Profiles of ARGs,MRGs, MGEs and pathogens between mouse and human feces
A total of 296 genes were detected among the 6 mixed fecal samples from mice (3) and human (3). The number of genes detected in the mouse and human feces were 157 and 98, respectively (Additional file 1: Table S1), and 71 genes were found in mouse and human feces simultaneously. Although the number of ARGs detected in the mouse feces was much higher than that in the human feces, the relative abundances of ARGs in the mouse feces were much lower than those in the human feces, which were 0.47 and 3.35, respectively (Fig. 1A).
Tetracycline resistance genes (tet genes) were the most abundance ARGs in both mouse and human gut microbiota, which occupied 75.73% and 39.31% of the total ARGs, respectively. The percentages of Macrolide-Lincosamide-Streptogramin (MLSB) and β-lactamase resistance genes in human feces were much higher than those in mouse feces, accounting for 38.57% and 18.29% in human feces, 2.11% and 2.77% in mouse feces, respectively (Fig. 1A and Additional file 3: Figure S1). The proportions of these three major ARGs in human gut microbiota were exactly similar to those of metagenome-wide analysis by Hu et al. [6]. The relatively higher abundance of MLSB and β-lactamase resistance genes in human feces might attribute to the widespread use of these two kinds of antibiotics in the treatment of human diseases. The dominant ARGs, including tetQ, tetW, tet(32), cfxA, ermB, ermF, aacA, strB and sul2, shared the same distribution characteristics (Fig. 1), which accounted for 84% and 94% of the total ARGs in the mouse and human feces.
The relative abundance of total MRGs was extremely lower than that of ARGs, which were 0.0012 and 0.0017 in mouse and human feces, respectively (Fig. 1B). Among the 10 detected metal resistance genes (MRGs), copper and zinc resistance genes, including pcoA, copA and czcA, were the major MRGs in both the mouse and human feces. The common MGEs in human and mouse feces mainly included Int1-1, IS613, tnpA-03 and tnpA-06, which class 1 integron gene (Int1-1) was detected in all the samples and occupied about 10% in both human and mouse feces (Fig. 1D and Additional file 5: Figure S3). In general, integrons have specific structures and ability to capture the ARGs by a site-specific recombination system [33, 34].
In addition, 6 species of pathogens genes, including Enterococci (22S rDNA), P. aeruginosa (ecfX), K. pneumonia (gltA), Staphylococci (mecA), A. baumannii (ompA) and E.coli (uidA), were determined among human and mouse samples. Enterococci, P. aeruginosa, K. pneumonia and E.coli were detected in both human and mouse feces. E.coli and K. pneumonia in the human samples were much higher than those in the mouse samples. These data suggested that the mice could be used as an ideal animal model to study the change of ARGs and 96 genes with relatively high abundance in both human and mice feces were selected from the total 296 genes for further analysis (Additional file 2: Table S2).
DIO mouse model
The body weight of mice administered with the Control and the HFD (high-fat diet) assays showed a significant difference, which those with the HFD assays gained around 38% body weight more than those with the Control assays after 16 weeks of feeding (p < 0.01, Fig. 2A). Food intake and energy intake were measured at the 1st, 5th, 9th, 13th, 16th week of the feeding intervention and the results showed in the Fig. 2B and 2C. These results demonstrated that energy intake of the HFD assays was significantly higher than that of the Control assays (p < 0.05). Histological analysis showed larger adipocyte size in the epididymal fat of the HFD groups than that of the control groups; and there were much more lipid droplets in the liver of the HFD groups (Fig. 2D). These data indicated that DIO mouse model was successful.
Effect of DIO on ARGs, MRGs, MGEs and pathogens
The abundance of total ARGs in the feces of HFD (ARGs/16S rRNA: 1.25) increased almost twice as much as that of the control groups (ARGs/16S rRNA: 0.47) (Fig. 3A). Among the 8 types of detected ARGs, the most significant increase one in the HFD groups was MLSB, which increased 54 times. The ARGs with the most significant increase was ermB (MLSB), which increased 142 times; mefA (MLSB) also went up sharply, which increased 33 times in the HFD groups. erm-type is typical acquired macrolide ARGs, which encode macrolide modifying enzymes. ermB, responsible for acquired resistance to macrolides, lincomycins and streptogramins, is one of the most widespread ARGs in bacterial hosts, such as E. coli, Klebsiella spp., and Shigella spp [35]. The data mentioned above proved that DIO alone could evidently increase the abundance of ARGs, especially ermB, which might reduce the susceptibility of potential pathogenic hosts to the corresponding antibiotics.
The ARGs of tetracycline showed a slightly increase in the HFD groups, which increased to 1.6 times. Although tetracycline resistance genes (tet) existed with the highest proportion in the feces of both human and mice, the effect of DIO seemed not to be significant on them. Since their introduction in the 1950s, tetracyclines have been widely used in human and veterinary medicine, as growth promoters in animal industry, and for prophylaxis in aquaculture [36, 37]. Therefore, tet genes have spread to almost all bacterial genera [36], and they were not significantly affected by the disturbance of intestinal microbiota caused by DIO.
The β-Lactamase enzymes are the foremost mechanism of antibiotic resistance in bacteria leading to the failure of clinical antibiotic treatment [38]. The enzymes, conferring resistance to extended-spectrum cephalosporins and extended-spectrum β-lactamase (ESBLs) are AmpC. In recent decades, there has been a significant increase in the prevalence of clinical strains of AmpC and ESBL-producing E.coli strains, which are usually multidrug-resistant [38]. In this study, β-Lactamase resistance genes increased 2.8 times in the HFD groups, among them, blaSFO gene increased nearly 31 times and ampC-01 gene increased almost 22 times as much as those in the control groups, indicating that DIO could increase ESBLs resistance and threaten clinical antibiotic treatment.
Colistin corresponding ARGs, mcr1, also significantly increased almost 27 times in the HFD groups. Colistin is considered as a last resort antibiotic used to treat bacterial infections caused by carbapenemases and ESBL-producing Enterobacteriaceae including E.coli [39, 40]. Colistin has been widely used in agricultural production and veterinary medicine for decades. Previous studies reported that most mcr1 positive strains were isolated from livestock samples and plasmid-mediated transmission of mcr1 from animal to human seriously threatened the clinical use of colistin [39, 41]. In general, the widely usage of colistin played an important role in dissemination and proliferation of mcr genes in bacteria. However, this study suggested that only obesity also significantly increased colistin resistance, without any exogenous antibiotics selection pressure.
Resistance to multiple drugs was first detected among enteric bacteria, such as Escherichia coli and Salmonella, in the late 1950s. Such strains posed severe clinical problems and cost lives, particularly in developing countries [42]. In this study, multidrug resistance genes amplified nearly 4 times in the HFD groups, especially yceL/mdtH-01, floR, acrA-03 and acrF, which all increased more than 10 times than those in the control groups (Fig. 3J). Therefore, all the results mentioned above suggested that DIO significantly increased the abundance of ARGs in gut microbiota, especially ermB, blaSFO, mcr1 and floR, which brought even greater challenges to clinical antibiotic treatment.
The relative abundance of MRGs nearly doubled in the HFD groups. The pathogenic bacteria increased significantly, especially Enterococci and P. aeruginosa, increasing to 23 times. Relatively, increase of MGEs in the HFD groups was not much evidently (2.4 times), but the dominant genes were changed obviously. In the control groups, IncQoriT gene was the most abundance MGEs, which accounted for 54.16% of the total MGEs. In the HFD groups, tnpA-03 became the most abundance MGEs, accounting for 50.59% of the total MGEs (Additional file 5: Figure S3).
Principal component analysis (PCA) of all the genes showed that only 4 ARGs, including sul2, strB, lnuC and tetG-01 clustered near the control group samples; most of the other ARGs clustered near the HFD group samples, such as ermB, vanHB, mexF, blaSFO, fox5 and so on. All of MRGs, especially pbrT, tcrB, merA and terW, also clustered around the HFD groups. Among the 10 detected MGEs, only IncQoriT showed a significantly positive correlation with the control groups, almost all of the rest assembled around the HFD groups.
Effect of DIO on microbiota community
Microbial community analysis was carried out to further figure out which bacteria were finally affected the fate of ARGs. On the whole, Firmicutes and Bacteroidetes were the most abundance phylum, occupying from 68.57% − 74.79% and 15.79% − 13.21% of all species in the mouse feces (Additional file 9: Figure S7), which were in accordance with those in the feces of healthy people [3]. Actinobacteria occupied much higher proportions in the control groups (11.05%) than that in the HFD groups (2.78%). At genus level, Faecalibaculum (Firmicutes) was the most abundance genus in the control groups, which occupied 63.46% and was much higher than that in the HFD groups (8.67%). Other genus, including Bifidobacteruim (5.82% and 0.45%) and Coriobacteriaceae_UCG-002 (5.89% and 0.76%), exhibited much higher levels in the control groups than those in the HFD groups. Dubosiella (18.90% and 1.57%) and Lactobacillus (13.34% and 2.90%) were the most abundant bacteria in the HFD groups, which exhibited much higher levels than those in the control groups (Fig. 4C).
Redundancy analysis (RDA) between microbial community and the genes showed that almost all of the genes had a significant positive correlation with the dominant microbial community of the HFD groups; only 5 genes (sul2, strB, lnuC, IncQoriT and tetG-01) had a positive correlation with the microbial community of the control groups (Fig. 5A). The ermB and tnpA-03 genes with the most obvious growth in the HFD groups mainly clustered next to the dominant bacteria of the HFD groups, including Lactobacillus, Ruminiclostridium_9 and Ruminococcaceae_UCG-014.
Effect of DIO on microbial metabolites
To further clarify whether the proliferation of ARGs in the HFD groups was affected by microbial metabolism, fecal metabolites of mice in the HFD and the control groups were analyzed by a specific and sensitive UHPLC-MS (ultra high performance liquid chromatography-mass spectrometry) method, which the details described in the supplementary information. Compared with the standards, 204 metabolites (including amino acids, carbohydrates, bile acids and organic acids) were detected and quantified in the mouse feces of both the HFD and the control groups (Additional file 8: Figure S6).
14 major metabolites were selected by interactive forward selection for the RDA with ARGs (Fig. 5B). The results showed that serine and tyrosine closely clustered with the HFD groups; and D-glucose and maltotriose closely clustered with the control groups. In the other words, serine and tyrosine were the dominant metabolites in the HFD groups (HFD: 2.04 and 1.23 µmol/g; Control: 0.98 and 0.58 µmol/g); and D-glucose and maltotriose were the dominant metabolites in the control groups (HFD: 3.84 and 0.13 µmol/g; Control: 20.18 and 2.48 µmol/g) (Additional file 8: Figure S6). Serine and tyrosine also showed a significant positive correlation with tnpA-03 (R2 = 0.594, 0.634, p < 0.05) and the dominant ARGs of the HFD groups (tyrosine and ermB: R2 = 0.622, p < 0.05; floR: R2 = 0.595, p < 0.05) (Fig. 5B).