The recombinant BpOmpW used in all experiments, except for the enzyme-linked immune absorbent spot (ELISpot) analysis of transgenic mice, was expressed, purified, and provided by Lionex GmBH (Braunschweig, Germany) in 20 mM ammonium bicarbonate. In the case of transgenic mouse studies, the pRSET_BpOmpW construct was transformed in BL21(DE3) cells and cultured in Luria–Bertani (LB) medium with 1 M d-sorbitol and 2.5 mM glycine betaine for 5 days at 22°C. The His-tag fusion protein was then purified by nickel affinity chromatography with endotoxin-free phosphate-buffered saline (PBS), 35 mM imidazole, and 2% Triton X-100 and eluted in endotoxin-free PBS containing 250 mM imidazole and 2% Triton X-100. The antigen was further purified by gel filtration chromatography. The affinity chromatography fraction containing the antigen (as identified by SDS-PAGE) was concentrated and loaded onto a HiLoad 16/600 GL Superdex 75 column (GE Healthcare, Chicago, IL, USA) pre-equilibrated in endotoxin-free PBS using an AKTA chromatography system (GE Healthcare). Fractions with the protein of interest were pooled and the protein was concentrated and stored at −80°C until its use. Protein concentration was determined using the bicinchoninic acid (BCA) protein assay kit (Thermo Fisher Scientific, Waltham, MA, USA) and used for the immunization of transgenic mice and in ELISpot assays.

All work involving animals was approved by the University College Dublin Ethics Committee (AREC-19-13-McClean), and mice were maintained according to the regulations of the Health Products Regulatory Authority (Directive 2010/63/EU and Irish Statutory Instrument 543 of 2012), with authorization number AE18982/P166.

For the human plasma samples from melioidosis patients and controls in Northeast Thailand (n = 15, each cohort), the study protocol was approved by the Ethics Committees of the Faculty of Tropical Medicine, Mahidol University (TMEC 12-014); Sunpasitthiprasong Hospital, Ubon Ratchathani (017/2559); and the Oxford Tropical Research Ethics Committee (OXTREC35-15). Blood samples were collected from in-patients with culture-confirmed melioidosis (M), diabetic patients (D), and healthy participants from endemic areas who were household contacts of the melioidosis cases (HH) at Sunpasitthiprasong Hospital, Ubon Ratchathani, Thailand, between 2015 and 2017. All the patients in the M cohort had diabetes. A group of non-endemic individuals (NE; n = 3) was also included in the study. Written informed consent was obtained from all patients enrolled in the study.

Cryopreserved primary PBMCs were isolated from whole blood donated by healthy volunteers. Whole blood was collected from healthy donors as described in the ethical protocol IXP-003_V1 (Belgian registration no. B707201627607) or protocol IXP-004_V1 (Netherlands, reg. no. NL57912.075.16). Written informed consent was obtained from all patients enrolled in the study. All blood samples were tested and found negative for hepatitis B virus (HBV), hepatitis C virus (HCV), and human immunodeficiency virus (HIV). PBMCs were separated from the blood by density gradient centrifugation and subsequently cryopreserved in fetal bovine serum (FBS), supplemented with 10% dimethyl sulfoxide, by controlled rate freezing. The PBMCs were kept in cryogenic storage (−180°C) until use.

Male C57BL/6J mice were used in these studies. Mice were given free access to food and water and subjected to a 12-h light/dark cycle. Groups of C57BL/6J male mice (n = 11) were immunized subcutaneously with 50 μg BpOmpW in Sigma Adjuvant System (SAS; Sigma, St. Louis, MO, USA) or SAS adjuvant alone as a negative control ( Figure 1A ). The antigen was mixed with SAS adjuvant at a 1:1 ratio (50:50 μl per mouse). The antigen was diluted in 20 mM ammonium bicarbonate in the same buffer used for purification. Two weeks later, mice were humanely killed by sedation with an isoflurane vaporizer followed by CO₂ exposure. Blood was withdrawn by cardiac puncture for serological analysis. Sera were extracted from blood and stored at −80°C. Spleens were also harvested and processed for splenocyte re-stimulation assays to assess antigen-specific cell-mediated response to BpOmpW.

Microtiter plates were coated with purified BpOmpW at 0.5 μg/ml in sodium bicarbonate buffer (pH 9.4) at 4°C overnight, after which the coating solution was removed and the plates blocked with 1% FBS (7524; Sigma) solution in PBS at room temperature for 1 h. The wells were washed three times with 0.05% Tween 20 in PBS using a plate washer (5165040; Thermo Scientific). All serum samples were serially diluted (fivefold) in PBS containing 1% FBS and 100 μl added to the wells in triplicate at room temperature (RT) for 2 h. The plates were then washed three times with 0.05% Tween 20 in PBS, as described above, before the addition of anti-mouse immunoglobulin G (IgG), IgG1, or IgG2a–horseradish peroxidase (HRP)-conjugated antibodies (ab97023, ab97240, ab97245, respectively; Abcam, Cambridge, UK) at RT for 1 h. Then, the TMB (3,3′,5,5′-tetramethylbenzidine) substrate was added and incubated until the color developed, at which time the reactions were stopped with 2 M sulfuric acid and the plates read at 450 nm. For the detection of BpOmpW-specific IgG in human plasma, the same protocol was applied using anti-human IgG–HRP antibody (ab6858; Abcam) in place of anti-mouse IgG antibodies. All the HRP-conjugated antibodies were diluted 1:5000 in PBS containing 1% FBS. The color intensity at OD₄₅₀ correlated with the amount of specific antibodies in the sera.

The spleen from each mouse was mechanically disrupted and filtered to obtain a single-cell suspension of splenocytes. Then, an ammonium–chloride–potassium (ACK) lysis buffer was used to remove red blood cells. The splenocytes from BpOmpW-immunized or control mice were then counted automatically using a Countless automated Cell Counter (Invitrogen, Carlsbad, CA, USA). One million cells in triplicate (per mouse) were plated per well and stimulated for 60 h with 50 μg/ml BpOmpW in a 96-well-plate using 10% FBS RPMI medium and 1% P/S (Penicillin-Streptomycin). Between 5 and 6 h before harvesting the cells, 5 μg/ml brefeldin A was added to block cell trafficking in order to increase the accumulation of intracellular cytokines. Cells were collected by centrifugation and stained for flow cytometry. Studies in non-insulin resistant and insulin resistant mice were performed independently at different times.

The stimulated splenocytes were incubated with Fc Block (anti-CD16/CD32; BD Biosciences, Franklin Lakes, NJ, USA) for 5 min on ice and labeled with Viakrome 808 (Beckman Coulter, Brea, CA, USA) and the fluorochrome-labeled antibodies against CD4, CD49b, CD45RB, CD8a, CD25, CD44, and CD3 surface markers (BD Biosciences) for 30 min on ice. Then, intracellular interleukin 2 (IL-2), IL-4, IFN-γ, IL-17, IL-9, tumor necrosis factor (TNF), and Foxp3 (BD Biosciences) were also analyzed with a BD Cytofix/CytoPerm™/Fixation/Permeabilization Solution Kit (BD Biosciences) according to the manufacturer’s instructions. The antibodies were all titrated for optimal performance, and Beckman Coulter VersaComp beads were used for compensation. A Beckman Coulter CytoFLEX LX (NUV full configuration) was used for the analysis of the samples with the gains for each one of the markers optimized by performing a gaintration. Quality control of the instrument was performed using Beckman Coulter Daily QC beads and IR Daily QC as per the manufacturers’ specifications. Fluorescence minus one (FMO) controls were used for every single marker in order to generate the gates, and several FMO controls were used daily for verification purposes. The gating strategy and the corresponding FMO controls for each gate are shown in Figures S3–S9 . Data analysis and reanalysis were performed with Beckman Coulter CytExpert v.2.4. The supplementary Excel file shows the full MIFlowCyt (the minimum information about a flow cytometry experiment). Data were extracted as the percentage of parental cells.

Seventy C57BL/6J male mice were fed a D12492i rodent diet comprising 60 kcal% from fat (Research Diets, Inc., New Brunswick, NJ, USA) or regular chow starting at 6–8 weeks of age until they were humanely killed. Insulin resistance was determined at 8 and 12 weeks by fasting the mice for 6 h, at which time 0.5 U/kg of insulin (Actrapid; Novo Nordisk, Bagsværd, Denmark) was administered subcutaneously. In order to alleviate any pain, EMLA cream was applied to the whole tail 10 min before blood was sampled from the tail vein using a 27-gauge needle, a drop of blood extracted at 15, 30, 45 and 60 min, and the glucose levels measured using the AlphaTRAK^® 2 blood glucose monitor. To characterize the model, we collected the spleen, liver, pancreas, and kidney from the mice and a hematoxylin–eosin (H&E) stain was applied to the cut tissue sections. The time required for individual mice to develop insulin resistance varied, and only mice that showed insulin resistance were considered eligible to be randomly selected for immunization studies.

The HLA-DR and HLA-DQ alleles are associated with melioidosis. This study used HLA class II transgenic mouse lines for the alleles HLA-DR1 (DRB1*0101), HLA-DR4 (DRB1*0401), and HLA-DQ8 (DQB1*0302), which were, in each case, maintained in the context of a homozygous knockout for murine H2-Ab. Mice were maintained in individually ventilated cages and were used in experiments as age- and sex-matched young adults. For CD4 T-cell epitope mapping studies, the mice were primed in one hind footpad with 25 μg antigen emulsified in Hunters Titermax Gold adjuvant (Sigma-Aldrich, St. Louis, MO, USA). On day 10, the draining popliteal lymph node was removed and disaggregated into a single-cell suspension for ELISpot assays. The frequency of cells producing IFN-γ in response to the antigen was quantified with ELISpot (Diaclone; 2B Scientific, Oxon, UK) performed in HL-1 serum-free medium (BioWhittaker; Lonza, Slough, UK) supplemented with l-glutamine and penicillin–streptomycin (Life Technologies, Paisley, UK). Cells plus antigen were added to the wells and plates and were incubated for 72 h at 37°C with 5% CO₂. Unless otherwise indicated, peptide was added to the wells at a final concentration of 25 μg/ml. Spots were counted on an automated ELISpot reader (Autoimmun Diagnostika, Strasbourg, France). Response frequencies were expressed as ΔSFC/10⁶ cells, with the presence of an epitope being confirmed when the majority of the mice in the immunized group responded with a magnitude greater than the mean number of spot-forming cells (SFCs) in medium-only control + 2 SD. The mean + 2 SD background SFCs for murine ELISpot data are indicated in each case by a dotted line in the figures. The ELISpot background range (per 10⁶ cells) was 0–30 SFCs.

PBMCs from 23 healthy human donors were thawed from ImmunXperts biobank and incubated for 7 days at 37°C with BpOmpW (50 µg/ml), SAS (1 µg/ml), and their combination in six replicates. T-cell proliferation was assessed by measuring the incorporation of 5-ethynyl-2′-deoxyuridine (EdU). EdU is a thymidine analogue incorporated into the DNA of dividing cells during the S-phase. On day 6 of culture, the cells were pulsed with EdU at 1 µM for 16 h. The next day, cell culture supernatants from human PBMC stimulation assays were collected and stored at −80°C for cytokine analysis. The remaining cells were fluorescently stained for T-cell surface markers (CD4 and CD8), fixed, permeabilized, and the incorporated EdU was stained with a fluorescent azide to finally perform the analysis using flow cytometry. The gating strategy for this experiment can be found in Figure S11 .

The supernatant was stored at −80°C and then analyzed for the quantification of IFN-γ with LEGEND MAX™ Human IFN-γ ELISA Kit (BioLegend, San Diego, CA, USA) following the manufacturer’s instructions.

The sample size in all cases was determined using G*Power software. The results are presented as mean ± SE, unless otherwise stated. Differences in the means between groups were tested using a t-test or ANOVA using GraphPad Prism, version 7. A p-value <0.05 was considered statistically significant.

T-cell-inducing vaccines are required to induce CD4⁺ and/or CD8⁺ T cells of sufficient effector function that directly participate in pathogen removal via cell-mediated effector mechanisms (25). Therefore, given that T-cell effector functions, especially IFN-γ responses, correlate with survival in acute melioidosis patients (16), we examined the T-cell responses associated with BpOmpW immunization. Groups of 11 mice were immunized once with recombinant BpOmpW ( Figure 1 ). Analysis of the mouse sera following BpOmpW immunization showed strong seroconversion (total IgG, IgG1, and IgG2a) at 2 weeks despite the mice receiving only one immunization ( Figures S1A–C ). Antigen-specific T-cell responses were then determined in splenocytes that had been re-stimulated (treatment group) to antigen in vitro by measuring the cytokine responses and a range of T-cell markers using flow cytometry. These were compared to antigen-exposed splenocytes from the adjuvant-only group (SAS control) ( Figure 1A ). Activation of BpOmpW-re-stimulated splenocytes was demonstrated by a significant decrease in CD45RB expression (p < 0.0001; Figure 1B ), while the levels of CD25 and CD44 were significantly increased in response to BpOmpW re-exposure in both CD4⁺ and CD8⁺ T cells when compared to splenocytes from SAS control mice (p < 0.0001; Figures 1C–F ). To further analyze the activation of T cells in immunized mice, we evaluated the relative expression of CD45RB and CD44 in both CD4⁺ and CD8⁺ T cells. A greater number of effector compared to naive CD4⁺ and CD8⁺ T cells were observed in splenocytes from BpOmpW-immunized mice compared to controls. Effector CD4⁺ CD45RB^loCD44^hi T cells were overrepresented in antigen-re-exposed splenocytes (p < 0.0001; Figure 1G ). Consistent with this, the naive CD4⁺ CD45RB^hiCD44^lo subset was significantly reduced in the BpOmpW-immunized group relative to SAS control splenocytes (p < 0.0001; Figure 1H ). Likewise, both effector CD8⁺ T cells CD45RB^hiCD44^hi ( Figure 1I ) and CD45RB^loCD44^hi ( Figure 1J ) were increased following BpOmpW immunization compared to SAS controls, whereas the number of naive CD45RB^hiCD44^lo CD8⁺ T cells were reduced ( Figure 1K ). These data show that immunization with BpOmpW results in the generation of strong antigen-specific CD4⁺ and CD8⁺ T-and B-cell responses.

We also investigated the CD4⁺ and CD8⁺ T-cell populations in more detail, and we observed that BpOmpW immunization induced the appearance of CD8^hi and CD4^hi populations ( Figure 2A ) that were absent in the control group splenocytes ( Figure 2B ). Further analysis of these populations demonstrated that they were predominantly CD8^hiCD45RB^veryhiCD44^lo ( Figure 2C , in red) and CD4^hiCD45RB^veryhiCD44^lo subpopulations ( Figure 2D , in blue), in contrast to SAS control splenocytes that did not show these subpopulations ( Figures 2E, F ). Furthermore, splenocytes from SAS control mice showed a CD4^hi population ( Figure 2B ), to a lesser extent than the BpOmpW group, although the population was predominantly CD45RB^loCD44^hi ( Figure 2F ). It is important to highlight that, including these CD45RB^veryhiCD44^lo populations as if they were CD45RB^hiCD44^lo naive CD4⁺ or CD8⁺ T cells, these subsets were still lower in the BpOmpW-immunized group compared to the SAS control group in both CD8⁺ ( Figure 2G ) and CD4⁺ ( Figure 2H ) T cells.

IFN-γ responses dominate the immune response induced by melioidosis infection in patients (26). Consequently, in order to elucidate the T-cell responses associated with BpOmpW immunization, we evaluated a range of cytokine levels by flow cytometry following re-stimulation in vitro. Although the autocrine growth factor IL-2 did not change in CD4⁺ T cells, it was upregulated in CD8⁺ T cells from the BpOmpW-immunized group relative to the SAS control group ( Figures 3A, B ). The antigen elicited high levels of IFN-γ, IL-4, and IL-17, indicating that CD4⁺ T cells differentiated to Th1, Th2, and Th17 cells, indicative of a mixed T helper (Th) response being elicited by BpOmpW (p < 0.0001, p < 0.0001, and p < 0.0001, respectively; Figures 3C–E ). Additionally, IFN-γ-producing CD8⁺ T cells were also more abundant following BpOmpW immunization (p = 0.011; Figure 3F ).

NK and NKT cells have been associated with patient responses to melioidosis ( 18, 19). Consequently, we evaluated the responses of NK and NKT cells in BpOmpW-re-exposed splenocytes. A population of T cells that were negative for both CD4 and CD8, i.e., double-negative (DN) cells, were elevated in the BpOmpW-immunized group relative to the SAS control group (p = 0.0001; Figure 3G ). The expression of CD49b⁺ indicated that NKT cells constituted virtually all these DN cells ( Figure 3H ). Moreover, the IFN-γ and IL-17 responses produced by NKT cells were increased in BpOmpW-immunized mice in comparison with the SAS-adjuvanted control mice, in contrast to IL-4 that remained comparable to SAS control splenocytes (p = 0.0062, p < 0.0001, and p = 0.942, respectively; Figures 3I–K ). Finally, IFN-γ-producing CD49b⁺ NK cells were also significantly higher in splenocytes from BpOmpW-immunized mice following re-exposure to antigen, relative to the adjuvant-only control group (p = 0.0025; Figure 3L ).

We also examined regulatory T cells as they can suppress pro-inflammatory damage produced by bacterial infections. CD4⁺CD25⁺Foxp3⁺ regulatory T cells were augmented in response to BpOmpW re-exposure in immunized mice relative to the spleen cells from adjuvant-only control mice (p = 0.0007; Figure 3M ). Finally, many studies have observed elevated expression of TNF during melioidosis infection in humans and animal models (27, 28). In this study, TNF was upregulated in both CD4⁺ and CD8⁺ T cells from BpOmpW-immunized mice (p < 0.0001 and p < 0.0001, respectively; Figures 3N, O ).

Due to the exquisitely enhanced susceptibility of people with diabetes to infection, and T2D being on the rise in tropical and subtropical regions, we needed to understand immune response in the context of diabetes. We developed a polygenic insulin-resistant mouse model by feeding C57BL/6J male mice with a high-fat diet (HFD) for up to 16 weeks as a model of T2D. Mice on HFD continuously gained more weight than did their littermates on a normal diet from the first 2 weeks of diet intervention (25.49 ± 2.12 vs. 29.38 ± 0.82, p = 0.00001) to 12 weeks ( Figure S2B ). Moreover, starting at week 8, HFD-fed mice began to develop hyperglycemia (HG = 13.17 ± 2.16 vs. 15.66 ± 2.77, p = 0.02) and insulin resistance (IR) (t ₄₅ = 46.52 ± 16.26 vs. 59.89 ± 23.20, p = 0.13), the latter being more apparent at 12 weeks of treatment with the diet (HG = 11.44 ± 1.89 vs. 13.82 ± 3.11, p = 0.044; t ₄₅ = 44.19 ± 8.20 vs. 98.32 ± 17.92, p < 0.00001) ( Figures S2C–E ). In addition, large lipid droplets were observed in the livers of insulin-resistant mice relative to control mice in the liver micrographs ( Figure S2F ).

To determine the impact of insulin resistance on the response to BpOmpW immunization, groups of 10 insulin-resistant mice ( Figure S3A ) were immunized with one subcutaneous injection of adjuvant alone or with SAS-adjuvanted BpOmpW as before. After 2 weeks, the splenocytes of both groups were exposed to BpOmpW and immunophenotyped. Unexpectedly, CD45RB was not decreased in the insulin-resistant model (p = 0.6198; Figure 4A ) despite the T-cell activation markers, CD25 and CD44, being consistently elevated in both CD4⁺ and CD8⁺ T cells in the BpOmpW-immunized group with respect to the SAS control group (p < 0.0001, p < 0.0001, p = 0.0001, and p = 0.0039, respectively; Figures 4B–E ), which was comparable to that in non-insulin-resistant mice. Insulin-resistant BpOmpW-immunized mice showed elevated levels of IL-2 in both CD4⁺ and CD8⁺ T cells following exposure to the vaccine antigen relative to the splenocytes from adjuvant control mice (p = 0.0004 and p = 0.0003, respectively; Figures 4F, G ), in contrast to non-insulin-resistant mice which only showed upregulation of IL-2 in CD8⁺ T cells ( Figure 3B ). Splenocyte cytokine responses from BpOmpW-immunized insulin-resistant mice showed upregulated Th1 and Th17 responses as determined by the levels of IFN-γ and IL-17 in CD4+ T cells, respectively (p < 0.0001 and p < 0.0001, respectively; Figures 4H, I ). In contrast to non-insulin-resistant mice, the Th2 response-associated IL-4 levels in CD4⁺ T cells were unaltered with respect to the control group (p = 0.2919; Figure 4J ). Remarkably, after antigen re-stimulation, IFN-γ-producing cytotoxic CD8 and NKT cells were also increased in BpOmpW-immunized insulin-resistant mice relative to the SAS control group (p = 0.0005 and p = 0.0013, respectively; Figures 4K, L ), with the exception of IFN-γ-producing NK cells which remained unchanged (p = 0.6148; Figure 4M ). Moreover, no significant changes were observed in IL-4- or IL-17-expressing NKT cells ( Figures 4N, O ). In contrast to non-insulin-resistant mice, the levels of regulatory T cells remained unmodified (p = 0.1729) in the splenocytes from BpOmpW-immunized mice relative to SAS control splenocytes ( Figure 4P ). The TNF in CD4 T cells was unchanged relative to the control group in the insulin resistance study ( Figure 4Q ), whereas the TNF from CD8⁺ T cells was more abundant in the BpOmpW group compared to control cells (p = 0.012; Figure 4R ). We also analyzed the antibody responses and the naive and effector T cells in the insulin-resistant model. We observed strong seroconversion ( Figures S3B–D ) and elevated effector T cells in the presence of BpOmpW ( Figures S3E–I ), as in non-insulin-resistant mice.

This and our previous work clearly demonstrated that BpOmpW antigen is immunogenic and elicits protective T-cell responses in mice; thus, as a precursor to its evaluation in human studies, we wanted to examine the responses in humanized HLA transgenic mice expressing different HLA alleles. The IFN-γ responses to the whole antigen were examined for epitopes in the context of HLA-DR1, HLA-DR4, and HLA-DQ8 alleles. In addition, a synthetic panel of overlapping peptides was generated covering the full coding sequence ( Table 1 ), and these were also evaluated in humanized transgenic mice. The lymph nodes of HLA class II transgenic mice (HLA-DR1, HLA-DR4, and HLA-DQ8) that had been immunized with recombinant BpOmpW antigen showed strong recall responses to the whole BpOmpW antigen and to immunodominant T-cell epitopes, P5, P7, and P21, as determined by the IFN-γ ELISpot assay ( Figure 5 ). When probed further by the priming of additional HLA-DR4 transgenic mice with P5 or P21, ELISpot analysis of drain lymph node cells confirmed that peptide 5 was a T-cell epitope. Subsequent priming of additional HLA-DQ8 mice with P7 confirmed that, in HLA-DQ8 transgenics, P7 was a T-cell epitope ( Figure 5 ).

We examined the T-cell proliferation and IFN-γ responses of human PBMCs from healthy donors in response to BpOmpW exposure. We interrogated human PBMCs from 23 different donors with different HLA alleles ( Figure 6 ) for CD4 and CD8 T-cell markers following exposure to BpOmpW with and without SAS adjuvant. T-cell proliferation in response to BpOmpW stimulation was confirmed in almost all donors ( Figure 6 and Figures S12–S19 ), as proliferating cells for the CD4 and CD8 populations were elevated in 87% and 70% of donors, respectively, following exposure to the antigen with or without SAS adjuvant relative to the untreated control group ( Figure 6 and Figures S12–S19 ). Eight donors out of 23 showed elevated IFN-γ responses in the presence of the antigen or when adjuvanted with respect to the control group ( Figure 6 ).

We have shown that BpOmpW stimulates strong serological responses in mice, both in non-insulin-resistant (10) and in insulin-resistant mice. In order to examine whether the antigen can stimulate a humoral response in melioidosis patients with diabetes, we examined the presence of BpOmpW-specific antibodies in sera from people with diabetes in Northeast Thailand who survived melioidosis infection ( Figure 7 ). We showed that the melioidosis survivor cohort with diabetes (M) had significantly higher BpOmpW-specific IgG responses than the non-melioidosis cohort with diabetes (D; p = 0.0289) and the non-endemic healthy control (NE; p = 0.0023) groups, indicating that BpOmpW is specifically recognized following melioidosis exposure and is immunogenic in people with diabetes. No significant difference in the IgG levels was observed between the melioidosis cohort and a cohort of healthy household contacts of people with melioidosis (HH; p = 0.0537), most likely reflecting the endemic nature of melioidosis in Thailand.