Total IgG responses to the vaccines in mice were assessed using sera collected four weeks after the second dose and about a week before challenge. ELISAs were conducted using the following antigens: irradiated whole cell B. pseudomallei K96243 (BpK), CPS, and Hcp1 ( Table 1 ; Supplementary Table 1 ). E555 ΔilvI generated a robust IgG response against the BpK, but little to no IgG response against purified CPS or Hcp1 ( Table 1 ). The two-component subunit vaccine, Hcp1 + CPS-CRM197, generated a similar IgG response to BpK as E555 ΔilvI, but a much stronger response against CPS and Hcp1. The mean anti-CPS titer of the Hcp1 + CPS-CRM197 group was greatly elevated compared to those of the two subunit vaccine groups which included E555 ΔilvI (p ≤ 0.0001 for both). The two combination vaccination approaches, 1) Hcp1 + CPS-CRM197 + E555 ΔilvI and 2) CPS-CRM197 + E555 ΔilvI, generated stronger IgG responses to BpK relative to the homologous subunit vaccine (p = 0.005 and p = 0.0002, respectively). The mean anti-BpK titer of the combined vaccine group CPS-CRM197 + E555 ΔilvI was also greater than that of the mice vaccinated with E555 ΔilvI alone (p = 0.0045). Both subunit vaccine formulations containing Hcp1 generated a strong IgG response to this antigen, p < 0.0001 ( Table 1 ).

Five groups of vaccinated C57BL/6 mice were challenged with an approximate dose of 4 LD₅₀s of B. pseudomallei K96243 by a whole-body aerosol route at 35 days post-boost and survival results are shown in Figure 1B . The mice were observed for 60 days post-challenge, and unexpectedly, 30% of the PBS control mice survived challenge ( Figure 1B ). The day 60 results suggest that the most efficacious vaccines were E555 ΔilvI (80% protection) and Hcp1 + CPS-CRM197 + E555 ΔilvI (70% protection); the survival rate of the mice receiving E555 ΔilvI was significantly greater than that of the PBS controls (p = 0.035). This preliminary study demonstrated efficacy of the novel B. thailandensis E555 ΔilvI vaccine candidate. However, the day 60 mortality rates did not allow us to determine whether combining the LAV and subunit vaccines could increase the protective efficacy against B. pseudomallei since there was incomplete mortality in the PBS control group at the completion of the study. Notably, CpG was not included with the subunit vaccines as it had been previously (31) which may have impacted their protective capacity compared to previous studies. Nevertheless, statistical analyses of the survival curves (which reflect differences in times to death or euthanasia [TTD]) by day 21 and day 60 post-challenge, and of survival rates by day 21, further differentiated the groups. By day 21, both the mortality rates and survival curves of all vaccine groups except those receiving Hcp1 + CPS-CRM197 alone were greater than that of the PBS controls (p = 0.035 and p = 0.010, respectively). By day 60, the survival curves of two vaccine groups were still significantly different from the control curve (p = 0.010 for the E555 ΔilvI group and p = 0.024 for the Hcp1 + CPS-CRM197 + E555 ΔilvI-vaccinated mice); the survival curve for the Hcp1 + CPS-CRM197 group approached significance with p = 0.059. The comparison of proportion succumbing to infection by day 60 was only significant when comparing the E555 ΔilvI group to the PBS control group (p = 0.035) ( Figure 1B ; Supplementary Tables 2A, B ).

The numbers of viable bacteria present in the blood, lungs, and spleens were determined on day 3 post-challenge for six animals in each group. Except for one animal in the Hcp1 + CPS-CRM197 + E555 ΔilvI immunized group (1/6 mice were bacteremic), the only mice with detectable bacteria in their blood were four of the six PBS control animals (4/6 mice were bacteremic). As shown in Figure 2A ; Supplementary Table 3A , bacteria were readily detected in the lungs in all groups; however, the four vaccinated groups had GM CFU values which were reduced by approximately 3.5- to 5-log (1.3 to 10.7 x 10³ CFU/g) than that of the PBS group (5.89 x 10⁷ CFU/g), with significances ranging from p = 0.005 to p < 0.0001 compared to the PBS mice. In contrast to the lung samples, spleens ( Figure 2B ) from mice in the three-subunit vaccine-immunized groups exhibited no detectable CFU in four to five of the six samples (p < 0.0001 compared to PBS). Most of the samples from the E555 ΔilvI-immunized groups had bacteria in the spleens, albeit low levels (GM of 24.5 CFU/g). The controls had a four log higher GM, 2.63 x 10⁵ CFU/g (p < 0.0001 compared to the E555 ΔilvI group). The results suggest that by three days post-challenge the vaccinated animals could clear or reduce dissemination of B. pseudomallei more effectively than the unvaccinated control animals.

The number of viable bacteria present in the lungs and spleens of all surviving mice were also determined on day 60 post-challenge. The lungs of all eight surviving E555 ΔilvI-immunized mice had no detectable CFU, whereas some mice in the other four groups had low levels of bacteria (3.1 CFU/g in one of the three surviving PBS mice, and GM values of 37 to 87 CFU/g in the other three vaccine groups, p = 0.0001 to p = 0.0012 compared to the PBS group ( Figure 2C ; Supplementary Table 3B ). Except for one mouse, none of the survivors in the five groups had detectable CFU in the spleen; one of five survivors in the CPS-CRM197 + E555 ΔilvI group had 1.36 x 10⁵ CFU/g (the group GM was 11 CFU/g) ( Figure 2D ). Thus, although the vaccines significantly reduced the levels of the challenge strain in the organs at both the early and end of study timepoints compared to unvaccinated controls, except for the E555 ΔilvI group, sterile immunity was not achieved.

Cellular immune responses were evaluated three days after challenge with B. pseudomallei K96243 in the lung homogenates from control and vaccinated mice ( Table 2 ; Supplementary Table 4 ). Relative to the PBS controls, the mice immunized with E555 ΔilvI alone displayed two-fold greater levels of IFN-γ in the lung homogenates at 3 days post-challenge, while the IFN-γ level of the Hcp1 + CPS-CRM197 group was lower than the controls (p = 0.001) and the other vaccine groups (p < 0.0001). IFN-γ is vital for host defense against B. pseudomallei infections; the basis for the reduced levels of IFN-γ in the group vaccinated with the Hcp1 + CPS-CRM197 conjugate alone is not clear but might be related to the absence of the CpG immunostimulant in this experiment (31). In contrast to the others, this group also exhibited depressed levels of IL-18 and IL-22 (p < 0.0001 and p = 0.0003, respectively, when compared to the PBS controls), which are involved in co-stimulation of IFN-γ and reduction of lung inflammation, respectively (49–51). This may also be partially attributed to the greater suppression of IL-17A in this vaccine group relative to others. As shown in Table 2 , animals that received any of the four vaccines displayed reduced levels of many pro-inflammatory cytokines, such as IL-1β, MIP-2α, IL-6, MIP-1α, MIP-1β, GRO-α, MCP-1, G-CSF, LIF, IL-1α, GM-CSF, TNF-α, IL-17A, and MCP-3 (p = 0.0421 to p < 0.0001). Spleen homogenates were also evaluated, and the relative responses of the vaccine groups compared to the controls were similar to those of the lung ( Supplementary Table 5 ). Depressed responses were observed for many cytokines in all four vaccine groups, such as G-CSF, IL-6, GRO-α, MCP-1, IL-1β, IL-13, MCP-3, LIF, or IL-1α (p = 0.0042 to p < 0.0001, Supplementary Tables 4 , 5 ). However, the IFN-γ levels were either not statistically different (the E555 ΔilvI group) or significantly reduced (the remaining vaccine groups, p = 0.0003 to p < 0.0001) relative to the PBS controls; and IL-17A levels in these four groups were also not significantly different compared to controls. These differences in cytokine results in the spleen compared to the lung might be attributed to the early stage after aerosol challenge with limited systemic bacterial dissemination.

Total IgG responses to the vaccines were assessed using blood collected 27 days after the boost dose, prior to challenge in vaccinated mice. ELISAs were conducted using the BpK, CPS, and Hcp1 antigens. The total serum IgG titers following homologous vaccination of mice were determined, as shown in Table 4A . The GM titer to the BpK antigen produced by group 2 (Hcp1 + CPS-CRM197) was nearly twice that of the anti-BpK titer elicited by the E555 ΔilvI vaccine. Nevertheless, both vaccine groups produced anti-BpK titers that were significantly greater than that of the PBS controls (p = 0.0006 and p = 0.0015, for groups 2 and 3, respectively). Antibody titers to CPS were high in group 2 and were negligible in the other two groups, as expected (p < 0.0001 comparing group 2 results to those of the other two groups). Only the vaccine containing Hcp1 produced significant responses to this antigen (p = 0.0012 compared to groups 1 and 3). Thus, overall, the E555 ΔilvI vaccine stimulated the lowest antibody levels with very low titers to BpK and no detectable antibodies to purified CPS and Hcp1; these results were similar to those described above in Table 1 . The findings suggest that there may be substantial surface antigen differences between B. pseudomallei and B. thailandensis, even though E555 ΔilvI produces a B. pseudomallei-like CPS.

In addition to total IgG titers, IgG1 and IgG2c responses to the irradiated whole-cell BpK antigen were assessed for the homologous vaccination groups ( Supplementary Table 6A ). The group 2 subunit vaccine induced IgG1 titers to BpK antigen that were much higher than the corresponding IgG2c levels (IgG2c/IgG1 GM ratio of 0.08). In contrast, E555 ΔilvI skewed titers toward IgG2c rather than IgG1 (IgG2c/IgG1 GM ratio of 6.99).

The IgG levels following heterologous vaccination of mice are shown in Table 4B . Groups 5 and 6, in which the LAV and Hcp1 + CPS-CRM197 were given as the prime and boost, or in the reverse order, exhibited titers to the BpK antigen that were the highest, p = 0.0371 and p = 0.0506 versus the controls, respectively. Elevated titers to CPS were elicited by all four vaccine groups (p = 0.0004 to p = 0.0027 compared to group 4), each of which included the CPS-CRM197 subunit. The anti-CPS titers were not significantly associated with the order of antigen delivery, i.e., prime or boost ( Table 4B ); however, boosting with CPS-CRM197 appeared to favor stimulation of greater anti-CPS titers compared to priming with the subunit formulation. High GM levels of anti-Hcp1 antibodies were observed when the two vaccines containing this antigen were used in groups 5 and 6 (p < 0.0001 compared to groups 4, 7, and 8). The higher level was elicited by group 6, which were immunized with the vaccine containing Hcp1 in the prime dose. It appears that the order of delivery of the vaccines potentially impacted the antibody responses to CPS and Hcp1, while no discernable differences were observed against BpK.

All four heterologous vaccine strategies stimulated IgG1 and IgG2c antibodies to the BpK antigen ( Supplementary Table 6B ). Groups 5–7 exhibited IgG2c/IgG1 ratios that were closely balanced, i.e., not Th1- or Th2-skewed (ratios of 0.69, 0.63, and 0.87, respectively). In contrast, the group 8 elicited a negligible level of IgG2c antibodies and its IgG2c/IgG1 ratio was 0.12. Thus, the subclass IgG responses of these mice, which were vaccinated with a prime dose of CPS-CRM197 in the absence of Hcp1, were Th2-skewed. All the heterologous vaccine combinations, regardless of the order of delivery (whether CPS-CRM197 conjugate was in the prime or the boost dose [groups 5 – 8]) elicited low anti-BpK titers, yet they were associated with survival rates of 50 – 90% ( Table 4B , Figure 3C ). This included one of the most protective vaccine formulations and regimens (CPS-CRM197 prime followed by a E555 ΔilvI boost) which induced a very low antibody titer to whole cell BpK antigen ( Table 4B ). There was no apparent association between pre-challenge anti-BpK titers elicited by the homologous vaccines and the extent of protection.

The various groups of vaccinated and control mice (n = 10/group) described in Table 3 were challenged by the aerosol route with B. pseudomallei K96243 on day 58 ( Figure 3A ). Homologously vaccinated mice received 2,400 CFU (approximately 6.0 LD₅₀s) ( Figure 3B ) and heterologously vaccinated mice received 1,360 CFU (approximately 3.4 LD₅₀s) ( Figure 3C ). The highest level of protection afforded by a homologous vaccine at day 60 post-challenge was conferred by a prime and boost of the LAV E555 ΔilvI (group 3, 50%) ( Figure 3B ), while the highest level of protection by a heterologous vaccine was produced by administering a CPS-CRM197 prime followed by a E555 ΔilvI boost (group 8, 90%), as shown in Figure 3C . Statistical analyses of the data further clarified these conclusions, as detailed below.

By 21 days post-challenge, the two homologously vaccinated groups exhibited greater survival rates compared to that of the PBS control mice, i.e., p = 0.006 for group 2 (Hcp1 + CPS-CRM197) and 0.057 for group 3 (E555 ΔilvI) ( Figure 3B ). Also, the mean TTDs at day 21 of both homologous vaccination groups were significantly greater than that of the PBS group (14.7 and 11.6 days and 6.7 days, respectively, p ≤ 0.0032). At the study endpoint (60 days post challenge), only the group 3 LAV-immunized mice exhibited a significantly greater survival rate compared to the PBS group (p = 0.033) ( Figure 3B ) while the survival rates of groups 2 and 3 were not statistically different. As observed at day 21, the TTDs at day 60 indicated that both vaccines extended the mean TTD compared to the controls (p = 0.0002 and p = 0.0006, respectively). The mean TTD of the controls was 7.8 days compared to 32.6 and 26.6 days for the groups 2 and 3, respectively. The results suggested that the LAV and Hcp1 + CPS-CRM197 subunit vaccines were efficacious, albeit neither was completely protective. Also, since differences in survival rates of the two vaccine groups could not be statistically distinguished, these results supported the evaluation of a heterologous vaccine scheme.

Four groups of mice were given heterologous prime and boost doses of vaccine, as shown in Table 3 and Figure 3C . By 21 days post-challenge, the ten control mice had succumbed, and all the vaccine groups exhibited greater survival rates than the controls (p = 0.0031 to p = 0.0001). Additionally, all four vaccine groups also had significantly extended mean TTDs at day 21 post-challenge compared to the control mice (p < 0.0001). Analyses of the survivors at the study endpoint (60 days post-challenge), showed that the four vaccines were more protective compared to the controls, with survival rates ranging from 50% to 90% ( Figure 3B ), with p = 0.0325 to p = 0.0001. Group 8 (prime CPS-CRM197/boost E555 ΔilvI) had the highest level of survival at day 60 (90%), which was significantly greater than the PBS group 4 (p < 0.0001). However, the three vaccines associated with 50% to 70% protection (groups 5, 6, and 7) were not statistically different compared to the mice receiving the prime CPS-CRM197/boost E555 ΔilvI vaccination strategy. The mean TTDs at day 60 again indicated that all four vaccines significantly extended the mean TTD compared to the controls (p < 0.0001).

Since the B. pseudomallei challenges of the homologous and the heterologous vaccine groups ( Figure 3 ) were performed on the same day, we compared the protection afforded across all eight groups. The survival rate of group 8 at 60 days post-challenge (90%) approached significance relative to the survival of homologous vaccine group 2, which induced 40% protection, with p = 0.057; but less when compared to the homologous vaccine that protected 50% of group 3 (E555 ΔilvI). Thus, the heterologous vaccine scheme identified a combination which was 90% protective; and all four heterologous vaccines protected ≥ 50% of mice, findings which were comparable to those of the homologous vaccines. Importantly, however, the challenge dose of the heterologously vaccinated mice was roughly half of the homologously vaccinated mice (3.4 versus 6.0 LD₅₀).

The cellular immune responses of vaccinated mice were first evaluated using ELISpot assays to measure IFN-γ secretion by splenocytes restimulated with Hcp1. Controls were incubated with medium alone and resulted in low levels of IFN-γ-secreting cells in all groups (data not shown). As illustrated in Figure 4 , the highest numbers of IFN-γ-secreting splenocytes after Hcp1 stimulation were produced by the homologous vaccine group 2 (p < 0.0001 compared to the PBS control group 1). The responses of homologous group 3 and the four heterologous vaccine groups were not significantly different from the control groups 1 or 4, and were less than those of homologous group 2, p < 0.0001 compared to groups 5-8.

Antibody responses to the vaccine constituents (CPS and Hcp1) were evaluated as potential correlates of protection ( Table 5 ). Except for the homologous LAV group, all vaccines stimulated significant titers to the CPS (p < 0.0001 when compared to controls) with the highest levels being produced by the group receiving two doses of Hcp1 + CPS-CRM197 (group 2). In agreement with Table 4 , anti-Hcp1 antibodies were only elicited in the vaccine containing this antigen (group 2).

As shown in Table 5 , four vaccine groups exhibited titers to the BpK antigen that were significantly greater than that of the control group (p = 0.0003 to p = 0.0295). As seen in the earlier experiments, the LAV homologous vaccine (group 3) elicited the lowest anti-BpK titer (p = 0.0588 compared to PBS) as well as low IgG1 and IgG2c titers to BpK ( Table 6 ). Except for group 3, all vaccines induced higher levels of IgG1 compared to IgG2c anti-BpK antibodies. While the IgG2c/IgG1 ratio of the LAV group (2.40) was highly Th1 polarized (albeit both titers were low), the ratios of the other groups, especially group 9, were distinctly Th2-skewed (IgG2c/IgG1 ratios of 0.01 to 0.14). These findings were again in agreement with those of the two studies described above ( Supplementary Tables 6A, B ).

To assess vaccine efficacy, six groups (groups 1, 2, 3, 7, 8 and 9) of C57BL/6 mice (n = 13 mice) were challenged with 3.4 LD₅₀s of B. pseudomallei K96243 via whole-body aerosol four weeks after the second immunization (day 56). Three mice from each group were euthanized three days post-challenge and tissues were cultured to compare the relative bacterial burdens at this early timepoint. The remaining ten mice were observed for 60 days post-challenge and survival results are shown in Figure 5B . All the PBS control mice in group 1 succumbed to disease or were euthanized in accordance with early endpoint euthanasia criteria by day 5 ( Figure 5B ) whereas the five vaccines all provided an extended TTD compared to the PBS control group.

By 21 days post-challenge, two groups (2 and 8) exhibited the highest survival rates (60%), which was significantly higher than that of the PBS controls (p = 0.011). In contrast, the survival rates of groups 3, 7, and 9 (20%) were not statistically greater than the controls. The mean TTDs by day 21 of all five vaccine groups were significantly extended compared to the controls (means of 3.9 days for PBS and 10.4 – 14.5 days for the vaccinated mice, p = 0.003 to p < 0.0001). At the end of the 60-day period, the highest survival rates were observed in homologous vaccination group 2 (60%, p = 0.011 vs group 1 controls) and group 8 (40%). Of the three heterologous groups, the group 8 vaccine was the most protective, whereas only 10% of the group 7 and 9 mice survived until day 60. None of the survival rates for any of the vaccine groups were statistically different from one another. Nonetheless, the mean TTDs of all five vaccine groups was greater than that of the controls at the study endpoint, (p = 0.003 for group 3, p = 0.0004 for group 7, and p < 0.0001 for the other three groups).

To determine bacterial loads in tissues, three mice from each group were euthanized three days after challenge, and lungs, spleens, and blood were cultured. All the vaccines assessed prevented bacteremia (p = 0.001) versus PBS controls, which were bacteremic. Vaccination was associated with a nearly 4-log GM CFU reduction in bacteria in the lungs ( Figure 6A ) and resulted in a ≥ 4-log reduction in bacterial dissemination to the spleens (p = 0.0044 to p = 0.0005) compared to controls ( Figure 6B ). No bacteria were recovered from any of the spleens obtained from group 2. The GM CFU in the lungs were significantly less than that of the controls for groups 2, 3, and 8 (p = 0.0001 to p = 0.003), almost significantly less for group 9 (p = 0.052), but not for group 7 (p = 0.088 compared to the PBS controls).

The number of viable bacteria present in the spleens and lungs from all surviving mice were also determined at end of study post challenge. A mouse succumbed to B. pseudomallei infection on the day of necropsy and was cultured (depicted as red data points), however these data were omitted from GM calculations due to the period between death and sample collection. All but one spleen from surviving vaccinated mice were sterile (limit of detection 5 CFU/spleen) and only 5 of 13 lungs had recoverable CFU with the positive samples including two from group 2 (15.6 CFU/g and 1.51 x 10⁷ CFU/g), one from group 3 (25 CFU/g), and one from group 8 (2.27 x10⁴ CFU/g) ( Figure 6C ; Supplementary Table 7 ).

Overall, the E555 ΔilvI vaccine stimulated the lowest levels of antibodies to BpK, CPS, and Hcp1 ( Tables 1 , 4A , 5 ). Masoud et al. reported that the majority of patients with different clinical manifestations of melioidosis had sera recognizing the CPS, which suggests that it is immunogenic and has the potential to be a vaccine or diagnostic antigen (68). Polysaccharides such as CPS are antigens that induce a T cell independent response and must be covalently attached to a carrier protein to elicit a strong adaptive immunity (32–34, 69). We hypothesize that the E555 ΔilvI vaccine might not persist long enough in the host to generate a robust CPS IgG response; and perhaps a larger booster dose might be needed to prevent rapid clearance. Additionally, the native CPS being produced by the LAV will be less immunogenic when compared to purified CPS conjugated to the carrier CRM-197; this fact may also explain the disparate anti-CPS titers noted in vaccine strategies that contain the LAV strain as the sole antigenic source or when used as a booster to the subunit vaccine. Regarding anti-Hcp1 titers, the recombinant Hcp1 antigen used for the ELISA was purified from the cloned B. pseudomallei antigen (33). Although B. thailandensis and B. pseudomallei CPS are highly similar, the Hcp1 proteins from B. thailandensis and B. pseudomallei are not highly conserved at the amino acid level (46, 72, 73), possibly explaining the low anti-Hcp1 response in mice vaccinated with E555 ΔilvI. In contrast, the subunit vaccines induced high titers to all antigens which were included in these vaccines ( Tables 1 , 4A , 5 ).

We also examined the immune response using irradiated B. pseudomallei K96243 (BpK) as the test antigen. The subunit vaccines elicited higher titers to BpK than did homologous LAV vaccines; these lower anti-BpK responses suggest the existence of important surface antigen differences between B. pseudomallei and B. thailandensis (74–76). The previously reported higher titers against BpK elicited by B. pseudomallei 668 ΔilvI support this possibility (31).

Subclass analysis of IgG responses to the killed BpK antigen was performed. Whereas Hcp1 + CPS-CRM197 (homologous group 2) induced much higher anti-BpK IgG1 titers than the corresponding IgG2c levels, the reverse was observed for the homologous LAV ( Supplementary Table 6 ; Table 6 ). These results confirmed our previous observations (31), in which CPS-CRM197 induced a Th2-skewed immune response whereas LAVs (B. pseudomallei 668 ΔilvI and B. thailandensis E555 ΔilvI) stimulated a more Th1-polarized response to whole cell inactivated B. pseudomallei.

The relative roles of the vaccine antigens, and of the specific immune responses to them, in protective efficacy is more difficult to dissect in the heterologously-vaccinated groups. All vaccines containing CPS-CRM197 in the prime or boost dose and CpG (groups 5 - 8), produced significant protection despite low antibody responses, including the most protective vaccination scheme, CPS-CRM197 prime followed by an E555 ΔilvI boost (group 8, Figures 3C , 8 ). The greater protection afforded by group 8 versus group 7 might be associated with their different anti-CPS responses ( Table 5 ).

Nonetheless, humoral responses have an important role in murine survival and vaccine-mediated protection, as shown in active and passive transfer studies (33, 53, 69, 77–83). In our previous report, the high antibody titers to Hcp1 and CPS elicited by the subunit vaccine, but not the LAV, were protective (31). Finally, the recovery of humans with melioidosis has been associated with B. pseudomallei-specific humoral responses (84, 85); and Pumpuang et al. reported high levels of antibodies to Hcp1 in survivors (86). Since the pathogenesis of melioidosis involves both extra- and intra-cellular phases of B. pseudomallei infection, a logical prophylactic approach employs a combination of vaccine antigens which can stimulate both protective humoral and cell-mediated immune responses (31–33).

Up-regulation of IFN-γ and downregulation of other pro-inflammatory cytokines appears to be important for protecting vaccinated animals against lethal Burkholderia infections. In the current study, the levels of IFN-γ in the lungs of mice were highest in those vaccinated with the LAV alone or combined with Hcp1 + CPS-CRM197 at three days after challenge ( Table 2 ). All vaccine groups exhibited reduced levels of many pro-inflammatory cytokines in lungs compared to controls (such as TNF-α, IL-1β and IL-6), indicating that the vaccines prevented excessive expression of cytokines that can cause host damage, as described previously (31). The most efficacious vaccine formulations were those containing CPS-CRM197 in the prime dose (groups 2 and 8) which resulted in more controlled levels of cytokine expression than did the less protective vaccines (groups 3, 7, and 9) ( Table 7 ). The cytokine and chemokine responses in sera, lungs, and reticuloendothelial organs of naïve mice infected with B. pseudomallei have been extensively characterized, as summarized recently (87). Although laboratory-associated differences in timing and specificity of cytokine production were apparent, most described a rapid increase of proinflammatory cytokine levels within one to three days in lungs, spleen, and sera, with a subsequent decline (40, 41, 88–91).

IFN-γ is a major Th1 cytokine involved in stimulating proinflammatory cytokines, and macrophage phagocytic activity is a likely correlate of survival in mouse models of melioidosis (30, 31, 33, 53, 92–95). The association of protection with strong IFN-γ responses in vaccinated mice is consistent with observations in humans with melioidosis (96, 97). Specifically, the secretion of IFN-γ by CD4+ T cells from melioidosis patients, when stimulated with Hcp1 and TssM, was associated with improved survival (98, 99). In contrast, elevated levels of anti-Hcp1 IgG were not associated with patient survival (98) which supports a role for IFN-γ and T cell immunity in recovery from melioidosis (32, 98, 100).