The gut microbiome of honey bees (Apis mellifera) is relatively simple, with 9 host-adapted bacterial phylotypes. The simplicity of this system and its similarity to mammal gut microbiomes makes it a tractable model to study host-associated microbial communities. Two bacterial species play a key role in bee’s digestion and health: Gilliamella apicola, a fermenter feeding on carbohydrates (glucose, fructose) and Snodgrassella alvi, micro-aerobe feeding on carboxylates (acetate). This remarkable niche partitioning is complemented by putative cross-feeding interactions: G. apicola releases acetate necessary for S. alvi growth, whereas S. alvi excretes amino-acids needed by G. apicola. The two species, which have evolved together since the origin of the pollen-basket bee family 80 million years ago, form a two-layer biofilm. The exact roles of these bacteria within the bees and the mechanisms maintaining the robustness of cross-feeding interactions are poorly known. Additionally, these species exhibit an extensive strain diversification that is not explained.
Here we develop a novel mathematical model of a bee gut to describe bacterial cross-feeding interactions in realistic conditions. We compare the results to in vivo experimental data and investigate the stability and robustness of the system. We then use genome-scale metabolic models to unravel the strain diversification mechanisms.
The model, calibrated with experimental data, correctly predicts the concentrations of carbohydrates and acetate, and the ratio of the two bacterial species at equilibrium.
Acetate is a byproduct of fermentation and is absorbed by the host but is toxic to the bee above a certain threshold. The model shows that, counterintuitively, if S. alvi is present, acetate concentration at equilibrium is determined only by S. alvi’s ability to metabolize it (its R* for acetate), independent of the S. alvi or G. apicola’s abundance. The main role of G. apicola may be to transform carbohydrates into acetate to promote S. alvi growth, especially during the first stage of gut colonization.
Depending on the bee feeding behavior, the carbohydrate input can vary drastically in time, but the effect of resource fluctuations may be buffered by the low bacterial mortality rate related to the biofilm lifestyle. Additionally, the robustness of the system is insured by the reciprocal control of the two bacteria, mediated by acetate and amino-acids, which renders it hard to invade and prevents cheating.
The flux balance metabolic models show that strain diversification is a dynamical process, probably driven by horizontal gene transfer, and strongly depends on the nutrient environment.