Metagenome meta-analysis reveals an increase in the abundance of some multidrug efflux pumps and mobile genetic elements in chemically polluted environments

The United Nations has identified chemical pollution as being one of the three greatest threats to environmental health, along with climate change and the loss of biodiversity (1). It is becoming increasingly clear that environmental degradation is resulting in new emerging infections, and managing this problem must be done through an integrative environment­ health sciences lens (2). Toxic organic or inorganic chemicals can inhibit the activities of microorganisms that carry out crucially important ecosystem functions, eroding their critical role in shaping the natural environment (3, 4). Some types of chemical pollution may be associated with the evolution of human pathogens in directions that make them a greater health risk. For example, clinical isolates of Pseudomonas aeruginosa carry genomic islands and integrative conjugative elements that were likely acquired from co-existing bacteria in the environment (5), such as P. aeruginosa clone ST395. This strain has caused important hospital outbreaks in France and the United Kingdom (6) and possesses a genomic island (GI-7) carrying six genes conferring copper resistance that is identical to that carried by non-pathogenic Pseudomonas spp. isolated from copper-contaminated environments. Petitjean et al. (6) hypothesized that GI-7 was selected for in copper-polluted environments and acquired by clone ST395 through horizontal gene transfer (HGT), allowing it to survive in copper-based hospital water distribution networks. Another example of the consequences of the environmental application of chemicals to human health is the strong association of the use of triazole fungicides for crop protection and the development of fungicide resistance in clinical isolates of Aspergillus fumigatus (7).

Human impacts on the environment may be accelerating the pace of microbial evolution by, for example, increasing the frequency of HGT and decreasing the fidelity of DNA replication (8). Anthropogenic activities that increase the abundance of antibiotic-resistance genes (ARGs) in the environment or that promote their genetic association with mobile genetic elements (MGEs) will increase the likelihood that humans or food animals are exposed to ARGs through contaminated water or foodstuffs and thus increase the probability that ARGs will be acquired by pathogens of human or animal health concern (9, 10). Such activities include environmental contamination with fecal wastes from humans or food-producing animals, antibiotic residues in pharmaceutical manufacturing effluents, and antibiotics used in aquaculture or crop production (11). Furthermore, the global bacterial metapangenome carries an enormous reservoir of ARGs (the “resistome”) and MGE (the “mobilome”) (12, 13), underscoring the potential for diverse ARGs to move into and among human pathogens.

The rapid pace of antimicrobial resistance (AMR) development, and the consequent erosion of the ability to treat and resolve infections, is a seminal public health challenge (14). Mandated authorities and public health policymakers recognize that any effective strategy to mitigate the pace of AMR development must deploy measures across the One Health continuum, including the environment (15, 16).

Numerous types of MGEs can facilitate the exchange of ARGs between environmental bacteria (17). Insertion sequences (IS) are the simplest type of MGE, consisting of an element that carries only the genes necessary for transposition (17). Notably, the IS IS26 plays a major role in the dissemination of ARGs in a wide range of Gram-negative bacteria including pathogens of human or animal health concern (18). In Gram-negative bacteria, IS26 recruits ARGs into the mobilizable gene pool by forming transposons carrying numerous varied ARGs (19). For example, multidrug resistance plasmids carrying IS26 have been detected in Escherichia coli from swine, avian wildlife, retail food vegetables, and agricultural soil, indicating that IS26 is widely disseminated across the One Health continuum (20–23). Integrons are genetic elements capable of capturing a variety genes, including ARGs, as part of gene cassettes (24). Integrons lack the capability to move themselves, however, relying upon other MGEs for mobility (25). Class 1 integrons can carry genes conferring resistance to multiple antibiotics and are widely found in human pathogens but also non-pathogenic environmental bacteria (24, 26). Class 1 integrons have been proposed as an indicator of anthropogenic impacts on the environment as they are more abundant in polluted than unpolluted environments (27).

Fecal pollution entrains enteric microorganisms, antibiotics, and other potentially co-selective chemical residues into the environment (28–31). In fecally polluted environments, the observed increase in ARGs can be adequately explained by the introduction of enteric bacteria, rather than an impact of (co)-selecting chemicals on environmental bacteria (32). However, the potential impact of chemical pollutants on the environmental resistome or mobilome is difficult to discern due to the overlapping presence of genes carried by enteric bacteria.

In the present study, it was hypothesized that various types of chemical pollution may increase the relative abundance of ARGs in microbial communities, either by increasing the frequency of HGT, or by skewing populations to favor individuals that carry ARGs. To explore whether chemical pollution impacts the abundance and composition of MGEs and ARGs, a collection of publicly available metagenome studies were analyzed. The relative abundance of genes related to MGEs and ARGs was examined in metagenomes sequenced from soil, water, or sediment environments polluted with antibiotics, coal- or metal-mining effluents, or petroleum hydrocarbons. These were compared to metagenomes from unpolluted terrestrial or aquatic environments to determine if pollution increased the abundance of ARGs or MGEs and, if so, which ones.

Results from the present study indicate that pollution of terrestrial and aquatic environments with various types of chemicals can increase the abundance of some ARGs and MGEs. In general, pollution increased the abundance of ARGs associated with multidrug efflux pumps and ß-lactam resistance (TEM181 and OXA-541) while genes associated with rifampicin resistance decreased (RNAP subunits rpoB and rpoB2). One possibility for the increase in these ARGs, especially efflux genes, is that these genes confer a fitness advantage in environments polluted with various chemicals, and the relative abundance of bacteria that carry them therefore increases. This is consistent with the composition of microbial communities being profoundly modified in environments impacted by unconventional oil and gas (fracking) wastewater (33), petroleum products (34–36), acid mine drainage (37), surface mining (38), and heavy metals such as copper, zinc, and nickel (39).

Exposure to petroleum in particular resulted in an increased number of unique ARGs encoding efflux pumps that confer resistance to various antibiotics (Table 1). Many organic solvents found in petroleum (e.g., toluene, n-hexane, xylene, and cyclohexane) are toxic to bacteria as a result of their ability to compromise the integrity of biological membranes (40). The toxicity of organic solvents increases with their lipophilicity (40). Intriguingly, all of the efflux genes whose abundance increased in petroleum-polluted sites encoded for several resistance-nodulation-division (RND) efflux systems found in Pseudomonas species, a genus known to be tolerant to organic solvents (40–42). These RND efflux systems included MexAB-OprM, MexEF-OprN, MexJK-OprM, and MexVW-OprM. In P. aeruginosa, the MexAB-OprM and MexEF-OprN efflux systems are major contributors to multidrug resistance in both lab and clinical isolates and also provide some level of tolerance to the organic solvents hexane and p-xylene (43). MexAB-OprM is a major contributor to intrinsic solvent tolerance in P. aeruginosa, while MexEF-OprN and MexCD-OprJ confer low-level tolerance to organic solvents (43). Intriguingly, in vitro exposure of P. aeruginosa to hexane was selected for nalB-type P. aeruginosa mutants overexpressing MexAB-OprM and thus exhibited a multidrug-resistant phenotype (44). Similarly, in Pseudomonas putida strain DOT-T1E, MexB (a homolog of the MexB transporter in P. aeruginosa) also contributes to high-level toluene tolerance (45). The gene mexB is essential for the growth of P. putida strain DOT-T1E in the presence of toluene (45). The biocidal chemical 2-hydroxybiphenyl is a major product of the bacterial desulfurization of dibenzothiophene, an important component of coal and crude oil (45). Pseudomonas azelaica strain HBP-1 is capable of completely mineralizing 2-hydroxybiphenyl (45). This bacterium uses the MexAB-OprM efflux system to adjust the internal concentration of 2-hydroxybiphenyl such that it is available for biodegradation but below the threshold of toxicity. Thus, it may be that enrichment of bacteria capable of metabolizing some of the toxic constituents of petroleum effluent may contribute to the increase in the abundance of genes encoding efflux pumps in the metagenome. While the MexVW (46) and MexJK (47–49) efflux systems have been implicated in antibiotic resistance in P. aeruginosa, their role in organic solvent tolerance is currently unknown and warrants further research.

The rpoB and rpoB2 genes were less abundant in petroleum-contaminated sites than in unpolluted sites. Mutations in rpoB, encoding the β subunit of the RNAP core enzyme, can facilitate adaptation to a variety of environmental and antibiotic stresses (50). Therefore, the identification of single nucleotide polymorphisms in the rpoB and rpoB2 genes would be needed to determine if alleles confer an ARG phenotype in the contaminated sites under study. As a core gene that occurs in all genomes, we have no reasonable explanation for why the relative abundance of these genes would be lower in petroleum-polluted sites compared to unpolluted sites, or pollution with antibiotics, coal- or metal-mining effluents.

The muxB and muxC efflux genes were elevated in metal and coal mining-polluted environments. In P. aeruginosa, the gene muxA encodes for the membrane fusion protein component and functions with two RND components, MuxB and MuxC, and an outer membrane protein OpmB to form the MuxABC-OpmB efflux system (51). Plasmid-mediated overexpression of this system in P. aeruginosa strain YM64 led to elevated minimum inhibitory concentrations (MICs) of aztreonam, macrolides, novobiocin, and tetracycline (51). Mutational inactivation of this efflux system in P. aeruginosa strain PAO1 resulted in elevated levels of a ß-lactamase with increased resistance to ß-lactam antibiotics and decreased virulence in plants and insects (52). More recently, mutational inactivation of the MuxABC-OpmB efflux system in mutants deficient in other RND efflux systems was shown to specifically alter the susceptibility to the aminocoumarin antibiotic novobiocin (53). Intriguingly, MuxABC-OpmB was implicated in the secretion of the coumarin-containing siderophore pyoverdine, suggesting that it plays a role in iron acquisition (53). At present, its potential role in providing a fitness advantage in metal- or coal-mining environments is unknown.

With respect to MGEs, pollution with antibiotic residues or petroleum, but not metal or coal-mining effluent, was associated with an increase in the relative abundance of intI1. The intI1 result was anticipated since it has previously been observed to increase in abundance under anthropogenic pressure (27). IS26, however, was detectably enriched in environments polluted with antibiotic residues, but not other types of chemical pollution. IS26 is associated with plasmids and other MGEs and implicated in the widespread dispersal of ARGs (17, 54). Our results suggest that IS26 is a more important contributor to the dissemination of ARGs that have a role as fitness determinants in antibiotic-polluted environments compared to petroleum, metal, or coal-mining effluent polluted environments. Thus, in these polluted environments, MGEs that carry IS26 are less likely to expand into the bacterial populations. Overall, these results are in agreement with the notion that MGEs are dynamic in response to a diversity of anthropogenic pressures and not only exposure to antibiotics (55). Thus, more research specifically directed at understanding the impacts of various anthropogenic pressures on the dynamics and selective advantages of environmental MGEs, independent of the resistome, is called for.

Except for blaOXA-541 and blaTEM-181, ARGs that increased in relative abundance in chemically polluted metagenomes were not identified in the CARD database as functionally carrying out either target modification or modification of the antibiotic molecule. These general mechanisms of antibiotic resistance are widely used by pathogens to defeat many classes of critically important antibiotics (55). They are often associated with MGEs and amenable to HGT, likely much more so than efflux pumps that are generally chromosomally encoded (55). Although we did not detect an increase in these types of ARGs, the detection limit for shotgun metagenomics is notoriously poor, relative to qPCR for example, and the approach used in the present study is arguably not sensitive enough to detect variations in relative abundance that might still be significant from a risk point of view. Furthermore, the short-read sequences analyzed here do not reveal the genetic context of ARGs, and thus events of concern such as their accrual into multidrug resistance plasmids. We emphasize therefore that our conclusions are limited to commenting on the impact of chemical pollution on the relative abundance of efflux pumps, but the dynamics of other ARGs with particular relevance to human or animal health may still be of concern.

A limitation of the retrospective approach used in the present study is that the polluted and unpolluted sites from which the metagenomes were obtained almost certainly vary in multiple aspects that were not captured by the associated metadata. These include, for example, variation in climate, physical and chemical properties of the aquatic or terrestrial environments sampled, vegetation and soil management in the case of terrestrial systems, history and extent of pollution, and so on. The toxicity of metals will vary with factors that determine their speciation. The toxicity of, and therefore selection pressures exerted by, antibiotics or components of petroleum will vary with factors that determine their sorption to organic or mineral components of soil or sediment and thus bioavailability. Likewise, there were variations in the methods for DNA extraction and sequencing and the bioinformatics tools employed in processing the sequences prior to database submission. Thus, the conclusions from this study must be challenged with prospective experiments. Most obviously by sampling chemically polluted environments alongside appropriate proximal unpolluted sites and comparing them using complementary methods that reveal the evolution of ARGs in the microbial communities, including shotgun metagenomics, transcriptomics, qPCR, or culturomics.

Research funding was obtained competitively through Agriculture and Agri-Food Canada (AAFC) and from the Canadian Federal Genomics Research and Development Initiative on Antimicrobial Resistance (GRDI-AMR). J. Subirats was funded by Juan de la Cierva Formación and Beatriu de Pinós grants.