After 5 days of DSS treatment, significant weight loss, increased DAI and reduced colon length indicated IBD mouse model was successfully established. Compared with d 5-model, the increased colon length and decreased DAI indicated the transition from progression phase to recovery phase in d 10-model.
Further, we screened response of colon bacterial flora to the DSS-induced inflammation changes. As rodents are coprophagic, gut microbiota of mice in the same cage will progressively become homogeneous over time. In order to avoid this, mice were randomly selected from multiple cages each time for sample collection. Completely different gut microbiota of IBD mouse model among d 0, d 5 and d 10 indicated gut microbiota alteration were accompanied by colitis initiation and recovery (Fig.3). Generally, commensal bacteria digested dietary fiber and produce short-chain fatty acids (SCFAs) therefore influencing gut epithelial cell and lymphocyte homeostasis and eventually contributed to the host intestine health. Germ-free mice have a higher mortality rate than wild-type mice when treated with DSS [25, 26], highlighting the important role of the flora in maintaining gut health.
However, the specific function of changed microbiota induced by inflammation remains unclear. To clarify this uncertainty, we transplant IBD mice colon bacterial flora to Abx-mice. As a result, FMT-mice showed symptoms similar to those of IBD mouse model while vehicle-mice remained healthy. This result implied that gut bacterial flora in IBD mouse shifts from normal commensal flora to harmful pathogenicity status and potentially elucidated the possible reason underlying the relapse of IBD.
We also analyzed the difference of microbiota composition between different phase of IBD mouse model and normal control. Both weighted and un-weighted unifrac PCoA separated these 3 group well (Fig.3), indicating that the detected community differences were not due to the presence and/or absence of rarer phylotypes. Also, the decrease of community richness was observed at both progressive phase and recovery phase (Fig.4, S1) but community diversity did not change significantly (Fig.4), suggesting the losses of particular taxa while some “new” taxa rise.
Human colonic microbiota is dominated by anaerobic members within phyla Bacteroidetes and Firmicutes [27]. The situation is the same for IBD mice, but the overall abundances of these two phyla and the taxa within them showed clear changes compared with normal control. The negative relationship between phylum Firmicutes and Bacteroidetes became weaker in d 5 and d 10-mouse model which may be an indicator of disordered microbiota flora in IBD mouse.
To be specific, there was a significant increase of phylum Firmicutes and a significant decrease of phylum Bacteroidetes in d10-model compared with normal control and d 5-model. Within phylum Firmicutes, Lactobacillus spp. and Bifidobacterium spp. have been widely used as probiotics. Unlike the low abundance of Bifidobacterium, the average percentage of Lactobacillus could reach up to 23.1%. In fact, this bacteria specie was the primary bacteria driving the calculated multivariate discrimination in the high level of phylum Firmicutes in d 10-model (Fig.S2). This result also indicated the abundance of Lactobacillus is negatively correlated with IBD pathogenesis. If this hypothesis is correct, the supplementation of Lactobacillus in IBD animals and patients might confer beneficial effects. Previous studies reported similar beneficial effects for protecting gut health of Lactobacillus. First, Lactobacillus could increase intestinal anti-inflammatory cytokine level and/or reduce the proinflammatory cytokines, thus altering the immune system [28-30]; it is worth noting that Lactobacillus could suppress TNF-α expression [31], which may explain the declined TNF-α in IBD mouse model on d 10 (Fig.2). Second, Lactobacillus directly competed with pathogenic bacteria therefore contributing to gut health [32, 33]. Though whether the change of Lactobacillus is the cause or consequence of the disease pathogenesis remains unclear, this understanding of the ecophysiology of the commensal and probiotic Lactobacillus in the dysbiosis disease states provide useful information for the successful treatment and prevention of chronic intestinal inflammation.
The second most abundant abundance bacteria, Muribaculaceae, changed in manner quite differently from Lactobacillus. The highest abundance was observed in normal control (82.7%) and nearly disappeared in IBD mouse. This phenomenon was similar to that observed in the context of feeding trials using high-calorie and/or carbohydrate-enriched diets [34-36]. This is probably the consequence of its ability to degrade particular types of polysaccharides: plant glycans, host glycans, and α-glucans [37, 38]. Also, the ability of degrading dietary carbohydrates produces succinate, acetate, and propionate [37], which may lead Muribaculaceae to occupy overlapped niches as Bacteroides does. Bacteroides also specialize in the fermentation of polysaccharides [39], and members of Bacteroides are known to produce succinate, acetate, and propionate by polysaccharides fermentation [40-42]. It may explain why Bacteroides exhibit opposite abundance trends compared with Muribaculaceae. The abundance of Bacteroides was low in both normal control (12%) and IBD mouse model recovery phase (17.3%) but became concentrated in IBD mouse model progressive phase (70.7%). This phenomenon may attribute to the ability of Bacteroides to sense and adapt to gut environmental changes and stress. Firstly, species within Bacteroides could utilize host cell surface glycoproteins and glycolipids as nutrients source at sites of infection [43-50], therefore have a superb ability to access nutrients; on the other hand, Bacteroides could modulate its surface polysaccharides to an “on” or “off” position [51], thus avoiding the host immune response. These systems used by Bacteroides may explain its high abundance in the gut inflammatory environment. Previous study also reported Bacteroides species were found in most of anaerobic infections [52]. Enterotoxigenic B. fragilis, a member of Bacteroides, could secret B. fragilis enterotoxin [53] and induce cyclooxygenase 2 and fluid secretion in intestinal epithelial cells [54], thus may be associated with occurrence of colorectal cancer and IBD [55, 56].
Another notable change occurred at Akkermansia. It has been isolated in 2004 [57] and classified as mucin-utilizing bacteria [58]. However, recent studies indicated that Akkermansia may also exert positive regulation on intestinal mucosal thickness and intestinal barrier integrity [59, 60] and be praised as “the next generation of probiotics” [61]. With the deepening of research, Akkermansia was thought to act as both “friend and foe”, so it would be unwise to advocate the efficacy of Akkermansia until more research and clinical data emerged. A research in China indicated that Akkermansia was positively correlated with the risk of Chinese type 2 diabetes [62]. Later, Ijssennagger et al found antibiotic treatment enhanced intestinal barrier function by eliminating sulphide-producing and mucus-degrading bacteria such as Akkermansia [63]. Besides, Akkermansia also triggers an excessive immune response that disrupts mucus secretion and damage intestinal barrier in IL-10-/- mice [64]. In the present study, our data suggested Akkermansia was highly enriched in d 5-model (68.4%), and 5 days recovery phase led to normalization of Akkermansia bacteria in d 10-model (15.2%), resulting in similar levels to those found in normal control (16.4%). Given this plus results of clinal signs, we could speculate reasonably that the gut microbiota of d10-model tended to restore homeostasis since the Akkermansia was considered a marker of gut homeostasis and restoration of the mucus layer in the gut [65].
Though transplantation of colon microbiota of d 5-model to Abx-mice could induce symptoms similar to colitis, it was not enough to infer that Akkermansia played as “foe” role in d 5-model. We inclined that pathogenicity was induced by a consortium of microbiota. The presence of Akkermansia or any other bacteria alone is not enough to induce these effects. In contrast, the composition and relative abundance of several specific microbes play an important role. The source or underlying cause of the increase of Akkermansia is still unclear. Previous study indicated that acetate and/or lactate produced by Bifidobacterium could stimulated the growth of Akkermansia [66]. Consistent with this, Bifidobacterium showed the same trend as Akkermansia in our result which may partly explain this elevation.
This study does have limitation because the microbiota composition during self-recovery phase (or return to initial weight) was not measured or the key bacteria may be highlighted to great extent. Secondly, the pathogenicity of altered microbiota community of IBD mouse model could be further interpreted if the microbiota composition in Abx-mice receiving FMT was monitored.