Mathematical modelling of microbial cross-feeding on hydrogen in the human colon: a thesis presented in partial fulfilment of the requirements for the degree of Doctor of Philosophy in Nutritional Science at Massey University, Palmerston North, New Zealand

Abstract

The microbial population of the human colon contains three subgroups that cross-feed on hydrogen produced during microbial metabolism of carbohydrates: methanogens, sulphate-reducing bacteria (SRB) and reductive acetogens. These microbes and their activities have been linked to various host physiological and health outcomes. This thesis aimed to construct mathematical models for the growth and metabolism of colonic hydrogenotrophs to investigate key factors in hydrogenotroph metabolism and population dynamics that would be difficult to study experimentally. Monoculture models based on Monod kinetics were developed for Methanobrevibacter smithii, Desulfovibrio vulgaris, and Blautia hydrogenotrophica, as representatives of colonic methanogens, SRB and reductive acetogens. The models were parameterised and validated using experimental data. The monoculture models were combined to examine interactions between these microbes, before incorporation into an existing microbial community model, microPop. Adaptations were made to microPop to enable simulation of the colonic environment, investigating the role of hydrogenotrophs in the colon. The D. vulgaris model provided similarly accurate predictions to an existing thermodynamics-based model. The B. hydrogenotrophica model estimated a hydrogen uptake threshold of 86 mM and provided supportive evidence for the confounding effect of growth media on reductive acetogenesis. Growth yield parameters for SRB and methanogenic strains were reduced in co-culture compared to monoculture, while tri-culture modelling identified conditions necessary for the survival of each hydrogenotroph. Substrate competition prevented survival of all three together in continuous culture. The community model predicted colonic pH, short chain fatty acid gradient and dominant microbial groups but could not accurately predict other experimental metabolite and microbial abundance measurements. Investigating the role of colonic sulphate availability showed contrasting predictions: sulphate availability positively correlated with SRB and sulphide concentrations and negatively correlated with methanogen abundance using a continuous representation of the colon, but no effect was predicted using a compartmental representation. This research demonstrates that modelling can extract additional information from existing experimental data. The community model provides a basis for the computational study of the microbiota and hydrogen cross-feeding dynamics in the colon, which can complement or even accelerate experimental research on the influences of the microbiome on the host.