Fig. 2. Enterotoxigenic B. fragilis (ETBF) is necessary but not sufficient to produce weight loss in recipient gnotobiotic mice.

(A) Gut microbial community composition, defined by COPRO-Seq, in mice colonized with either of the two unmanipulated culture collections or the derived manipulated versions. Mean values for relative abundances ± s.e.m. are plotted using aggregate data generated from fecal samples collected from mice colonized with a given community. Taxa present at abundances lower than 1% are not represented in the stacked barplots. (B) The proportional representation of Bacteroides taxa in unmanipulated culture collections installed in gnotobiotic mice do not differ significantly as a function of the diet embodiments animals were fed. Mean values ± s.e.m. for data generated from feces are shown (n=5–6/group; one-way ANOVA). (C) Schematic illustrating the different groups of gnotobiotic mice generated by manipulating the presence/absence of ETBF and NTBF within the stunted/underweight or healthy donors’ culture collections, and the questions addressed by the indicated comparisons. (D) Removal of ETBF prevents weight loss in mice colonized with the stunted/underweight donor’s culture collection. In contrast, addition of ETBF with the simultaneous removal of NTBF does not significantly impact weight in mice colonized with the culture collection derived from the healthy child (n=5–6 mice/treatment group). Mean values ± s.e.m. are plotted. *, p<0.05 (paired two-tailed Student’s t-test and linear mixed effects model as above). (E) Addition of NTBF to the stunted/underweight donor’s culture collection ameliorates ETBF-associated weight loss in gnotobiotic mice fed embodiment 2 of a representative Bangladeshi diet. (n=6 mice/treatment group). Mean values ± s.e.m. are plotted. *, p<0.05 (paired two-tailed Student’s t-test and linear mixed effects model as above).