Search

Although pathogenic bacteria are the targets of antibiotics, these drugs also affect hundreds of commensal or mutualistic species. Moreover, the use of antibiotics is not only restricted to the treatment of infections but is also largely applied in agriculture and in prophylaxis. During this work, we tested the hypothesis that there is a correlation between the number and the genomic location of antibiotic resistance (AR) genes and virulence factor (VF) genes. We performed a comprehensive study of 16,632 reference bacterial genomes in which we identified and counted all orthologues of AR and VF genes in each of the locations: chromosomes, plasmids, or in both locations of the same genome. We found that, on a global scale, no correlation emerges. However, some categories of AR and VF genes co-occur preferentially, and in the mobilome, which supports the hypothesis that some bacterial pathogens are under selective pressure to be resistant to specific antibiotics, a fact that can jeopardize antimicrobial therapy for some human-threatening diseases.

ABSTRACT Human beings have used large amounts of antibiotics, not only in medical contexts but also, for example, as growth factors in agriculture and livestock, resulting in the contamination of the environment. Even when pathogenic bacteria are the targets of antibiotics, hundreds of nonpathogenic bacterial species are affected as well. Therefore, both pathogenic and nonpathogenic bacteria have gradually become resistant to antibiotics. We tested whether there is still cooccurrence of resistance and virulence determinants. We performed a comparative study of environmental and human gut metagenomes from different individuals and from distinct human populations across the world. We found a great diversity of antibiotic resistance determinants (AR diversity [ARd]) and virulence factors (VF diversity [VFd]) in metagenomes. Importantly there is a correlation between ARd and VFd, even after correcting for protein family richness. In the human gut, there are less ARd and VFd than in more diversified environments, and yet correlations between the ARd and VFd are stronger. They can vary from very high in Malawi, where antibiotic consumption is unattended, to nonexistent in the uncontacted Amerindian population. We conclude that there is cooccurrence of resistance and virulence determinants in human gut microbiomes, suggesting a possible coselective mechanism. IMPORTANCE Every year, thousands of tons of antibiotics are used, not only in human and animal health but also as growth promoters in livestock. Consequently, during the last 75 years, antibiotic-resistant bacterial strains have been selected in human and environmental microbial communities. This implies that, even when pathogenic bacteria are the targets of antibiotics, hundreds of nonpathogenic bacterial species are also affected. Here, we performed a comparative study of environmental and human gut microbial communities issuing from different individuals and from distinct human populations across the world. We found that antibiotic resistance and pathogenicity are correlated and speculate that, by selecting for resistant bacteria, we may be selecting for more virulent strains as a side effect of antimicrobial therapy.

Human metagenomes with a high diversity of virulence genes tend to have a high diversity of antibiotic-resistance genes and vice-versa. To understand this positive correlation, we simulated the transfer of these genes and bacterial pathogens in a community of interacting people that take antibiotics when infected by pathogens. Simulations show that people with higher diversity of virulence and resistance genes took antibiotics long ago, not recently. On the other extreme, we find people with low diversity of both gene types because they took antibiotics recently—while antibiotics select specific resistance genes, they also decrease gene diversity by eliminating bacteria. In general, the diversity of virulence and resistance genes becomes positively correlated whenever the transmission probability between people is higher than the probability of losing resistance genes. The positive correlation holds even under changes of several variables, such as the relative or total diversity of virulence and resistance genes, the contamination probability between individuals, the loss rate of resistance genes, or the social network type. Because the loss rate of resistance genes may be shallow, we conclude that the transmission between people and antibiotic usage are the leading causes for the positive correlation between virulence and antibiotic-resistance genes.