Male and female spontaneously hypertensive rats (SHR) and the Wistar-Kyoto (WKY) control strain from Charles River Laboratories in Wilmington, Massachusetts, were used in this study. The animals were housed in a controlled environment with a temperature of 24±1 and a 12-hour light-dark cycle. After a week of acclimatisation, the rats were randomly divided into two experimental groups, regardless of the strain. The groups were formed based on similar body weights and lengths at four weeks of age. One group was fed a standard chow diet (SCD) for 13 weeks starting from week 17. This group served as the control group (SHR, n=12 per sex; WKY, n=12 per sex), with food intake restricted to the baseline levels. The alternate group was given unrestricted access to a high-fat solid diet (HFD) consisting of 40% w/w bovine and porcine fats added to standard chow, as previously described [29]. Supporting information provides complete details of the experimental model, biochemical determinations, histological analysis of the liver tissue, and measurement of liver triglyceride content.

The microbiota composition of male and female rats with MASLD and control animals was analysed using 16S rRNA gene sequencing. DNA was extracted from fresh-frozen intestinal samples, including the small intestine and descending colon. Intestinal segments, approximately 1.5 cm in length, were emptied of the luminal contents and gently washed with PBS (10 mM NaH₂PO₃, pH 7.4, 0.9% NaCl). DNA was extracted from tissues using a manual protocol outlined in our previous study, which consisted of two mechanical lysis steps [13]. The Supporting file contains comprehensive details.

Subsequently, a stepwise analysis workflow was implemented, comprising data processing, visual exploration of rarefaction curves, community profiling, and clustering and correlation network and pattern searches. The workflow involves filtering, dereplication, sample inference, chimera identification, and merging of paired-end reads [30]. The Greengenes database (13.8) [31] was used for taxonomic annotations. The Supporting file contains comprehensive details.

The supporting information file provides full details on the analysis of complex metadata, including single and multivariate analyses, general linear models in the MaAsLin2 R package, linear and nonlinear microbial correlations, Sparse Estimation of Correlations among microbiomes (SECOM), LEfSe (linear discriminant analysis Effect Size), and linear discriminant analysis (LDA) score. The false discovery rate (FDR) was integrated to correct all gathered data for multiple tests, with statistical significance set at P<0.05. Analyses were conducted using the bioinformatics pipeline provided by the online platform Microbiome Analyst version 2.0 (https://www.microbiomeanalyst.ca/MicrobiomeAnalyst/) [32]. The Supporting information file contains comprehensive details.

The Supporting information file contains comprehensive details.

IHC was used to explore the abundance and localisation of LPS utilising an antibody specific to the Lipid A endotoxin region of LPS from Escherichia coli, as previously described [13]. To assess the liver protein levels of Tlr4 we used a specific antibody (OASG07216; Aviva Systems Biology, San Diego, Ca, USA). Additional information is available in the Supporting Information file.

Following dietary intervention, irrespective of the strain, both female and male rats developed microvesicular and macrovesicular hepatic steatosis (Supplementary Fig. 1A, B). SHR exhibited markedly higher scores of portal inflammation compared to the WKY strain (ordinal logistic regression: coef: 2.32±0.47, P=0.00000072). There were significant differences in lobular inflammation scores between male and female animals (P=0.031) (Table 1). Rats with metabolic syndrome traits such as heightened blood pressure and insulin resistance were more prone to develop visceral fat (Table 1). Serum triglyceride levels were considerably higher in the HFD groups (WKY rats: 211.7±18.2 and SHR: 216.6±19.8 mg/dL, respectively) compared to SCD-controls (WKY rats 88.5±19.4 and SHR 153.3±19.0 mg/dL, respectively); P=0.0000036. The changes in body weight over time, stratified by gender and strain, are presented in Supplementary Figure 1C, D. In addition, HFD was associated with a significant increase in visceral fat (either absolute mass or relative to body weight), irrespective of rat strain and sex. However, there were substantial differences between rat strains and sex, i.e., male and female SHR animals had significantly higher visceral fat mass than male and female WKY animals, particularly after HFD feeding. On the other hand, female animals had significantly more visceral fat relative to body weight than male animals. In terms of hepatic triglyceride content, the amount was significantly higher in the HFD groups (WKY rats 21.2±2.3 and SHR 21.4±2.5 μg/mg liver, respectively) compared to SCD controls (WKY rats 9.8±2.4 and SHR 12.3±2.4 μg/mg liver, respectively); P=0.0001. Table 1 presents sex-disaggregated data on biochemical, histological, and metabolic features. In this model, the animals do not develop liver fibrosis. This is a relevant consideration, as the development of liver fibrosis in rats necessitates the introduction of dietary manipulation or other interventions that may alter the microbiota, thereby rendering the analysis more complex and less reflective of human conditions.

The hierarchical structures of taxonomic features in MASLD and control samples in the jejunum and descending colon were compared using heat tree analysis (Fig. 1A, B).

Analysis of jejunum samples from rats with MASLD in comparison to control animals at the genus level indicate an increase in the abundance ratios of Proteobacteria (Gammaproteobacteria, Alphaproteobacteria, Betaproteobacteria and Epsilonproteobacteria), Actinobacteria, Acidobacteria and a decrease in the abundance ratios of Turicibacteriales, among other taxa (Fig. 1A). Examination of samples extracted from the descending colon of animals that developed MASLD indicates a significant increase in the abundance ratio of Vivrionales, Verrucomicrobiae, and Actinobacteria and a notable decline in the abundance ratios of Bacteroidetes (see Fig. 1B).

Supplementary Figure 2A, B illustrates the taxonomic features of jejunum samples from rats that developed liver steatosis compared with those that did not, according to the rat strain. The results demonstrate that animals with liver steatosis and features of metabolic syndrome (SHR rats) exhibit, among many other taxa, an increase in Gammaproteobacteria, particularly Xanthomonadales, Alphaproteobacteria, and Actinobacteria (Bifidobacterium). The control strain (WKY rats) shows changes to a lesser extent. Supplementary Figure 2C illustrates representative taxa abundance according to rat strain in MASLD and control animals showing that in the case of Prevotella, changes are in the opposite direction.

The taxonomic features illustrated in Supplementary Figure 3A, B compare the microbiota of the descending colon samples from rats, according to the rat strain, that developed liver steatosis with those from rats that did not. SHR and WKY with MASLD exhibit a marked reduction in the amount of the bacterial genus Prevotella. Nevertheless, as illustrated in Supplementary Figure 3A, B, WKY rats exhibit a disparate taxonomic profile, particularly with regard to the relative abundance of taxa expressed as median ratio (Log2) rather than the presence of a given taxon. Supplementary Figure 3C provides an illustrative example of the deregulated taxa in MASLD compared to the control animals in the descending colon samples.

As a primary objective, we analysed the composition of the intestinal microbiota in male and female rats. The results showed significant differences in the community profiles of females and males in the small intestine. While jejunum samples from female rats showed substantially higher alpha diversity (Fig. 2A), richness in the descending colon did not account for significant differences between the sexes (Fig. 2B).

Furthermore, we conducted PERMANOVA to examine the diversity of intestinal microbial communities between male and female animals. Analysis of beta diversity using the Bray-Curtis index highlighted significant gender-associated differences in the microbial composition of both the small and large intestines (Fig. 2C, D). Compared with male rats, female rats demonstrated increased beta diversity in the jejunum and colon microbiome (P=0.001).

The hierarchical structure of taxonomic features in the jejunum and descending colon stratified according to sex was compared using heat tree analysis, as shown in Figure 3A, B, respectively.

Correlation analysis was conducted to identify biologically significant associations between taxa and sex. To estimate the linear and nonlinear relationships between the bacterial taxa, we generated a microbiome interaction network for the small and large intestines using the SECOM approach. This approach accounts for both sample- and taxon-specific biases while maintaining sparsity. We identified significant positive correlations among the taxa, with increased abundance in both the small and large intestines of female rats. Additionally, we observed positive correlations between taxa enriched in both the jejunum and descending colon of male rats. In contrast, a negative correlation was observed between taxa that were more abundant in the female intestinal microbiome and those that were more abundant in the male intestinal microbiome (Supplementary Fig. 4), irrespective of the sample location. For example, the orders Erysipelotrichales, Deferribacteres, and Chlamydiales, which were enriched in the jejunum of female rats, were negatively correlated with the orders Nitrospirales and Rickettsiales which were enriched in the jejunum of male rats (Supplementary Fig. 4). Conversely, the orders Brachispirales and Deferribacteres, which were abundant in the descending colon of female rats, were negatively correlated with the orders Turicibacteriales, Campylobacterales and Actinomycetales, which were abundant in the colon samples of male rats (Supplementary Fig. 4). Additionally, we employed Spearman’s correlation analysis to rank relationships between pairs, which yielded the same outcome, indicating a significant difference between the sexes.

The core taxonomic heatmap across jejunum samples derived from female and male rats is shown in Supplementary Figure 5 panel I A, B. At the family taxa level, jejunum microbiome profiling in female rats showed 13 most abundant features at a sample prevalence higher than 20%, including Clostridiaceae (100%), Enterobacteriaceae (98%), Lactobacillaceae (96%), Ruminococcaceae (94%), Helicobacteraceae (89%), Bifidobacteriaceae (79%), Erysipelotrichaceae (75%), Alcaligenaceae (61%), Bacteroidaceae (59%), Verrucomicrobiaceae (38%), Lachnospiraceae (34%), Turicibacteraceae (32%), and Prevotellaceae (24%). Supplementary Figure 5A (panels I and II) presents further detail. In male rats, jejunal microbiome profiling revealed that the most abundant taxa were Lactobacillaceae and Clostridiaceae, present in all samples. Additionally, Ruminococcaceae was found in 93% of the studied samples, Helicobacteraceae in 86%, Enterobacteriaceae in 71%, Bifidobacteriaceae in 53%, Turicibacteraceae in 51%, Erysipelotrichaceae in 42%, Thermodesulfovibrionaceae in 28%, Bacteroidaceae in 24%, and Verrucomicrobiaceae in 22%.

The analysis focused on the liver steatosis variable (MASLD vs. control animals) after adjusting for sex and showed a significant increase at the class level for Epsilonproteobacteria (logarithm of fold change ratios [Log2FC]: 1.32, FDR P-value: 0.005, Wilcoxon Rank Sum test), Alphaproteobacteria (1.0, P=0.029), and Actinobacteria (0.624, P=0.038) in animal samples that developed MASLD versus control animals. At the order level, five significant features associated with MASLD were identified after adjusting for sex. These included an increase in the abundance of Campylobacterales (1.31, P=0.005), Streptophyta (1.31, P=0.013), Rickettsiales (1.63, P=0.025), and Chlamydiales (0.671, P=0.043), as well as a decrease in the abundance of Ellin 6513 (–0.522, P=0.025). At the family level, significant changes were identified in the abundance of the four taxa in the MASLD rats. Specifically, an increase was observed in Helicobacteraceae (1.31, P=0.0045) and Rhabdochlamydiaceae (0.671, P=0.049), as well as changes in the abundance of Paraprevotellaceae (–0.845, P=0.034) and Aerococcaceae (–0.669, P=0.04).

Subsequently, we conducted an analysis using sex as the main variable, adjusted for the presence of MASLD (liver steatosis) (Supplementary Table 1). Taxonomic analysis at the genus level identified 52 significant features associated with sex after the MASLD adjustment (Supplementary Table 1). For example, the most remarkable changes occurred in the Mucispirillum (Log2FC: 3.23), Bradyrhizobium (Log2FC: 3.11), Corynebacterium (Log2FC: 2.0), Streptococcus (Log2FC: 1.85), and Escherichia (Log2FC: 1.78) taxa, which were significantly increased in the jejunum samples of female rats compared to their male counterparts (see the complete list in Supplementary Table 1). In contrast, the number of Lardizabala (Log2FC: –3.14), Gemella (Log2FC: –1.28), and Streptomyces (Log2FC: –1.06), along with various other taxa, showed a considerable increase in the jejunum samples of male rats.

Furthermore, we utilised LDA to identify the taxa that were most likely responsible for the differences between the respective phenotypes and biological conditions of interest, as determined by LDA scores. Using LEfSe approach we discovered that 24 significant taxon features at the genus level were strongly linked to MASLD as a dichotomous trait (Supplementary Fig. 6A). Analysis of the relative abundance of features in samples from control animals compared to animals that developed MASLD revealed that three taxa, Aggregatibacter, Turicibacter and Gemella were significantly decreased in animals with MASLD. Additionally, 21 features, including Escherichia, Pseudomonas, and Bifidobacterium among other relevant taxa, were significantly increased in animals with MASLD (Supplementary Fig. 6A). Finally, after adjusting for sex, LEfSe analysis revealed 40 significant taxon features at the genus level associated with MASLD (see Supplementary Fig. 7). LDA scores for the top 15 features are shown in Figure 4A.

Supplementary Figure 8 shows the abundance profiles between animals with MASLD and the control group according to the respective sex in a quantitative (using the median abundance) and statistical manner (using the nonparametric Wilcoxon Rank Sum test) in the jejunum, along with some representative taxa abundance. For example, the Clostridia class demonstrated an increase in MASLD males, but not in females. Conversely, the Erysipelotrichaceae exhibited a significant increase only in female animals with MASLD.

The core taxonomic heatmaps for the samples obtained from the descending colon of female and male rats are shown in Supplementary Figure 5 (panels I and II, C, D). At the family taxa level, profiling of the female rat large intestine microbiome revealed 21 of the most prevalent characteristics, with a sample relative abundance greater than 20%. Helicobacteraceae (100%), Ruminococcaceae (96%), Deferribacteraceae (92%), Alcaligenaceae (84%), and Clostridiaceae (77%) were among the key families. Bifidobacteriaceae (73%), Enterobacteriaceae (71%), Bacteroidaceae (67%), Desulfovibrionaceae (61%), Brachyspiraceae (59%), Lachnospiraceae (57%), Veillonellaceae (53%), Paraprevotellaceae (53%), and Erysipelotrichaceae (49%) also played significant roles (Supplementary Fig. 5C, panels I and II). In contrast, the descending colon of male animals contained 17 prevalent taxa at a rate higher than 20%. These included Helicobacteraceae (100%), Ruminococcaceae (95%), Enterobacteriaceae (87%), Clostridiaceae (82%), Erysipelotrichaceae (76%), Lactobacillaceae (73%), Veillonellaceae (69%), Turicibacteraceae (58%), Paraprevotellaceae (54%), and Prevotellaceae (48%) (Supplementary Fig. 5D panels I and II).

Further, we conducted a statistical comparison of taxon abundance using a multiple linear regression analysis with covariate adjustment. First, we examined the liver steatosis variable (MASLD vs. control animals), which was adjusted for sex. We discovered a significant increase in Deferribacteres at the class level (log 2FC: 1.13, FDR: P=0.0291) in animals developing MASLD compared with non-MASLD animals. At the family level in MASLD rats, we observed a noteworthy increase in the proportion of four types, which encompassed Vibrionaceae (log 2FC: 1.74, FDR: P=0.0034), Deferribacteraceae (log 2FC: 1.12, FDR: P=0.03), and Aerococcaceae (log 2FC: 0.91, FDR: P=0.033), in conjunction with Corynebacteriaceae (log 2FC:1.56, FDR: P=0.001). Conversely, we detected a significant decrease in the abundance of Prevotellaceae (log 2FC: –2.18, FDR: P=0.0036) and Paraprevotellaceae (log 2FC: –1.75, FDR: P=0.02). Second, using LEfSe, we identified 13 genus-level taxon features significantly linked to MASLD, as shown in Supplementary Figure 6B. Analysis of the relative abundance in the control animals revealed that nine taxa were significantly increased in animals with MASLD. These taxa included Mucispirillum, Catenibacterium, Streptococcus, Sulfurimonas, Vibrio, Anaerobaculum, Bifidobacterium, Faecalibacterium, Acinetobacter and Desulfovibrio. In contrast, three genera showed significant decreases, including Prevotella, CF231 and SMB53.

Additionally, we analysed the sex variable (female versus male) adjusted for MASLD as our primary data. Analysis at the genus level revealed 26 significant characteristics associated with sex adjusted by MASLD (Supplementary Table 2). Finally, we identified 31 taxonomic features at the genus level that were significantly associated with both MASLD and sex (Supplementary Figure 9). The LDA scores for the top 15 features are shown in Figure 4B.

Supplementary Figure 10A, B illustrates the genus-level taxon features of the LEfSe that were significantly correlated with MASLD in the two intestinal compartments, the jejunum samples (a) and the descending colon (b), stratified according to the rat strain (WKY and SHR).

Supplementary Figure 11 illustrates the abundance profiles of animals with MASLD and the control group, stratified by sex, in a quantitative and statistical manner. The abundance profiles are presented in descending colon and include representative taxa abundance graphs. The quantitative analysis employs the median abundance, while the statistical analysis employs the non-parametric Wilcoxon Rank Sum test. For instance, the Desulfovibrionales increased in MASLD females but not in males. Furthermore, although Escherichia was increased in MASLD regardless of sex, the abundance in male animals with MASLD was significantly higher.

Ethical considerations preclude the use of human samples in the validation of our hypothesis. Therefore, an experimental model of steatotic liver disease that mimics human MASLD was employed to analyse the microbiome of the upper small intestine and colon. The main challenge is the translation of preclinical results to complex human populations and the intricacies associated with human diseases. A notable feature of our investigation is the use of an experimental model that conforms to the recently established diagnostic criteria for MASLD [1]. Because metabolic syndrome and MASLD are complex conditions that involve both environmental and genetic factors, finding suitable experimental models to study them is challenging. Meeting the current diagnostic criteria for MASLD, including liver fat accumulation, metabolic dysregulation, and cardiovascular risk factors, requires careful attention. However, it is worth noting that the SHR strain has a genetic predisposition to a range of metabolic syndrome features, such as hypertension, glucose intolerance, and dyslipidemia, despite the absence of overt obesity, the lack of which was compensated for by HFD. Nevertheless, after HFD challenge, SHR rats accumulated significantly higher levels of visceral fat.

Our findings demonstrate that distant anatomical regions of the gut have a unique microbiome profile, which is consistent with the results of previous studies [35-37]. Therefore, stool cannot be substituted by the entire gut microbiome. However, the observed heterogeneity in distant regions of the gut may affect the biology of MASLD, which largely depends on sex-specific patterns of microbial abundance. This is particularly important because microbiome-related studies on MASLD rarely test sex as the target variable of interest.

Finally, it can be argued that it is challenging to extrapolate conclusions from experimental studies to humans. Interestingly, we found that the most notable features observed in animals that developed hepatic steatosis in the metabolic syndrome-related model, such as Gammaproteobacteria, particularly Xanthomonadales, Alphaproteobacteria and Actinobacteria (Bifidobacterium), mirror the findings observed in humans [13].

In conclusion, our study offers novel insights into the pathogenic mechanisms of MASLD, which represents a pivotal initial step in translating knowledge to the clinic. In this context, we consider that the generation of new knowledge on the potential for heterogeneity in MASLD disease manifestations based on sexual differences in the composition of the gut microbiome represents one of the key steps in implementing personalised medicine and subsequently, its direct applicability to a clinical setting.