1. Antibiotic treatment and transfer of gut microbiota by cohousing effectively change the gut microbial community composition in HFHC diet-fed LDLR −/− hamsters.
Recently, the central role of gut microbiota in the development of metabolic syndrome has been reported in numerous clinical and experimental studies[4]. Firstly, to examine whether intermittent antibiotic treatment altered the profile of gut microbiota in our hamster model with CHL, we treat high-fat and high-cholesterol (HFHC) diet-fed male LDLR−/− hamsters with the antibiotics or placebo via a separate or cohousing approach (Fig. 1A). 16S rRNA gene sequencing analysis in feces revealed the changes in gut microbial ecology under different conditions. After denoising Amplicon Sequence Variants (ASVs) using DADA2 and annotating each ASV species using Naive Bayes classifier, we found that in total, 1,911,588 clean tags were clustered into operational taxonomic units based on 100% identity. Rarefaction analysis showed that observed number of ASVs reached saturation and was sufficient to describe the bacterial diversity in the samples (Fig. 1B). For clustering of species shown, there were differences in species content among samples (Fig. 1C). The Venn diagram showed that the PS, AS, PM and AM groups contained 5,986, 6,403, 6,007 and 6,202 ASVs, respectively (Fig. 1D). As expected, α-diversity was not different among these four groups on the Simpson indexes (p > 0.05) (Fig. 1E).
Weighted unifrac distances of 16S rRNA gene sequences, a measurement for β-diversity, demonstrated a clustering between the AS and AM group (Fig. 1F). Analogously, we also observed a clustering between the PS and PM group (Fig. 1F). Moreover, β-diversity analysis was conducted using principal coordinates analysis (PCoA) and non-metric multidimensional scaling (NMDS) methods. The results showed that although the microbiota cluster in LDLR−/− hamsters of the AS and AM groups, or of PS and PM groups, was relatively similar, the microbial community structure was significantly different between the PS and AS groups (Fig. 1G and 1H). Consistently, a clear separation was observed between PM and AM (Fig. 1G and 1H).
Further analysis at the phylum level indicated that a top 10 phyla were defined in our study, in which Bacteroidetes and Firmicutes were the primary phyla among all the groups. Intermittent antibiotic treatment remarkably decreased the relative abundance of Firmicutes, but had no effect on Bacteroidetes. Moreover, the relative abundances of Proteobacteria, Euryarchaeota, and Fusobacteriota were obviously increased in LDLR−/− hamsters treated with antibiotics (Fig. 1I, Supplementary Figure. 1A), while Patescibacteria was significantly reduced (Fig. 1I, Supplementary Figure. 1A).
Specifically, the relative abundance data showed that at the genus level, LDLR−/− hamsters treated with antibiotics had significantly lower relative abundances of Lactobacillus, Allobaculum, Candidatus-Saccharimons, and Ileibacterium (Fig. 1J, Supplementary Figure. 1B). Meanwhile, antibiotics markedly increased the relative abundances of Blautia, Fusobacterium, Parasutterella, Enterococcus, and Coprobacillus, which were restored by a transfer of gut microbiota through cohousing approach (Fig. 1J, Supplementary Figure. 1B). These results indicated that intermittent antibiotic treatment and transfer of gut microbiota through cohousing approach could effectively change gut microbial community composition in LDLR−/− hamsters fed a HFHC diet.
2. Altered gut microbiota by antibiotics and transfer of gut microbiota by cohousing generate a favorable lipid profile to protect from HFHC diet induced obesity and hyperlipidemia in LDLR −/− hamsters.
Next, we assessed whether remodeling of gut microbiota and transfer of gut microbiota improve HFHC diet-induced obesity and dyslipidemia in LDLR−/− hamsters. No significant differences of initial weight were observed among these four groups. However, after 12-week HFHC-diet feeding, the body weight gain of the animals from AS group was obviously decreased, and transfer of gut microbiota shown in the AM group moderately reduced body weight gain when compared to the PS control group (Fig. 2A). In addition, our results from the body temperature and the biological parameters of blood, measured at the endpoint of our entire experiment, indicated that the alteration of gut microbiota didn't influence the systemic immune function of LDLR−/− hamsters (Fig. 2B, Supplementary Figure. 1C-1I). Interestingly, although there were no differences in total cholesterol (TC) levels among the four groups (Fig. 2C), the plasma triglyceride (TG) levels in AS, PM and AM groups were significantly reduced relative to PS group (Fig. 2D), accompanied by elevated high density lipoprotein cholesterol (HDL-C) content in AS and AM groups (Fig. 2E) and reduced non-esterified fatty acid (NEFA) level in only AS group (Fig. 2F) when compared to PS group. Further analysis by Western blots showed that plasma ApoB100, ApoB48 and ApoE levels were markedly reduced, but ApoA1 level was increased in LDLR−/− hamsters of AS and AM groups compared with PS and PM groups (Fig. 2G and 2H). Lipid distribution analyzed by FPLC demonstrated that cholesterol content was increased in the HDL fractions in LDLR−/− hamsters of AS, PM, AM groups compared with LDLR−/− hamsters of PS group, with no obvious changes in VLDL and LDL fractions (Supplementary Figure. 2A). Moreover, the TG concentration in the VLDL fractions was visibly reduced in the animals of AS and AM groups, while the peak of VLDL fractions was modestly reduced in PM group compared with PS control group (Supplementary Figure. 2B). Consistent with the observations from plasma samples, Western blots of FPLC fractions revealed that the amounts of ApoE carried on TG-rich lipoproteins were decreased and the ApoA1 concentration in HDL fractions was elevated in LDLR−/− hamsters of AS compared to PS, PM and AM groups (Supplementary Figure. 2C).
3. LDLR −/− hamsters with antibiotic treatment and transfer of gut microbiota by cohousing are protected from HFHC diet-induced adipocyte hypertrophy.
Since diet-induced obesity is defined as an abnormal expansion and excessive accumulation of fat mass in white adipose tissue (WAT)[15] and the body weight was significantly reduced in HFHC diet-fed LDLR−/− hamsters after antibiotic treatment, we explored whether this beneficial effect of antibiotics on diet-induced obesity was attributable to maintaining the homeostasis of adipose tissue. Therefore, we examined the weight of adipose tissues in LDLR−/− hamsters. As expected, the fat mass and the ratio of fat weight/body weight in subcutaneous WAT (sWAT) (Fig. 3A and 3B) and epididymal WAT (eWAT) (Fig. 3C and 3D) were markedly decreased after antibiotic treatment, but there were no differences in inguinal WAT (iWAT) (Fig. 3E and 3F) and brown adipose tissue (BAT) (Fig. 3G and 3H). Histological analysis showed that eWAT from LDLR−/− hamsters of AS and AM groups contained normal mature adipocytes, which were characterized by the presence of unilocular lipid droplets (Fig. 3I, 3M and 3O). In contrast, adipocytes from LDLR−/− hamsters of PM groups showed adipocyte hypertrophy, and adipocytes from LDLR−/− hamsters of PS were highly hypertrophic (Fig. 3I, 3L and 3N).
Strikingly, brown adipocytes from LDLR−/− hamsters of AM and PM groups displayed giant form with white adipocyte like lipid droplets (Fig. 3J). In LDLR−/− hamsters of PS, brown adipocytes were uncovered to contain much larger lipid droplets, implying BAT whitening (Fig. 3J). Although the expression levels of lipolysis and thermogenesis-related genes were unchanged in four groups (Supplementary Figure. 3A and 3B), the uncoupling protein 1 (UCP1) protein content was more in AS and AM groups compared with PS and PM groups (Fig. 3K).
4. Antibiotic treatment and transfer of gut microbiota by cohousing alleviate HFHC diet-induced NASH and atherosclerosis in LDLR −/− hamsters.
Since obesity and abnormal lipid metabolism are associated with liver disease, we then analyzed the parameters reflecting liver function in LDLR−/− hamsters fed with HFHC diet. Firstly, we examined bacteria contents in liver translocated from intestine and found that the levels of 16S rRNA gene remained no difference among four indicated groups (Supplementary Figure. 4A). Subsequently, we observed that the liver weight and the liver weight to body weight ratio were both reduced in AS group hamsters compared with hamsters of the PS group (Fig. 4A and 4B); however, there were no obvious difference in these indicators between AM and PM groups (Fig. 4A and 4B). The reduced levels of plasma alanine aminotransferase (ALT) and aspartate transaminase (AST) demonstrated that the liver injury caused by HFHC diet was significantly improved in LDLR−/− hamsters of AS and AM groups compared with PS and PM groups, respectively (Fig. 4C and 4D). Furthermore, ALT and AST levels were distinctly lower in the PM group than PS group, indicating that transfer of gut microbiota by cohousing defended hepatic injury induced by HFHC diet (Fig. 4C and 4D). Nevertheless, both of plasma LPL activity (Supplementary Figure. 4B) and hepatic VLDL secretion rate (Supplementary Figure. 4C and 4D) were greatly decreased, the amounts of hepatic TC and TG were markedly reduced in LDLR−/− hamsters administrated with antibiotics (Supplementary Figure. 4E and 4F). Compared with PS control group, there was a mild decrease in TC and TG contents in hamster livers of the PM group (Supplementary Figure. 4E and 4F). Additionally, antibiotic treatment dramatically alleviated the phenotypes of NASH (steatosis, inflammation and ballooning) (Fig. 4E) with the decreased NAFLD activity score (NAS) (Fig. 4J), and consistently, those animals also had reduced accumulation of lipid stained by oil red O dye (Fig. 4F and 4K). Sirius Red staining, which indicates the fibrosis, showed that there was abundant abnormal deposition of collagen matrixes surrounding the macrovesicular lipid droplets in PS group hamsters, and these collagens were dramatically reduced by antibiotics (Fig. 4G and 4L). The Kupffer cells aggregated to form crown-like structures surrounding these macrovesicular lipid droplets in PS group hamsters, which could be improved by antibiotics (Fig. 4H and 4M). In agreement with the finding of oil red O staining, BODIPY staining also revealed more neutral lipid droplet accumulation in the hamster livers of PS group compared with AS group (Fig. 4I and 4N). Moreover, histological analysis exhibited that transfer of gut microbiota by cohousing obviously relieved NASH in LDLR−/− hamsters of the PM group compared with PS group (Fig. 4E-4N). Consistent with the pathological observations, qPCR analysis indicated that the transcriptional expression of genes involved in inflammation and fibrosis (Supplementary Figure. 5A), lipid uptake, synthesis and transport (Supplementary Figure. 5B) in the liver was obviously downregulated in antibiotic-treated hamsters, while the expression levels of genes regulating fatty acid synthesis and oxidation were reduced to a certain extent (Supplementary Figure. 5C), which could be due to the suppression of adipocyte hypertrophy by antibiotic treatment. Meanwhile, the AKT phosphorylation (p-AKT) level was raised in LDLR−/− hamsters of AS and AM groups compared with PS and PM groups, respectively (Supplementary Figure. 5D and 5E), demonstrating that antibiotic application and transfer of gut microbiota enhanced hepatic insulin sensitivity.
As LDLR−/− hamsters with hypercholesterolemia was also an ideal small rodent animal model used for human atherosclerosis, and metabolic disorders are highly linked to atherosclerotic CVD (ASCVD), we also investigated whether antibiotic treatment or transfer of gut microbiota could mitigate the atherosclerotic development. Our oil red O staining data disclosed that the atherosclerotic lesion areas in both the aortic roots (Fig. 4O and 4P) and whole aortas (Fig. 4Q and 4R) were markedly reduced in LDLR−/− hamsters of AS and AM groups compared with PS and PM groups, respectively. Compared with PS control group, the atherosclerotic lesion areas presented a significantly decrease in aortic roots and a mild reduce in whole aortas in hamster of the PM group. Collectively, these results suggest that antibiotics and transfer of gut microbiota by cohousing have prophylactic effects on both NASH and atherosclerosis in the preclinical hamster model with CHL.
5. Intestinal lipid absorption is inhibited in HFHC diet-fed LDLR −/− hamsters after antibiotic treatment and transfer of gut microbiota by cohousing.
Antibiotics have been declared to possess functions on intestinal epithelial cells and maintaining barrier integrity[16, 17]. In the gut, epithelial absorption of dietary lipids is a known regulator of metabolic syndrome[18]. We therefore examined the effects of altered gut microbiota on intestinal lipid absorption and explored the potential molecular mechanism.
We found that postprandial plasma triglyceride was significantly reduced (Fig. 5A and 5B), whereas the lipids were strikingly retained in ileum tissue in LDLR−/− hamsters of AS and AM groups compared with PS and PM groups, respectively (Fig. 5C and 5E). As expected, the contents of TC and TG in feces were sharply increased in LDLR−/− hamsters of AS and AM groups compared with PS and PM groups, respectively (Fig. 5F and 5G). Besides, we observed the crypts height was reduced in AS group (Fig. 5D and 5H), but villus length and mucosal thickness were not altered among four groups (Fig. 5D, 5I and 5J). Additionally, there was no difference in the expression level of tight junction protein ZO-1, a marker indicating the permeability of the intestinal epithelium in all groups, indicating that antibiotics and transfer of gut microbiota by cohousing didn’t affect the integrity and the permeability of the intestinal epithelium in our animal model (Fig. 5K). However, the expression levels of genes involved in lipid absorption (CD36 and NPC1L1) and transport (FABP1 and FATP4) in ileum tissue were significantly decreased after antibiotic treatment (Fig. 5K), but the expressions of FXR and SHP, two critical genes regulating intestinal BA metabolism, were not changed in our study (Fig. 5L).
6. Effects of antibiotic treatment and transfer of gut microbiota by cohousing on the gut microbiota function in LDLR −/− hamsters on HFHC diet.
In order to explore the potential involvement of the gut microbiota in mediating HFHC diet-induced metabolic syndrome, we performed 16S rRNA gene sequencing on stools from four indicated groups. PCoA and NMDS analyses also showed that the microbiota composition clustered distinctly for LDLR−/− hamsters from four indicated groups (Fig. 1G and 1H). Co-occurrence network analysis revealed that the non-random co-occurrence patterns among dominant commensal taxa and potential pathogens of abundant LDLR−/− hamsters gut genera are disparate (Fig. 6A-6D). To verify the different bacteria among PS, AS, PM, and AM groups, we then performed LEfSe analysis and selected genera based on the LDA score > 6. Notably, we observed an overrepresentation of distinct bacterial genus Blautia in AS group compared with other groups (Fig. 7A and 7B). Using the Correlation analysis to determine the potential association of bacterial abundance with hamster phenome, we observed that Lactobacillus, Clostridia-UCG-014, Bifidobacterium and Eubacterium_coprostanoligenes_group enriched in fecal samples from LDLR−/− hamsters of PS group were positively correlated with the HFHC diet-induced metabolic disorder phenome (Supplementary Figure. 1B; Fig. 7C, 7D); however, Blautia, Fusobacterium, Parasutterella, and Methanosphaera enriched in LDLR−/− hamsters of AS group was negatively correlated with HFHC diet-induced metabolic disorder phenome (Supplementary Figure. 1B; Fig. 7C, 7D), indicating that diet-induced metabolic disorders are strongly correlated with gut microbiota composition, especially the level of Blautia. LEfSe and Similarity percentage analysis combined with Student’ s t-test further exhibited the detailed differences and contributions of different bacteria species between PS and AS group (Supplementary Figure. 6A, 6C and 6E), PM and AM group (Supplementary Figure. 6B, 6D and 6F), PS and PM group (Supplementary Figure. 7A, 7C and 7E), and AS and AM group (Supplementary Figure. 7B, 7D and 7F), respectively. These findings implied that antibiotics and cohousing approach had different impacts on systemic metabolism via the changes of gut microbiota composition.
To better understand the effect of antibiotics and cohousing approach on gut microbiota function based on KEGG orthology database, we conducted a Venn diagram and weighted unifrac principal component analysis and found that gut microbiota compositional discrimination contributed to the changes in the gut microbiota function (Fig. 8A and 8B). Cluster analysis of relative abundance further revealed that metagenomic functions were different between PS and AS groups, and PM and AM groups, but metagenomic functions were similar between PS and PM groups, and AS and AM groups (Fig. 8C). Importantly, we discovered that the transportation of glucose, nucleic acids, amino acids and metal ions was altered in gut microbiota of HFHC diet-fed LDLR−/− hamsters after treated with antibiotics (Fig. 8C) together with the changes of nutrient transportation by gut microbiota that were strongly associated with gut microbiota and host metabolism (Fig. 8D). Taken together, our results demonstrated that antibiotic treatment and cohousing approach modulated systemic metabolic homeostasis in LDLR−/− hamsters with CHL through influencing gut microbiota composition and function.