Mutation in an Unannotated Protein Confers Carbapenem Resistance in Mycobacterium tuberculosis

ABSTRACT

β-Lactams are the most widely used antibacterials. Among β-lactams, carbapenems are considered the last line of defense against recalcitrant infections. As recent developments have prompted consideration of carbapenems for treatment of drug-resistant tuberculosis, it is only a matter of time before Mycobacterium tuberculosis strains resistant to these drugs will emerge. In the present study, we investigated the genetic basis that confers such resistance. To our surprise, instead of mutations in the known β-lactam targets, a single nucleotide polymorphism in the Rv2421c-Rv2422 intergenic region was common among M. tuberculosis mutants selected with meropenem or biapenem. We present data supporting the hypothesis that this locus harbors a previously unidentified gene that encodes a protein. This protein binds to β-lactams, slowly hydrolyzes the chromogenic β-lactam nitrocefin, and is inhibited by select penicillins and carbapenems and the β-lactamase inhibitor clavulanate. The mutation results in a W62R substitution that reduces the protein's nitrocefin-hydrolyzing activity and binding affinities for carbapenems.

INTRODUCTION

In 2014, tuberculosis (TB), a disease resulting from infection with Mycobacterium tuberculosis, killed an estimated 1.5 million people globally (1). An increasing incidence of drug-resistant TB is gradually leaving fewer options for treating this disease. β-Lactams, arguably the most successful class of antibacterials, are seldom considered for treatment of TB due to innate resistance, which is attributable in part to the potent β-lactamase BlaC (2). Among β-lactams, carbapenems are unique because they not only inhibit penicillin-binding proteins and are relatively resistant to the hydrolytic activity of BlaC (3) but also inhibit l,d-transpeptidases, the enzymes that generate the majority of transpeptide linkages in the M. tuberculosis peptidoglycan layer (4–6). Emerging data from the bench (3, 7–11) and the bedside (12–14) demonstrate promising activity of carbapenems in treating drug-resistant TB.

As the consideration of carbapenems for use in TB regimens is slowly increasing, it is only a matter of time before M. tuberculosis strains resistant to carbapenems will emerge. Anticipating the inevitable, we screened M. tuberculosis for genotypic resistance to carbapenems. Meropenem exhibits high potency against M. tuberculosis (3) and has been included in regimens to successfully treat drug-resistant TB (12). Emerging evidence suggests that biapenem is potent against M. tuberculosis: recent reports have described high in vitro potency of biapenem against drug-susceptible (11) and drug-resistant (15) M. tuberculosis strains and antituberculous activity in a mouse model of TB (16). We used meropenem and biapenem as representative carbapenems and selected M. tuberculosis spontaneous mutants in their presence. Following identification of the mutations that conferred M. tuberculosis resistance to these drugs by whole-genome sequencing, we characterized the identified allele.

RESULTS

A single nucleotide polymorphism (SNP) is common among meropenem- and biapenem-resistant mutants.

We used meropenem and biapenem at 10× the respective MIC₉₀s to select spontaneous mutants resistant to these β-lactams. Plates containing Middlebrook 7H11 enriched medium with agar produced distinct colonies without any sign of satellite colonies after 5 weeks of incubation at 37°C. Although we expected these colonies to grow robustly in Middlebrook 7H9 enriched broth supplemented with 10× the MIC of the carbapenem that was used to select them, we were able to rescue these mutants at only 4× the MIC of the drug (see Table S1 in the supplemental material). Carbapenems are known to undergo slow hydrolysis in the presence of water (17). In addition to water, both media used for growth of M. tuberculosis were supplemented with oleic acid-albumin-dextrose-catalase. Whether components of the medium contributed to the degradation or sequestration of the carbapenems is not known. We hypothesize that the carbapenems slowly degraded over time in the selection medium and that their actual concentration was closer to 4× MIC than to the initial concentration of 10× MIC. An observation where initial inhibition of Mycobacterium abscessus in Middlebrook 7H9 enriched broth containing carbapenems was followed by growth (11) provides precedent to our hypothesis.

Analysis of the whole-genome sequences of the resistant mutants revealed a single nucleotide polymorphism (SNP) (T → A) in a region between the divergently transcribed genes Rv2421c and Rv2422 (Fig. 1a and b; Table S2). This identical mutation occurred in all 12 resistant isolates, including six resistant to meropenem and six resistant to biapenem. We further confirmed this SNP in each strain by PCR and Sanger sequencing.

FIG 1.

Locus of crfA in the M. tuberculosis H37Rv genome and its expression. (a and b) The proposed coordinates for crfA DNA and the DNA sequences of the parent and mutant strains are shown. The mutation is highlighted with an asterisk. The coordinates in this figure are for the parent M. tuberculosis strain (H₃₇Rv-JHU). (c) Purified CrfA on SDS-PAGE. (d) Western blotting using polyclonal antibodies against recombinant CrfA. Lane 1, recombinant CrfA with two additional residues (glycine and histidine) at the N terminus (20-pmol load); lanes 2 and 5, culture filtrates; lanes 3 and 6, cytosolic fraction; lanes 4 and 7, cell wall fraction. (e) CrfA protein sequences with secondary structure for the parent and mutant strains.

According to the current annotation (18), a 274-bp intergenic region exists between Rv2421c and Rv2422. Rv2421c, an essential gene (19), encodes NadD, a putative nicotinate-nucleotide adenylyltransferase, and Rv2422, a nonessential gene, encodes a hypothetical protein of unknown function. Further study of this region revealed an open reading frame (ORF) that is 393 bases long and partially overlaps the 5′ coding region of Rv2421c (Fig. 1a). A recent study (20) focusing on previously unidentified transcripts and small proteins reported a transcription start site 164 bases upstream of this ORF, providing additional evidence for the existence of a previously unidentified gene. We refer to this ORF as carbapenem resistance factor A (crfA).

Next, we assessed if CrfA was expressed in M. tuberculosis. For this task, we used antibodies raised against CrfA expressed and purified from Escherichia coli (Fig. 1c) and observed a dominant band for the membrane and cytosolic fractions, but not the secreted fraction, for M. tuberculosis at the exponential and stationary phases of growth (Fig. 1d). Therefore, crfA encodes a protein that is abundantly expressed during both growth phases and readily detectable in membrane and cytosolic fractions of M. tuberculosis. The T184 → A SNP produces a tryptophan62-to-arginine (W62R) substitution in a putative α-helical region of CrfA (Fig. 1e). crfA is highly conserved not only in all M. tuberculosis strains, including extensively drug-resistant clinical isolates, but also in M. bovis and M. bovis BCG; however, the T → A SNP is not present in >100 M. tuberculosis genomes deposited in GenBank.

CrfA interacts with β-lactams.

We hypothesized that CrfA either interacts directly with meropenem and biapenem or affects binding of these drugs to their target(s). To test these hypotheses, we measured thermodynamic parameters associated with protein-ligand interactions and observed the heat of complex formation for both the wild-type (CrfA_wt) and mutant (CrfA_W62R) proteins interacting with biapenem (Fig. 2a; Table S3). This interaction is consistent with a single-site binding model. The binding isotherm for biapenem-CrfA revealed a favorable binding enthalpy (−ΔH), indicating that the binding is driven by polar interactions. These data also showed that the W62R substitution reduced the favorable binding enthalpy (−ΔH) 1.6-fold and, consequently, the overall affinity for biapenem 3.4-fold. This confirms that W62 affects binding of biapenem to CrfA. The W62R mutation also changes the entropy from negative to positive, which suggests a conformational change in the active site of the CrfA_W62R mutant.

FIG 2.

Profile of β-lactam interaction with CrfA. Isothermal titration calorimetric measurements were made with biapenem (a) and meropenem (b) and the wild-type (CrfA_wt) and mutant (CrfA_W62R) proteins. (c) Comparison of rates of nitrocefin hydrolysis by the wild-type (CrfA_wt) and mutant (CrfA_W62R) proteins. (d and e) Comparisons of rates of nitrocefin hydrolysis by CrfA (2 μM), BlaC (0.01 μM), and Ldt_Mt2 (10 μM). (f) Assessment of inhibition of nitrocefin hydrolysis activity of CrfA by clavulanate, a β-lactamase inhibitor. (g and h) Comparisons of inhibition of nitrocefin hydrolysis activity of CrfA by various β-lactams at 80 μM. The control included CrfA (2 μM), nitrocefin, and no drug. (i and j) Rates of nitrocefin hydrolysis in the presence of various concentrations of biapenem and methicillin. The control included CrfA (2 μM), nitrocefin, and no drug.

Binding of meropenem to CrfA_wt was readily detectable (Fig. 2b; Table S3): this interaction was not enthalpy driven but entropy driven (+ΔS), suggesting hydrophobic interactions of meropenem with the protein. The CrfA_W62R mutant did not show defined isotherm values with meropenem, suggesting a lack of binding. Next, we assessed if CrfA interacts with other subclasses of β-lactams in addition to carbapenems. The penicillins ampicillin, amoxicillin, methicillin, and oxacillin showed a moderate binding isotherm with CrfA which is thermodynamically driven, but the cephalosporins cefoxitin, ceftriaxone, cefotaxime, and cephalexin showed no significant binding. We also assessed the binding of tebipenem, a carbapenem, and of faropenem, a penem, and neither showed any detectable interaction with CrfA (Fig. S1).

CrfA hydrolyzes a β-lactam substrate, and the W62R mutation decreases this activity.

We assessed if CrfA also has β-lactamase activity. We used nitrocefin, a chromogenic β-lactam that permits simple monitoring of β-lactamase activity based on β-lactam ring opening (21). While CrfA hydrolyzed nitrocefin with a weak catalytic efficiency of ∼71.8 M⁻¹ s⁻¹, this rate was further reduced to ∼9.96 M⁻¹ s⁻¹ for the CrfA_W62R mutant protein (Fig. 2c; Table S4), giving a >7-fold reduction in catalytic efficiency. In concordance with the iTC₂₀₀ calorimetry data, the W62A mutation affected the rate of β-lactam hydrolysis by CrfA. We compared the nitrocefin hydrolysis activity of CrfA with those of β-lactam binding proteins from M. tuberculosis, namely, β-lactamase (BlaC) and an l,d-transpeptidase (Ldt_Mt2), and found that BlaC was ∼10,000-fold more catalytically efficient than CrfA and Ldt_Mt2 in hydrolyzing nitrocefin (Fig. 2d and e; Table S4).

Clavulanate and select β-lactams inhibit CrfA.

Clavulanate is a β-lactamase inhibitor: unlike that of other β-lactams, following acylation of the catalytic serine, the rate of clavulanate adduct hydrolysis to regenerate the protein is extremely low (22). Incubation of CrfA with clavulanate completely abolished its ability to hydrolyze nitrocefin (Fig. 2f). Next, we studied if β-lactams, by virtue of their ability to interact with CrfA, can inhibit this protein. We measured the level of nitrocefin hydrolysis by CrfA_wt in the presence of β-lactams of the penicillin class (ampicillin, amoxicillin, methicillin, and oxacillin), cephalosporin class (cefotaxime, cefoxitin, ceftriaxone, and cephalexin), carbapenem class (meropenem, biapenem, and tebipenem), and penem class (faropenem). In general, carbapenems (with the exception of meropenem) consistently inhibited CrfA, and the cephalosporins failed to do so. To our surprise, methicillin and oxacillin, which are classified as penicillinase-resistant penicillins, inhibited CrfA, but ampicillin and amoxicillin, which belong to the aminopenicillin class, showed no inhibition at all (Fig. 2g and h). Because members of the carbapenem and penicillin classes inhibited CrfA, we chose biapenem and methicillin as representatives to validate their inhibitory potencies. Increasing concentrations of biapenem and methicillin reduced the nitrocefin-hydrolyzing activity of CrfA, thereby demonstrating a dose-dependent inhibition of CrfA (Fig. 2i and j).

DISCUSSION

Carbapenems and other β-lactams are known to bind to and inhibit d,d-transpeptidases, penicillin-binding proteins, β-lactamases, and l,d-transpeptidases (6, 23, 24). In addition to these targets, multidrug efflux pumps, such as the one encoded by Rv0194 (25), and the thick hydrophobic mycolic acid layer in the cell wall of M. tuberculosis (26) are known to contribute to β-lactam resistance. A study that used synthetic lethality to identify M. tuberculosis genes associated with resistance to imipenem revealed 74 genes, the majority of which were involved in cell wall biosynthesis (27). Changes in outer membrane permeability and the presence of efflux pumps leading to resistance to β-lactams are also well documented for other bacteria unrelated to M. tuberculosis (23, 28), suggesting the existence of a conserved type of molecular machinery in bacteria to withstand β-lactam insult. Therefore, we expected carbapenem resistance in M. tuberculosis to arise from mutation in these protein classes or drug efflux systems. The mutation in CrfA and the lack of mutation in the aforementioned protein classes in all meropenem- and biapenem-resistant mutants were surprising. Generating the CrfA_W62R mutant in wild-type M. tuberculosis by using targeted mutagenesis to demonstrate resistance to carbapenems and complementing this strain with crfA_wt to restore sensitivity would provide additional validation. However, the whole-genome sequence data were very compelling and sufficient to attribute carbapenem resistance to the W62R mutation. Also, Rv2421c, whose 5′ coding region and untranslated region (UTR) overlap the 5′ region of crfA, is an essential gene. Therefore, generating an M. tuberculosis strain with a mutation in this region likely requires further characterization of the locus and was therefore beyond the scope of the current study. β-Lactams resemble the d-alanyl–d-alanine end of the pentapeptide substrate of peptidoglycan biosynthesis (23). It is possible that peptides whose structures are mimicked by β-lactams are native substrates of CrfA. Western blotting using antisera against CrfA revealed two bands, one that corresponds to CrfA and a weaker one of higher molecular weight only in the cell wall fractions (Fig. 1d). It is possible that the latter band represents the fraction of CrfA that is covalently bound to peptide substrates. Therefore, we hypothesize that the β-lactam-binding and -hydrolyzing activities of CrfA are incidental and that its native function is related to an unknown vital aspect of M. tuberculosis cellular physiology. It is likely that by reducing the affinity for the carbapenems meropenem and biapenem, the W62R mutation allows CrfA to escape binding by these drugs while maintaining its primary cellular function, rendering the mutant M. tuberculosis more resistant to these carbapenems. The lack of sequence similarity to l,d-transpeptidases, penicillin-binding proteins, and BlaC suggests a convergent evolution of CrfA in relation to binding to β-lactams. Further genetic and molecular studies are needed to validate CrfA as a target of β-lactams and also to understand its native function in M. tuberculosis.

MATERIALS AND METHODS

Bacterial strains and growth and selection conditions.

M. tuberculosis strain H37Rv, whose genome has been sequenced and annotated, was used as the parent strain. M. tuberculosis H37Rv was grown at 37°C with constant shaking in Middlebrook 7H9 broth supplemented with 10% (vol/vol) oleic acid-albumin-dextrose-catalase to exponential phase (A₆₀₀, ∼0.8). Spontaneous mutants were selected by plating 1 ml of the H37Rv culture per plate on Middlebrook 7H10 agar supplemented with 100 μg/ml meropenem or 50 μg/ml biapenem. These concentrations represent 10 times the respective MIC₉₀s determined in the same growth medium lacking agarose (11). Colonies appeared on the plates between 4 and 5 weeks following incubation. The frequencies of spontaneous mutation conferring resistance to meropenem or biapenem were 1 × 10⁻⁸ and 5.5 × 10⁻⁸, respectively. One clone, a derivative of H37Rv referred to as H₃₇Rv-JHU, was included as the parent control. H₃₇Rv-JHU was isolated from a carbapenem-free control plate at the same time as the mutants. Axenic cultures of each isolate were prepared, and the MIC₉₀ was determined using a standard broth microdilution assay (11) to verify that the selected mutants were stably resistant to the respective drugs. Isolates from distinct plates were chosen for whole-genome sequencing.

Whole-genome sequencing and analysis.

Genomic DNA was extracted from six meropenem-resistant isolates, six biapenem-resistant isolates, and the parent strain M. tuberculosis H₃₇Rv-JHU as described previously (29) and subjected to whole-genome sequencing using an Illumina MiSeq 2 × 300 platform (Eurofins Genomics). The raw sequences of H₃₇Rv-JHU were mapped onto the reference M. tuberculosis H37Rv genome (GenBank accession no. NC_000962) by using BWA-mem. The properly aligned read pairs were then used to generate a parent genome for the mutants. Next, raw sequences belonging to the mutants were mapped onto the parent genome and analyzed for insertions, deletions, inversions, and nucleotide polymorphisms by using FreeBayes. The read coverage means for our strains were between 82 and 115. The inferred variations with solid quality scores (>100) and high fractions (>0.80) are listed in Table S1 in the supplemental material.

Cloning, expression, and purification.

crfA was amplified from the parent strain H37Rv and the resistant strains by use of primers ATTGCGGTACCGTGCTCGGCGGCGGAGACCTG and CAATACTCGAGTCACCGGTAGGCCATGCCG and cloned into the KpnI and XhoI sites of pET32a+TEV. The resulting plasmids, p1601KA and p1603KA, allow expression of wild-type CrfA (CrfA_wt) and the W62R mutant (CrfA_W62R), respectively. Similarly, fragment 29-307 of BlaC was amplified from the H37Rv genome by use of primers ATTGCCATATGAGCGGGGCCCGTCCGGCATC and CAATACTCGAGTTACTATGCAAGCACACCGGCAACGCA and cloned into the NdeI and XhoI sites of pET28a+TEV, resulting in the expression plasmid p175KA. E. coli DH5α and BL21δε3 (NEB Labs) were used for cloning and manipulation of plasmids and for overexpression of proteins, respectively. For CrfA_wt and CrfA_W62R overexpression, BL21δε3 cells were grown for 2 to 2.5 h at 37°C to an A₆₀₀ of 0.5 to 0.6 and induced with 0.2 mM isopropyl-β-d-1-thiogalactopyranoside (IPTG) and grown overnight at 16°C. For protein purification, cells were lysed in 50 mM Tris, pH 8.0, 400 mM NaCl, 5% glycerol, 1.0 mM phenylmethylsulfonyl fluoride (PMSF), 1 mM dithiothreitol (DTT), and protease inhibitor cocktail (Sigma-Aldrich). The lysed cells were centrifuged at 13,000 rpm for 45 min to clear the cell debris, and the supernatant was passed through a 5-ml HiTrap Ni-nitrilotriacetic acid (Ni-NTA) column to purify the TRX-His₆-tagged CrfA protein. The TRX-His₆ tag from the purified protein was cleaved with tobacco etch virus (TEV) protease and again passed through the Ni-NTA column to remove the tag and uncleaved tagged protein. The protein was further purified by Superdex 200 10/300 size exclusion chromatography, concentrated to 2 mg/ml, and stored at −80°C.

Verification of CrfA in M. tuberculosis.

The presence of CrfA in M. tuberculosis was studied using Western blot analysis with polyclonal antibodies generated in rabbits (Spring Valley Laboratory) against purified recombinant CrfA_wt. An M. tuberculosis culture grown in 7H9 broth was used to obtain cells at the exponential (A₆₀₀ = 0.4) and stationary (A₆₀₀ = 1.5) phases of growth. The M. tuberculosis pellet and culture supernatant were obtained by centrifuging cultures at 3,500 rpm for 15 min at 4°C. The culture supernatant was passed through a 0.22-μm filter, and the culture filtrate was concentrated using 3-kDa-cutoff Centricone devices (Vivaspin 20). The M. tuberculosis cell pellet was resuspended in lysis buffer (phosphate-buffered saline [PBS], pH 7.2, 5% glycerol, 0.2 mM PMSF, 5 mM β-mercaptoethanol, and EDTA-free inhibitor cocktail [Sigma-Aldrich]), lysed by bead beating for 10 cycles of 20 s per cycle at 400 rpm, and centrifuged at 16,000 rpm for 15 min to obtain the cytosolic and cell wall fractions. The cell wall fractions were further washed with lysis buffer containing 1% Triton X-100 and centrifuged at 25,000 rpm for 15 min to remove the remaining cytosolic proteins. The culture filtrate, cytosolic, and cell wall fractions from both the exponential and stationary growth phases were analyzed by Western blot analysis using the anti-CrfA polyclonal antibodies (1:1,000). A horseradish peroxidase-conjugated anti-rabbit secondary antibody (Cell Signaling) at a 1:5,000 dilution was used for detection with an enhanced chemiluminescence (ECL) substrate per the manufacturer's protocol (Santa Cruz Biotechnology). Purified recombinant CrfA was used as a control.

Isothermal titration calorimetry studies.

Both CrfA_wt and CrfA_W62R were dialyzed against 50 mM Tris, pH 8.0, 150 mM NaCl, and 1 mM DTT. Ligands were also solubilized in the same buffer, and 45 μM protein and 1 mM ligands were used. Calorimetry experiments were performed using a microcalorimeter (iTC₂₀₀; MicroCal, MA) at 24°C for titration of the ligand (1 0.5-μl and 16 2.5-μl injections at 200-s intervals, with a stirring speed of 750 rpm). The MicroCal data analysis software Auto-iTC200 was used to determine the thermodynamic properties of ligand binding, using nonlinear least-squares fitting assuming a single-site model.

Nitrocefin hydrolysis assay.

Nitrocefin (CalBioChem) is a chromogenic β-lactam whose hydrolysis can be monitored at 496 nm. The CrfA protein (2 μM) was incubated with 2 to 800 μM nitrocefin in a 100-μl reaction volume at room temperature in 50 mM HEPES-morpholineethanesulfonic acid (MES)-diethanolamine, 300 mM NaCl, pH 6.0. The concentration (c) of hydrolyzed nitrocefin was calculated from its absorbance (A), molar extinction coefficient (ε) (20,500), and path-length (l) (0.5 cm) values by using Beer's law, i.e., A = εcl. To assess if β-lactams inhibited nitrocefin hydrolysis by CrfA, a 2 μM concentration of this protein was added to a reaction mixture containing 2 to 600 μM nitrocefin and 80 μM various β-lactams, and the rate of hydrolyzed nitrocefin was determined spectroscopically. For comparative nitrocefin hydrolysis studies, CrfA (2 μM), BlaC (0.01 μM), and Ldt_Mt2 (10 μM) were used. Data were fitted by nonlinear regression using the Michaelis-Menten equation, with 95% confidence.