3.1 Identification and taxonomy of ARGs
The metagenomic library was constructed for evaluating ARGs reservoir in gut of yak, beef and dairy cattle. 40 fecal samples were tested using Illumina platform and obtained 548.3662Gbp of high-quality and average on each sample was 13.0564Gbp. And the detailed data summary was exhibited in Table S2.
According to the annotation results compared with CARD, the species information corresponding to the drug-resistant genes can be analyzed and the dominant flora carrying the drug-resistant genes can be reflected. In comparison of the distribution of the bacterial gene sets and ARGs at phylum level, Firmicutes, Bacteroidetes and Proteobacteria were dominant, accounting for 53% to 33%, 15% to 11% and 5% to 9% in the total amount of bacteria of yak, respectively (Fig.1). For beef cattle, Firmicutes, Bacteroidetes were dominant, accounting for 39% to 32% and 34% to 15% in the total amount of bacteria. And for dairy cattle, Firmicutes, Bacteroidetes were dominant, accounting for 34% to 28% and 41% to 16% in the total amount of bacteria. These asymmetric relationships suggest that Firmicutes are more likely to carry resistant genes in bovine fecal samples.
3.2 Diversity and distribution of ARGs in gut
Approximately 85-99% of microbes cannot be cultivated in the laboratory, which limits our understanding of microbes, including those with ARGs(31). Therefore, the metagenomic approach was used to investigate the distribution and diversity of drug resistance genes in the intestinal tract of bovine. A total of 5701,582 predictive genes were annotated after the original redundancy, and 1688 genes could be annotated by CARD database, including 734 ARG types. To eliminate the differences in the number of ARGs caused by the differences in sample data, the number of ARGs annotated to each Gb of data in different groups was calculated. To be specific, the number of ARG/Gb in yak group(17.83±2.67) was much lower than in group beef(18.28±2.56) and dairy cattle(19.25±1.77)( p<0.001). And the detailed information of ARGs in each sample was shown in Table S3 (a&b&c).
According to the abundance information of ARG in each sample, the top 30 ARGs are used to draw the heatmap (Fig.2). And it's seen that heat map hierarchical clustering based on the relative abundance of each ARGs showed that yak samples were clustered individually and weren't clustered with beef and dairy cattle samples.
3.3 The relative abundance of ARGs in gut
There was some difference of ARGs number in gut among yak, beef and dairy cattle. Specially, the number of ARGs in yak group was lower than in other groups (Fig.3). Additionally, the relative abundance of ARGs was significantly higher in dairy cattle and beef than that in yak (p<0.001). And the detailed relative abundance of ARGs in each sample was shown in Table S4.
Starting from the relative abundance table of resistance genes, the ARGs content and percentage in each sample were calculated, and the ARGs results of the maximum abundance ranking top 20 were screened as diagram in Fig.4a&b. It's seen that there was a difference on the relative abundance of ARGs among yak group and other groups. Specifically, the yak of the relative abundance and relative percentage of the yak were different from those of beef and dairy cattle.
Fig.4 (a) represents the relative abundance of ARGs in each sample, and the unit PPM is the result of amplifying the original relative abundance data by 106 times; (b) represents the relative abundance of top20 ARGs in all ARGs, and others represent the total relative abundance of non-top 20 ARGs.
Fig.5 The top 30 ARG abundance clustering heat map. The right vertical axis is the ARG name, the left vertical axis is the ARG clustering tree, and the corresponding value of the intermediate heat map is the Z value of ARG relative abundance in each row after standardized processing.
It showed that yak samples were clustered individually and weren't clustered with beef and dairy cattle samples in the Fig.5&6. Interestingly, the abundance of tetracycline resistance genes (tetX, tetQ, tet44, tet40, tetO, tetW, tetW/N/W) were much higher in beef and dairy cattle than in yak. And the abundance of aminoglycoside resistance genes (VanRI, VanYA, VanRA, DHA-19), β-lactam resistance genes (OXA-363, PER-7) and multidrug resistance genes (abcA, AcrS) were much higher in yak than in beef and dairy cattle.
Fig.6 Heatmap variations of the relative abundance level of each top30 ARGs subtype among the group of yak, beef and dairy cattle. The right vertical axis is the name of ARG subtype, and the bottom of horizontal axis is the group name. The left vertical axis is cluster tree.
3.4 Shared ARGs among yak faces, beef faces and dairy cattle faces
To detect the distribution of shared ARGs among yak, beef and dairy cattle, the Venn diagram and Ternary plot were constructed. A sum of 318 ARGs were shared by faeces from yak, beef and dairy cattle (Fig.7). And the differences in the abundance of 318 ARGs for different drugs in three groups were also analyzed in ternary plot (Fig.8). The percentage of each certain ARGs in each group of gut is equal to its corresponding abundance which are divided by the sum abundance of ARGs in three groups of gut. Among these shared ARGs, the abundance of tetracycline genes, quinolone genes, β-lactam genes and macrolide genes were much higher in beef and dairy cattle than those in yak.
Fig.7 The Venn diagram showing the number of shared ARGs in yak gut, beef gut and dairy cattle gut
Fig.8 Ternary plot showing the abundance comparison of 318 shared ARGs in yak gut, beef gut and dairy cattle gut. The sum of the abundance for one species ARG in these three types of gut was set as 100%. The percentage of each certain ARG in each gut is equal to its corresponding abundance which is divided by the abundance sum of this ARG in the three groups of gut.
Fig.9 The relative abundance of shared ARGs types assigned to each major antibiotic class among the group of yak, beef and dairy cattle.
Moreover, the results showed that two antibiotic classes resistance genes consisting tetracyclines and β-lactams accounted for >50% of the total ARGs in beef and dairy cattle (Fig.5). While, the results also showed that multidrug accounted for nearly 50% in yak (Fig.5).
3.5 Occurrence and Abundance of MGEs
By comparing with the IS finder database, a total of 153,981 MGEs were annotated, and the differences existed among three groups. Observational studies of mobile transfer elements showed that the abundance of integron in yak group was much higher in beef and dairy cattle (P<0.0001) (Fig.10). But it is shown that there was no obvious difference between beef cattle and dairy cattle (P>0.05). Moreover, the top ten most abundant integron varied from yak to beef and dairy cattle, the AP011957 were the most abundant type respectively in three groups (Fig.10b).
Fig.10 (a)The difference of relative abundance of mobile genetic elements (integron) among yak, beef and dairy cattle is shown in the sigbox. (**represents p<0.005, ***represents p<0.001, **** represents p<0.0001) (b)The difference of the abundance of top ten mobile genetic elements (integron) in the yak, beef and dairy cattle.