The Ribosomal S10 Protein Is a General Target for Decreased Tigecycline Susceptibility

Abstract

Tigecycline is a translational inhibitor with efficacy against a wide range of pathogens. Using experimental evolution, we adapted Acinetobacter baumannii, Enterococcus faecium, Escherichia coli, and Staphylococcus aureus to growth in elevated tigecycline concentrations. At the end of adaptation, 35 out of 47 replicate populations had clones with a mutation in rpsJ, the gene that encodes the ribosomal S10 protein. To validate the role of mutations in rpsJ in conferring tigecycline resistance, we showed that mutation of rpsJ alone in Enterococcus faecalis was sufficient to increase the tigecycline MIC to the clinical breakpoint of 0.5 μg/ml. Importantly, we also report the first identification of rpsJ mutations associated with decreased tigecycline susceptibility in A. baumannii, E. coli, and S. aureus. The identified S10 mutations across both Gram-positive and -negative species cluster in the vertex of an extended loop that is located near the tigecycline-binding pocket within the 16S rRNA. These data indicate that S10 is a general target of tigecycline adaptation and a relevant marker for detecting reduced susceptibility in both Gram-positive and -negative pathogens.

INTRODUCTION

Tigecycline is an FDA-approved antibiotic used to treat complicated skin and skin structure infections (cSSSI), complicated intra-abdominal infections (cIAI), and community-associated bacterial pneumonia (CABP). Increasingly, tigecycline and other drugs of last resort, such as vancomycin and daptomycin, are used “off label” to combat infections arising from multidrug-resistant pathogens (1). As infections with multidrug-resistant pathogens increase, they pose a major threat to public health, with at least two million infections and 23,000 deaths annually in the United States (2). With increased usage, tigecycline resistance will inevitably develop. By understanding how drug resistance emerges, we can be more proactive in identifying the spread of alleles associated with nonsusceptibility or even move toward a preemptive strategy to block the emergence of resistance through new classes of “anti-evolution” drugs that could work with current antibiotics.

The antibiotic tigecycline (TGC) has good in vitro activity against a broad spectrum of multidrug-resistant pathogens and is often used as part of a combination regimen in severe infections caused by multidrug-resistant isolates (3). TGC is a semisynthetic drug belonging to the tetracycline-derived glycylcycline family. Like tetracycline, TGC inhibits the elongation step of translation by binding reversibly to the 16S rRNA and blocking the entry of tRNAs into the A site. The therapeutic success of TGC is due, in part, to its high affinity for the ribosome, which is about 20-fold higher than that of tetracycline (4). TGC has a bulky t-butylglycylamido modification at the 9 position that has been proposed to limit the activity of common tetracycline-resistance mechanisms, such as ribosomal protection (tetM, tetO, etc.) and efflux pumps (tetK, tetA, etc.) (3).

Despite the promising in vitro activity of TGC and its spectrum, resistance to this drug has been observed in clinical strains due to chromosomal mutations that upregulate efflux pumps, including the resistance nodulation-cell division (RND) pumps, the AcrAB-TolC pump, and the multidrug and toxin extrusion (MATE) pumps in Acinetobacter baumannii, Escherichia coli, and Staphylococcus aureus, respectively (5–7). While efflux pumps are major contributors to TGC resistance, there have been a few reports, including one from our group, suggesting that mutations in rpsJ, the gene that encodes the ribosomal S10 protein, may also confer reduced susceptibility to TGC, often in conjunction with other resistance alleles (8–10) Interestingly, these reports included both Gram-positive and -negative pathogens, suggesting that mutations in rpsJ may provide a general mechanism for reduced TGC susceptibility.

Due to the emergence of rpsJ mutations in response to TGC exposure among a variety of Gram-positive and -negative species, we hypothesized that rpsJ is a general target of TGC resistance across many species. To explore this hypothesis, we selected several Gram-positive and -negative strains and used experimental evolution to adapt the populations in vitro to TGC. We used Sanger sequencing of the rpsJ gene of individual clones isolated from the evolved populations to evaluate the frequency and position of TGC-associated S10 mutations within the three-dimensional structure of the ribosome. We determined that 35 out of 47 replicate populations had clones with a mutation in rpsJ, showing that mutations in rpsJ commonly occur in response to TGC exposure.

MATERIALS AND METHODS

The clinically isolated strains Enterococcus faecalis S613 (GCA_000163795.1) (11), E. faecalis R712 (GCA_000163815.1) (11), E. faecalis S613(S10^{R53Q-Δ54-57ATHK}) (SAMN03164156) (10), Enterococcus faecium 105 (12), E. faecium R499 (GCA_000294875.1) (13), methicillin-resistant S. aureus MRSA131 (GCA_000187145.1), and A. baumannii AB210 (GCA_000189655.2) (14) and the lab strains E. faecalis OG1RF (GCA_000172575.2) and E. coli BW25113 (GCA_000750555.1) were used for this study. Enterococcal strains were cultured in 80% lysogeny broth–20% brain heart infusion (LBHI) medium at 37°C with or without agitation. AB210, MRSA131, and BW25113 were cultured in lysogeny broth (LB) on a shaker at 37°C. Before adaptation experiments were carried out, the TGC MICs were determined using the broth dilution method and the breakpoints described by the Clinical and Laboratory Standards Institute (CLSI) (15). The TGC, minocycline (MIN), tetracycline (TET), and daptomycin (DAP) MICs against S613 and S613(S10^{R53Q-Δ54-57ATHK}) were measured using the agar dilution method described by CLSI (15).

For each strain, seven to 11 replicate populations were adapted in parallel to increasing TGC concentrations. At the start of each experiment, an individual colony was used to inoculate 10 ml of broth without TGC. After overnight growth, 10 μl of culture was transferred to 10 ml of broth supplemented with a subinhibitory concentration of antibiotic (typically 0.5× MIC). To determine the next increase in TGC concentration, 10 μl of the culture was tested for growth against two higher concentrations of TGC (typically 1.25 and 1.5 times the current [TGC]). If growth at the higher concentration was absent or poor, the populations were transferred to either the current or only 1.25 times the current concentration. Once a specific population achieved growth at a concentration of TGC at least 5-fold higher than the starting MIC of the strain, eight colonies were selected from each population, and colony PCR was used to amplify rpsJ. The PCR product was treated with ExoSap-IT (Affymetrix, USA), and the samples were sequenced by the Sanger method.

The growth curves of S613 and S613(S10^{R53Q-Δ54-57ATHK}) at five different concentrations of TGC were evaluated (0, 0.0313, 0.063, 0.125, and 0.25 μg/ml TGC). Triplicate cultures of both strains at each drug concentration were measured in parallel. A 100-μl portion of stationary-phase culture was added to 5 ml of LBHI broth and TGC. Cultures were then placed on a shaker at 225 rpm and 37°C. The optical density of the cultures was checked every 30 min for 12 h using a DEN-1 McFarland densitometer (Grant-Bio). The averages from three replicates were plotted, with the standard deviations (see Fig. 3).

FIG 3.

The S10^{R53Q-Δ54-57ATHK} mutation in E. faecalis confers improved growth in the presence of TGC. The growth of S613 and S613(S10^{R53Q-Δ54-57ATHK}) were measured for 12 h at 0 μg/ml TGC (A), 0.031 μg/ml TGC (B), 0.063 μg/ml TGC (C), 0.125 μg/ml TGC (D), and 0.25 μg/ml TGC (E). The averages for three replicates run in parallel were plotted; error bars show standard deviations.

RESULTS

To determine whether acquisition of mutations in rpsJ in response to in vitro exposure to TGC occurs broadly in Gram-positive and -negative species, we used experimental evolution to select for populations with reduced TGC susceptibility. Sanger sequencing of rpsJ was then used to identify the variety and frequency of adaptive mutations. In each experiment, at least eight colonies were selected from each replicate population. Within rpsJ, mutations were identified only in a small loop of the S10 protein comprised of residues 53 to 60. Remarkably, of the 47 replicate populations across five species, 22 had at least one clone with a mutation at position 57, suggesting that this position within S10 is strongly selected for adaptation to TGC (Fig. 1 and 2).

FIG 1.

S10 mutations identified in strains that underwent TGC exposure. The symbol “Δ” indicates a deleted residue. Eight colonies were sampled from each population. For some populations, multiple types of rpsJ mutations were identified between different clones, and therefore the number of observations among the different mutations is sometimes greater than the total number of populations.

FIG 2.

TGC selects for mutations on a loop of S10 that is in close proximity to the 16S rRNA and composes the TGC binding pocket. (A) Image of the crystal structure of TGC bound to the Thermus thermophilus ribosome (PDB 4G5T) (21). Labels on amino acids indicate positions on the loop of S10 where mutations were identified in Gram-positive species (red) or both Gram-positive and -negative species (blue). For clarity, the amino acid numbering corresponds to the E. coli S10 sequence, and only rRNA structure (gray) proximal to the loop is shown. The inset shows the entire structure of S10 with colored spheres at the α-carbon of residues where mutations where identified in Gram-positive (red) or both Gram-positive and -negative (blue) organisms. (B) Alignment of the T. thermophilus S10 sequence and the susceptible S10 sequences for species where mutations in S10 have been identified in response to TGC adaptation. The sequence logo shows the S10 consensus sequence. The black box outlines the residues that compose the loop structure (positions 53 to 61). The blue and red boxes outline the residues where mutations were identified in Gram-positive species (red) or both Gram-positive and -negative species (blue).

Adaptation of E. faecium R499 to TGC.

Ten populations of E. faecium R499 with an MIC of 0.0313 μg/ml TGC were adapted to growth at 0.18 μg/ml TGC over the course of 30 days. All colonies selected from six different populations were found to have a D60Y mutation in S10 (Fig. 1). Interestingly, all colonies from one population had a twelve-nucleotide deletion in rpsJ (S10^{R53Q-Δ54-57ATHK}) that is identical to a deletion identified in strains of E. faecalis S613 that were also adapted in vitro to TGC resistance in a previous study (Fig. 1) (10). The net effect of the deletion is to remove residues 54 to 57 but restore the reading frame of the S10 protein and change arginine 53 to glutamine.

TetM is a ribosomal protection protein that dislodges tetracyclines from the ribosome and is one of the most common tetracycline resistance mechanism found among enterococci (16). To date, clinical isolates expressing tetM have not been shown to confer resistance to TGC, although we showed recently that overexpression of tetM or amplification of gene copy number can produce TGC resistance in E. faecalis during experimental evolution (10). Using PCR, we established that E. faecium R499 does not carry any tetracycline resistance genes. Additionally, MIC testing confirmed that R499 is susceptible to tetracycline (MIC = 0.25 μg/ml), and no genes with sequence homology to tetM were identified in the R499 reference sequence (GCA_000294875.1).

Adaptation of E. faecium 105 to TGC.

To determine if rpsJ mutations arise during TGC exposure when TetM is present, we selected a strain of E. faecium with tetM. We identified the presence of tetM using PCR and confirmed that E. faecium 105 is resistant to tetracycline (MIC = 64 μg/ml). Seven populations of E. faecium 105 with a TGC MIC of 0.0313 μg/ml were adapted to growth at 0.18 μg/ml TGC over the course of 18 days. Colonies isolated from six of the seven populations had a mutation in rpsJ, which demonstrates that rpsJ mutations can arise in E. faecium strains that carry tetM (Fig. 1). All colonies selected from one population retained the ancestral rpsJ sequence.

Adaptation of S. aureus MRSA131 to TGC.

Eleven populations of S. aureus MRSA131 with an MIC of 0.5 μg/ml TGC were adapted to growth at 11.2 μg/ml TGC over the course of 17 days. Among the sampled colonies amino acid K57 of S10 was the most frequently mutated position (Fig. 1). The second most frequently mutated position was D60 of S10 (Fig. 1). Interestingly, two populations had colonies with a double mutation in S10 (Fig. 1).

Adaptation of A. baumannii AB210 to TGC.

Ten populations of A. baumannii AB210 with an MIC of 0.5 μg/ml TGC were adapted using batch cultures to growth at 19 μg/ml TGC over the course of 15 days. Two of the 10 populations had colonies with S10 mutations (Fig. 1). All colonies selected from one population carried a V57L allele, and all colonies from another population carried a V57I mutation (Fig. 1) suggesting that all cells in these two populations had acquired an adaptive mutation in S10. Colonies selected from the remaining populations retained the ancestral rpsJ sequence.

Adaptation of E. coli BW25113 to TGC.

Nine populations of E. coli BW25113 with an MIC of 1 μg/ml TGC were adapted to growth at 5.6 μg/ml TGC over the course of 24 days. All colonies selected from two populations had a V57L mutation and all colonies sampled from another population had a V57D mutation (Fig. 1). Colonies selected from the remaining populations retained the ancestral rpsJ sequence.

Validation of rpsJ mutation in conferring reduced susceptibility to TGC.

While the data presented in Fig. 1 and those reported by Cattoir et al. and Villa et al. show a correlation between mutations in rpsJ and reduced TGC susceptibility, there has been no direct confirmation that mutations in rpsJ alone are able to significantly decrease susceptibility (8, 9). To confirm the causal link between adaptive mutations in rpsJ to reduced TGC susceptibility, we used whole-genome sequencing to isolate a strain of E. faecalis S613 that contained only a mutation in rpsJ (10) Using comparative whole-genome sequencing between S613 and clones isolated from an adapted bioreactor population, a strain with a single mutation in rpsJ was identified, referred to here as S613(S10^{R53Q-Δ54-57ATHK}). We used the agar dilution technique to measure the MICs of TGC, tetracycline (TET), minocycline (MIN), and daptomycin (DAP) against S613 and S613(S10^{R53Q-Δ54-57ATHK}) (Table 1). The TGC MIC against S613(S10^{R53Q-Δ54-57ATHK}) was 4-fold higher than that of S613 (Table 1). Importantly, the TGC MIC against S613(S10^{R53Q-Δ54-57ATHK}) was greater than the FDA nonsusceptibility breakpoint (>0.25 μg/ml TGC) and equal to the EUCAST resistance breakpoint (≥0.5 μg/ml TGC) (Table 1). S613(S10^{R53Q-Δ54-57ATHK}) also had higher MICs than S613 for TET and MIN, suggesting that mutation of S10 confers resistance to multiple classes of tetracyclines (Table 1).

TABLE 1.

MICs against four E. faecalis strains, as determined by the agar dilution method

Strain	MIC (μg/ml)
	TGC	MIN	TET	DAP
OG1RF	0.125	2	4	0.5
R712	0.125	8	32	8
S613	0.125	8	32	0.5
S613(S10R53Q-Δ54-57ATHK)	0.5	16	64	0.5

Mutations to rpsJ did not alter growth rates in the absence of antibiotic, suggesting a low fitness cost.

To further characterize the impact of the S10^{R53Q-Δ54-57ATHK} allele on growth in the presence of TGC, the growth rates of S613 and S613(S10^{R53Q-Δ54-57ATHK}) were evaluated in the presence of different TGC concentrations (0, 0.031, 0.063, 0.125, and 0.25 μg/ml) (Fig. 3). In the absence of antibiotic, S613(S10^{R53Q-Δ54-57ATHK}) and S613 had similar growth rates, suggesting that the S10^{R53Q-Δ54-57ATHK} allele does not confer a serious fitness cost (Fig. 3A). In the presence of TGC S613(S10^{R53Q-Δ54-57ATHK}) was able to grow better than S613 (Fig. 3B to E). At 0.125 and 0.25 μg/ml TGC, no growth was detected for S613, while S613(S10^{R53Q-Δ54-57ATHK}) achieved a high level of growth (Fig. 3D and E). These data demonstrate that mutations in rpsJ can increase resistance above the clinical breakpoints and suggest that such mutations would be persistent in the absence of antibiotic, as the S10^{R53Q-Δ54-57ATHK} allele does not burden the growth rate of the cell under nonselective conditions.

DISCUSSION

By using in vitro experimental evolution, the species included in this study were all adapted to growth in TGC concentrations well above their respective resistance or nonsusceptibility cutoffs. We have demonstrated that changes in rpsJ alone are able to confer increased resistance to TGC in E. faecalis. Additionally, we report the first identification of rpsJ mutations associated with decreased TGC susceptibility in S. aureus, A. baumannii, and E. coli and an expanded spectrum of novel rpsJ mutations in E. faecium. Including data from previous studies, mutations in rpsJ have been identified in a total of six different species that underwent TGC adaptation; three Gram-negative (E. coli, A. baumannii, and K. pneumoniae) and three Gram-positive (S. aureus, E. faecalis, and E. faecium) species (Fig. 1) (8, 9). Comparable mutations in S10 have also been found in tetracycline-resistant Bacillus subtilis (17), Neisseria gonorrhoeae (18), and S. aureus (19). Combined, these data strongly suggest that the ribosomal S10 protein is a general target for decreased TGC susceptibility across many species of bacteria.

We made several unsuccessful attempts to perform allelic replacement of resistance-associated rpsJ alleles into the susceptible ancestral strains of E. coli BW25113 and A. baumannii AB210. In addition, Cattoir et al. reported that attempts to perform allelic replacement at this locus in E. faecium Aus0004 also failed (8). This could be due to the essential role of S10 in translation and transcription; S10 is both a component of the 30S ribosomal subunit and an important transcription factor involved in lambda N-mediated antitermination (20). In the absence of successful allelic replacement, we tested the effect of rpsJ mutations directly using an evolved strain with only a single mutation in rpsJ, as determined by whole-genome sequencing and showed that it had an elevated MIC of 0.5 μg/ml, compared to 0.125 μg/ml for the susceptible ancestral strain (Table 1) (10). This shows that a mutation in S10 alone can reduce susceptibility to TGC. In addition, the high reproducibility of mutant S10 alleles across strains and replicate populations in response to TGC exposure strongly suggests that these mutations commonly play a role in conferring reduced susceptibility to TGC. Overall, we passaged four species (five strains) against TGC, and we identified mutations in rpsJ in 35 of 47 populations and observed six cases where different rpsJ mutants were isolated from a single population (Fig. 1).

Importantly, all of the identified mutations occurred on the tip of an extended loop of the S10 protein that is in close proximity to the 16S rRNA TGC binding site (Fig. 2) (21). The S10 protein does not appear to make any direct contacts to the TGC binding pocket, as none of the atoms are close enough to make either van der Waals contacts or hydrogen bonds. S10 does contact the 16S rRNA that comprises part of the TGC binding pocket. Thus, altering the structure of the S10 loop likely affects the conformation or conformational dynamics of the 16S rRNA, which in turn could reduce the binding affinity of TGC for the ribosome. Alternatively, the adaptive mutations in S10 that reduce TGC susceptibility might favor tRNA entry and binding to the ribosome, reducing the translational inhibition produced by TGC. In either model, mutations in rpsJ that reduce TGC effectiveness appear to be broadly observed in both Gram-positive and -negative organisms, suggesting that it is a general target for the evolutionary selection of mutations leading to reduced TGC susceptibility.

Amino acid position 57 of S10 was a commonly mutated site among both Gram-positive and -negative species after TGC exposure (Fig. 1). Interestingly, in the Gram-negative species, V57 was usually mutated to either a leucine or an isoleucine, with the exception of one population of E. coli BW25113, where a S10^V57D allele was observed. In contrast, the mutations affecting K57 of the Gram-positive species showed greater variability, with no clear pattern (Fig. 1). Also, in the Gram-negative species, the S10 mutations identified all affected position 57, whereas the Gram-positive species displayed more genetic flexibility, with the identification of different mutations at the very tip of the extended loop (positions 53 to 60) (Fig. 2). Particularly, in the Gram-positive species, D60 of S10 was frequently mutated to a tyrosine. In this study, the S10^D60Y allele was independently isolated from 10 different populations, and the same allele was also identified in a previous study, where E. faecium was also adapted in vitro to TGC (8) (Fig. 1). Together, these data suggest that mutations at position 57 of S10 in both Gram-negative and -positive species and the S10^D60Y mutation in Gram-positive species are important for reduced TGC susceptibility.

TGC maintains high efficacy against strains that carry the tetracycline resistance determinant and ribosomal protection protein TetM. However, overexpression of tetM has been linked to TGC resistance in E. faecalis (10). Therefore, we adapted two strains of E. faecium to TGC: one that carries tetM (E. faecium 105) and one without tetM (E. faecium R499). Both E. faecium strains had the same MIC before adaptation (0.0313 μg/ml TGC), but the strain that lacked tetM required 12 passages more than the tetM-carrying strain to reach growth at 0.18 μg/ml TGC. This observation suggests that the presence of tetM could play a potentiating role in the adaptation of E. faecium to TGC nonsusceptibility.

Importantly, we present novel data showing that mutation of rpsJ alone can confer nonsusceptibility to TGC and that rpsJ mutations in S. aureus, A. baumannii, and E. coli emerge repeatedly after exposure to TGC. Additionally, we report an expanded spectrum of novel rpsJ mutations in E. faecium exposed to TGC. The emergence of these alleles in a diverse range of species strongly suggests that mutation of rpsJ is a general strategy to achieve decreased susceptibility to TGC. Previous studies identified mutations affecting the loop of S10 in E. faecium and K. pneumoniae strains that developed nonsusceptibility to TGC in patients undergoing TGC therapy, which highlights the relevance of these mutations to the clinical setting (8, 9). Our observation that the growth rates of E. faecalis S613(S10^{R53Q-Δ54-57ATHK}) are comparable to those of the original susceptible strain in the absence of antibiotic suggests that there is little fitness cost incurred by this adaptive mutation in the organism and that this would be a stable and persistent allele within the population once established. Therefore, we propose that the identification of rpsJ mutations could potentially serve as a useful tool for detecting TGC nonsusceptibility in a variety of pathogens.