Introduction

Bacterial natural transformation, first described in 19281, is the process in which exogenous DNA is taken from the environment by a recipient for nutrients, DNA repair, or genetic diversification2. Since then, this process has been described in 82 bacterial species including both gram-positive and gram-negative bacteria3. The molecular machinery that facilitates natural transformation is homologous to the type II secretion system (T2SS) and the type IV pilus (T4P)2,3. In gram-negative bacteria, DNA is transported across the outer membrane through the PilQ channel that houses the pseudopilus, PilE3, and is then shuttled across the periplasm by ComE4. An unknown nuclease generates a single-stranded DNA molecule that is transported into the cytoplasm through the ComA channel in the inner membrane3. The single-stranded DNA molecule is further broken down into nucleotide components or recombined into the chromosome of the recipient organism2,5. The ability of bacteria to be naturally transformable impart an evolutionary advantage and has driven diversification of species over time2,6.

Burkholderia pseudomallei (Bp) is the causative agent of the tropical disease melioidosis that presents in patients with diverse symptoms and clinical outcomes7. Bp is endemic to tropical regions around the world and readily isolated from the environment8. Different clinical and environmental isolates show a significant level of genetic diversity in part due to frequent recombination9,10,11. Clinical isolates have evolved within a host by removal of virulence loci and stress response regulators leading to asymptomatic infection12. Another example of recombination that occurred within the host is the evolution of a single Bp isolate into B. mallei (Bm), the causative agent of glanders13. Although glanders is primarily an equine disease, it also affects humans14 and is a public health concern due to its past use as a bioterrorism agent15,16. Beyond melioidosis and glanders, other members of the Burkholderia genus also cause severe diseases in humans. The Burkholderia cepacia complex comprises of many species within the Burkholderia genus that cause a rapid degradation of pulmonary function leading to high mortality rates in cystic fibrosis patients17,18,19. Burkholderia species also encode multiple forms of antimicrobial resistance mechanisms20,21,22 further complicating treatment of these diseases and highlighting the need for increased investment in basic research of these organisms at the genetic level.

Among all forms of useful genetic manipulations, techniques have been developed for the rapid generation of chromosomal deletion mutants23,24,25. Although these techniques are less cumbersome than traditional allelic-replacement strategies, they rely on the natural transformative properties of the background strain, limiting the utility of these methods23,24,25. To our knowledge, there are no naturally competent strains of B. mallei or within the B. cepacia complex yet described. Thus, the ability to make these Burkholderia species uptake DNA would be very significant in manipulating their genomes. Some Bp strains are naturally transformable (i.e., ~50% of Bp strains)5,24. Bp strain 1026b is able to naturally uptake extracellular DNA allowing for easy genetic manipulation23,24,25. Prototype strains Bp K96243, Bm ATCC23344 and B. cenocepacia (Bc) K56-2 are commonly used in the field but are non-naturally competent making genome manipulation tedious and requiring many steps. To further investigate the natural transformation mechanisms of some Bp strains and increase the utility of natural transformation-based genetic manipulation techniques, our lab has sought to identify specific genetic properties that confer this phenotype5.

Four fosmids containing genomic regions from naturally competent Bp 1026b were previously isolated and able to confer natural transformation to non-naturally competent Burkholderia species5. The fosmids isolated encode ~30–40 genes each, which allowed the non-naturally competent Bp K96243, Bm ATCC23344 and Bc K56-2 to uptake gfp DNA and grow on DNA as a sole carbon source5. A bioinformatics analysis of each genomic region revealed several candidate genes for natural competency5. Mutation in five of these genes, in the naturally competent Bp 1026b background, led to a reduction in growth on DNA and gfp DNA uptake indicating their involvement in natural transformation5. In the present work, we pursued further characterization of the genetic regions of these fosmids and identified the minimal components necessary for natural competency for the purpose of creating a possible broad-species-range strategy for genome manipulation. Additionally, we exploited these genetic elements to expand the λ-Red-recombineering system for rapid chromosomal manipulation into non-naturally competent Burkholderia species.

Results

Downsized fosmids identify genes responsible for Burkholderia natural competency

We initially planned to digest each fosmid (~50 Kbp5,), maintained in non-competent Bp K96243 or Bc K56-2, into smaller genetic fragments in order to identify the minimal number of genes necessary for natural competency. However, working with 50 Kbp inserts on fosmids to pinpoint a subset of genes responsible for natural competency is an arduous task. Therefore, we reintroduced and passaged the fosmid clones in the Bc K56-2 background, while maintaining selective pressure for DNA utilization, in anticipation that natural downsizing of the fosmid clones would occur. Upon digestion of the fosmids Bp1 and Bc175, after passage and re-transformation into Escherichia coli, it was discovered that each fosmid from E. coli had been significantly reduced in size (data not shown). The downsized fosmids were re-introduced into Bc K56-2 and growth on DNA was confirmed. The downsized fosmids Bp1 and Bc17 contained two open reading frames, BP1026B_I0804 and BP1026B_II2056, respectively. Interestingly, BP1026B_I0804 is highly similar to the known competence protein ComE, which has a high amino acid similarity (56%) and identity (38%) to ComEA of Neisseria meningitidis, a model organism for natural competency. Mutation in BP1026B_I0804 (comE) significantly reduced the ability of naturally competent Bp 1026b to grow on DNA as a sole carbon source and to uptake exogenous DNA5. BP1026B_II2056 (crp) is a putative transcriptional regulator of the Crp/Fnr family that also showed critical involvement in DNA uptake and catabolism in Bp 1026b5. Looking at the available genomes of Bp, both comE and crp of Bp 1026b exist in 99 and 198 available Bp genomes, respectively, even non-naturally competent strains. Bm ATCC23344 has comE and crp homologs to Bp 1026b at 100% identity while Bc K56-2 has homologs with 66.67% and 73.98% identity, respectively. Moving forward, the natural downsizing event of these fosmids, down to comE and crp, led us to further investigate the possible transfer of this heritable trait in other Burkholderia species.

comE and crp allow DNA uptake, utilization, and expression

To further investigate the roles of comE and crp in Burkholderia natural transformation and competency, we constructed strains that conditionally express each of these proteins, individually and in combination, under the control of a rhamnose-inducible promoter. To reduce the chance that the introduced version of comE, crp, or comE-crp recombine with the native copy of comE or crp, we exchanged codons throughout each gene to ensure that the nucleotide sequence differed significantly while the amino acid sequence remained unchanged (Supplementary Fig. S1). These combinations were inserted into the attTn7 site in a diverse group of non-naturally competent pathogenic Burkholderia species including Bp K96243, Bm ATCC2344, and Bc K56-226,27. No growth differences were observed when these engineered strains were tested in LB or M9 minimal glucose (MG) media containing rhamnose (Fig. 1a,b). Additionally, empty vector controls (Prha) of non-naturally competent strains Bp K96243, Bm ATCC23344, and Bc K56-2 were unable to grow in media containing DNA as a sole carbon source (red lines in Fig. 1c). However, single copy expression of comE, crp, or comE-crp enabled these non-naturally competent Burkholderia strains to grow in DNA to various degrees (Fig. 1c). Although the expression of comE, crp, or comE-crp in these strains did not allow growth to similar levels as the naturally competent Bp 1026b, they afforded these non-naturally competent Burkholderia strains the ability to significantly uptake DNA as a carbon source and sustain observable growth.

Figure 1
figure 1

In vitro growth characteristics of Bp K96243, Bm ATCC23344 and Bc K56-2. Site-specific recombination at the attTn7 site was used to insert comE, crp, or comE-crp driven by the rhamnose-inducible promoter (Prha). All strains were tested at 37 °C while shaking in LB (a), M9 minimal media supplemented with 20 mM glucose (b), or minimal media supplemented with 0.1% purified salmon sperm DNA (c). All media contained 0.2% rhamnose to express genes inserted in the attTn7 site. The naturally competent Bp 1026b is shown as a point of reference for growth on DNA as a sole carbon source (c).

Full size image

Beyond DNA catabolism, we tested DNA uptake and expression characteristics provided by comE, crp, or comE-crp to these non-naturally competent Burkholderia species (Fig. 2). Each engineered strain was incubated with linear gfp DNA24 and the level of GFP uptake and transient expression of GFP was quantitated using flow cytometry. To establish GFP detection parameters, an E. coli strain constitutively expressing GFP was compared to an E. coli strain with no GFP expression. The E. coli strain constitutively expressing GFP showed 96.8% of cells GFP+, while the E. coli strain that is gfp-, showed no GFP expression (Fig. 2a). When incubated with gfp DNA, 75.7% and 63.7% of naturally competent Bp 1026b and Burkholderia thailandensis (Bt) E264 were able to uptake and express GFP, respectively (Fig. 2a). Non-naturally competent Bp K96243, Bm ATCC23344, and Bc K56-2 strains displayed no GFP expression (Fig. 2b, WT and Prha columns). In contrast, these non-naturally competent strains that express comE, crp, or comE-crp showed GFP expression ranging from 39.7% to 73%, demonstrating that comE or crp is able to confer DNA uptake and subsequent expression of exogenous DNA (Fig. 2b). The capability of DNA uptake enabled by comE and crp made them promising candidates for testing in conjunction with genetic manipulation techniques that rely on natural transformation in non-naturally competent strains. In the presence of comE-crp, the high frequencies of cells taking up gfp DNA by Bc K56-2 (69.9% of cells) and Bm ATCC23344 (73% of cells) were comparable to wildtype Bp strain 1026b (75.7% of cells) and Bt strain E264 (63.7% of cells, Fig. 2). Because manipulating and modifying the genomes of Bp 1026b and Bt E264 has been highly dependent upon the ability to uptake DNA by natural competency24,25 the high frequencies of this heritable trait (Fig. 2) indicate that genetic manipulation of the genome of Bm and Bc is very much possible. As proof-of-concept, we next utilized comE-crp to manipulate the genome of four different Bm strains. A codon-altered version of comE-crp was synthesized to prevent recombination between comE-crp on plasmid and the native genomic comE and crp copies.

Figure 2
figure 2

Linear gfp DNA uptake assay of strains expressing comE, crp, or comE-crp. For all plots side scatter (SSC) is plotted against GFP fluorescent intensity (GFP). (a) Naturally competent Bp 1026b and Bt E264 show 63–75% of cells expressing GFP after incubation with gfp. E. coli constitutively expressing gfp (GFP+) shows 96.8% of cells expressing GFP in contrast to wildtype E. coli (GFP-) showing no GFP expression. (b) Wildtype (WT) and attTn7 controls (Prha) of Bp K96243, Bm ATCC23344 and Bc K56-2 show no GFP expression indicating their inability to uptake gfp. However, expression of comE, crp or comE-crp empowered natural competency, showing 39–73% of cells expressing GFP.

Full size image

Genetic manipulation of non-naturally transformable Burkholderia spp

To expand the use of genetic manipulation techniques that rely on natural transformation23,24,25, we created pKaKa3, pKaKa4, and pKaKa5 where each only differ in the resistance marker (Fig. 3). These vectors expand the applicability of the λ-red recombineering system24 to non-naturally competent Burkholderia species, including type strains Bp K96243, Bm ATCC23344, and Bc K56-2. Incorporation of the codon-altered version of comE-crp into the λ-Red-recombineering system will allow DNA uptake and rapid generation of mutants in strains that are non-naturally competent. As proof of concept, we tested this newly designed recombineering system in the non-competent select-agent Bm using pKaKa4. The gene encoding aspartate-semialdehyde-dehydrogenase (asd) was targeted to generate potentially attenuated strains that could also be useful to the research field28. The vector pKaKa4 was introduced into a variety of Bm strains (ATCC23344, Ivan, China 5, and 2002721278), and the chromosomal asd gene was deleted by incubating with a DNA fragment containing a gat-pheS-FRT cassette flanked by 45 bp regions homologous to Bm asd. Glyphosate-resistant colonies of Bm strains were purified and their diaminopimelate (DAP) requiring phenotype verified (data not shown). The pKaKa4 plasmid with the sacB gene was cured by counter-selection on sucrose. Recombinant efficiencies varied among different Bm strains but generally, 10–50 colonies were obtained from a typical experiment when approximately 5 × 108 to 1 × 109 CFU were used. The introduction of comE-crp enabled Bm to uptake and recombine DNA, which was previously impossible. Although the focus of the present study was to conditionally attenuated Bm strains, we have also utilized these genetic tools successfully in non-naturally competent Bc K56–2 and Bp K96243 to manipulate their genomes with similar frequencies of recombinants.

Figure 3
figure 3

Plasmid maps of pKaKa3, pKaKa4, and pKaKa5 to expand the utility of λ-red recombineering to non-naturally competent Burkholderia species. Abbreviations: araC on pKaKa3 and pKaKa5, activator of the arabinose-inducible promoter (Para) from E. coli; araBCDEFGHI on pKaKa4, B. thailandensis arabinose utilization operon38; gam-exo-bet, λ-red recombineering genes39; mob; RP4-dependent conjugal origin of transfer of B. bronchiseptica cryptic plasmid pBBR1; ori-rep; bhr replicon of B. bronchiseptica pBBR1 plasmid40; nptII, encodes kanamycin resistance41; Para, arabinose inducible promoter42; Prha, rhamnose inducible promoter43; PCS12, constitutive promoters of B. pseudomallei and B. cenocepacia rpsL gene44; pheS, engineered gene encoding a mutant version of α-subunit of phenylalanyl tRNA synthase45; rhaR and rhaS, regulators of the rhamnose inducible promoter43; sacB, encoding for a modified levansucrase counter-selectable marker46. TcR, tetracycline resistance.

Full size image

Attenuation of Bm Δasd mutants in intracellular replication and acute glanders models

To determine the level of attenuation of the four Bm Δasd strains produced using the natural transformation properties of comE-crp, we first tested them in a RAW264.7 murine macrophage model of infection28. RAW264.7 cells were infected with wildtype Bm ATCC23344, Ivan, China5, 2002721278 and the Δasd mutants of each, at an MOI of 1:1 in a modified kanamycin protection assay in order to assess each mutants ability to infect intracellularly. Wildtype Bm strains ATCC23344 (Fig. 4a), Ivan (Fig. 4b), China 5 (Fig. 4c), and 2002721278 (Fig. 4d) were able to replicate to high levels intracellularly, while all Bm Δasd strains behaved as expected, showing no replication within the intracellular environment where no DAP was present. Single-copy complementation of each Bm Δasd strain recovered this defect, indicating that the defect in intracellular replication was due to the deletion of the asd gene (Fig. 4, triangles).

Figure 4
figure 4

In vitro attenuation of Bm Δasd strains in RAW264.7 murine macrophages. Bm strains ATCC23344 (a), Ivan (b), China 5 (c), 2002721278 (d) were able to replicate well within the intracellular environment (circles) while the Δasd counterparts showed complete abolishment of the ability to replicate intracellularly (squares). Complementation of the Δasd gene in each Bm strain rescued the intracellular replication defect (triangles).

Full size image

In addition to in vitro attenuation, we sought to test the Bm Δasd strains in an acute glanders model. To best mimic inhalation glanders, we infected BALB/c mice via intranasal inoculation with an intentionally high-dose of each strain that leads to acute pneumonic glanders. Groups of five mice were inoculated with 1 × 107 CFU of wildtype Bm strain or its Δasd counterpart and survival was monitored. Mice infected with wildtype Bm strains ATCC23344 (Fig. 5a), Ivan (Fig. 5b), China 5 (Fig. 5c), and 2002721278 (Fig. 5d) rapidly deteriorated showing severe symptoms of acute glanders and had to be euthanized within the first four days of the trial. In contrast, BALB/c mice inoculated with Bm Δasd strains showed no signs or symptoms of disease and survived until the study was terminated at day 63 (Fig. 5, squares). Bacterial burdens from the lungs, liver, and spleen were assessed in surviving mice to determine any level of Bm Δasd mutant persistence within the host. Organs were homogenized and plated onto LB agar containing DAP. Bm Δasd were not detected in any organ, indicating that the mutant strains were not able to persist within the host (Fig. 5).

Figure 5
figure 5

In vivo attenuation of Bm Δasd strains in BALB/c intranasal challenge. BALB/c mice (n = 5) were challenged intranasally with 1 × 107 CFU of Bm strains ATCC23344 (a), Ivan (b), China 5 (c), 2002721278 (d) and their Δasd counterparts. Survival was monitored for 63 days (left panels). Surviving mice were sacrificed and bacterial burdens from the lungs, liver, and spleen were determined by serial dilution and plating (right panels).

Full size image

Discussion

Natural transformation is a complex process that drives genetic diversification, DNA repair, and DNA catabolism in bacteria2. The amount of bacterial species identified as naturally transformable is increasing as the molecular mechanisms that drive this activity are better understood in model organisms3. Although the exact mechanism and relationship between DNA up-take and catabolism is yet to be determined, we summarized as previously depicted that one strand of the double-stranded DNA is broken down for catabolism and the other strand can enters the cell for transformation5. Only some strains of Bp are naturally competent5,24, a phenomenon that is not unique to Bp, but also found in the emerging plant pathogen Xylella fastidiosa29. The mechanism by which certain strains of Bp are naturally competent is not well understood. Crp has been implicated as a regulatory factor in the process of natural competency in many organisms including Vibrio cholerae30 and Haemophilus influenzae31. We therefore hypothesize that Crp in Burkholderia species plays a critical role in the regulation of competence, supported by the data presented here. The regulation network of Crp in Burkholderia is of critical interest for future studies and could reveal valuable insights into the complex mechanism of natural competency. ComE has been known to be involved in natural competency of many gram-negative organisms3 including N. gonorrhoeae and N. meningitides, which have shown a direct correlation with the copy number of comE and the level of competency32. This supports the conclusion that an additional copy of comE expressed in non-naturally competent Burkholderia would lead to increase levels of DNA uptake and catabolism above the threshold of detection. Although, no significant additive affect was observed when comE and crp were introduced in combination to the non-naturally competent backgrounds, our data did indicate that overall the comE-crp gave rise to a higher ability for DNA catabolism and uptake, compared to comE or crp individually (Figs 1 and 2).

The fosmids previously isolated5 narrowed down the genetic elements for DNA uptake and utilization to ~50Kbp. The fosmids themselves were naturally downsized further during selection in Burkholderia and maintenance in E. coli, leading us to investigate the role of the remaining genetic elements, comE and crp. Individually and in combination, comE and crp conferred the ability for non-naturally competent Bp K96243 to grow on DNA as a sole carbon source. In addition to DNA catabolism, the ability to uptake and express gfp DNA was also observed, solidifying the role that comE and crp play in Burkholderia natural transformation. To broaden the scope of these findings, we investigated a diverse range of non-naturally competent Burkholderia species that are of public health concern. These include the closely related but distinct Bm, the etiological agent of glanders, and the more distantly related Bc, one of the agents that cause the cepacia syndrome in cystic fibrosis patients. The expression of comE and crp in both Bm ATCC23344 and Bc K56-2 conferred the ability to catabolize DNA and uptake and express exogenous gfp DNA, indicating that the competence machinery in Burkholderia species is likely similar.

Researchers have gravitated toward utilizing Bp 1026b and Bt E264 because these are naturally competent and, therefore, easier to manipulate genetically. However, Bp K96243 was one of the first Bp genomes sequenced and is a prototype strain. The genetic manipulation of Bp K96243 has been limited due to its inability to uptake DNA efficiently. Likewise, the genetic manipulation of Bm strains and Bc K56-2 has been tedious because of the inefficiency in DNA uptake. We showed here Bp K96243, Bc K56-2, and Bm strains can inherit the high frequencies of gfp DNA uptake comparable to wildtype Bp 1026b and Bt E264 (Fig. 2), alleviating the difficulty in manipulating the genomes of these bacteria to knock-out and pull-out genomic sequences24.

In the present study, we also developed genetic tools to expand the λ-Red recombineering system to include non-naturally competent Burkholderia species and strains. Three different λ-Red recombineering vectors were constructed based on various antibiotic and non-antibiotic selective markers, as well as counter-selective markers for curing of the vectors, making them broad-host-range. As a proof of concept, pKaKa4 was used successfully to mutate the asd gene from four strains of Bm. These mutants show complete attenuation in cell culture and BALB/c models of infection and a request for the exclusion from the select-agent list has been submitted and approved by the CDC (https://www.selectagents.gov/SelectAgentsandToxinsExclusions.html). The exclusion of these Bm strains will help accelerate the study of glanders and could be of great interest to the research field. Furthermore, development of these novel genetic tools significantly simplifies the genetic manipulation in many other non-naturally competent Burkholderia species/strains, allowing high-throughput targeted chromosomal manipulation.

Methods and Materials

Bacterial strains, media and culture conditions

All manipulation of Bp and Bm were conducted in a CDC-approved and -registered BSL3 facility at the University of Hawaii at Manoa (UHM). All select agent experiments were approved by the Institutional Biosafety Committee of UHM (reference number: 16–07–004–585–1 R) and were performed using BSL3 practices following recommendations set forth in the BMBL, 5th edition33. Escherichia coli strain EPMax10B (BioRad), E1869, and E1354 were routinely used for cloning or plasmid mobilization into Bp, Bm and Bc as described previously27,34. Luria-Bertani (LB) medium (Difco) or 1x M9 minimal medium supplemented with 20 mM glucose (MG) or 0.1% salmon sperm DNA was used to culture all strains. Induction of genes controlled by the rhamnose-inducible promoter (Prha) was done as previously described24. Selection of the gat gene in E. coli and Bp strains was performed as previously described34. Fosmid downsizing: fosmids5 were isolated from E. coli and introduced into Bc K56–2 and selected on plates with DNA as a sole carbon source. Colonies of Bc K56–2 containing the fosmids from DNA plates were grown up in liquid media with DNA, fosmids were re-purified from the liquid cultures and re-transformed into E. coli and tested for a downsizing event. Downsized fosmids were tested for growth on DNA in Bc K56-2 as previously described5.

Molecular methods and reagents

Molecular methods and reagents were carried out as described previously24,28,34,35. Versions of both comE and crp genes were designed to avoid recombination between the introduced copy of comE-crp and the native genomic comE and crp. To achieve this, codons were swapped throughout each gene to change the nucleotide sequence without altering the amino acid sequence (Supplementary Fig. S1). Newly designed comE and crp genes were synthesized through Genscript®. Strains conditionally expressing comE, crp, and comE-crp were constructed utilizing mini-Tn7 integration vectors26,27,34. Briefly, the rhamnose inducible promoter fragment was PCR amplified from pFlpe423 using oligos 5′-CATATGCATTTAATCTTTCTGCGA-3′ and 5′-CGACTAGTGGATATCGAACTGGCTCATG-3′, digested with NsiI and SpeI, and cloned into mini-Tn7-gat34 digested with the same enzymes, yielding mini-Tn7-gat-Prha. Newly synthesized comE, crp, and comE-crp were cloned into mini-Tn7-gat-Prha as BamHI/HindIII, HindIII/SpeI-blunted, and BamHI/SpeI-blunted fragments, respectively. These plasmids were conjugated into non-naturally competent Burkholderia strains and insertion into the attTn7 site was screened as previously described26,27.

Growth analysis of Burkholderia species

All strains were first grown overnight in LB at 37 °C, bacteria were harvested and washed twice with 1xM9 minimal media and subcultured 1:200 into fresh LB, M9 minimal media supplemented with 20 mM glucose, or minimal media supplemented with 0.1% purified salmon sperm DNA (Fig. 1). All media contained 0.2% rhamnose to express genes inserted in the attTn7 site. Growth curves were done using the BioTek ELx808IU by measuring OD630 every 30 minutes for the duration of the time course. Growth analysis was done in triplicate and average ODs were shown.

Gfp uptake assays

Gfp uptake assays were performed as previously described5,24, with the exception of the detection method. Briefly, gfp-DNA was amplified by PCR from pPS74736 and 250 ng of the gfp-DNA was incubated with various strains for 30 min at room temperature5. After 45 min recovery in LB broth with shaking, bacteria were fixed in 1% paraformaldehyde in 1x phosphate buffered saline (PBS) for 45 min for fluorescent analysis. After fixation, bacteria were harvested and resuspended gently with 1xPBS + 0.1% Triton X-100 to reduce clumping, and then washed twice with 1xPBS to remove detergent. Fixed bacteria were analyzed using flow cytometry to detect transient expression of GFP-protein indicating that cells were able to uptake extracellular gfp-DNA, along with fixed E. coli wildtype strain DH5α and DH5α/attB::Gm-gfp as negative and positive controls, respectively.

λ-Red knockout recombineering in Bm with pKaKa4

Generation of mutants was done as previously described24 with slight modifications. Briefly, pKaKa4 was introduced into various Bm strains via conjugation and selection on M9 minimal media containing 40 mM arabinose as the sole carbon source. Bm strains harboring pKaKa4 were streaked out on M9 + arabinose plates and grown for 3 days at 37 °C, then harvested from plates by gentle scraping and resuspended in fresh LB containing 0.2% rhamnose. Bacteria were then concentrated by centrifugation and resuspended in 20 μl LB + 0.2% rhamnose, and incubated with 2 μg of DNA containing a gat-pheS-FRT cassette flanked by 45 bp regions homologous to Bm asd. After incubation at room temperature for 30 min, bacteria were recovered in fresh LB for 2 hours at 37 °C, and selected on MG medium containing 200 μg/ml DAP, 0.4% GS, and 1 mM each of lysine, methionine, and threonine (these 3 amino acids are required for the specific asd mutation).

Intracellular replication assays

RAW264.7 murine macrophages were grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% (v/v) fetal bovine serum (FBS) at 37 °C in 5% CO2. Antibiotic/antimycotic (Gibco) containing 100 U/mL penicillin, 100 µg/mL streptomycin and 250 ng/mL of amphotericin B was added to media at a 1X concentration during cell growth but omitted during infection trials. A modified kanamycin protection assay was used test intracellular replication37. RAW264.7 cells were seeded into 24-well Corning CellBIND culture plates to 80% confluence, allowed to attach overnight, and were washed twice with 1XPBS before infection. Bm Δasd strains were used to infect macrophage monolayers at an MOI of 1:1. After 1 hour, infected monolayers were washed with 1XPBS and then DMEM supplemented with 10% FBS, 700 µg/mL amikacin and 700 µg/mL kanamycin were added to kill any extracellular bacteria. At 2, 6, 12, and 24 hours post-infection, infected monolayers were lysed with 0.1% Triton X-100. Serial dilutions of lysates were plated on LB containing 200 µg/mL DAP and colony forming units (CFU) per well were determined.

Animal studies

BALB/c mice between 4 and 6 weeks of age were purchased from Charles River Laboratory. All infections with Bm strains were administered via the intranasal (i.n.) inoculation route. Mice were anesthetized with 100 mg of ketamine/kg of body weight plus 10 mg/kg xylazine. The challenge dose (1 × 107 CFU) of each Bm strain was suspended in 20 µl of 1XPBS and used to inoculate each mouse via the i.n. route. Each strain was used to inoculate 5 mice. Animals were monitored for disease symptoms daily and euthanized at predetermined humane end points. Lungs, liver, and spleen of surviving mice were harvested, homogenized, serially diluted, and plated on LB containing 200 µg/mL DAP to determine bacterial burdens. Survival characteristics were plotted using Prism software (GraphPad, La Jolla, CA) and statistical analysis was done by Kaplan-Meier curves.

Ethics statement

All animal studies described in this manuscript were approved by the Institutional Animal Care and Use Committee at the University of Hawaii at Manoa (Protocol No. 10-1073-8), and conducted in compliance with the NIH (National Institutes of Health) Guide for the Care and Use of Laboratory Animals.