Molecular Medicine Israel

LVS ΔcapB-vectored multiantigenic melioidosis vaccines protect against lethal respiratory Burkholderia pseudomallei challenge in highly sensitive BALB/c mice


Melioidosis, caused by the intracellular bacterial pathogen and Tier 1 select agent Burkholderia pseudomallei (Bp), is a highly fatal disease endemic in tropical areas. No licensed vaccine against melioidosis exists. In preclinical vaccine studies, demonstrating protection against respiratory infection in the highly sensitive BALB/c mouse has been especially challenging. To address this challenge, we have used a safe yet potent live attenuated platform vector, LVS ΔcapB, previously used successfully to develop vaccines against the Tier 1 select agents of tularemia, anthrax, and plague, to develop a melioidosis vaccine. We have engineered melioidosis vaccines (rLVS ΔcapB/Bp) expressing multiple immunoprotective Bp antigens among type VI secretion system proteins Hcp1, Hcp2, and Hcp6, and membrane protein LolC. Administered intradermally, rLVS ΔcapB/Bp vaccines strongly protect highly sensitive BALB/c mice against lethal respiratory Bp challenge, but protection is overwhelmed at very high challenge doses. In contrast, administered intranasally, rLVS ΔcapB/Bp vaccines remain strongly protective against even very high challenge doses. Under some conditions, the LVS ΔcapB vector itself provides significant protection against Bp challenge, and consistent with this, both the vector and vaccines induce humoral immune responses to Bp antigens. Three-antigen vaccines expressing Hcp6-Hcp1-Hcp2 or Hcp6-Hcp1-LolC are among the most potent and provide long-term protection and protection even with a single intranasal immunization. Protection via the intranasal route was either comparable to or statistically significantly better than the single-deletional Bp mutant Bp82, which served as a positive control. Thus, rLVS ΔcapB/Bp vaccines are exceptionally promising safe and potent melioidosis vaccines.


Melioidosis, a major neglected disease caused by the intracellular bacterial pathogen Burkholderia pseudomallei, is endemic in many tropical areas of the world and causes an estimated 165,000 cases and 89,000 deaths in humans annually. Moreover, B. pseudomallei is categorized as a Tier 1 select agent of bioterrorism, largely because inhalation of low doses can cause rapidly fatal pneumonia. No licensed vaccine is available to prevent melioidosis. Here, we describe a safe and potent melioidosis vaccine that protects against lethal respiratory challenge with B. pseudomallei in a highly sensitive small animal model—even a single immunization is highly protective, and the vaccine gives long-term protection. The vaccine utilizes a highly attenuated replicating intracellular bacterium as a vector to express multiple key proteins of B. pseudomallei; this vector platform has previously been used successfully to develop potent vaccines against other Tier 1 select agent diseases including tularemia, anthrax, and plague.


Melioidosis, a disease endemic in many tropical areas of the world, with ~165,000 cases and 89,000 deaths per year (12), is caused by the intracellular bacterial pathogen Burkholderia pseudomallei (Bp). Infection with Bp occurs via inhalation, ingestion, and entry through broken skin (13). Most natural disease is thought to occur via percutaneous inoculation with contaminated soil or water and also via inhalation (34). Melioidosis can present as an acute infection (85% of cases); a chronic infection with symptoms lasting >2 months (11% of cases); or re-activation from latency (4% of cases) (5). Even in a high-resource setting, mortality from naturally acquired melioidosis is ~10%, and where resources are more limited, mortality is ~40%. For acute pneumonia with septic shock, mortality is very high (up to 90%). Prolonged treatment is required—a minimum of 10–14 days of intravenous antibiotics, followed by 3–6 months of oral antibiotics to prevent relapse (5). In addition to its significant public health burden, Bp is categorized as a Tier 1 select agent by CDC. Bp is easily aerosolized and, given the high mortality of pneumonic melioidosis, transmission via inhalation is the route of greatest concern in a bioterrorist attack; inhalation of even low doses of Bp is rapidly fatal in animal models.

In view of Bp’s significant public health burden and potential for weaponization, a potent vaccine against Bp is needed, but currently there are no licensed vaccines. Our approach to developing a safe, effective Bp vaccine that provides long-lasting immunity, is to use a live attenuated intracellular bacterial vector, LVS ΔcapB, to express Bp antigens. LVS ΔcapB is derived from Live Vaccine Strain or LVS, a tularemia vaccine developed in the early 1900s that has been administered to ~60 million people including ~5,000 U,S, laboratory workers. LVS is derived from Francisella tularensis subsp. holarctica, a less virulent subspecies of F. tularensis than the Tier 1 select agent F. tularensis subsp. tularensis. Attenuated by serial passage on artificial medium, LVS has two major attenuating deletions and several minor ones, and yet retains significant toxicity. LVS ΔcapB, with a third major attenuating deletion resulting from knockout of the capB gene, has minimal toxicity; it is >10,000-fold less virulent than LVS when administered to highly sensitive mice intranasally (6). Despite its low virulence, LVS ΔcapB retains the capacity to invade and multiply in host macrophages among other antigen-presenting cells (6). Consequently, in addition to safety, a major advantage of the LVS ΔcapB vector is its ability to induce both potent humoral and broad T-cell (CD4+ and CD8+) responses to expressed antigens (711).

Utilizing the LVS ΔcapB vector platform to express immunoprotective antigens of target pathogens, our laboratory has developed potent vaccines against tularemia, anthrax, plague, and COVID-19 (711). In the current study, to construct vaccines against melioidosis, we selected four promising Bp antigens—Hcp1, Hcp2, Hcp6, and LolC—for expression in LVS ΔcapB; these antigens were selected based upon their (i) documented protective capacity as subunit vaccines in mouse studies (1214); (ii) capacity to generate an immune response in melioidosis patients (Hcp1 and LolC) (121516); and (iii) high sequence conservation in Bp strains. We generated rLVS ΔcapB/Bp vaccines expressing individual Bp antigens as well as fusion proteins comprising 2, 3, or 4 Bp antigens. We adapted the Electra cloning system (17) to facilitate the construction of a large number of fusion protein variants so that we could optimize for heterologous expression of the antigens by the LVS ΔcapB vector. We varied the coding regions (native vs codon-optimized), the order of the coding regions, and the linkers fusing the coding regions in order to identify the best-expressed fusion proteins. By doing so, we achieved good expression of fusion proteins consisting of two or three antigens, with molecular weights of 42–67 kDa; however, the expression of fusion proteins comprising all four antigens (~87 kDa) was relatively poor.

We assessed the efficacy of our rLVS ΔcapB/Bp vaccines in BALB/c mice, a strain of mice especially sensitive to Bp (1820). Moreover, we challenged these mice with lethal doses of highly virulent Bp via the respiratory (intranasal) route, the most difficult route to protect against. This challenge model set a high bar for our vaccines, as very few vaccines have demonstrated significant efficacy against lethal respiratory challenge in BALB/c mice (21). In a series of six independent experiments, we demonstrate that rLVS ΔcapB/Bp vaccines can induce high levels of protective immunity against lethal respiratory challenge in BALB/c mice when the vaccines are administered intradermally or intranasally. When mice were immunized by the intradermal (ID) route, the protective efficacy of the rLVS ΔcapB/Bp vaccines was overwhelmed at high challenge doses, but when mice were immunized by the intranasal (IN) route, protection remained strong even at high challenge doses. Intranasal vaccination resulted in protective efficacy greater than the positive control vaccine, Bp82, a live attenuated Bp strain with a single major attenuating deletion (22) and hence unsuitable for clinical use, due to safety concerns regarding reversion to virulence of a vaccine comprising an attenuated bacterial pathogen when only one major attenuating mutation is present (23). While most of our studies utilized a homologous prime-boost vaccination strategy (three doses, 4 weeks apart), even a single intranasal immunization was highly protective. Surprisingly, the LVS ΔcapB vector by itself was sometimes capable of eliciting potent protection, particularly when delivered by the intranasal route. rLVS ΔcapB/Bp vaccines expressing three-antigen fusion proteins Hcp6-Hcp1-Hcp2 or Hcp6-Hcp1-LolC were among the most potent vaccines tested.


Bacterial strains and media

The bacterial strains used in this study are listed in Table S1. Escherichia coli was grown on Luria-Bertani or YT agar and Luria-Bertani broth at 37°C. Ampicillin (100 µg/mL) and kanamycin (30 or 50 µg/mL) were included as appropriate. Bp82 was grown on Luria-Bertani (Lennox) agar plates containing 0.6 mM adenine at 37°C. LVS ΔcapB and rLVS ΔcapB/Bp strains were grown on chocolate agar plates (CA; Difco GC Medium Base + 1% [wt/vol] bovine hemoglobin + 1% [vol/vol] IsoVitaleX enrichment) and modified medium T broth (2425) at 37°C (we substituted N-Z-Amine A [enzymatic digest of casein] for the Casamino Acids [acid hydrolysate of casein] component of the formula). Kanamycin (7.5 µg/mL) was included for rLVS ΔcapB/Bp strains maintaining pFNL plasmids. Bp 1026b was grown on BHI plates and colonies were harvested in BHI broth 24 h after plating. The Bp 1026b suspension was vortexed to a uniform appearance, glycerol was added with mixing to 15%, and 0.5 mL aliquots were frozen rapidly and stored at −80°C.

Construction of Electra compatible E. coli-Francisella shuttle plasmids

To facilitate cloning and expression analysis of Bp antigens (individually or as fusion proteins), we first constructed three Electra compatible DAUGHTER plasmids. Electra cloning (17) is similar to fragment exchange (FX) cloning (26) and uses SapI, a type IIS restriction enzyme that produces 3 bp overhangs, together with T4 DNA ligase in a single-tube reaction.

To generate Electra compatible DAUGHTER plasmids, we first removed the four SapI restriction sites from the E. coli-Francisella shuttle expression vector pFNL/pbfr-SD-iglA (10), using the QuikChange Lightning Multi Site-Directed Mutagenesis Kit (Agilent; Santa Clara, CA). We used PrimerX software (27) to design four mutagenesis primers and, as all four SapI restriction sites occurred within open reading frames (ORFs), we ensured that the mutations did not affect the protein sequence from the mutated ORF’s translation (see Table S2 for primers). Next, we replaced the iglA coding sequence with three different Electra compatible sacB cassettes, to generate three Electra compatible DAUGHTER plasmids, pFNL-bfr-D1 (sacB), pFNL-bfr-D2[N3F-8H] (sacB), and pFNL-bfr-D3[C8H-3F] (sacB) (Tables S3 and S4). Using Electra cloning, ORFs from Electra compatible MOTHER plasmids can be readily cloned in frame into the three different DAUGHTER vectors, with expression of the ORF driven from the strong Francisella bacterioferritin promoter. Cloning into the D1 vector results in no fusion with the ORF, cloning into the D2 vector results in an N-terminal fusion of a dual 3×FLAG-His8 tag (MDYKDHDGDYKDHDIDYKDDDDKHHHHHHHHGGGS) with the ORF, and cloning into the D3 vector results in a C-terminal fusion of the ORF with a dual His8-3×FLAG tag (GGSHHHHHHHHDYKDHDGDYKDHDIDYKDDDDK). ORFs excised from Electra MOTHER plasmids with SapI have a 5′ overhang of ATG on the coding strand (coding for Met) and an ACC overhang on the complementary strand (compatible with the GGT overhang of the vector, coding for Gly). Thus, ORFs cloned into the D1 or D2 DAUGHTER vectors, in most cases, will have an extra Gly residue on the C-terminus of the protein, unless the native sequence ends in a Gly residue. The presence of the sacB cassette in these vectors (which is replaced by the desired ORF in the cloning reaction), permitted us to use undigested DAUGHTER vectors in Electra reactions instead of gel-purified, linearized vector. Selection of transformants on agar plates with kanamycin and 5% sucrose was highly successful in selecting for recombinant DAUGHTER plasmids as transformants that take up unmodified DAUGHTER plasmids are killed in the presence of sucrose due to expression of the sacB gene.

We used in vivo assembly (IVA) (28) to construct several additional Electra compatible DAUGHTER plasmids derived from the original three plasmids. We first created a new series of Electra plasmids, in which the unnecessary ampicillin resistance gene is removed, by amplifying the original Electra plasmids using primers pFNL-delAmp-F and pFNL-delAmp-R and transforming the polymerase chain reaction (PCR) products into E. coli to generate pFNLdA-bfr-D1 (sacB), pFNLdA-bfr-D2[N3F-8H] (sacB), and pFNLdA-bfr-D3[C8H-3F] (sacB). We also used IVA to construct additional derivatives of these new pFNLdA DAUGHTER plasmids with modified SapI overhangs to allow for cloning of ORFs from MOTHER plasmids which have different SapI overhangs (see further details below). Derivatives with a modified SapI overhang at the 5′ end of the cloning site (GCA or TCT, instead of ATG) have an ATG codon immediately preceding the SapI overhang for translation of the cloned ORF.

Recombinant plasmids were confirmed to be correct by restriction analysis and DNA sequencing.

Electra cloning to construct expression plasmids with a single Bp antigen ORF

We purchased two Electra MOTHER plasmids from DNA2.0 (now ATUM; Newark, CA) containing an ORF encoding the Bp Hcp6 antigen (hcp6, BPSL3105), with one version codon-optimized for Francisella tularensis subsp. holarctica LVS (, and one version codon-optimized for Listeria monocytogenes (, cloned into pM268 and pM264 MOTHER plasmids, respectively.

For construction of additional Electra MOTHER plasmids containing ORFs for other Bp antigens (hcp1 [BPSS1498, hcp2 [BPSS0518], and lolC [BPSL2277]), we first generated plasmid pM264-sacB, a derivative of pM264 with a sacB cassette inserted in between the SapI restriction sites (Tables S2 and S5). This allowed us to use undigested pM264-sacB, rather than gel-purified, linearized vector, in Electra reactions to clone PCR products and synthetic DNA fragments. Transformation of Electra reactions into E. coli and plating on agar plates with ampicillin and 5% sucrose was highly successful in selecting for recombinant MOTHER plasmids, as transformants that take up residual pM264-sacB are killed in the presence of sucrose due to expression of the sacB gene.

For hcp2hcp6, and lolC, we used GeneDesigner 2.0 software (DNA2.0/ATUM) to generate coding sequences codon-optimized for LVS and L. monocytogenes, purchased Electra compatible synthetic DNA fragments for these codon-optimized genes from SGI-DNA (San Diego, CA), and cloned them into the pM264-sacB. In the case of lolC, we only cloned the sequence encoding the periplasmic region (amino acid residues 51–273) (13). For hcp1hcp2, and hcp6, we also amplified the native genes from Bp K96243 gDNA (gift from Christopher T. French) and cloned the PCR products into pM264-sacB. We were unsuccessful in amplifying the native Bp lolC.

For a typical cloning reaction, we would mix a MOTHER plasmid containing a Bp antigen ORF (Table S6), a pFNL DAUGHTER plasmid (Table S3), and Electra Cloning reagents (DNA2.0/ATUM) containing buffer, SapI, and T4 DNA ligase in a final volume of 5 µL and incubate at room temperature for 1 h. We would then use 1 µL of the cloning reaction to transform 10 µL of competent E. coli and select for recombinant clones on YT plates containing kanamycin and 5% sucrose.

Recombinant plasmids were confirmed to be correct by restriction analysis and DNA sequencing.

Electra cloning to construct expression plasmids with multiple Bp antigens joined as a fusion protein

For our first set of fusion constructs, we joined hcp6 or lolC to either hcp1 or hcp2 with a sequence coding for a flexible peptide linker (GSSG-GSSG) in between the two ORFs. To accomplish this, we first amplified hcp6 and lolC by PCR, using a primer for the pM264 vector (pM264-FP) and a primer specific for the gene (hcp6_coLm-L-R, lolC_coLm-L-R, or lolC_coLVS-L-R) which included the sequence for the peptide linker in the primer tail. The PCR products were reamplified in a second PCR using primers pM264-FP and Linker-3gca-R, which allowed the resulting PCR products to be cloned by Gibson Assembly (29) into the SapI cloning sites of the pM264 MOTHER vector, replacing the 3′GGT SapI overhang with a GCA overhang. We then amplified hcp1 and hcp2 by PCR, using a primer specific for the gene (hcp1_nat-5gca-F, hcp1_coLm-5gca-F, hcp1_coLVS-5gca-F, hcp2_nat-5gca-F, hcp2_coLm-5gca-F, or hcp2_coLVS-5gca-F) and a primer specific for the pM264 vector (pM264-RP). The gene-specific primer included 30–33 nt of homology to the pM264 vector allowing the PCR product to be cloned by Gibson Assembly into the pM264 SapI cloning sites, replacing the 5′ATG SapI overhang with a GCA overhang. Fusions of two ORFs (hcp6 plus hcp1 or hcp2; and lolC plus hcp1 or hcp2) were done by combining pM264 MOTHER plasmids with compatible GCA overhangs (Table S6) together with a pFNL DAUGHTER plasmid (Table S3) using Electra cloning to generate the desired expression plasmid (Table S7).

To facilitate the construction of subsequent fusion proteins, we made additional modifications to the Electra cloning system to simplify the in-frame joining of two or three ORFs. Using Gibson assembly or IVA (primer pairs listed in Table S2), we constructed five modified pM264-sacB MOTHER plasmids which vary in the SapI overhangs (Table S5), with the 5′ ATG overhang on the coding strand (coding for Met) and/or the GGT overhang for the 3′ end of the ORF (coding for Gly) being replaced by GCA (Ala) or TCT (Ser). We then cloned ORFs into the modified pM264-sacB plasmids. Finally, we performed Electra reactions combining two or three modified MOTHER plasmids (each with an ORF) and a DAUGHTER plasmid to construct expression plasmids with fusion protein genes. Fusions of two ORFs were done by combining pM264 MOTHER plasmids with compatible GCA or TCT overhangs (see GCA example above). Likewise, fusions of three ORFs were done by combining three MOTHER plasmids with compatible GCA and TCT overhangs together with a pFNL/pFNLdA DAUGHTER.

Sign up for our Newsletter