Characterization of Ceftazidime Resistance Mechanisms in Clinical Isolates of Burkholderia pseudomallei from Australia

Burkholderia pseudomallei is a Gram-negative bacterium that causes the serious human disease, melioidosis. There is no vaccine against melioidosis and it can be fatal if not treated with a specific antibiotic regimen, which typically includes the third-generation cephalosporin, ceftazidime (CAZ). There have been several resistance mechanisms described for B. pseudomallei, of which the best described are amino acid changes that alter substrate specificity in the highly conserved class A β-lactamase, PenA. In the current study, we sequenced penA from isolates sequentially derived from two melioidosis patients with wild-type (1.5 µg/mL) and, subsequently, resistant (16 or ≥256 µg/mL) CAZ phenotypes. We identified two single-nucleotide polymorphisms (SNPs) that directly increased CAZ hydrolysis. One SNP caused an amino acid substitution (C69Y) near the active site of PenA, whereas a second novel SNP was found within the penA promoter region. In both instances, the CAZ resistance phenotype corresponded directly with the SNP genotype. Interestingly, these SNPs appeared after infection and under selection from CAZ chemotherapy. Through heterologous cloning and expression, and subsequent allelic exchange in the native bacterium, we confirmed the role of penA in generating both low-level and high-level CAZ resistance in these clinical isolates. Similar to previous studies, the amino acid substitution altered substrate specificity to other β-lactams, suggesting a potential fitness cost associated with this mutation, a finding that could be exploited to improve therapeutic outcomes in patients harboring CAZ resistant B. pseudomallei. Our study is the first to functionally characterize CAZ resistance in clinical isolates of B. pseudomallei and to provide proven and clinically relevant signatures for monitoring the development of antibiotic resistance in this important pathogen.


Introduction
Burkholderia pseudomallei, the etiologic agent of melioidosis, is a saprophytic bacterium that is commonly found in surface waters and soil of Australia and Thailand. There is no effective vaccine against melioidosis, and reducing mortality from infection is based on effective antimicrobial therapy combined with supportive care. B. pseudomallei has a large accessory genome [1,2] and is intrinsically resistant to many antibiotics, including gentamicin, streptomycin, rifampicin, erythromycin and many b-lactams [3,4]. There are several different mechanisms of antibiotic resistance in B. pseudomallei, including multi-drug efflux pumps [5,6], enzymatic inactivation [7,8], impermeability of the bacterial cell membrane [9] and mutations in the antibiotic target site [10]. This impressive array of intrinsic resistance and broad-spectrum mechanisms limits the number of treatment options for melioidosis. Successful treatment of melioidosis is protracted and typically involves two stages comprising an intravenous (IV) phase followed by prolonged oral eradication therapy [11]. In Australia, the IV drug of choice for treating melioidosis is ceftazidime (CAZ), although the carbapenem drugs meropenem or imipenem are used for severe infection or in the event of treatment failure. The oral eradication phase consists of trimethoprim-sulfamethoxazole (TMP-SMX) (in combination with doxycycline in Thailand) or amoxicillinclavulanate (AMC), and is given for up to six months because of the frequency of relapse upon termination of treatment with shorter therapy [12]. In Thailand, CAZ is the IV antibiotic of choice [11]. Thus, CAZ is the single most important antibiotic for the treatment of melioidosis.
The vast majority of B. pseudomallei strains are susceptible to CAZ, imipenem, meropenem, TMP-SMX, doxycycline and AMC, although a small percentage of isolates display primary resistance [4]. Of great concern to clinicians is the potential for this bacterium to develop resistance during the course of chemother-apy, especially to the first line therapy, CAZ. Although primary resistance of B. pseudomallei to CAZ is rare, the prolonged nature of melioidosis treatment increases the likelihood that acquired resistance can develop, especially if monotherapy is used or if the infection relapses and CAZ is employed multiple times in the same patient. Such acquired resistance has important ramifications due to the high morbidity and mortality associated with this infectious disease and the paucity of alternate treatment options.
Determining the molecular basis of CAZ resistance (CAZ R ) ultimately provides the genetic targets needed for improved treatment outcomes for melioidosis patients by allowing clinicians to rapidly and inexpensively monitor the emergence of resistant populations. It has been previously shown that mutations in the B. pseudomallei class A b-lactamase (encoded by the gene, penA) may confer CAZ R [13][14][15]. These studies identified mutations in the penA gene of CAZ R strains that caused amino acid alterations around conserved motifs. However, functional characterization of penA in clinical isolates of B. pseudomallei has not yet been explored. Thus, there is a need to pinpoint the precise molecular mechanisms behind CAZ R in clinical B. pseudomallei isolates that correlate with the CAZ R phenotypes observed by clinicians.
In the current study, we determined the molecular mechanisms for CAZ R in B. pseudomallei strains from two Australian melioidosis patients who temporally developed resistant CAZ R strains during CAZ therapy. To confirm that CAZ R developed in vivo and was not the result of re-infection with a resistant strain, we subjected the patient isolates to multilocus variable-number tandem repeat analysis (MLVA) and multilocus sequence typing (MLST). In addition, we screened a large panel of clinical and environmental B. pseudomallei for CAZ R mechanisms using allele-specific real-time PCR to determine the rate of primary CAZ R in this bacterium. Last, we tested a panel of b-lactams to determine the suitability of these alternate antibiotics for treating CAZ R B. pseudomallei clinical isolates.

Ethics statement
Ethics approval was granted by the Human Research Ethics Committee of the Northern Territory Department of Health and the Menzies School of Health Research (HREC 04/09), with written informed consent obtained from patients.

B. pseudomallei clinical isolates used in this study
Melioidosis ''Patient 21''. Three isolates from Patient 21 (P21) [16,17] were used for this study (Table 1). The first two isolates were susceptible to CAZ (CAZ S ), whereas the most recent isolate, MSHR 99, displayed CAZ R (16 mg/mL; Table 1). P21, a 63 y.o. male with Type 2 diabetes, chronic renal disease and hazardous alcohol use from Darwin, Australia, was diagnosed with melioidosis following B. pseudomallei isolation from blood cultures (isolate MSHR 73). The patient was treated with IV CAZ and TMP-SMX for two weeks and was discharged on doxycycline. The patient had recrudescence of disease three months later, and B. pseudomallei was again isolated from blood cultures (MSHR 95). Following further treatment with CAZ the patient was placed on oral AMC, but subsequently deteriorated (MSHR 99 from blood culture) and died 4 months after his initial admission.
Melioidosis ''Patient 337''. Six B. pseudomallei isolates derived from an individual (P337) suffering from relapsing melioidosis were obtained for this study. Although the earliest two isolates obtained from this patient were CAZ S , four latter isolates harbored a high-level CAZ R phenotype ($256 mg/mL; Table 1). P337, a 61 y.o. male with Type 2 diabetes and metastatic bronchogenic carcinoma, likely contracted melioidosis following environmental exposure with B. pseudomallei-contaminated soil. Upon initial admission and B. pseudomallei isolation (isolate MSHR 1141), P337 was placed on IV CAZ for four weeks, followed by oral TMP-SMX and doxycycline. In addition to antimicrobial therapy, the patient required immunosuppressive therapy for their malignancy. A four-month follow-up revealed that P337 was still culture-positive for B. pseudomallei (MSHR 1225) despite being on oral doxycycline. Intravenous CAZ was re-administered for three weeks followed by maintenance therapy comprising oral doxycycline and chloramphenicol. P337 remained culture positive for B. pseudomallei and strains isolated between five to seven months after the initial diagnosis displayed CAZ R (MSHR 1226 onwards). P337 remained on oral antibiotics but succumbed to the bronchogenic carcinoma shortly thereafter.

Antibiotic sensitivity testing
A previous study has shown that B-lactam MIC testing in B. pseudomallei is independent of salt concentrations [15]. Therefore, MICs were determined solely using E-tests (bioMérieux, Durham, NC) according to the manufacturer's instructions.
PCR amplification, sequencing and cloning of penA PCR amplification of the Burkholderia penA gene was performed using HotStarTaq Master Mix (Qiagen, Valencia, CA) with the addition of betaine (Sigma Aldrich, St. Louis, MO) to a final concentration of 1.8 M, using the following primers (59-39); penA_F (CGCCACAAATTCGCACGCAC) and penA_R (GCG-ACTCGCGCTCCGTGAAC) (IDT, Coralville, IA). The thermocycling conditions were 95uC for 15 min followed by 35 cycles of 30 s at 95uC, 30 s at 65uC, and 1 min at 72uC, with a final extension of 10 min at 72uC. Cloning of penA PCR products was performed using the pGC TM Blue Vector (Lucigen) according to the manufacturer's instructions. Sanger sequencing was used to identify polymorphisms, to determine insert orientation and to verify presence of insert. Nucleotide sequences of penA from CAZ S and CAZ R B. pseudomallei isolates from P21 and P337 have been submitted to NCBI.

Allelic exchange of penA
Allelic exchange was performed using the suicide vector pMo130 [18], which is compliant for use with Select Agent bacteria such as B. pseudomallei. We used this allelic exchange system to remove penA from both CAZ S and CAZ R isolates, enabling us to determine whether this single gene was responsible for the CAZ R . Briefly, both the upstream and downstream regions of penA were amplified using the following respective primers (59-39); penA_US_R (AAGCGGTCAGATCTTCCGCGTTGTGC-TGGA) and penA_US_F (GCATATCTGCTAGCTCTGTTG-CGGCATCGCTTT), penA_DS_R (CCGAGATCTTCACG-GAGCGCGAGTC) and penA_DS_F (GACAAGCTTGAAAAA-CAGGGCGAACGCACAGG). These primers amplify approximately 800-1000 bp regions flanking penA but do not amplify the gene itself. Underlined nucleotides represent deliberately introduced restriction enzyme (RE) sites, which are required for ligation into pMo130. Amplification was performed using slowdown PCR as previously described [19] without the addition of the altered dGTP analog. Following PCR, amplicons were purified using a PCR purification kit (Qiagen). The upstream product was digested with NheI (Promega, Madison, WI) and BglII (Promega), and the downstream fragment was digested with BglII and HindIII (Promega) according to the manufacturer's instructions. It is important to note that target amplicons did not contain these RE sites to ensure intact, full length amplicons upon digestion. Following digestion, products were ligated into NheI and HindIII digested pMo130 to create pMo130-US-DS. Correct orientation and incorporation of PCR products in the final construct were verified by multiple RE analyses. The remaining allelic exchange procedure was performed as previously described [18].

Cis complementation of penA
The penA gene was reintroduced into DpenA strains according to previously described methods [18] that are in compliance with Select Agent Rule 42 CFR Part 73 (http://www.selectagents.gov/ Regulations.html). First, the upstream region was amplified using the primers described previously. Following amplification, the upstream region was digested with NheI and BglII and was introduced into pMo130 digested with NheI and BglII to create pMo130-US. Second, penA was amplified from MSHRs 663 (penA + ), 99 (penA minus(2)21A; 16 mg/uL CAZ R ), 1226 (penA 281A; $256 mg/uL CAZ R ) or 1300 (penA -21A, 281A; $256 mg/ uL CAZ R ) depending on the desired construct (Supplemental Table S1). Amplification of penA was performed using the following primers (59-39); penA_comp_F (GTTCAGCAGATC-TAACAGATCGCCGAGATGG) and penA_comp_R (GCAC-CGCGATATCTCGCGCTCCGTGAACCTT) with underlined nucleotides representing deliberately introduced BglII and EcoRV restriction sites, respectively. Third, the downstream (DS) region of penA was amplified using the following primers; penA_DScompF (59-CTTCCGGATATCTCACGGAGCGCGAGTC) and pe-nA_DScompR (59-CGACGACAAGCTTGAAAAACAGGGC-GAACGCACAGG). Underlined nucleotides represent EcoRV and HindIII sites, respectively. Last, the amplified fragment containing penA was digested with BglII and EcoRV and the DS fragment was digested with EcoRV and HindIII, and both DS and penA fragments were ligated into pMo130-US digested with BglII and HindIII. The final construct was verified by restriction analyses (data not shown). Importantly, the penA amplicon included 200 bp of the penA upstream region, such that promoter sequences were incorporated into the pMo130-US-penA-DS vectors to ensure expression when re-introduced into B. pseudomallei. Following creation, the pMo130-US-penA-DS construct was introduced into the B. pseudomallei DpenA strains and the gene deletion was reversed following previously published protocols [18].
penA SNP characterization using PCR Genomic DNA obtained from approximately 2,400 genetically and geographically diverse Burkholderia spp. isolates (Price EP et al., manuscript in preparation) was used to screen for the presence of the mutant penA 281A SNP in these isolates. These samples largely comprise clinical and environmental Burkholderia isolates from Australia and Thailand, but also other isolates collected from around the world. DNA was extracted using the DNeasy tissue extraction kit (Qiagen), 5% chelex-100 heat soak [20] or the Wizard genomic DNA purification kit (Promega). DNA was normalized to 1 ng/mL using the NanoDrop 8000 spectrophotometer (Thermo Scientific, Waltham, MA) prior to PCR analysis.
We developed a SYBR Green-based mismatch amplification mutation assay (SYBR MAMA) protocol for interrogating the mutant and wild-type variants at the penA -21A and penA 281A SNPs in B. pseudomallei and Burkholderia mallei. Due to the highly conserved nature of penA among near-neighbor Burkholderia spp., these assays also produce amplicons for B. oklahomensis, B. thailandensis, B. vietnamiensis, B. humptydooensis sp. nov. and B. ubonensis, albeit at lowered efficiency due to primer-template mismatches. SYBR MAMA exploits the differential amplification efficiency of allele-specific primers for SNP interrogation [21,22]. In real-time PCR, this differential efficiency is observed by measuring the cycles to threshold (C T ) of both allele-specific primers to determine the nucleotide present at the SNP. For penA -21A interrogation, two allele-specific primers, 21promA_99_F (59-CACTCCTGTGACGAGAGCTGATTCA) and 21promG_wt_F (59-CACTCCTGTGACGAGAGCTGATTCG) and one common reverse primer 21prom_comR (59-GGCGACGTTTTTC-GCTTGG) were used to interrogate the SNP. For penA 281 primers, penA_281-G (59-GGCGACGAGCGTTTCCCGTTA-TG) and penA_281-A (59-TTTTTTTTTTTTTTTCGACGAG-CGTTTCCCGTTATA) were used in combination with penA_ 281_R (59-CGCAGCGCAAAGCATCAT) to interrogate the SNP. PenA_281-A amplifies the mutant allele, which confers an enhanced ability to degrade CAZ, whereas penA + is preferentially amplified by penA_281-G. PCRs consisted of one allele specific primer and the appropriate reverse primer per well. All samples were run in duplicate. A total 0.3 uM of each primer (IDT), 16 SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA), and molecular grade water (GIBCO, Carlsbad, CA) were added to a volume of 9 mL. One mL of DNA template was added to the reaction. All PCRs were conducted using an ABI PRISM 7900HT real-time PCR instrument (Applied Biosystems) and default cycling conditions that comprised 2 min at 50uC, 10 min at 95uC, followed by 40 cycles of 15 sec at 95uC and 1 min at 60uC. A dissociation curve was performed following amplification to confirm amplicon specificity. MSHR 1300 was used as a positive control for the mutant alleles.

16S rDNA PCR
To verify bacterial DNA quality, we ran a control 16S PCR against all DNA samples using previously published primers [23]. All 16S PCRs were performed on the ABI PRISM 7900HT realtime PCR instrument (Applied Biosystems) using default cycling conditions.

Multiple locus variable number tandem repeat analysis (MLVA)
MLVA was performed on the P21 and P337 isolates as previously described [24], with the exclusion of locus 20 k, to rule out re-infection with different strains.

Sequencing of penA from P21 and P337 isolates
We sequenced penA of B. pseudomallei isolates obtained from two melioidosis patients (P21 and P337; Table 1; Genbank accession numbers JQ364927 through JQ364935) where CAZ R appeared to have developed directly in response to CAZ chemotherapy. All isolates from P21 (MSHRs 73, 95 and 99) demonstrated the same PenA amino acid composition and were identical to the wild-type PenA of CAZ S B. pseudomallei K96243. However, analysis of the promoter region uncovered a novel G to A nucleotide transition (referred to herein as penA -21A) in the latter isolate, MSHR 99, which was not present in either MSHRs 73 or 94 (Table 1) (Note that mutation numbering was determined from the whole genome annotation of B. pseudomallei 1106a due to the absence of penA signal peptide annotation in B. pseudomallei K96243). Importantly, the G to A transition corresponded to a 10-fold increase in CAZ R (16 mg/mL), suggesting this SNP is involved in up-regulation of penA expression.
The majority of CAZ R isolates from P337 also showed the same putative regulatory mutation in the promoter region (MSHRs 1298, 1300 and 1302). In addition, we identified a mutation in penA at position 281 in most of the latter isolates that resulted in a cysteine to tyrosine (C69Y) substitution adjacent to the 70 SXXK 73 conserved motif (Ambler's numbering scheme) [26]. This SNP has been previously identified in CAZ R B. pseudomallei isolates [14], and functionally characterized in a Select-Agent exempt strain of B. pseudomallei [15]. The mutated penA (referred to herein as penA 281A) directly corresponded to high-level CAZ R ($256 mg/mL; Table 1), resulting in a .170-fold increase in CAZ hydrolysis, similar to a previous report [15]. Isolates MSHRs 1141 and 1225 contained penA + , whereas isolates subsequently collected from P337 (Table 1), with the exception of MSHR 1226, contained both penA -21A and penA 281A mutations. The dual-mutant penA isolates from P337 also yielded CAZ MICs of $256 mg/mL. Interestingly, MSHR 1226, an isolate collected mid-infection, contained only the penA 281A mutation and not the promoter region -21A SNP.

Comparison of native and heterologous expression of penA variants
We used a heterologous cloning and expression approach to better understand the link between penA mutations, CAZ R and substrate specificity and compared these data with penA behavior in the native host, as a previous study has demonstrated that heterologous hosts can be useful for approximating the activity of B. pseudomallei PenA towards b-lactam substrates [13]. We used an E. cloni system to express penA amplified from MSHRs 99 (-21A mutant), 1226 (281A mutant) and 1300 (-21A and 281A dual mutant) and a CAZ S penA+ control strain (MSHR 663). MICs of the E. cloni host were compared to E. cloni expressing the penA variants to determine background CAZ hydrolysis ( Table 2).
Neither the heterologously expressed PenA + or the -21A penA mutant altered the CAZ MIC in E. cloni despite a 16-fold increase in CAZ MIC in B. pseudomallei -21A penA, indicating lowered sensitivity of the heterologous system. MSHR 1226 penA (containing the 281A mutation) increased degradation of CAZ by 8-fold and MSHR 1300 (dual penA mutant) by 16-fold, which mimicked the increases observed in the native B. pseudomallei host, albeit with substantially lower fold differences. Interestingly, increased activity towards CAZ by both the dual and C69Ymutated PenA mutants was accompanied by a decrease in hydrolytic activity for amoxicillin (AMX), and to a lesser extent ampicillin (AMP) ( Table 2). In contrast, we saw an increase in MICs in penA -21A against a range of b-lactams. This phenomenon was most evident when comparing AMX MICs, where E. cloni expressing penA -21A yielded a 1.5-fold increase compared with penA + . In B. pseudomallei, both penA -21A and penA + gave AMX MICs of $256 mg/mL, indicating that higher MIC detection would be required to confirm these heterologous AMX MIC differences in the native host. These results demonstrate that the amino acid mutation in PenA is highly favorable for CAZ hydrolysis but causes a substantial reduction in hydrolytic activity towards at least two other b-lactams, as previously shown [15]. The penA -21A mutant mutation also enhances hydrolysis of CAZ, although not to the level of the C69Y mutant. However, penA -21A causes upregulation of penA, which enhances hydrolysis towards other b-lactam substrates, including antibiotics containing clavulanic acid.

Knockout and complementation of penA
Following confirmation of CAZ R by heterologous expression of penA, this gene was removed from mutant and w.t. B. pseudomallei strains (CAZ R MSHR 99, CAZ S MSHR 1141 and CAZ R MSHRs 1226 and 1300) to determine CAZ MICs in the original host compared with its penA knockout. Following penA removal, we screened for CAZ MICs in penA knockouts. All DpenA strains possessed a CAZ S phenotype, with MICs of approximately 1 mg/ mL (Table 2). To further verify the role of penA in CAZ R , we complemented all strains with penA + to test for restoration of wildtype CAZ S and AMX R phenotypes. Finally, we reinserted the original penA genes back into CAZ R strains to examine reproducibility of CAZ R phenotypes ( Table 2). When penA + was introduced into the MSHR 99, MSHR 1226 or MSHR 1300 DpenA strains, the wild-type resistance profile was restored indicating that these mutations were the sole cause of CAZ R in these strains. Further evidence of a single-gene phenotype was confirmed following restoration of the CAZ R MICs in CAZ R strains complemented with their native penA ( Table 2).
Prevalence of penA mutants in Burkholderia spp.
Once it was established that penA 281A and penA -21A were responsible for conferring CAZ R in the B. pseudomallei strains from our study, we determined the frequency of these mutations across clinical and environmental Burkholderia isolates. We obtained the most robust allelic discrimination using the SYBR MAMA format ( Figure S1). Following verification of assay performance, the penA SYBR MAMA assays were screened across approximately 2400 Burkholderia spp. isolates derived from clinical and environmental sources. Screening for these mutations across our Burkholderia DNA collection revealed that only MSHR 1226, 1298, 1300 and 1302, all isolates from P337, contained the mutant penA 281A allele. In contrast, we identified two additional isolates carrying the penA -21A mutation. One isolate was from a clinical case in Malaysia and the second was an environmental isolate from undisturbed soil in Northeast Thailand.

Clonality of P21 and P337 infection
We were interested in determining if the isolates obtained from P21 and P337 were clonal, suggestive of a relapsed infection and in vivo development of CAZ R rather than re-infection with a different CAZ R strain. To determine the clonality of infection, we carried out 22-locus MLVA [24] on the nine isolates from the two patients. MLVA targets rapidly mutable loci throughout the genome; therefore, unrelated isolates are highly likely to display distinct MLVA profiles, making this method indispensable for determining in vivo infection clonality [24,27].
In P21, MLVA failed to show any variability among the three strains across all 22-loci. In P337, MLVA demonstrated 12 mutations among all six strains, ranging from a two-repeat insertion to a five-repeat deletion (data not shown). These mutation rates are within the realm of expected in vivo mutation rates of clonal B. pseudomallei isolates within a single host, as determined in previous studies [24,27]. Further, there was no evidence of a temporal distinction between MLVA mutants, with mutations occurring across all timepoints (data not shown).
MLST was also performed on the patient isolates to consolidate our conclusion of clonality from the MLVA profiles. As expected, MLST genotypes were identical within patients, with P21 isolates being sequence type (ST) 135 and P337 isolates genotyping as ST-330. These results are unsurprising as infection relapse is more common than re-infection [28][29][30].

Discussion
Melioidosis is a serious disease without an effective vaccine that requires prolonged antibiotic therapy to eradicate. Due to the intrinsic resistance of B. pseudomallei to a wide range of antibiotics, the treatment options for melioidosis are unfortunately limited to a small number of antimicrobial agents. Primary treatment usually consists of IV CAZ followed by prolonged oral antibiotic therapy with a secondary drug such as trimethoprim-sulfamethoxazole, doxycycline or AMC [11]. Although primary resistance to these clinical drugs is rare [4], development of resistance can result as a consequence of the prolonged therapy typically needed for treating melioidosis, especially in cases of recurrent melioidosis, which afflicts approximately 10% of patients [31]. Due to the heavy reliance on CAZ as first line therapy for melioidosis, both primary and secondary CAZ R pose a significant challenge in treatment and play a critical role in patient outcomes.
Most cases of melioidosis are treated with IV CAZ monotherapy in the initial eradication phase, followed by a change in antimicrobial drugs once the patient starts oral therapy. The switch in treatment probably abrogates the selective pressure on CAZ R mutants to arise and become dominant in vivo. However, clinicians may employ CAZ multiple times or for an extended length during the course of melioidosis, particularly in recurrent cases, a strategy that leads to an increased potential for CAZ R to develop. In the current study, we observed B. pseudomallei develop both low-level and high-level CAZ R in direct response to chemotherapy with CAZ in two separate cases of recurrent melioidosis. Both patients suffered relapse within months of initial infection and were treated with IV CAZ as the primary treatment in both instances. In the first patient, P21, B. pseudomallei evolved low-level resistance due to a SNP located -21 bp upstream of the putative penA start codon, which resulted in an approximate 10fold up-regulation of the class A b-lactamase PenA. This promoter mutation caused resistance to not only the first line treatment, CAZ, but alarmingly, the follow-up AMC chemotherapy. In the second patient, P337, high-level CAZ R developed due primarily to an amino acid substitution in PenA that altered the substrate specificity of this enzyme, increasing CAZ R by at least 170-fold. Interestingly, the same promoter mutation altering PenA expression was also observed in many isolates from P337, suggesting continued selection pressure for increased penA expression. We strongly suspect that the repeated treatment with CAZ in these relapsed melioidosis patients likely provided the prolonged selective pressure needed for CAZ R mutations to develop and become dominant within the in vivo bacterial population. Although more recurrent melioidosis cases would need to be investigated to confirm this hypothesis, our study demonstrates that there is a risk for treatment failure associated with repeated CAZ chemotherapy in relapsing melioidosis patients that is worthy of further study.
Although high-level CAZ R is very uncommon in B. pseudomallei, it has been previously reported. Sam and colleagues [14] isolated a CAZ R strain (24 mg/mL) from a patient who later harbored B. pseudomallei with high-level CAZ R ($256 mg/mL), indicating a potential stepwise mutation progression in CAZ R . In the current study, we did not detect a low-level CAZ R isolate from P337 (all isolates obtained over the course of infection were either CAZ S or showed high-level CAZ R ), suggesting that the single penA 281A mutation (C69Y) was responsible for the high-level resistance phenotype. Alteration of this amino acid yielded a CAZ R MIC of $256 mg/mL in a Select Agent exempt strain of B. pseudomallei [15]. Our study confirmed these previous results based on isolate MSHR 1226, which contained this single mutation and was resistant to CAZ at $256 mg/mL. However, the small sample size (n = 6) used in the current study renders the possibility that the lowlevel CAZ R phenotype was missed during sampling. In addition, Sam and co-workers [14] saw an increased resistance to AMC for their initial isolates that we did not identify in P337. However, we observed a similar resistance profile towards other b-lactams in the P21 isolate, MSHR 99. It is interesting to speculate whether the initial isolates from the Sam et al. study possessed an alteration in the penA promoter region, similarly to MSHR 99, as the MIC values for both AMC and CAZ are identical between studies.
Having demonstrated both heterologously and in the native host that the 281A and -21A penA mutants were responsible for CAZ R in these isolates, we were interested in determining the frequency of these mutations over a large collection of B. pseudomallei from both clinical and environmental origins. PCR screening of over 2400 samples (of which none are known to be from other recurrent melioidosis patients) showed that no other isolates with the PenA C69Y mutation were found, indicating that this mutation is fortunately rare. We propose several reasons for the low frequency of C69Y in B. pseudomallei. First, B. pseudomallei is a soil dwelling organism, yet CAZ is a synthetically manufactured antibiotic that does not naturally occur in the environment [32] and thus there is no selection pressure to develop CAZ R in the environment. Second, multiple molecular mechanisms likely exist for generating CAZ R in Burkholderia spp. [13,14]. Third, there appears to be a trade-off for high-level resistance to CAZ in the form of increased susceptibility to the other b-lactams (Table 2), which provides a heavy selective disadvantage to B. pseudomallei in the face of blactamases produced by other soil-borne microbes. In other words, the penA281A mutation is only favorable to B. pseudomallei during an in vivo infection that includes CAZ as a chemotherapeutic agent.
Unlike penA281A, our screening efforts did identify two additional isolates with the penA -21A mutation, totaling approximately 0.1% prevalence within our B. pseudomallei collection. One of these mutants belonged to a Malaysian melioidosis case. We lack clinical data on this patient so cannot determine whether this adaptation was acquired in vivo and as a result of treatment with CAZ. The penA -21A mutation was not restricted to clinical isolates, being found in an environmental sample from an undisturbed soil location in Northeast Thailand. Unlike penA281A, the penA -21A mutation caused cross-resistance to all the b-lactams tested, including AMC, which contains a b-lactamase inhibitor. The infrequency of both penA mutations identified in this study supports the continued usefulness of CAZ as a first line treatment option for melioidosis. However, caution should be exercised when administering any antibiotic to melioidosis patients, particularly in light of our study. Ideally, clinicians should ascertain the MIC status of strains during the course of treatment and adjust therapy accordingly. In particular, B. pseudomallei strains that possess lowlevel CAZ R , such as the P21 penA -21A mutants, can result in treatment failure to other b-lactams, including those that contain b-lactamase inhibitors. Alternately, treatment with CAZ can supply sufficient selective pressure for B. pseudomallei to develop high-level CAZ R , as observed in P337.
The system of heterologous cloning and expression used in our study has been important in expediting the identification of genes responsible for antibiotic resistance. Due to the Select Agent classification of B. pseudomallei, genetic manipulation is often highly laborious and the number of genetic tools available to researchers is limited. Using DNA derived from B. pseudomallei and a Biosafety Level 1 heterologous screening host (such as E. coli K-12) has allowed us to screen larger numbers of genes for antibiotic resistance than would otherwise be possible working with the native hosts. However, a secondary method for candidate gene identification, such as allelic exchange, is still required to verify heterologous expression results as expression profiles can vary markedly between the heterologous and native bacteria. For instance, G+C content, codon usage, alterations in expression levels from non-native promoters or vector copy numbers, differences in protein folding or other post-translational modifications can result in changes of native function, leading to erroneous expression results [33,34]. A recent study describes the approval and use of a Select Agent exempt strain of B. pseudomallei as a heterologous host for virulent B. pseudomallei counterparts [15]. This new strain will enable researchers to rapidly identify single gene candidates and allow multiple gene, random cloning approaches to be undertaken, greatly accelerating functional genomics of B. pseudomallei.
In conclusion, we have demonstrated that antibiotic administration in cases of chronic and recurrent bacterial infections can have a profound impact on treatment efficacy. We provide functional evidence for direct selection of B. pseudomallei mutants with enhanced antibiotic resistance following administration of the first line of defense, CAZ, and for one patient, both the first and second lines of defense (CAZ and AMC). Sufficient information regarding mechanisms of resistance and the development of robust PCR assays will one day allow clinicians to monitor bacterial populations in real-time, with alteration of treatment as bacterial populations are identified that develop or alter their resistance profiles. We have shown that although B. pseudomallei is able to develop high-level CAZ R it does so infrequently and at the cost of becoming more sensitive to other antimicrobials, providing an avenue for future research in combating recurrent and chronic melioidosis. Other potential mechanisms of CAZ R in B.
pseudomallei, such as efflux pumps, alterations in other b-lactamases or changes in the cell wall, remain to be characterized. Elucidation of these resistance mechanisms will allow rapid characterization of a B. pseudomallei infection and appropriate treatment to be administered, reducing the morbidity and mortality of melioidosis. Figure S1 B. pseudomallei real-time SYBR MAMA penA 281A SNP assay. The left real-time PCR amplification plot demonstrates preferential amplification of penA + in a non-mutated B. pseudomallei strain (K96243), whereas the right amplification plot shows the mutant polymorphism (penA 281A) from B. pseudomallei MSHR 1300. Blue, penA + ; red, mutant penA 281A allele; green, notemplate controls. (TIF)