The contribution of microbiota to host fat accumulation and consequent weight gain has been demonstrated in several studies using mice models and microbiota transplantation[20–23], supported by the results of the present study as well. FMT mice gained more weight than GF mice (Fig. 1 and Supplementary Fig. S1). The subcutaneous adipose tissue weight (Fig. 2a), blood glucose (Fig. 4a), insulin (Fig. 4d), and leptin (Fig. 4h) levels, along with the fat droplet size in subcutaneous adipose tissue (Fig. 3a) and liver (Fig. 3b), were all greater in FMT mice than in GF mice. The extent to which the host was affected by the gut microorganisms varied with diet (Fig. 1–4), and no difference in appetite or food preference was observed between FMT and GF mice (Fig. 1h), suggesting that the existence of gut microorganisms directly affects host metabolism.
Several functions of gut microbiota that affect host energy balance have been proposed[24], including derivation of energy from food[14] via fermentation and degradation of indigestible polysaccharides, increase in gene expression related to nutrient transport and storage, production of intestinal hormones that contribute to body fat accumulation[25, 26], and induction of metabolic abnormalities such as insulin resistance by endotoxin[27]. Data from our study suggest that the ability of gut microorganisms to break down food ingested by the host and provide surplus energy from the resulting metabolites may contribute to weight gain. The BWs of mice fed diets rich in monosaccharides, such as fructose, showed a minor difference depending on the presence or absence of gut microorganisms (Fig. 1d and 1g), while that of mice fed diets rich in disaccharides, such as sucrose, or polysaccharides, including starch, showed large differences (Figs. 1b, 1c, & 1g). Major carbon sources that provide energy for bacteria growing in the human intestine include dietary starches, plant cell wall polysaccharides (fiber), and oligosaccharides[28]. Therefore, a diet containing substances preferentially metabolized by gut microorganisms, such as polysaccharides and disaccharides, will help them produce more metabolites that serve as energy sources for the host, leading to host weight gain. In contrast, ingestion of monosaccharides, which are less utilized by microorganisms, will provide less additional energy to the host, resulting in less impact on weight. Such an effect of the gut microbiota metabolism was also observed in a comparison of mice fed two different HF diets. The additional weight portion of mice fed USaHF-rich diet was greater than that of those fed an SaHF-rich diet (Fig. 1e, 1f, and 1g). Alcock and Lin [29] reported that the composition of gut microbiota of mice fed saturated fatty acids and unsaturated fatty acids differs considerably, based on which the generated bacterial metabolites are different, resulting in different gene expression levels in the host and consequent physiological outcomes. Thus, as the function of gut microorganisms in host physiology depends on the type of energy source for both the microorganisms and host, the process of identifying obesity-associated microorganisms may require consideration of the nutritional substances that affect both.
We analyzed the effects of six diets on the BW (Fig. 1a–g) and gene expression (Table 1, Table 2, and Supplementary Fig. S3) of GF and FMT mice, along with their gut microbiota (Fig. 5a–c). DEG analysis in a total of seven tissues identified host genes affected by gut microorganisms under each diet, and DEGs that were common across all six diet groups were determined (Table 1). Based on the correlation coefficients between BW and the amount of gene expression fluctuation associated with gut microorganisms, seventeen DEGs were identified as host genes associated with gut microorganism-mediated weight regulation (Table 2). The identification of gut microbiota involved in BW regulation (Fig. 5d and 5e) was performed by evaluating the correlation coefficients between the amount of fluctuation in the expression of these 17 genes by the gut microorganisms and abundance of gut microbiota in the same diet group. Finally, we validated whether the identified gut microbiota contributed to body weight and found a positive correlation between the abundance of EU006300_s, EF097240_s, E. caecimuris, KE993550_s, C. cocleatum, and EU505160 _s and individual mouse BW (Fig. 6a). Notably, of all levels of microbiota analyzed (phylum, class, order, family, genus, and species), the family-level microbiota member Ruminococcaceae showed the highest positive correlation with BW, being more abundant in heavier mice and less so in lighter mice (Fig. 6b).
It is well known that a characteristic of the intestinal microbiota of individuals with obesity is a high proportion of the phylum Firmicutes[30, 31]. Moreover, the abundance of certain Firmicutes, such as Mollicutes[32] at the class level; Lactobacillus[33] and Staphylococcus[34] at the genus level; and Faecalibacterium prausnitzii[35], Ruminococcus bromii[36], Lactobacillus reuteri[37], and Staphylococcus aureus[38] at the species level has been reported to increase with obesity traits. Considering that Ruminococcaceae belongs to the phylum Firmicutes (Firmicutes; Clostridia; Clostridiales; Ruminococcaceae), its identification as a gut microbiota member involved in BW regulation in the present study is reasonable. Furthermore, species belonging to the family Ruminococcaceae, such as EU006300_s, EU 505160_s, and KE993550_s, were identified as gut bacteria involved in BW regulation; however, the association that was expressed as a correlation coefficient between bacterial abundance and BW was weaker than that for the Ruminococcaceae family with the sum of several species (R2 = 0.51, 0.49, 0.54 versus R2 = 0.61). Therefore, when developing an anti-obesity treatment targeting intestinal bacteria, it may be more effective to include several species belonging to the family Ruminococcaceae rather than targeting a single species.
The family Ruminococcaceae and the genus Ruminococcus are involved in alcohol metabolism[39], adipokine metabolism[40], cirrhosis[41], acute-on-chronic liver failure[42], allergy[43, 44], antibiotic biosynthesis[45], inflammation[46], and cardiovascular risk[47]. However, the involvement of these bacteria in BW changes is controversial, with reports stating that individuals with obesity have increased[48–50] or decreased[51–53] abundance of Ruminococcaceae or that it varies with species[54]. As for the mechanism by which Ruminococcaceae members are involved in host BW regulation, it has been suggested that they degrade various polysaccharides to produce short-chain fatty acids (SCFAs)[55]. Supplementation of dietary SCFAs has been reported to inhibit BW gain associated with changes in the expression of G-protein coupled receptor 43 (GPR43) and GPR41[56]. Nonetheless, the direct causal relationship among Ruminoccaceae, BW, and SCFAs is yet unclear. To explore the reason why Ruminococcaceae are considered BW-associated bacteria, we focused our discussion on Cldn22, which had the highest frequency of occurrence as a gene, showing a correlation coefficient > 0.8 (Fig. 5d) between the gut microbiota abundance and GEL of the host genes associated with gut microorganism-mediated BW regulation (Table 2).
Although the function of CLDN22 has not yet been elucidated, it is inferred to represent a component of the tight junction chain[57–59] as it encodes a member of the claudin family[60, 61]. Tight junctions are thought to function as physical and chemical barriers that prevent food components and intestinal bacteria from freely passing through intercellular spaces between the epithelial and endothelial cell sheets[62, 63]. For example, the loss of intestinal Cldn7 expression results in epithelial cell sloughing and spontaneous inflammation[64]; Cldn2-deficient mice have a defective water and sodium reabsorption function at the tight junctions of kidney proximal tubules[65], and decreased Cldn1 expression in the skin disrupts skin barrier function[66]. As claudin, a tight junction protein, can function as a regulator of paracellular barrier permeability in organs such as the intestine[67], we speculate that in visceral adipose tissue, eWAT, it exchanges certain nutrients with the adherent intestinal tissue using its barrier function. Our hypothesis regarding the mechanism of involvement of Ruminococcaceae and Cldn22 in obesity is illustrated in Fig. 6c. Dietary fat components downregulate Cldn22, resulting in the disruption of the barrier function, and nutrients stored in adipose tissue leak into the interstitial space, getting absorbed in the intestinal tract. If Ruminococcaceae members are present, the bacteria produce metabolites from these absorbed nutrients and they are transported to the liver and other tissues of the host body, generating additional energy, which in turn causes weight gain. The metabolites transported to eWAT further downregulate Cldn22, resulting in a cycle of additional disruption of the barrier function, leakage of nutrients, digestion and absorption of those nutrients in the intestine, as well as consequent weight gain. Our results showed that Cldn22 expression in eWAT was decreased when the host consumed HF diet, regardless of the presence or absence of gut microorganisms. In addition, the extent of decreased expression was greater in FMT mice than in GF mice (Supplementary Fig. S3b). These results may be consistent with the finding that the extent of BW gain in FMT mice, in which the barrier function is presumed to be repeatedly disrupted, was greater than that in GF mice, in which barrier disruption presumably did not occur.
Will improving tight junction and barrier function by manipulating the proportion of Ruminococcaceae lead to weight loss and weight gain inhibition? The answer is: it probably depends on diet. Cldn22 expression is decreased when the host consumes an HF diet, even in the absence of gut microorganisms. The HF diet-induced downregulation of Cldn22 expression is considerably greater than the decrease caused by gut microorganisms in a low-fat diet (Supplementary Fig. S3b). Thus, as long as the host consumes an HF diet, manipulating the proportion of Ruminococcaceae to suppress the downregulation of Cldn22 will not overcome the downregulation caused by fat intake; therefore, it will not inhibit weight gain. However, if the host diet mainly includes carbohydrates such as sucrose and starch and is low in fat, the decrease in Cldn22 expression caused by diet does not occur in the first place. Hence, if the downregulation of Cldn22 due to gut microorganisms is prevented by manipulating the proportion of Ruminococcaceae, it may be possible to inhibit weight gain. Accordingly, the extent to which the host dietary components alter both the abundance of intestinal microorganisms and host gene expression should be considered to appropriately judge their effect on host physiology. In this way, obesity-associated intestinal microorganisms can be identified and used for developing obesity prevention and intervention strategies.
In this study, we focused on all of the above. However, important questions remain regarding the direct causal link between Cldn22 and Ruminococcaceae. Future studies should examine this relationship in Cldn22 transgenic mice.