Bacteria Subsisting on Antibiotics

See allHide authors and affiliations

Science  04 Apr 2008:
Vol. 320, Issue 5872, pp. 100-103
DOI: 10.1126/science.1155157


Antibiotics are a crucial line of defense against bacterial infections. Nevertheless, several antibiotics are natural products of microorganisms that have as yet poorly appreciated ecological roles in the wider environment. We isolated hundreds of soil bacteria with the capacity to grow on antibiotics as a sole carbon source. Of 18 antibiotics tested, representing eight major classes of natural and synthetic origin, 13 to 17 supported the growth of clonal bacteria from each of 11 diverse soils. Bacteria subsisting on antibiotics are surprisingly phylogenetically diverse, and many are closely related to human pathogens. Furthermore, each antibiotic-consuming isolate was resistant to multiple antibiotics at clinically relevant concentrations. This phenomenon suggests that this unappreciated reservoir of antibiotic-resistance determinants can contribute to the increasing levels of multiple antibiotic resistance in pathogenic bacteria.

The seemingly unchecked spread of multiple antibiotic resistance in clinically relevant pathogenic microbes is alarming. Furthermore, an important environmental reservoir of antibiotic-resistance determinants, termed the antibiotic resistome, has been discovered (1, 2). The primary microbial antibiotic-resistance mechanisms include efflux pumps, target gene-product modifications, and enzymatic inactivation of the antibiotic compound (3, 4). Many of the mechanisms are common to several species of pathogens and spread by lateral gene transfer (5). Although enzymatic inactivation is often sufficient to annul the antimicrobial activity of these chemicals, the biochemical processing of these compounds is unlikely to end there, and we hypothesize that the soil microbiome must include a substantial reservoir of bacteria capable of subsisting on antibiotics. Although many bacteria growing in extreme environments (6) and capable of degrading toxic substrates (7) have been previously reported, only a few organisms have been shown to subsist on a limited number of antibiotic substrates (810).

We cultured clonal bacterial isolates from 11 diverse soils (table S2) that were capable of using one of 18 different antibiotics as their sole carbon source. These antibiotics included natural, semisynthetic, and synthetic compounds of different ages and from all major bacterial target classes. Every antibiotic tested was able to support bacterial growth (Fig. 1A and fig. S1). Six out of 18 antibiotics supported growth in all 11 soils, covering five of the eight classes of antibiotics tested. Appropriate controls were performed to ensure that carbon source contamination of the source media or carbon fixation from the air were insignificant (11).

Fig. 1.

Clonal bacterial isolates subsisting on antibiotics. (A) Heat map illustrating growth results from all combinations of 11 soils with 18 antibiotics, where blue squares represent the successful isolation of bacteria from a given soil that were able to use that antibiotic as their sole carbon source at an antibiotic concentration of 1 g/liter. Soil samples labeled F1 to F3 are farm soils and those labeled U1 to U3 are urban soils. P1 to P5 are pristine soils, collected from non-urban areas that had minimal human exposure over the past 100 years (table S2). (B) HPLC traces at 214 nm of representative penicillin- and carbenicillin-catabolizing clonal isolates and corresponding un-inoculated media controls for different time points over 20 or 28 days of growth, respectively.

We obtained clonal isolates capable of subsisting on penicillin and carbenicillin from all the soils tested, and isolates from 9 out of 11 soils that could subsist on dicloxacillin. Representative isolates capable of growth on penicillin and carbenicillin were selected for subsequent analysis by high-performance liquid chromatography (HPLC) (11). Removal of the antibiotics from the media was observed within 4 and 6 days, respectively (Fig. 1B). Mass spectrometry analysis of penicillin cultures has indicated the existence of a penicillin catabolic pathway (9) that is initiated by hydrolytic cleavage of the beta-lactam ring. Beta-lactam ring cleavage, followed by a decarboxylation step, is the dominant mode of clinical resistance to penicillin and related beta-lactam antibiotics (fig. S3) (11).

We isolated bacteria from all the soils tested that grew on ciprofloxacin (Fig. 1A), a synthetic fluoroquinolone and one of the most widely prescribed antibiotics. Clonal isolates capable of catabolizing the other two synthetic quinolones tested, levofloxacin and nalidixic acid, were also isolated from a majority of the soils (Fig. 1A). Previous studies have highlighted the strong parallels between antibiotic-resistance determinants harbored by soil-dwelling microbes and human pathogens (5, 12, 13). The lateral transfer of genes encoding the enzymatic machinery responsible for subsistence on quinolone antibiotics from soildwelling microbes to human pathogens could introduce a novel resistance mechanism so far not observed in the clinic.

Phylogenetic profiling of the clonal isolates (11) revealed a diverse set of species in the Proteobacteria (87%), Actinobacteria (7%), and Bacteroidetes (6%) (Fig. 2 and fig. S2). These phyla all include many clinically relevant pathogens. Of the 11 orders represented, Burkholderiales constitute 41% of the species isolated. The other major orders (>5%) are Pseudomonadales (24%), Enterobacteriales (13%), Actinomycetales (7%), Rhizobiales (7%), and Sphingobacteriales (6%).

Fig. 2.

Phylogenetic distribution of bacterial isolates subsisting on antibiotics. 16S ribosomal DNA (rDNA) was sequenced from antibiotic-catabolizing clonal isolates with the use of universal bacterial rDNA primers. High-quality nonchimeric sequences were classified using the Greengenes database (16), with consensus annotations from Ribosomal Database Project II (17) and National Center for Biotechnology Information taxonomies (18). Phylogenetic trees were constructed using the neighbor-joining algorithm in the ARB program package (19) using the Greengenes-aligned 16S rDNA database. Placement in the tree was confirmed by comparing automated Greengenes taxonomy to the annotated taxonomies of the nearest neighbors of each sequence in the aligned database. Branches of the tree are color-coded by bacterial orders, and clonal isolates are represented as squares.

One explanation for the widespread catabolism of both natural and synthetic antibiotics may relate to organic substructures found in nature. Microbial metabolic mechanisms exist for processing those substructures and may allow for the utilization of the parent synthetic antibiotic molecule. It is interesting that more than half of the bacterial isolates identified in this study belong to the orders Burkholderiales and Pseudomonadales. Organisms in these orders typically have large genomes of approximately 6 to 10 megabases, which has been suggested to be positively correlated to their metabolic diversity and multiple antibiotic resistance (14). These organisms can be thought of as scavengers, capable of using a large variety of single carbon sources as food (15).

We determined the magnitude of antibiotic resistance for a representative subset of 75 clonal isolates (table S3). Each clonal isolate was tested for resistance to all 18 antibiotics used in the subsistence experiments at antibiotic concentrations of 20 mg/liter and 1 g/liter in rich media (11). The clonal isolates tested were on average resistant to 17 out of 18 antibiotics at 20 mg/liter and 14 out of 18 antibiotics at 1 g/liter (Fig. 3). Furthermore, for 74 of the 75 isolates, we found that if a bacterial isolate was able to subsist on an antibiotic, it was also resistant to all antibiotics in that class at 20 mg/liter.

Fig. 3.

Antibiotic-resistance profiling of 75 clonal isolates capable of subsisting on antibiotics. (A) Heat map illustrating the resistance profiles of a representative subset of 75 clonal isolates capable of using antibiotics as their sole carbon source (table S3). Resistance was determined as growth after 4 days at 22°C in Luria broth media containing antibiotic at concentrations of 20 mg/liter (top) and 1 g/liter (bottom). (B) Percentage of clonal isolates resistant to each of the 18 antibiotics. Antibiotics are color-coded by class; the full height of each bar corresponds to the percentage of clonal isolates resistant at a concentration of 20 mg/liter and the solid colored section of each bar corresponds to the percentage of clonal isolates resistant at 1 g/liter. (C) Histogram depicting the distribution of the number of antibiotics that the clonal isolates were resistant to at 20 mg/liter (top) and 1 g/liter (bottom).

Previous work showing that strains from the genus Streptomyces are on average resistant to seven to eight antibiotics at 20 mg/liter has highlighted the importance of producer organisms as a reservoir of antibiotic resistance (2). Here we describe bacteria subsisting on antibiotics as a substantial addition to the antibiotic resistome in terms of both phylogenetic diversity and prevalence of resistance. The bacteria isolated in this study were super-resistant, because they tolerated concentrations of antibiotics >1 g/liter, which is 50 times higher than the antibiotic concentrations used to define the antibiotic resistome (2).

The Greengenes database (16) identified isolates among the bacteria subsisting on antibiotics that are closely related to known pathogens, such as members of the Burkholderia cepacia complex and Serratia marcescens. In principle, relatedness allows for easier transfer of genetic material, because codon usage, promoter binding sites, and other transcriptional and translational motifs are likely to be similar. It is therefore possible that pathogenic microbes can more readily use resistance genes originating from bacteria subsisting on antibiotics than the resistance genes from more distantly related antibiotic producer organisms.

To date, there have been no reports describing antibiotic catabolism in pathogenic strains. However, because most sites of serious infection in the human body are not carbon-source–limited, it is unlikely that antibiotic catabolism confers a strong selective advantage on pathogenic microbes as compared to just resisting them. Therefore, it is likely that only the resistance-conferring part of the catabolic machinery would be selected for in pathogenic strains.

In addition to the finding that bacteria subsisting on natural and synthetic antibiotics are widely distributed in the environment, these results highlight an unrecognized reservoir of multiple antibiotic-resistance machinery. Bacteria subsisting on antibiotics are phylogenetically diverse and include many organisms closely related to clinically relevant pathogens. It is thus possible that pathogens could obtain antibiotic-resistance genes from environmentally distributed super-resistant microbes subsisting on antibiotics.

Supporting Online Material

Materials and Methods

Figs. S1 to S3

Tables S1 to S3


References and Notes

View Abstract

Navigate This Article