. ( 1989b) J. Cell BioI., 109, 225-234 
Molecular interactions in paracrystals of a fragment corresponding to the a-helical 
coiled-coil rod portion of glial fibrillary acidic protein: Evidence for an anti parallel 
packing of the molecules and polymorphism related to intermediate structure 
Stick, R., Angres, B., Lehner, c.P. and Nigg, E.A. ( 1988) 1. Cell BioI., 107, 
397-406 
The fates of chicken nuclear lamin proteins during mitosis: Evidence for a 
reversible redistribution of lamin B2 protein between inner nuclear membrane and 
elements of the endoplasmic reticulum 
Stone, D., Sodek, J., Iohnson, P. and Smillie, L.B. ( 1974) Proc. 9th FEBS Meeting 
, (Budapest), 31, 125- 136 
Tropomyosin: Correlation of amino acid sequence and structure 
BmUOORAPHY page 186 
Tainer, J.A., Getzoff, E.D., Alexander, H., Houghten, R.A., Olson, A.J., Lemer, 
R.A. and Hendrickson, W.A. ( 1984) Nature (London), 312, 1 27- 1 34 
The reactivity of anti-peptide antibodies is a function of the atomic mobility of sites 
in a protein 
Thompson, M.A. and Ziff, E.B. (1989) Neuron, 2, 1043-1053 
Structwe of the gene encoding peripherin, an NGF-regulated new-onal-specific 
type III intermediate filament protein 
Torchia, D.A. and van der Hart, D.L. ( 1976) J. Mol. BioI., 104, 3 1 5-321 
Carbon-13 magnetic resonance evidence for anisotropic molecular motion in 
collagen fibrils 
Traub, P. and Vorgias, C.E. ( 1983) J. Cell Sci., 63, 43-67 
Involvement of the N-terminal polypeptide ofvimentin in theformation of 
intermediate filaments 
Trachtenberg, S.  (1987) Biochem. Biophys. Acta, 923, 327-332 
Assemblies of psoriatic keratin and their relation to normal intermediate filament 
structure 
Trachtenberg, S . ,  Steinert, P.M., Trus, B.L. and Steven, A.C. ( 1986) in "Electron 
Microscopy and Alzheimer's Disease", ed J. Metuzals, San Francisco Press, 
56-59 
Electron microscopy and image analysis of paracrystalline filament bundles 
obtained from psoriatic epidermal keratin 
Trinick, J., Cooper, J., Seymour, J. and Egelman, E.H. (1986) J. Microsc. (Oxford), 
141 ,  349-360 
Cryo-electron microscopy and three-dimensional reconstruction of actin filaments 
Troncoso, J.C., Haner, M., March, J.L., Reichelt, R., Engle, A. and Aebi, U. 
( 1989) in "Cytoskeletal and Extracellular Proteins: Structure, Interactions and 
Assembly", eds U. Aebi and J. Engel, Springer Series in Biophysics, Springer­
Verlag, Berlin, 3, 33-38 
Structure and assembly of neurofilaments (NF): Analysis of native NF and of 
speCific NF subunit combinations 
Tyner, A.L. and Fuchs, E. ( 1986) J. Cell BioI., 103, 1945- 1955 
Evidence for posttranscriptional regulation of the keratins expressed during 
hyperproli/eration and malignant transformation in human epidermis 
BmUOORAPHY page 187 
Tyner, A.L., Eichman, M.I. and Fuchs, E. ( 1985) Proc. Natl. Acad. Sci. USA, 82, 
4683-4687 
The sequence of a type II keratin gene expressed in Juunan skin: Conservation of 
structure among all intermediate filament genes 
Venetianer, A., Schiller, D.L., Magin, T. and Franke W.W. ( 1983) Nature (London), 
305, 730-733 
Cessation of cytokeratin expression in a rat hepatoma cell line lacking differentiated 
functions 
Weber, K. ( 1986) Nature (London), 320, 402 
Link between lamins and intermediate filaments 
Weber, K. and Geisler, N. (1982) EMBO J., 1, 1 155- 1 160 
The structural relation between intennediate filament proteins in living cells and the 
a-keratins of sheep wool 
Weber, K. and Geisler, N. ( 1984) in "Cancer Cells I ,  The Transformed Phenotype", 
eds A.J. Levine, G.F. Vande Woude, W.C. Topp and J.D. Watson, Cold Spring 
Harbour, New York, 1, 153-159 
Intermediatefilaments -from wool a-keratins to neurofilaments: A structural 
overview 
Weber, K. Plessmann, U., Dodemont, H. and Kossmagk-Stephan, K. ( 1988) EMBO 
J., 7, 2995-3001 
Amino acid sequences and homopolymer-fonning ability of the intennediate 
filament proteins from an invertebrate epithelium 
Westhof, E., Altschuh, D., Moras, D., Bloomer, A.C., Mondragon, A.,  Klug, A. 
and Van Regenmortel, M.H.V. ( 1984) Nature (London), 311,  123- 126 
Correlation between segmental mobility and the location of antigenic determinants 
in proteins 
Winkles, J.A., Sargent, T.D., Parry, D.A.D., Jonas, E. and Dawid, I.B . ( 1985) Mol. 
Cell. BioI. ,  5, 2575-2581 
Developmentally regulated cytokeratin gene in Xenopus laevis 
Wolin, S .L., Krohne, G. and Kirschner, M.W. ( 1987) EMBO J., 6, 3809-38 1 8  
A new lamin in Xenopus somatic tissue displays strong homology to human 
lamin A 
BmUOORAPHY page 188 
Woods, E.F. ( 1983) Biochem. Int., 7, 769-774 
Structural studies on the microfibrillar proteins of wool: Characterization of the 
a-helix-rich particle produced by chymotryptic digestion 
Woods, E.F. and Gruen, L.C. (1981)  Aust. J. Bioi. Sei., 34, 5 15-526 
The number of polypeptide chains in the rod domain of bovine epidermal keratin 
Woods, E.F. and Inglis, A.S. (1984) Int. J. Bioi. MaeromoI., 6, 277-283 
Organization of the coiled-coils in the wool microfibril 
Worman, H.J., Yuan, 1., Blobei. G. and Georgatos, S.D. (1988) Proc. Natl. Aead. 
Sci. USA, 85, 853 1-8534 
A lamin B receptor in the nuclear envelope 
Zackroff, R.V., Goldman, A.E., Jones, J.C.R., Steinert, P.M. and Goldman, R.D. 
( 1984) J. Cell BioI., 98, 123 1 - 1237 
Isolation and characterization of keratin-like proteins from cultured cells with 
fibroblastic morphology 
Zehner, Z.E. and Paterson, B.M. ( 1985) Ann. N.Y. Aead. Sei., 455, 79-94 
The chicken vimentin gene: Aspects of organization and transcription during 
myogenesis 
Zhou, X.-M., Idler, W.W., Steven, A.C., Roop, D.R. and Steinert, P.M. ( 1988) 1. 
BioI. Chem., 263, 15584- 15589 
The complete sequence of the human intennediate filament chain keratin 10: 
Subdomainal divisions and model for folding of end domain sequences 
PUBUCATIONS page 189 
PUBLICATIONS 
The research detailed in  the following publications fonns much of the basis for this 
thesis. Papers included (in chronological order) are: 
1)  Parry, D.A.D., Conway� I.F. and Steinert, P.M. (1986) Biochem. J., 238, 
305-308 
Structwal studies on lamin: Similarities and differences between lamin and 
intermediate-jilamenJ proteins 
2) Parry, D.A.D., Conway, J.F., Goldman, A.E., Goldman, R.D. and Steinert, 
P.M. ( 1987a) Int. J. BioI. MacromoI., 9, 1 37-145 
Nuclear lamin proteins: Common structures for paracrystalline,jilamentous and 
lattice forms 
3) Conway, J.F. and Parry, D.A.D. (1988) lnt. J. BioI. MacromoI., 10, 79-98 
InJermediatefilamenJ structwe: 3. Analysis of sequence homologies 
4) Conway, J.F. and Parry, D.A.D. ( 1989) in "Cytoskeletal and Extracellular 
Proteins: Structure, Interactions and Assembly", eds U. Aebi and J. Engel, 
Springer Series in Biophysics, Springer-Vedag, Berlin, 3, 140- 149 
Structwal and spatial organization of intermediate filament and nuclear lamin 
proteins 
5) Conway, J.F., Fraser, R.D.B., MacRae, T.P. and Parry, D.A.D. ( 1989) in "The 
Biology of Wool and Hair", eds G.E. Rogers, P.J. Reis, K.A. Ward and R.e. 
Marshall, Chapman and Hall, London and New York, 127-144 
Protein chains in wool and epidermal keratin IF: Structural featwes and spatial 
arrangement