Versatile Dual-Technology System for Markerless Allele Replacement in Burkholderia pseudomallei

Abstract

Burkholderia pseudomallei is the etiologic agent of melioidosis, a rare but serious tropical disease. In the United States, genetic research with this select agent bacterium is strictly regulated. Although several select agent compliant methods have been developed for allelic replacement, all of them suffer from some drawbacks, such as a need for specific host backgrounds or use of minimal media. Here we describe a versatile select agent compliant allele replacement system for B. pseudomallei based on a mobilizable vector, pEXKm5, which contains (i) a multiple cloning site within a lacZα gene for facile cloning of recombinant DNA fragments, (ii) a constitutively expressed gusA indicator gene for visual detection of merodiploid formation and resolution, and (iii) elements required for resolution of merodiploids using either I-SceI homing endonuclease-stimulated recombination or sacB-based counterselection. The homing endonuclease-based allele replacement system is completed by pBADSce, which contains an araC-P_BAD-I-sceI expression cassette for arabinose-inducible I-SceI expression and a temperature-sensitive pRO1600 replicon for facile plasmid curing. Complementing these systems is the improved Δasd Escherichia coli mobilizer strain RHO3. This strain is susceptible to commonly used antibiotics and allows nutritional counterselection on rich media because of its diaminopimelic acid auxotrophy. The versatility of the I-SceI- and sacB-based methods afforded by pEXKm5 in conjunction with E. coli RHO3 was demonstrated by isolation of diverse deletion mutants in several clinical, environmental, and laboratory B. pseudomallei strains. Finally, sacB-based counterselection was employed to isolate a defined chromosomal fabD(Ts) allele that causes synthesis of a temperature-sensitive FabD, an essential fatty acid biosynthesis enzyme.

Burkholderia pseudomallei is the etiologic agent of melioidosis (3, 35). While the bacterium and disease are typically endemic to tropical and subtropical regions of the world (5), historical precedent for use in bioweapon development programs, low infectious doses, high morbidity and mortality, and arduous therapy caused B. pseudomallei to be listed as a category B select agent by the Centers for Disease Control and Prevention. In the United States, transport, possession, and use of select agents is regulated by strict federal guidelines. These guidelines restrict the use of antibiotic resistance markers in research to those that do not compromise the use of the respective drugs in humans, veterinary medicine, or agriculture (27). The paucity of selection markers approved for this bacterium has led to development of genetic manipulation strategies that allow the isolation of unmarked mutants. These include fragment mutagenesis, where a linear DNA fragment containing the mutation, assembled in vitro by PCR, is transferred to the host strain and selection for the antibiotic resistance encoded by the fragment results in gene replacement in the homologous region of the chromosome (4, 32). When the selection markers are flanked by Cre or Flp recombinase target sites, they can be removed in vivo by temporary expression of the respective site-specific recombinase, resulting in markerless mutants (4).

Additionally, allelic replacement schemes driven by genetically engineered pheS- (1, 20), sacB- (9, 15), and rpsL- (19, 30) based counterselection markers have been developed for use in B. pseudomallei. With these technologies, regions of homology containing markerless mutations are cloned into a nonreplicative plasmid. Transfer of the recombinant plasmid into the bacterial host followed by selection of an antibiotic resistance encoded by a gene located on the plasmid backbone leads to integration of the nonreplicative plasmid by regions of homology. Loss of plasmid sequences by homologous recombination results in a population in which a significant portion of the survivors of the appropriate counterselection will have undergone the desired gene replacement (Fig. 1, lower right).

FIG. 1.

Schematic of allele exchange procedures. For plasmid-based allelic exchange, a PCR-assembled chromosomal segment containing a deletion of orfY with flanking orfX and orfZ sequences is cloned into an appropriate vector, e.g., pEXKm5. The nonreplicative plasmid is delivered to the host strain by conjugation (or electroporation), followed by kanamycin resistance selection. This step results in integration of the allelic replacement construct into the chromosome by homologous recombination between cloned and chromosomal sequences and can be visualized by the appearance of blue colonies on Km- and X-Gluc-containing medium. The two different merodiploid resolution strategies enabled by pEXKm5 are illustrated. For I-SceI-catalyzed resolution (illustrated on the left side), the merodiploid is transformed with the I-SceI expression construct, which results in double-stranded cleavage of the chromosome and release of most of the plasmid backbone. This event can be monitored by the appearance of white colonies on X-Gluc-containing medium. Repair of the double-stranded break by homologous flanking repeat sequences leads to formation of a wild-type strain (event denoted by the circled number 1) or a mutant strain (event denoted by the circled number 2). The two events are distinguished by phenotypic analyses and/or PCR. In a final step that is not illustrated in this figure, purification of colonies with the desired mutant genotype/phenotype at 42°C leads to loss of the pBADSce expression vector in 100% of the colonies. For sacB-mediated counterselection (illustrated on the right side), the merodiploid strain is plated on medium containing sucrose. This counterselection will either result in a wild-type strain (event denoted by Δ1) or in a mutant strain (event denoted by Δ2). These events can be monitored by the appearance of white colonies on X-Gluc- and sucrose-containing medium and are distinguished by phenotypic analyses and/or PCR. Abbreviations: gusA, Escherichia coli glucuronidase-encoding gene; ori, pMB9-derived narrow-range origin of replication; sacB, Bacillus subtilis levansucrase-encoding gene optimized for expression and localization in B. pseudomallei (9).

An alternative to counterselection schemes involves the use of the intron-encoded homing endonuclease I-SceI (18). This enzyme recognizes a specific 18-bp sequence which is absent from all eukaryotic genomes (except the source, Saccharomyces cerevisiae) and prokaryotic genomes sequenced to date. The basic principle of the method is that cleavage of the bacterial chromosome at an artificially introduced I-SceI site(s) stimulates recombination (21). In this scheme, a nonreplicative plasmid containing cloned regions of homology and an I-SceI site(s) is integrated into the chromosome by homologous recombination between cloned and chromosomal sequences. This integration event is selected by an antibiotic resistance encoded by a gene on the plasmid backbone and results in merodiploid formation. Next, expression of the I-SceI enzyme, either encoded by the integrated plasmid or by a separately introduced plasmid, results in cleavage at the I-SceI site(s) within the integrated vector sequences (Fig. 1, lower left). The resulting double-strand break is repaired by the host recombination machinery by recombination of the regions of sequence homology flanking the break in the merodiploid. Loss of plasmid sequences by homologous recombination results in a mixed population in which a certain percentage will have undergone the desired gene replacement, given the gene is nonessential under the experimental conditions. Although cleavage of the chromosome induces the host's SOS response, this does not result in an increased mutation rate (21). The use of I-SceI for promoting allelic exchange in Escherichia coli (21), Bacillus anthracis (11), Burkholderia cenocepacia (8), Corynebacterium glutamicum (31), and Pseudomonas aeruginosa (36) has been reported.

Biparental or triparental mating is the most efficient way to introduce nonreplicative plasmids into B. pseudomallei for purposes of allele replacement. An obvious disadvantage of this method is that a donor strain must be available that is compatible with the conjugative plasmid, e.g., it cannot contain antibiotic resistance markers that interfere with those encoded by the conjugative plasmid. And herein lie the problems. The most commonly used E. coli mobilizer strains, S17-1 (29) and SM10 (29) and its derivatives, e.g., SM10(λpir) (17), contain chromosomally integrated RP4 sequences and, therefore, different resistance markers, depending on how they were engineered. For example, the most versatile mobilizer strain, SM10(λpir), is kanamycin resistant (Km^r) and can therefore not be used with genetic elements containing an nptII Km^r-encoding gene, one of the few approved selection markers that work with all clinical and environmental B. pseudomallei strains tested to date. Conjugation experiments require counterselection against the donor and untransformed recipient strains. This is usually achieved by utilizing an antibiotic or growth medium that precludes growth of the donor strain but does not affect the recipient strain. For instance, E. coli is highly susceptible to polymyxin B but Burkholderia spp. are naturally resistant to this antimicrobial. Similarly, Pseudomonas aeruginosa can utilize citrate but E. coli cannot. To avoid the use of antibiotics or for situations where other intrinsic properties cannot be exploited, donor strains have been engineered that require nutritional supplements for growth, but they either still contain antibiotic resistance markers or are based on nutritional requirements that preclude the use of rich media (1, 6).

Here we describe a versatile select agent compliant allele replacement system for B. pseudomallei based on a single vector which contains the features required for resolution of merodiploids using either I-SceI-driven recombination or sacB-based counterselection. Complementing this system is an improved E. coli mobilizer strain susceptible to all antibiotics and counterselectable on rich media. The versatility of these methods was demonstrated by isolation of deletion mutants as well as a temperature-sensitive allele in an essential fatty acid biosynthesis gene.

MATERIALS AND METHODS

Bacterial strains, media, and growth conditions.

Bacterial strains used in this study are listed in Table 1 and Table S1 in the supplemental material. The construction of E. coli mobilizer strains RHO1, RHO2, and RHO3 is described in the supplemental material. Bacteria were routinely grown at 37°C in Luria broth Lennox (LB) (28) (previously referred to as low-salt LB [4, 22]) or on LB agar purchased from (MO BIO Laboratories, Carlsbad, CA). Yeast extract-tryptone (YT) medium contained 10 g liter⁻¹ yeast extract (Difco, Detroit, MI) and 10 g liter⁻¹ tryptone (Fisher Scientific, Fairlawn, NJ). Strains containing temperature-sensitive (TS) plasmid derivatives or TS alleles were grown at 30°C (permissive temperature) or 42°C (nonpermissive temperature). M9 medium (16) with 10 mM glucose was used as the minimal medium and was supplemented with 0.6 mM adenine to support growth of purM mutants. LB medium was supplemented with 400 μg ml⁻¹ diaminopimelic acid (DAP; ll-, dd-, and meso-isomers; Sigma, St. Louis, MO) to support growth of E. coli Δasd mutants. Antibiotics were added at the following concentrations: 100 μg ml⁻¹ ampicillin (Ap), 25 μg ml⁻¹ chloramphenicol (Cm), 35 μg ml⁻¹ kanamycin (Km), and 25 μg ml⁻¹ zeocin (Zeo) for E. coli; 1,000 μg ml⁻¹ Km and 2,000 μg ml⁻¹ Zeo for wild-type B. pseudomallei. For Δ(amrAB-oprA) B. pseudomallei efflux pump mutants, Km and Zeo were used at 35 μg ml⁻¹ and 100 μg ml⁻¹, respectively. Antibiotics were either purchased from Sigma (ampicillin, chloramphenicol, and kanamycin) or Invitrogen, Carlsbad, CA (zeocin). β-Galactosidase or β-glucuronidase indicator media contained 50 μg ml⁻¹ 5-bromo-4-chloro-3-indolyl-β-d-glactopyranoside (X-Gal) or 50 μg ml⁻¹ 5-bromo-4-chloro-3-indolyl-β-d-glucuronide (X-Gluc). Both chromogens were purchased from Gold Biotechnology (St. Louis, MO).

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Relevant genotype or propertiesa	Reference or source
E. coli strains
DH5α	F− φ80lacZΔM15 Δ(lacZYA-argF)U169 deoR recA1 endA1 hsdR17(rK− mK+) phoA supE44	14
SM10(λpir)	Kmr; thi-1 thr-1 leuB26 tonA21 lacY1 supE44 recA integrated RP4-2 Tcr::Mu aphA+ (RP4-2 is RP4 ΔTn1)	17
RHO1	Kms; SM10(λpir) ΔaphA::FRT	This study
RHO2	Kmr; SM10(λpir) Δasd::FRT	This study
RHO3	Kms; SM10(λpir) Δasd::FRT ΔaphA::FRT	This study
B. pseudomallei strains
1026b	Wild-type strain, clinical isolate	7
K96243	Sequenced prototype strain, clinical isolate	10
Plasmids
pBAD-I-sceI	Apr; source of araC-PBAD-I-sceI expression cassette	33
pBADSceb	Zeor; I-SceI expression vector with pRO1600(Ts) replicon	This study
pEX100T	Apr; source of I-SceI sites, lacZα, and pMB9 ori	25
pEXKm4b	Kmr; pEX100T with pFKM2 nptII gene, E. coli gusA reporter gene, and a synthetic multiple cloning site	This study
pEXKm5b	Kmr; pEXKm4 with B. pseudomallei optimized sacB gene region from pMo130	This study
pFKM2	Apr Kmr; source of nptII gene	4
pMo130	Kmr; source of sacB gene optimized for B. pseudomallei	9
pPS2382	Apr Kmr; pCR2.1 TA cloning vector (Invitrogen) with B. pseudomallei fabD gene	This study
pPS2383	Apr Kmr; pPS2382 with fabD(Ts) allele	This study
pPS2592	Kmr; pEXKm5 with fabD(Ts) allele fragment from pPS2383	This study

^aAbbreviations: Ap, ampicillin; Km, kanamycin; r, resistant; s, susceptible; Tc, tetracycline; Tp, trimethoprim; Zeo, zeocin.

^bDetails on the construction of pBADSce, pEXKm4, pEXKm5, and intermediary plasmids are provided in the supplemental material.

DNA methods and transformation.

Routine procedures were employed for manipulation of DNA and transformation of E. coli cells (23). Plasmid DNAs were isolated from E. coli using the Fermentas GenJet plasmid miniprep kit (Fermentas, Glen Burnie, MD). DNA fragments were purified from agarose gels by utilizing the Fermentas DNA extraction kit. Replicative plasmids were introduced into B. pseudomallei strains by using a rapid electroporation procedure or via conjugation (4, 22). Chromosomal B. pseudomallei mutations were either verified by colony PCR (4) or by PCR using purified chromosomal DNA (Gentra Puregene DNA purification kit [Qiagen, Valencia, CA]) as template. Custom oligonucleotides were purchased from Integrated DNA Technologies (Coralville, IA), and DNA sequences were determined by the Proteomics and Metabolomics Facility at Colorado State University. DNA maps were constructed using Gene Construction kit 2.5 (Textco, West Lebanon, NH) and exported to Microsoft PowerPoint for final annotation.

Construction of pEXKm4, pEXKm5, and I-SceI expression vector pBADSce.

The construction of the pEXKm4, pEXKm5, and I-SceI expression vector pBADSce and intermediary plasmids is detailed in the supplemental material. Plasmid pEXKm4 combines the following: (i) the narrow-host-range pMB9 origin of replication, an origin for conjugal transfer, a lacZα gene, and I-SceI sites from pEX100T (25) and a synthetic multiple cloning site; (ii) the kanamycin resistance gene from pFKM2 (4), approved for use in B. pseudomallei; (iii) the E. coli gusA indicator gene driven from the strong P1 promoter. Plasmid pEXKm5 is pEXKm4 with sacB, a Bacillus subtilis levansucrase-encoding gene optimized for expression and localization in B. pseudomallei (9). The conditional I-SceI expression vector pBADSce contains the arabinose-inducible I-SceI structural gene from pBAD-I-sceI (33), a pRO1600 broad-host-range replicon derivative that is TS in Burkholderia spp. (4), and a zeocin resistance gene approved for use in B. pseudomallei.

General methods for derivation of allele replacement plasmids.

Downstream and upstream homologous DNA fragments of approximately 500 to 800 bp in length (exemplary extents of upstream and downstream DNA) listed in Table 2 were PCR amplified using Taq HiFi DNA polymerase (Invitrogen) and gene-specific primers (primer sequences are available from the authors upon request) from purified chromosomal DNA. These segments were then assembled in a second reaction mixture by using overlap extension PCR, purified from an agarose gel, and ligated into the TA cloning vector pCR2.1 (Invitrogen). Correct inserts, as determined by restriction enzyme digests, were ligated into pEXKm4 or pEXKm5 digested with appropriate restriction enzymes, and the ligation mixtures were transformed into DH5α. Correct recombinant pEXKm plasmids from Km^r transformants that were white on X-Gal indicator medium were then transformed into RHO3 for conjugal transfer to B. pseudomallei.

TABLE 2.

B. pseudomallei mutants generated with the I-SceI method

Allele	Host straina	Mutant	Homology (bp)		Size of deletion (bp)
			Upstream	Downstream
ΔpurM	1026b	Bp318	742	689	114
ΔpurM	K96243	Bp190	742	689	114
Δ(bpeEF-oprC)	1026b	Bp253	769	813	4,465
ΔBPSS2307	1026b	Bp254	824	763	1,517

^aAll strains were generated using pEXKm4, except for Bp318 which was engineered using pEXKm5.

Conjugations using E. coli mobilizer strain RHO3.

RHO3 harboring the recombinant mobilizable plasmid was grown overnight at 37°C in LB supplemented with 400 μg ml⁻¹ DAP and the antibiotic required for plasmid maintenance. The B. pseudomallei recipients were grown overnight at 37°C in LB medium. A 0.1-ml aliquot of each donor and recipient culture was added to a microcentrifuge tube with 0.6 ml of 10 mM MgSO₄, and the cells were pelleted by centrifugation at 7,000 × g for 2 min. Cells were gently resuspended in 1 ml of 10 mM MgSO₄ by pipetting, centrifuged as before, and then resuspended in 30 μl of 10 mM MgSO₄. The conjugation mixture was applied to a cellulose acetate membrane (13-mm diameter; 0.45-μm pore size) on a prewarmed LB plate containing 400 μg ml⁻¹ DAP and incubated overnight at 37°C. The membrane was transferred to a 1.5-ml microcentrifuge tube containing 1 ml LB and the tube was centrifuged for 30 s at 7,000 × g to dislodge the cells. The supernatant was removed, and cells were resuspended in 1 ml of LB to remove residual DAP and centrifuged as before. The pellet was resuspended in 300 μl of LB, and 200-μl samples of the undiluted cell suspension and 200 μl of a 10⁻² dilution were plated on selective LB plates without DAP to select for exconjugants and counterselect against RHO3.

Homing endonuclease I-SceI-mediated gene replacement.

Merodiploids were obtained by selection of the chromosomally integrated pEX recombinant plasmid on LB plates containing 1,000 μg ml⁻¹ Km and 50 μg ml⁻¹ X-Gluc (LB-Km1000-XGluc). A blue Km^r colony was picked and inoculated into 3 ml LB-Km1000 medium and grown overnight. Electrocompetent cells were prepared and transformed with 50 ng of pBADSce DNA by electroporation as previously described (4). Several 25-μl aliquots were plated on LB plates with 2,000 μg ml⁻¹ zeocin, 50 μg ml⁻¹ X-Gluc with and without 0.5% l-arabinose. After 36 h at 30°C, plates with arabinose contained on average five, mostly white, colonies whereas plates without arabinose contained on average 50 predominantly blue colonies. White colonies from the arabinose-containing plates were screened either phenotypically and/or by colony PCR (4). Curing of pBADSce was then achieved by streaking colonies with the desired genotype and/or phenotype on an LB plate at 42°C.

Gene replacement using sucrose counterselection.

Merodiploids were obtained by selection of the chromosomally integrated pEX recombinant plasmid on LB-Km1000-XGluc plates. A blue colony was inoculated into 1 ml YT broth and shaken at 37°C for at least 4 h. Serial dilutions were prepared in YT broth, and 100-μl aliquots of these dilutions (typically 10⁻¹, 10⁻², and 10⁻³) were plated on YT agar plates containing 15% sucrose and 50 μg ml⁻¹ X-Gluc and incubated at 30°C for 48 h or until colonies were well grown. White colonies arising on these plates at a frequency of >90% were then picked for further characterization. An alternative to growing colonies in YT broth and plating dilutions is to streak cells from a large blue merodiploid colony or patch directly on YT-sucrose-X-Gluc plates.

Isolation of a B. pseudomallei fabD(Ts) mutant.

The B. pseudomallei fabD gene (BPSL2441 in the K96243 annotation [10]) was PCR amplified from purified 1026b genomic DNA using Hi-Fi Taq polymerase and primers 1702 (5′-GCTCGAATCGATTCAATTGGG) and 1703 (5′-TCGGTTGAATCGGCAAGCTCG). The resulting 994-bp fragment was gel purified and ligated to pCR2.1 to yield pPS2382. Codon 255 (TGG) was changed to CAG using the Stratagene QuikChange kit and phosphorylated mutagenic oligonucleotide 1723 (5′-Phos-CCCCGTGCGCCAGGTCGAGTGCGTGCAGCACATC). The insert was sequenced to verify the presence of the 2-bp change, and a clone (pPS2387) containing the mutation was retained for further study. A 1,010-bp EcoRI fragment containing the mutated fabD gene was excised from pPS2387 DNA and ligated into the EcoRI site of pEXKm5 to form pPS2592. This plasmid was transformed into E. coli mobilizer strain RHO3 and transformants were selected on LB medium containing 400 μg ml⁻¹ DAP and 35 μg ml⁻¹ Km. The plasmid was then transferred to B. pseudomallei 1026b by biparental filter mating as described above, except that all steps were performed at 30°C. Conjugation mixtures were plated on LB-Km1000-XGluc plates. A blue Km^r colony was picked and purified on an LB-Km1000-XGluc plate. Merodiploids were resolved using sucrose counterselection as described above. Four white colonies growing on the YT-sucrose-X-Gluc plates were purified on the same plates and then patched on two LB plates. One plate was incubated at 30°C and the second plate at 42°C. Colonies growing at 30°C but not 42°C were putative fabD(Ts) mutants. The presence of the fabD(Ts) mutation in these strains was verified by colony PCR using primers 1702 and 1703 and sequencing of the resulting PCR products.

Nucleotide sequence accession numbers.

The sequences of pBADSce, pEXKm4, and pEXKm5 were deposited in GenBank and assigned accession numbers FJ797515, FJ797516, and GQ200735, respectively.

RESULTS AND DISCUSSION

Derivation of an improved E. coli mobilizer strain.

To enable the conjugal transfer of plasmids with a Km^r selection marker from E. coli to B. pseudomallei utilizing rich media, we constructed E. coli strain RHO3, a Δasd ΔaphA SM10(λpir) derivative. Aspartate β-semialdehyde dehydrogenase-deficient strains do not grow on rich or minimal media without DAP, as this amino acid is not present in the ingredients used for medium preparation. The ability of RHO3 to mobilize replicative and nonreplicative plasmids with an oriT was assessed by using E. coli, B. thailandensis, and B. pseudomallei recipient strains. To counterselect against RHO3, mating mixtures were spread on LB plates without DAP but with the antibiotic for which resistance is conferred by markers transferred by the plasmid. RHO3 is the most versatile mobilizer strain constructed to date, because it is susceptible to all commonly used antibiotics and is easy to counterselect on rich media because of its DAP auxotrophy, and as RHO3 supports replication of plasmids with the conditional ori_R6K, it can be used for genetic manipulation of enteric and nonenteric bacteria.

Use of I-SceI homing endonuclease for markerless gene replacement in B. pseudomallei.

The I-SceI homing method uses two plasmids, a I-SceI site delivery vector, pEXKm4 or pEXKm5, and the I-SceI expression vector, pBADSce (Fig. 2). To demonstrate the utility of the approach, we initially focused on pEXKm4 and used it to introduce a number of mutations into B. pseudomallei strains 1026b and K96243 (Table 2).

FIG. 2.

Maps of plasmids used for allelic exchange. The I-SceI delivery plasmids pEXKm4 and pEXKm5 contain the following common features: gusA, gene encoding E. coli glucuronidase; lacZα, E. coli LacZα peptide-encoding gene for X-Gal-based blue-white screening with the indicated unique cloning sites; nptII, gene coding for neomycin phosphotransferase, which confers kanamycin resistance; ori, E. coli pMB9 origin of replication; oriT, RK2-derived origin for conjugal plasmid transfer; P1, Pseudomonas integron promoter; P_lac, E. coli lac operon promoter; T₀T₁, transcriptional terminators T₀ and T₁ from bacteriophage λ and E. coli rrnB operon, respectively. pEXKm5 additionally contains sacB, a Bacillus subtilis levansucrase-encoding gene optimized for expression and localization in B. pseudomallei by inclusion of the predicted B. pseudomallei groES promoter (P_groES) and the B. mallei sacB leader sequence (9). The I-sceI source vector pBADSce also contains the following features: I-sceI, gene encoding I-SceI endonuclease; ble, bleomycin resistance gene, which confers zeocin resistance; ori₁₆₀₀, pRO1600 origin of replication requiring the rep(Ts)-encoded replication protein, which confers a temperature-sensitive phenotype in Burkholderia spp. at temperatures of 37°C and above; P_BAD, the arabinose-inducible E. coli araBAD operon promoter controlled by the araC-encoded regulatory protein; P_EM7, synthetic prokaryotic promoter.

The B. pseudomallei purM locus was used as an initial test of this method because (i) purM mutants are adenine auxotrophs and can thus be easily screened by growth on M9 glucose minimal medium with and without 0.6 mM adenine and (ii) we are studying purM as a candidate for attenuated mutant construction. Using the conditions outlined in Materials and Methods, an average of 22% of the white colonies obtained after I-SceI-stimulated merodiploid resolution were adenine auxotrophs. The presence of the ΔpurM alleles was verified by PCR (Fig. 3) and sequencing of the PCR products. Mutant generation was dependent on I-SceI expression, since no purM mutants were obtained in the absence of arabinose induction. The method was also used to generate a B. pseudomallei 1026b efflux pump deletion mutant (Table 2) that had the correct genotype (as determined by PCR) (Fig. 3) and phenotype [Δ(bpeEF-oprC) mutants became susceptible to trimethoprim and chloramphenicol, two substrates of the BpeEF-OprC pump (12; T. Mima and H. P. Schweizer, unpublished observations)]. In search of possible β-lactamase activity of the BPSS2307 gene product, we deleted this gene from B. pseudomallei 1026b and verified the presence of the correct mutant allele by PCR (Fig. 3) and sequencing of the PCR product. The β-lactam susceptibility profile of the ΔBPSS2307 mutant, Bp254, was the same as that of 1026b (data not shown). Finally, we tested the utility of pEXKm5 as an I-SceI delivery vector by isolating a ΔpurM derivative of 1026b which, as expected, was readily obtained.

FIG. 3.

PCR analysis of selected B. pseudomallei mutant strains isolated in this study. PCR was performed on colony boiling preparations (4) of the indicated strains with primers flanking the mutant allele (primer sequences available upon request). Lane 1, K 96243 wild-type purM locus; lane 2, Bp190 (K96243 ΔpurM); lane 3, 1026b wild-type bpeEF-oprC locus; lane 4, Bp253 [1026b Δ(bpeEF-oprC)]; lane 5, 1026b wild-type BPSS2307 locus; lane 6, Bp254 (1026b ΔBPSS2307). PCR products were separated on a 1% agarose gel and stained with ethidium bromide. Sizes of the expected PCR fragments (in bp) are indicated above the respective bands (the size of the expected DNA fragment for the wild-type bpeEF-oprC locus is indicated in parentheses, since it could not be amplified under the PCR conditions used here). Lanes M contained the Hi-Lo DNA ladder (Minnesota Molecular, Minneapolis, MN), and the sizes of pertinent fragments are indicated on the left.

The I-SceI-based allele replacement procedure described here functions efficiently in B. pseudomallei. While we have only demonstrated its use in generation of markerless deletion mutants ranging from 114 bp to 4,465 bp, I-SceI-based methods have been previously shown to achieve allele replacement with marked or unmarked mutations, as well as single base pair changes, allowing effective transfer of nucleotide changes or deletions ranging from 1 bp (11) to >60 kb (21) into the chromosomes of various bacteria. Use of our I-SceI method for transferring potential TS alleles to the genome is complicated, because pBADSce curing is most efficiently achieved at temperatures of >37°C. While we have not investigated this in detail, we have, however, observed that pBADSce can be cured by growing cells in the absence of zeocin at 30°C, presumably because plasmids with the TS pRO1600 replicon exist in single copy in Burkholderia spp. cells at the permissive temperature (4). We have used the I-SceI method with B. pseudomallei strains that are refractory to PCR fragment mutagenesis, e.g., K96243, as well as clinical and environmental strains. Sequential iterations allow rapid engineering of strains containing multiple unmarked mutations (31). Because I-SceI expression is tightly regulated in the absence of arabinose, the araC-P_BAD-I-sceI expression cassette could theoretically be engineered into the pEXKm4 vector to result in a one-plasmid system. However, we opted for a two-plasmid system to (i) facilitate cloning of homologous DNA segments by using X-Gal-based blue-white screening in a vector with a versatile multiple cloning site uncorrupted by sites contained elsewhere on the plasmid and (ii) enable use of electroporation as an alternative to conjugation for plasmid delivery, because pEXKm4 is considerably smaller in size without the expression cassette. The new method has significant advantages over previously described I-SceI-based allelic replacement systems: (i) inclusion of a gusA reporter gene for visualization of merodiploid formation, as well as resolution, and (ii) unlike some previously described systems which rely on an unstable plasmid for curing (8), inclusion of a TS replicon in the I-SceI expression vector promotes loss of the vector in 100% of the colonies grown at the nonpermissive temperature.

The usefulness of the method described here extends to other Burkholderia spp. For instance, we successfully used it to derive several B. thailandensis E264 mutants (2), including Δ(amrAB-oprA) and Δ(bpeAB-oprB) single and double efflux pump mutants, which allowed elucidation of the substrate profiles of these pumps (Trunck and Schweizer, unpublished). Although not tested in this study, the newly described method can most likely be applied to more distantly related Burkholderia spp. with some caveats: (i) the kanamycin resistance selection marker contained on pEXKm5 may have to be replaced with another marker applicable to the target bacterium, and/or (ii) the pBADSce helper plasmid may have to be custom designed for the intended target bacterium by replacing the zeocin selection marker and perhaps the replicon. We previously demonstrated that the pRO1600(Ts) replicon exhibited a TS phenotype only in B. pseudomallei or closely related Burkholderia spp. (e.g., B. thailandensis [4]). Although plasmids containing this conditional replicon replicated in Pseudomonas aeruginosa (and presumably other bacteria, such as B. cenocepacia [26]) at 30°C, they did not exhibit a TS phenotype in this bacterium (4).

Use of sacB-based sucrose counterselection for markerless gene replacement in B. pseudomallei.

Two previous studies indicated that under certain conditions (e.g., low salt, low temperature) and with proper modifications (e.g., SacB secretion via a Burkholderia leader sequence), the widely used sacB-based counterselection method can also be applied to most, if not all, B. pseudomallei strains irrespective of whether the endogenous sacB gene is expressed (9, 15).

We tested the efficacy of sucrose counterselection afforded by pEXKm5 by deriving ΔpurM derivatives of five clinical and five environmental B. pseudomallei isolates. Using the conditions detailed in Materials and Methods and the same allele with flanking sequences used to construct Bp190 and Bp318 with the I-SceI method (Table 2), such mutants were readily obtained with all isolates. Rates of resolution of merodiploids to the ΔpurM genotype varied between 4.3% and 60%, with the average and median being 38% and 40%, respectively.

The ultimate test for any allele replacement system is its utility for generating markerless mutations in essential genes, either deletions or, even more challenging, nucleotide substitutions. To test the utility of pEXKm5 for transfer of nucleotide substitutions to the B. pseudomallei genome, we attempted to generate a TS allele of fabD. Bacterial fabD genes encode malonyl-coenzyme A:acyl carrier protein transacylase, an essential enzyme of de novo fatty acid biosynthesis (reviewed in reference 24). We previously used this strategy successfully to generate a P. aeruginosa fabD(Ts) mutant by introducing a W258Q change (13), which corresponded to a previously identified W257Q substitution in an E. coli fabD(Ts) mutant (34). The putative B. pseudomallei 307-amino-acid FabD protein (BPSL2441 in the annotated K96243 genome [10]) contains a conserved tryptophan at position 255. We changed tryptophan 255 to a glutamine residue by site-specific mutagenesis, which introduced a TG-to-CA mismatch that resulted in a W255Q change in the FabD amino acid sequence. This putative fabD(Ts) allele was then returned to B. pseudomallei 1026b chromosome 1 as described in Materials and Methods. Upon sucrose counterselection, plates contained small and large colonies. Whereas small colonies were 100% FabD(Ts) mutants, including Bp321, none of the large colonies had a TS phenotype and they were probably wild type. This demonstrated, for the first time, the utility of the sacB system for transfer of nucleotide substitutions to the B. pseudomallei genome and also the essentiality of FabD in B. pseudomallei. The newly developed method complements recently described sacB-based methods for gene replacement in B. pseudomallei (9, 15). It does, however, possess distinct advantages over these methods: (i) to facilitate discrimination of true merodiploids from background growth and to visualize merodiploid resolution events, the most versatile sacB-based method developed to date (9) uses a xylE (catechol-2,3-dioxygenase) indicator gene for visualization of merodiploid formation and resolution events. As the catechol substrate cannot be added to plates, the method requires misting of plates with the substrate solution to detect yellow colonies, which is cumbersome and potentially hazardous when working under biosafety level 3 conditions. In contrast, X-Gluc is a readily available indicator substrate that can be incorporated directly into growth media and offers a considerably safer and more sensitive method for detection of merodiploid formation and resolution. (ii) The pEXKm5 vector possesses a multiple cloning site within the lacZα coding region, allowing blue-white screening of recombinants in E. coli on media containing the indicator X-Gal. (iii) pEXKm5 contains a select agent compliant Km^r marker. In contrast, the system described by Logue et al. (15) is not select agent compliant and cannot be used by U.S. researchers because it uses a chloramphenicol resistance marker. To date we have tested pEXKm5 only with B. pseudomallei, but sacB-mediated gene replacement has been previously demonstrated in B. mallei (9), and there is no obvious reason why pEXKm5 should not be useful for genetic manipulation of other Burkholderia spp.

Conclusion.

Although several allele replacement methods have recently been described for B. pseudomallei, they all entail some shortcomings, for instance, a requirement for an antibiotic resistance marker for selection of chromosomal integration of the mutated sequences (4, 32), defined minimal medium for selection (1, 20), lack of versatile merodiploid detection systems (1, 9), or utilization of a select agent noncompliant selection marker (15). To overcome these shortcomings, we constructed a new allele replacement vector, pEXKm5. This vector provides ultimate versatility in terms of (i) cloning of recombinant DNA fragments using X-Gal-based blue-white screening, (ii) detection of merodiploid formation using X-Gluc-based blue-white screening, and (iii) choice of method for merodiploid resolution by combining I-SceI site delivery potential (and thus implementation of homing endonuclease-based allele replacement) with the familiarity and ease of use of sacB-based plasmids. The sacB-based method is more efficient for mutant generation than the I-SceI-based method, e.g., with purM the efficiencies resolving merodiploids to the ΔpurM genotype were ∼22% and 38% with I-SceI and sacB, respectively. An advantage of the sacB-based method over the homing endonuclease-based method is that it requires only one plasmid and is thus more expeditious than the latter method. While the I-SceI site delivery feature of pEXKm5 may look superfluous and may not be routinely used, it provides flexibility in those instances where sucrose counterselection fails for whatever reason(s). Both methods circumvent the need for specific mutant strains that are sometimes required for counterselection-based allele replacement strategies. For example, streptomycin-based counterselection (30) only works in mutants with chromosomal rpsL mutations, and in B. pseudomallei such mutants can only be obtained in an AmrAB-OprA mutant background, as wild-type strains expressing this pump are resistant to high levels (>1 mg/ml) of streptomycin (19). The necessity for a Δ(amrAB-oprA) rpsL double mutant does not allow facile mutant generation in various B. pseudomallei strain backgrounds, most notably clinical and environmental isolates, which are generally more refractory to genetic manipulation. The pheS-based counterselection for Burkholderia spp. requires addition of p-chloro-phenylalanine and use of a minimal medium lacking phenylalanine (1), which complicates use of this method with some strains, including genetically engineered or natural auxotrophs and slow-growing strains. To facilitate and enable the routine use of pEXKm5 in allele replacement experiments, an improved E. coli mobilizer strain was developed that is susceptible to all known antibiotics and allows nutritional counterselection on rich media. Although we had some success in transferring recombinant pEX plasmids to B. pseudomallei when employing a rapid electroporation procedure (4), the transformation method is much less efficient and reliable than the conjugation method.

In summary, the combination of E. coli mobilizer strain RHO3 and pEXKm5 provides the most versatile and expeditious allele replacement system developed for B. pseudomallei to date. The I-SceI- and sacB-based allele replacement strategies have been shown to work with all B. pseudomallei strains tested thus far.