High-fat and low-fiber diet elevates the gut resistome: a comparative metagenomic study

Diet altered the gut microbial composition in mice

The study design are shown in Fig. 1. Prior to the transition from the basic maintenance (normal) diet, the dominant genera in the three groups of mice (normal, high-fat, and high-Fiber) were Alistipes, Mucispirillum, Lactobacillus, and Bacteroides, which together constituted 80% of the gut microbiota (Fig. 2A and Supplementary Data 1). After switching from the normal diet to either a high-fat or high-fiber diet, both groups exhibited significant changes in alpha diversity, as evaluated by the Shannon index, and in beta diversity, as determined by principal coordinate analysis (PCoA) (Fig. 3A and Supplementary Fig. 1A). In the high-fat group, the relative abundance of Alistipes decreased from 28.71% to 4.85% and Bacteroides from 7.81% to 2.88%. Conversely, Lactococcus, Enterococcus, Anaerotruncus, and Escherichia increased from 0% to 20.55%, 0% to 0.04%, 4.88% to 5.58%, and 0.07% to 0.25%, respectively (Fig. 2B and Supplementary Data 2). In the high-fiber group, Alistipes decreased from 22.74% to 0.8%, while Parabacteroides and Bacteroides increased from 3.07% to 40.37% and from 7.12% to 14.03%, respectively (Fig. 2B and Supplementary Data 3). Similar trends were observed when comparing the high-fat group to the normal diet group, and the high-fiber group to the normal diet group. Lactococcus, Enterococcus, Anaerotruncus, and Escherichia were identified as biomarkers in the high-fat group, while Parabacteroides and Bacteroides were identified as biomarkers in the high-fiber group (Supplementary Fig. 2). These findings underscored the ability of high-fat and high-fiber diets to alter the taxonomic composition of the gut microbiota.

Microbiota (A, B), ARGs (C, D), VGs (E, F), and MGEs (G, H). Significant between-group differences were detected by LefSe analysis with an LDA threshold score of 2 and a significance level of 0.05. Data are presented as average in bar plots and median with IQR (interquartile range) in box plots; horizontal lines within the boxes represent the first quartile, median, and third quartile, respectively. Whiskers denote the range of values within the first quartile – 1.5× the interquartile range and the third quartile + 1.5× the interquartile range. The Kruskal-Wallis rank sum test was used to determine significance between groups, with ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, and NS no significance.

Alpha and beta diversity of the microbiota (A), ARGs (B), VGs (C), and MGEs (D) were accessed before and after switching from a normal diet to a high-fat or high-fiber diet. Data are presented as median with IQR in box plots; horizontal lines within the boxes represent the firstquartile, median, and third quartile, respectively. Whiskers represent the range of values within the first quartile—1.5× the interquartile range and the third quartile + 1.5× the interquartile range. The Wilcoxon rank sum test was used to detect significance in Shannon index differences, with ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, and NS no significance. Circles represent the 95% confidential interval for the corresponding group in beta diversity, with significant differences detected using PERMANOVA (permutations = 999).

High-fat diet increased the abundances of the resistome, virulome, and mobilome in mice

Prior to dietary transition, the gut resistome in mice mainly comprised genes encoding resistance to tetracycline, vancomycin, macrolide−lincosamide−streptogramin (MLS), bacitracin, and multidrug classes (Fig. 2C, Supplementary Data 4–6). Twenty-one days after switching to a high-fat diet, the total abundance of the resistome increased significantly from 0.14 to 0.25 (ARG/16S rRNA gene ratio; p < 0.001, Supplementary Data 5). In contrast, the high-fiber diet led to a decrease in resistome abundance from 0.14 to 0.09 (p < 0.05) (Fig. 2D and Supplementary Data 6). Distinct patterns in alpha and beta diversity were observed across the groups, indicating significant differences in resistome composition after the dietary change (Fig. 3B and Supplementary Fig. 1B). Notably, the relative abundance of vancomycin resistance genes (vanD, vanG, vanR, and vanS) in the high-fat group increased significantly from 0.019 to 0.071 ARG/16S rRNA gene ratio (p < 0.01, Fig. 2D, Supplementary Fig. 3A, and Supplementary Data 5). Conversely, the high-fiber diet resulted in significant decreases across most ARG categories, including resistance to bacitracin (bacA and bcrA), chloramphenicol (cat), MLS (lsa, vatB, and vatC), and vancomycin (vanD, vanG, vanR, and vanS) (Fig. 2D, Supplementary Fig. 3B, and Supplementary Data 6). These findings suggested that the high-fat diet promoted an increase in resistome abundance, whereas the high-fiber diet reduced it.

Similarly, the virulome—comprising 13 main categories of virulence genes—was significantly affected by diet. In the high-fat group (Supplementary Data 7–9), the virulome abundance increased from 0.56 to 0.91 VG/16S rRNA gene ratio (p < 0.001, Supplementary Data 8), whereas in the high-fiber group, it decreased from 0.58 to 0.50 (p < 0.05) (Fig. 2E, F, Supplementary Fig. 1C, and Supplementary Data 9). Alpha diversity increased significantly in the high-fat group but not in the high-fiber group (Fig. 3C). PCoA revealed distinct changes in beta diversity in the virulome induced by the high-fiber diet (Fig. 3C and Supplementary Fig. 3B). Functional category analysis showed increased abundances of genes associated with adherence, effector delivery system, motility, and immune modulation following the change to a high-fat diet (p < 0.01), along with the emergence of corresponding virulence systems (Fig. 2F, Supplementary Fig. 3C, and Supplementary Data 8). In contrast, the high-fiber diet was associated with decreases in genes related to adherence, biofilm, and stress survival, along with their corresponding virulence systems (Fig. 2F, Supplementary Fig. 3D, and Supplementary Data 9). Overall, these results suggested that the high-fat diet largely altered the virulome by increasing its abundance, while the high-fiber diet exerted a modest reductive effect.

The mobilome, which encompasses all MGEs including plasmids, transposons, and integrons in the microbiome, also showed large changes in response to diet (Fig. 2G, Supplementary Data 10–12). Following the switch from the normal diet, the total relative abundance of the mobilome increased 8-fold (from 0.20 to 1.66 ratio of MGE/16S rRNA gene ratio, Supplementary Data 11) on the high-fat diet, while it decreased from 0.22 to 0.13 on the high-fiber diet (Fig. 2H and Supplementary Data 12). Specifically, the high-fiber diet did not affect plasmid abundance, whereas the high-fat diet increased transposon abundance from 0.09 to 1.26 (MGE/16S rRNA ratio; p < 0.001, Fig. 2H). Further analyses revealed increases in the abundances of intl1, int2, Tn916-orf6, Xis-Tn916, and IS91 in the high-fat group, with corresponding decreases in the high-fiber group (Supplementary Fig. 3E, F, Supplementary Data 11 and 12). Although alpha diversity of the mobilome did not show significant differences after the diet change (Fig. 3D), PCoA of beta diversity indicated distinct gene clustering in the high-fat group compared to the high-fiber and normal groups (Fig. 3D and Supplementary Fig. 1D). Collectively, these findings highlighted the profound impact of a high-fat diet on the mobilome.

Changes in ARGs and VGs were closely related to changes in host bacteria in mice

We used assembled contigs to evaluate the host bacteria carrying specific ARGs and VGs. Prior to the diet change, Bacteroides and Alistipes were identified as hosts for both fosmidomycin resistance gene rosA and tetracycline resistance gene tet37. Anaerotruncus hosted vancomycin resistance genes, while Lactobacillus was the primary host for the multidrug resistance gene mdtG (Fig. 4A and Supplementary Data 13). Most host bacteria–ARG relationships remained unchanged following high-fat diet feeding, however, Lactobacillus, Lactococcus, and Parabacteroides emerged as new hosts for certain ARGs (Supplementary Fig. 4A and Supplementary Data 14). Notably, vanG and vanY were absent in Anaerotruncus after high-fiber diet feeding (Supplementary Fig. 4B and Supplementary Data 15), suggesting a potential elimination of these host bacteria under the high-fiber diet (Fig. 2A). Changes in both the relationship and abundance of host bacteria and ARGs were observed in the high-fat and high-fiber groups (Fig. 4B, C). Regarding VGs, Alistipes was associated with capsular polysaccharide, type III secretion system effectors, and type VI secretion system-related VGs, while Anaerotruncus was linked with capsule-related VGs, and Bacteroides was associated with capsular polysaccharide and capsule-related VGs prior to the diet change (Fig. 4D and Supplementary Data 16). Similar to ARGs, both increased and decreased abundances of VGs and their related host bacteria were observed following high-fat and high-fiber diet feeding, respectively (Fig. 4E, F, Supplementary Data 17 and 18). Additionally, the presence or absences of VGs in their corresponding host bacteria were also noted in both experimental diet groups (Supplementary Fig. 4C, D).

Bacterial taxonomy and ARGs (A–C), bacterial taxonomy and VGs (D–F), MGEs and ARGs (G–I), MGEs and VGs (J–L). Chord plots depict the distribution of ARGs/VGs linked with taxonomic genera or MGEs, with arch size representing the total number of linked items. In heatmap plot, the filled/empty box represents bacteria carrying/non-carrying corresponding ARGs; the top bar plot represents the relative abundance of MGEs or bacterial taxonomy, and the right bar plot represent the relative abundance of ARGs or VGs.

Increased MGEs were associated with increases in the corresponding ARGs and VGs in mice

Given the important role of HGT in enriching the resistome and virulome, we examined the networks between MGEs and ARGs/VGs. The ARGs were mainly associated with intl1, IS91, ISBf10, Tn916-orf6, tnpA-related transposon, and Xis-Tn916 (Fig. 4G and Supplementary Data 19). Most ARGs and their corresponding MGEs exhibited increased abundances following high-fat diet feeding but decreased after high-fiber diet feeding (Fig. 4H, I, Supplementary Fig. 5A,B, Supplementary Data 11, 12, 20 and 21). Notably, Tn916-orf6 was mainly associated with genes encoding resistance to vancomycin and tetracycline, and its abundance increased significantly after high-fat diet feeding (Fig. 4H, Supplementary Fig. 3C, Supplementary Data 5, 11, and 20). The abundance of tnpA-related transposons dramatically increased in the high-fat group, while the associated ARGs only slightly increased, indicating a potential but unclear association between tnpA and other enriched ARGs (Fig. 4H, Supplementary Data 5, 11, and 20). In the high-fiber group, both MGEs and their associated ARGs generally decreased in abundance, except for class A β-lactamase and its host vector ISBf10, which showed increased abundances (Fig. 4I, Supplementary Data 6, 12, and 21).

We also explored the relationship between MGEs and VGs before and after the dietary change (Fig. 4J, Supplementary Fig. 5C, D, Supplementary Data 22–24). Although the relationship between MGEs and VGs appeared more complex than those between MGEs and ARGs, similar trends were observed following the change to high-fat and high-fiber diet feeding (Fig. 4K, L, Supplementary Data 22–24). Notably, the abundances of capsule-related VGs increased after high-fiber diet feeding, while the abundances of their four associated host vectors (Xis-Tn916, tnpA, Tn916-orf6, and IS91) decreased, with the exception of an increase in ISBf10. These results suggested that ISBf10 may serve as the primary host vector for capsule-related VGs (Fig. 4L, Supplementary Data 9, 12 and 24).

Distinct energy source use by gut microbiota in high-fat and high-fiber groups in mice

We analyzed gene function within the microbiome to further explore the relationship between diet and microbial metabolic activity. Genes related to membrane transport, including phosphotransferase system and ABC transporters involved in carbohydrate uptake, were enriched in response to either the high-fat or high-fiber diet intervention (Supplementary Fig. 6). After high-fat diet feeding, the abundances of genes associated with fatty acid degradation and starch and sucrose metabolism significantly increased, likely due to the high lard and maltodextrin content in the high-fat diet (Supplementary Fig. 6A). In contrast, after high-fiber diet feeding, genes involved in glycan degradation and starch and sucrose metabolism became more abundant, likely reflecting the higher starch and cellulose content in the high-fiber diet (Supplementary Fig. 6B).

High-fat diet and obesity increase human gut resistome and mobilome

We investigated the resistome and mobilome in the human population from the perspectives of dietary habits and BMI (Fig. 1). In terms of resistome, genes encoding MLS, beta-lactam resistance and multidrug resistance were predominant in the human gut (Fig. 5A, E, Supplementary Data 25–30). Notably, the total abundance of the resistome in the high-fat population was significantly higher than in the high-fiber and normal diet populations (Supplementary Fig. 7A). Specifically, the relative abundances of genes conferring resistance to cephalosporin (bla_TME-136, 7.31 × 10⁻⁶ vs 0.00, p < 0.05), MLS (lsa, 2.42 × 10⁻³ vs 9.56 × 10⁻⁴, p < 0.05) and aminoglycoside (aph(3”‘)-III, 7.19 × 10⁻³ vs 1.90 × 10⁻³, p < 0.05) were higher in individuals consuming a high-fat diet compared to those on a normal diet (Supplementary Data 31). A similar trend was observed in the obesity group (Supplementary Fig. 7B), where the relative abundance of genes associated with resistance to cephalosporin (bla_TME-127, 3.25 × 10⁻⁶ vs 0.00, p < 0.05), phenicols-lincosamides-oxazolidinones-pleuromutilins-streptogramin A (cfr, 1.54 × 10⁻⁴ vs 1.80 × 10⁻⁵, p < 0.05), and tigecycline (tet(X), 2.68 × 10⁻³ vs 5.95 × 10⁻⁴, p < 0.05) were higher in individuals with obesity compared to those with a healthy BMI (Supplementary Data 32). Although Shannon diversity of ARGs showed no significant difference among the groups (Fig. 5B, F), the Bray-Curtis distance revealed distinct patterns based on dietary habits (Fig. 5C). Specifically, genes encoding resistance to tetracycline, vancomycin, and aminoglycoside resistance were enriched in the high-fat diet/obesity population (Fig. 5D, H). Regarding the mobilome, the total abundance of MGEs was higher in the high-fat/obesity population compared to the normal diet/healthy BMI population (Supplementary Fig. 8).

Dietary habits (A–D) and BMI (E–H). Data are presented as average in bar plots and median with IQR in box plots; horizontal lines within the boxes represent the first quartile, median, and third quartile, respectively. Whiskers denote the range of values within the first quartile—1.5× the interquartile range and the third quartile + 1.5× the interquartile range. The Wilcoxon rank sum test was used to detect significance between groups, with * p < 0.05, ** p < 0.01, *** p < 0.001, and NS no significance. Circles represent the 95% confidential interval for the corresponding group, and PERMANOVA (permutations = 999) was used to detect significant differences between groups.

ARG in dietary habits (A) and BMI categories (B), relationship in dietary habits (C) and BMI categories (D). The Bar plot at the top shows the relative abundance of ARG hosts at genus level. The bar plot on the right shows the relative abundance of ARGs. A correlation value of 1 indicates that the bacteria carry the corresponding ARGs, while a value of 0 indicates the absence of these ARGs.

We also detected the ESKAPE pathogens (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter species) in each group. The results showed that the total abundance of the genera Enterococcus, Staphylococcus, Klebsiella and Pseudomonas (ESKP) was higher in the high-fat and obesity group compared to the normal and healthy group (Supplementary Fig. 9). In terms of ARGs, the ESKP pathogens were the main host of ARGs, and the total abundance of ESKP-carrying ARG contigs in the high-fat population was significantly higher than in the high-fiber and normal diet populations, with an average abundance of 710.25, 89.37 and 133.48 TPM, respectively. Similarly, the total abundance of ESKP-carrying ARG contigs in the obesity population was higher than in the healthy population (1028.72 versus 158.96 TPM) (Fig. 6A, B). Notably, Klebsiella sp., which had the highest abundance among ESKP in different diet populations, mainly carried fosfomycin (fosA) and beta-lactam (bla_SHV and penA) resistance genes (Fig. 6C, D, Supplementary Fig. 9, Supplementary Data 33 and 34). Staphylococcus sp. was associated with tetracycline resistance genes (tetM and tetL), while Enterococcus sp. was associated with tetracycline (tetM, tetL, and tetW), pleuromutilin–lincosamide–streptogramin A (lsa), and florfenicol (fexB) resistance genes. All these ARGs were more abundant in the high-fat diet/obesity population compared to the normal diet/healthy BMI population (Fig. 6C, D, Supplementary Fig. 9, Supplementary Data 33 and 34).