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 µg/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.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Burkholderia pseudomallei isolates obtained from two relapsed melioidosis patients, Patient 21 and Patient 337, and their corresponding ceftazidime MICs.

https://doi.org/10.1371/journal.pone.0030789.t001

Bacterial growth conditions

B. pseudomallei isolates were grown on Luria-Bertani agar (LBA) (Beckton-Dickinson, Franklin Lakes, NJ) at 37°C for 24 h to 48 h. All Escherichia coli strains (Supplemental Table S1) including “Escherichia cloni” 10G (Lucigen, Middleton, WI) were grown on LBA at 37°C for 24 h. Antibiotic susceptibility testing was performed on Mueller-Hinton agar (Hardy Diagnostics, Santa Maria, CA) at 37°C.

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 (5′-3′); penA_F (CGCCACAAATTCGCACGCAC) and penA_R (GCGACTCGCGCTCCGTGAAC) (IDT, Coralville, IA). The thermocycling conditions were 95°C for 15 min followed by 35 cycles of 30 s at 95°C, 30 s at 65°C, and 1 min at 72°C, with a final extension of 10 min at 72°C. Cloning of penA PCR products was performed using the pGC™ 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 (5′-3′); penA_US_R (AAGCGGTCAGATCTTCCGCGTTGTGCTGGA) and penA_US_F (GCATATCTGCTAGCTCTGTTGCGGCATCGCTTT), penA_DS_R (CCGAGATCTTCACGGAGCGCGAGTC) and penA_DS_F (GACAAGCTTGAAAAACAGGGCGAACGCACAGG). 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 ΔpenA 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(−)21A; 16 µg/uL CAZ^R), 1226 (penA 281A; ≥256 µg/uL CAZ^R) or 1300 (penA -21A, 281A; ≥256 µg/uL CAZ^R) depending on the desired construct (Supplemental Table S1). Amplification of penA was performed using the following primers (5′-3′); penA_comp_F (GTTCAGCAGATCTAACAGATCGCCGAGATGG) and penA_comp_R (GCACCGCGATATCTCGCGCTCCGTGAACCTT) 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 (5′-CTTCCGGATATCTCACGGAGCGCGAGTC) and penA_DScompR (5′-CGACGACAAGCTTGAAAAACAGGGCGAACGCACAGG). 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 ΔpenA 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/µL 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 (5′-CACTCCTGTGACGAGAGCTGATTCA) and 21promG_wt_F (5′-CACTCCTGTGACGAGAGCTGATTCG) and one common reverse primer 21prom_comR (5′-GGCGACGTTTTTCGCTTGG) were used to interrogate the SNP. For penA 281 primers, penA_281-G (5′-GGCGACGAGCGTTTCCCGTTATG) and penA_281-A (5′-TTTTTTTTTTTTTTTCGACGAGCGTTTCCCGTTATA) were used in combination with penA_281_R (5′-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), 1× SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA), and molecular grade water (GIBCO, Carlsbad, CA) were added to a volume of 9 µL. One µL 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 50°C, 10 min at 95°C, followed by 40 cycles of 15 sec at 95°C and 1 min at 60°C. 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 real-time 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.

Multi-locus sequence typing (MLST)

MLST was performed as described elsewhere [25].

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 µg/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 ⁷⁰SXXK⁷³ 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 µg/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 µg/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 β-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).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. MICs of B. pseudomallei isolates with and without penA and heterologously expressed genes in “E. cloni”.

https://doi.org/10.1371/journal.pone.0030789.t002

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 C69Y-mutated 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 β-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 µg/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 β-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 β-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 ΔpenA strains possessed a CAZ^S phenotype, with MICs of approximately 1 µg/mL (Table 2). To further verify the role of penA in CAZ^R, we complemented all strains with penA⁺ to test for restoration of wild-type 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 ΔpenA 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]–[30].