Single vector platform vaccine protects against lethal respiratory challenge with Tier 1 select agents of anthrax, plague, and tularemia

Bacillus anthracis, Yersinia pestis, and Francisella tularensis are the causative agents of Tier 1 Select Agents anthrax, plague, and tularemia, respectively. Currently, there are no licensed vaccines against plague and tularemia and the licensed anthrax vaccine is suboptimal. Here we report F. tularensis LVS ΔcapB (Live Vaccine Strain with a deletion in capB)- and attenuated multi-deletional Listeria monocytogenes (Lm)-vectored vaccines against all three aforementioned pathogens. We show that LVS ΔcapB- and Lm-vectored vaccines express recombinant B. anthracis, Y. pestis, and F. tularensis immunoprotective antigens in broth and in macrophage-like cells and are non-toxic in mice. Homologous priming-boosting with the LVS ΔcapB-vectored vaccines induces potent antigen-specific humoral and T-cell-mediated immune responses and potent protective immunity against lethal respiratory challenge with all three pathogens. Protection against anthrax was far superior to that obtained with the licensed AVA vaccine and protection against tularemia was comparable to or greater than that obtained with the toxic and unlicensed LVS vaccine. Heterologous priming-boosting with LVS ΔcapB- and Lm-vectored B. anthracis and Y. pestis vaccines also induced potent protective immunity against lethal respiratory challenge with B. anthracis and Y. pestis. The single vaccine platform, especially the LVS ΔcapB-vectored vaccine platform, can be extended readily to other pathogens.

Lm-vectored vaccines were assessed for expression and secretion of B. anthracis and Y. pestis fusion proteins after growth in broth and in infected mouse macrophage-live J774A.1 cells by evaluating culture filtrates by Western blotting using antibodies specific to B. anthracis PA or to Y. pestis LcrV. As shown in Fig. 1c, the antibody to PA or LcrV detected major protein bands of 59-kDa ActAN-BaLFnPAc (ActAN-Ba) (Fig. 1c,  Initial in vivo studies examined dissemination, clearance, and plasmid stability of the newly constructed B. anthracis and Y. pestis vaccines. Our results showed that rLVS ΔcapB/Ba grew and disseminated similarly to the parental LVS ΔcapB, while rLVS ΔcapB/Yp showed delayed growth and dissemination (Figs S4 and S5). The shuttle plasmid-encoded antigen expression cassettes in rLVS ΔcapB/Ba and rLVS ΔcapB/Yp were stably maintained in mouse liver, spleen, local skin (after i.d. administration), and lung (after i.n. administration) up to 14 days post vaccination (data not shown); The rLm/Ba and rLm/Yp vaccines also showed systemic dissemination, similar to the parental Lm vector, and all were cleared by Day 7 post vaccination (Fig. S6).
Subsequently, we explored the efficacy of three homologous (rLVS ΔcapB/Ba or rLm/Ba) or heterologous (one rLVS ΔcapB/Ba prime + two rLm/Ba boosts) immunizations, both mucosally (i.n.) and systemically (i.d. for rLVS ΔcapB/Ba and i.m. for rLm/Ba), and compared them with that of sham or AVA immunization (s.q.), and with one rLVS ΔcapB/Ba prime + one rLm/Ba boost vaccination, as depicted in Fig. 3a. The immunized animals were bled, subsequently challenged with 371,000 B. anthracis Ames spores (~10 LD 50 ), and monitored for 3 weeks (Fig. 3a). As shown on the left side of Fig. 3b, mice homologously primed-boosted with rLVS ΔcapB/Ba i.n. (Group C) or i.d. (Group D) or with rLm/Ba i.n. (Group E) or i.m. (Groups F) produced significantly greater amounts of B. anthracis PA and/or LF antigen-specific serum IgG antibody, dominated by subtype IgG2a, than sham-immunized mice (Fig. 3b, top left two panels), consistent with the results from Experiment I ( Fig. 2a and b) and additional experiments ( Fig. S7a-d, Group C). Of note, as in the previous experiment, mice immunized with AVA produced PA-specific antibody, but did not produce LF-specific serum antibody. Upon challenge, mice homologously primed-boosted with rLVS ΔcapB/Ba or rLm/Ba, systemically or mucosally, had greater survival of protein standards; on the right border are listed the proteins of interest. Each blot was processed by using the Bio-Rad imaging system (ChemiDoc XRS) and Quantity One software, which allows the overlap of a white-light image, for visualization of the protein standards  than sham-and AVA-immunized mice; the survival of mice immunized systemically with rLVS ΔcapB/Ba and rLm/Ba and mucosally with rLm/Ba was significantly greater than that of the sham-immunized mice (Fig. 3b, bottom left panel).
As shown on the right side of Fig. 3b, mice heterologously primed-boosted with rLVS ΔcapB/Ba -rLm/Ba, mucosally (i.n./i.n.) (Group G & H) or systemically (i.d./i.m.) (Group I & J) also produced significantly greater amounts of B. anthracis PA and LF antigen-specific serum IgG antibody than sham-immunized mice, dominated by subtype IgG2a to PA and LF (Fig. 3b, top right two panels; Fig. S7b-e, Groups E and F), and elevated levels of serum antibodies that neutralized anthrax toxin (assayed in mouse macrophage cell line J774A.1), as did AVA-immunized mice (Fig. S7f, Groups B, E, and F). After challenge, these heterologously primed-boosted mice (Groups G, H, I, J) showed significantly increased survival compared with the sham-immunized mice (P < 0.05 or P < 0.01), whether boosted only once (Groups G and I) or twice (Groups H and J) with rLm/Ba (Fig. 3b, bottom, right panel). In contrast, survival of mice immunized with the AVA vaccine was not significantly different from that of sham-immunized mice (P = 0.3) (Fig. 3b, bottom panels). These results indicate that both systemic and mucosal homologous priming-boosting with rLVS ΔcapB/Ba or rLm/Ba and both systemic and mucosal heterologous priming-boosting with rLVS ΔcapB/Ba -rLm/Ba induce strong protective immunity against lethal respiratory challenge with B. anthracis spores. As in the previous challenge experiment, mean survival time 3 weeks post-challenge correlated with pre-challenge serum antibody to LF but not to PA (Fig. 3c) or to toxin neutralizing antibody (Fig. S8).
To investigate T-cell mediated immune responses induced by rLVS ΔcapB and rLm vaccines, we immunized mice, observed them for signs of discomfort or weight loss, and assayed their lung and spleen cells for cytokine secretion and intracellular cytokine staining in response to in vitro stimulation with B. anthracis and F. tularensis antigens (Figs 4a and S9). After vaccination with rLVS ΔcapB/Ba i.n., rLm/Ba i.n. or i.m., or AVA s.q., mice did not show signs of significant discomfort (e.g. ruffled fur) or weight loss (data not shown). As expected, mice immunized with the rLVS ΔcapB/Ba and rLm/Ba anthrax vaccines developed T-cell mediated immune responses, dominated by Th1 responses (Figs 4 and 5). Lung and spleen cells of mice immunized with rLVS ΔcapB/Ba i.n. twice (Group C) or once (Group D) or heterologously primed-boosted with rLVS ΔcapB/Ba -rLm/ Ba i.n./i.n. (Group E) or i.n./i.m. (Group F) secreted much higher levels of IFN-γ and IL-4 in response to LF and HI-LVS than sham-(statistically significant) or AVA-immunized mice (Fig. 4b). In response to LF and HI-LVS, these mice had elevated levels of lung and spleen CD4+ T cells expressing IFN-γ, TNF-α, IL-2, and IL-17A compared with sham-and AVA-immunized mice and greater frequencies of polyfunctional CD4+ T cells expressing IFN-γ, TNF-α, IL-2, and/or IL-17 than sham-immunized mice (Figs. 4c, 5); the heterologously immunized mice had especially high levels of cytokine-expressing CD4+ T cells (Fig. 4c, leftmost and rightmost upper and lower panels), and modestly increased levels of CD8+ T-cells secreting IFN-γ in the lung when boosted i.n. and in the spleen when boosted i.m. (Fig. 4d). In response to PA, CD4+ T-cells from mice heterologously primed-boosted by the i.n. route also had increased cytokine-producing cells in the lung and to a much less extent in the spleen; not unexpectedly, when the booster was instead administered systemically (i.m.), there were increased cytokine positive CD4+ T-cells in the spleen (Fig. 4c, middle upper and lower panels). With a few minor exceptions, the AVA vaccine induced very poor cell-mediated immune responses ( Fig. 4b-d). Mice immunized i.d. once with rLVS ΔcapB/Ba or heterologously primed-boosted with rLVS ΔcapB/Ba i.d. -rLm/Ba i.m. also showed enhanced IFN-γ secretion by lung and spleen cells and increased frequencies of lung and spleen CD4+ T cells expressing IFN-γ, TNF-α, IL-2 and/or IL-17A in response to LF and HI-LVS compared with sham-immunized mice and mice immunized with AVA (data not shown). These results indicate that both homologous priming-boosting with rLVS ΔcapB/Ba and heterologous priming-boosting with rLVS ΔcapB/Ba -rLm/Ba vaccines induce F. tularensis and B. anthracis antigen-specific Th1-type cytokine secretion and polyfunctional CD4+ T cells.
Vaccine immunogenicity and protective immunity against pneumonic plague. To evaluate the protective efficacy of rLVS ΔcapBand rLm-vectored Y. pestis vaccines administered by the mucosal and systemic routes, we sham-immunized mice, or immunized them once with the unlicensed EV76 vaccine s.q. (its standard route of administration), or twice homologously with rLVS ΔcapB/Yp i.n. or i.d. or heterologously primed-boosted with rLVS ΔcapB/Yp (i.n. or i.d.) -rLm/Yp (i.n. or i.m., respectively), as indicated in Fig. 6a. All mice were bled at Week 8; challenged at Week 9 with 1900 CFU Y. pestis CO92 strain (~8 LD 50 ; the pre-determined LD 50 for Y. pestis CO92 was ~250 CFU), and monitored for signs of illness, weight change, and survival for three weeks. Mice immunized with rLVS ΔcapB/Yp i.d or with rLm/Yp i.m. did not show signs of significant discomfort (e.g. ruffled fur) or weight loss (data not shown); in one experiment, mice immunized with rLVS ΔcapB/ Yp i.n. showed transient mild weight loss (~5%) but rapidly recovered such that their weights matched that of sham-immunized controls by at least Day 6 post-vaccination (data not shown). Homologous priming-boosting with rLVS ΔcapB/Yp i.n. or i.d. (Groups C, D) or heterologous priming-boosting with rLVS ΔcapB/Yp -rLm/ Yp i.n. or i.m. (Groups E, F) induced significantly elevated serum antibody titers to LcrV protein and to F1/LcrV ***P < 0.001; ****P < 0.0001. (c) Survival after vaccination and challenge. The survival curve of each vaccinated group is compared with that of Group A (Sham) or Group B (AVA) by the log-rank test (Mantelcox); P values for individual vaccine groups significantly different from the Sham or AVA group are marked with asterisks and " §", respectively, color-coded to the color of the vaccine symbol; **P < 0.01; ***P < 0.001 vs. Sham group; § P < 0.05 vs. AVA group. (d) Correlation between serum antibody and mean survival time. Linear regression was used to obtain values for the slope and intercept and the correlation coefficient (R 2 ) between prechallenge serum antibody and post-challenge mean survival time at 21 days post-challenge. Two-tailed P values were calculated for the correlation. Mice (n = 8/group) were immunized two or three times homologously with 10 6 CFU rLVS ΔcapB/Ba (rLVS/ Ba) or rLm ΔactA ΔinlB prfA/Ba (rLm/Ba) or heterologously first with rLVS/Ba and subsequently with rLm/ Ba, as indicated. Controls were sham-immunized with PBS i.d. or with AVA s.q. three times. All mice were bled at week 11; challenged at week 12 with B. anthracis Ames spores (371,000 CFU, ~10 LD 50 ); and monitored for survival for 3 weeks post-challenge. (b) Serum antibody prior to challenge and survival post challenge. Top panels. Sera were assayed for IgG or IgG subtypes IgG1 and IgG2a to B. anthracis PA and LF proteins, as indicated, after homologous (left two panels) and heterologous (right two panels) prime-boost vaccination. Values are mean + SEM of serum antibody endpoint titer for n = 8 per group. Differences in serum endpoint titer among individual groups were analyzed by two-way ANOVA with Tukey's corrections. *P < 0.05; Scientific REPORTS | (2018) 8:7009 | DOI:10.1038/s41598-018-24581-y fusion protein, balanced between IgG1 and IgG2a, compared with sham-immunization ( Fig. 6b) (Fig. S10a,b, upper panel, Groups B, E, & F). The majority of the IgG antibody induced by LVS ΔcapBand Lm-vectored vaccines was directed to LcrV. In contrast, mice immunized with EV76 produced very little antibody to LcrV (small amounts of IgG1 and IgG2a were evident) but substantial amounts of IgG to F1 and F1-LcrV (Fig. 6b).
Mice immunized three times i.d. with rLVS ΔcapB/Yp (Group D) survived significantly longer than sham-immunized mice (P = 0.03); mice primed-boosted with rLVS ΔcapB/Yp -rLm/Yp (Group F) also survived longer than sham-immunized mice, although the difference was not statistically significant (Fig. 7c). As in the previous experiment shown in Fig. 6, the amount of serum antibody specific to F1, but not to LcrV, was highly correlated with the mean survival time three weeks post-challenge (P < 0.001, P < 0.0001 and P < 0.0001 for IgG, IgG1 and IgG2a, respectively) (Fig. 7d). These results indicate that systemic homologous priming-boosting with rLVS ΔcapB/Yp induces strong protective immunity against Y. pestis CO92 respiratory challenge and that vaccine efficacy is correlated with F1-specific antibody.
Potent protective immunity against pneumonic tularemia. In previous studies, we have shown that heterologous priming-boosting with LVS ΔcapB or rLVS ΔcapB overexpressing IglC as the prime vaccine and rLm expressing F. tularensis IglC (rLm/iglC) as the booster vaccine induces potent protective immunity in mice against virulent F. tularensis Schu S4 respiratory challenge 18 . We have also shown that immunization with rLVS ΔcapB/iglABC is highly safe and induces greater protective immunity than the parental LVS ΔcapB vector against F. tularensis Schu S4 respiratory challenge 19 . However, the efficacy of homologous priming-boosting with this vaccine has not been investigated. To evaluate the efficacy of homologous priming-boosting with rLVS ΔcapB/iglABC, both systemically and mucosally, we immunized BALB/c mice once or twice i.d. or i.n. with this vaccine, as indicated in Fig. 8a. Control mice were sham-immunized with PBS, immunized i.d. with the unlicensed LVS vaccine (which is highly lethal by the i.n. route), or i.d. with the LVS ΔcapB vector. Mice were challenged with 10 CFU F. tularensis Schu S4 (~10 LD 50 ) at Week 10 and monitored for survival for 3 weeks. All immunized mice survived significantly longer than the sham-immunized mice (P = 0.03 for LVS ΔcapB and P = 0.001 or 0.0001 for all other groups vs. sham-immunized mice). Mice immunized i.d. once (Group D) or twice (Group E) or i.n. twice (Group F) with rLVS ΔcapB/iglABC survived significantly longer than mice immunized once with the LVS ΔcapB vector (P = 0.006, P = 0.0001, and P = 0.0005, respectively); mice immunized i.d. or i.n. twice with rLVS ΔcapB/iglABC (Groups E and F) had survival times equivalent to or greater than that of mice immunized with LVS (Group B); differences were not statistically significant (Fig. 8b).
In a subsequent experiment comparing two vs. three doses of rLVS ΔcapB/iglABC, we immunized mice i.d. or i.n. twice or three times with rLVS ΔcapB/iglABC; control mice were immunized i.d. once with PBS (Sham) or LVS. All the mice were bled, challenged i.n. with 2 or 6 CFU (LD 50 ≈ 1 CFU) of F. tularensis Schu S4, and monitored for survival for 3 weeks (Fig. 8c). Consistent with the experiment described above, all the immunized mice survived significantly longer than the sham-immunized mice (P < 0.001) (Fig. 8d, upper panels). Notably, 100% of mice immunized with rLVS ΔcapB/iglABC i.n. either twice or three times survived challenge with 2 or 6 LD 50 , the highest survival rate among the groups; and the survival of rLVS ΔcapB/iglABC immunized mice was significantly greater (P < 0.05) than that of LVS-immunized mice after the 6 LD 50 challenge (Fig. 8d upper **P < 0.01; ****P < 0.0001 vs. Sham group. Bottom panels. The survival curve of each vaccinated group, color-coded as indicated in panel b, after homologous (left) and heterologous (right) prime-boost vaccination and challenge is compared with that of the Sham group by the log-rank test (Mantel-cox); P values for vaccine groups that are significantly different from the Sham group are marked with one or more asterisks color-coded to the color of the vaccine symbol. *P < 0.05 and **P < 0.01. (c) Correlation between serum antibody and mean survival time. The correlation coefficient (R 2 ) and one-tailed P values were obtained as described in legend to  . Interestingly, when the rLVS ΔcapB/iglABC vaccine was administered by the i.d. route, three immunizations were not more efficacious than two immunizations; differences in survival between these mice and LVS-immunized mice were not statistically significant. All the immunized mice had F. tularensis antigen-specific IgG antibody, balanced between IgG1 and IgG2a, at levels significantly greater than sham-immunized mice (P < 0.0001) (Fig. 8d, lower panels). Mice immunized i.d. three times with rLVS ΔcapB/iglABC (group D) had antibody (IgG and IgG1) level significantly greater than LVS-immunized mice (P < 0.05 and P < 0.0001, resp.). Mice immunized with rLVS ΔcapB/iglABC i.n. twice (group E) or three times (group F) had antibody (IgG, IgG1, and IgG2a) levels significantly higher than mice immunized i.d. once with LVS (P < 0.0001) or i.d. twice or three times with rLVS ΔcapB/iglABC (P < 0.05 -P < 0.0001) (Fig. 8d, lower panel). While not examined herein, a previous study demonstrated that rLVS ΔcapB/iglABC also induces strong cell-mediated immune responses, which play a dominant role in host defense against F. tularensis 19 . These results show that systemic and especially mucosal homologous priming-boosting with rLVS ΔcapB/iglABC induces strong protective immunity against virulent F. tularensis Schu S4 respiratory challenge.

Discussion
In this study, we report single platform vaccines including homologous LVS ΔcapB-vectored vaccines and heterologous LVS ΔcapB and Lm-vectored vaccines against three Tier I pathogens, the causative agents for anthrax, plague and tularemia. We show that LVS ΔcapBand Lm-vectored vaccines express recombinant B. anthracis, Y. pestis, and F. tularensis immunoprotective proteins on solid agar or in broth, and in macrophage-like cells in vitro; the shuttle plasmids for antigen expression in LVS ΔcapB vectored vaccines are stable after passage in broth, macrophages, and mice. All vaccines are safe in mice after systemic (i.d. or i.m.) or mucosal (i.n.) immunization. Homologous priming-boosting with LVS ΔcapB-vectored B. anthracis, Y. pestis, or F. tularensis vaccines administered mucosally or systemically induces potent antigen-specific humoral and T cell-mediated (including both CD4+ and CD8+) immune responses, known to be important for long-lasting potent immunoprotection against the target pathogens of anthrax [34][35][36] , plague 20,37,38 , and tularemia [39][40][41][42][43] in animal models, and potent protective immunity against respiratory challenge with lethal doses of these pathogens. Protection against anthrax was far superior to that obtained with the licensed AVA vaccine and protection against tularemia was comparable to or greater than that obtained with the toxic and unlicensed LVS vaccine. Heterologous priming-boosting with LVS ΔcapBand Lm-vectored B. anthracis and Y. pestis vaccines also induced potent protective immunity against respiratory challenge with virulent B. anthracis spores and Y. pestis.
Studies on the immunity induced by anthrax vaccines, mostly comprising PA and LF antigens or their immunodominant domains, have focused on serum antibody (mostly Th2 biased) and toxin neutralization antibody; the role of T cell mediated immunity has not been widely investigated 12,44 . Thus, the cumbersome immunization regimen for the licensed AVA vaccine (consisting primarily of PA with some LF protein) entails repeated boosting to induce sufficient toxin neutralizing activity for protection. With respect to a potential role for cell-mediated immunity, Glomski et al. reported that IFN-γ-producing CD4+ lymphocytes, but not humoral immune responses, mediate spore-induced immunity to capsulated B. anthracis 36 . Also, in this regard, Altman has reported that patients recovered from cutaneous anthrax, who anecdotally exhibit long-term protection from subsequent infection, show high frequencies of CD4+ T cells in response to PA and LF 34 . In our study, the AVA vaccine, which gave very poor protection against respiratory challenge with B. anthracis, induced strong antibody responses and toxin neutralization activity but essentially no T-cell responses to PA, and neither humoral nor T-cell responses to LF. In contrast, homologous prime-boost vaccination with rLVS ΔcapB/Ba and heterologous prime-boost vaccination with rLVS ΔcapB/Ba -rLm/Ba expressing an LFnPAc fusion protein, which gave strong protection against B. anthracis respiratory challenge, not only induced PA-and LF-specific serum IgG, dominated by Th1-type IgG2a, and toxin neutralizing antibody, but also induced strong T-cell responses, including IFNγ-, TNFα-, and/or IL2-expressing multifunctional CD4+ T cells, suggesting a role for T-cell mediated immunity in addition to humoral immunity in protection against anthrax. In our study, the antibody titer to LF, but not to PA, correlated with protection. The lack of correlation between protection and anti-PA antibody in our study is consistent with results of some studies 45,46 but contrasts with the results of others [47][48][49] .
Currently there is no licensed vaccine against Y. pestis. Vaccines studied preclinically include live attenuated Y. pestis vaccines, such as EV76; subunit vaccines based primarily on F1 and LcrV antigens; and live attenuated heterologous bacterial (Salmonella and Yersinia pseudotuberculosis) or viral (adenovirus, modified vaccinia Ankara, or bacteriophage) vectors expressing F1 and/or LcrV [50][51][52][53][54][55] . These vaccines have their advantages and disadvantages. Generally speaking, live attenuated Y. pestis vaccines, while effective, induce serious local and systemic reactions 55 ; subunit vaccines are effective but do not induce high levels of cell-mediated immunity, likely important for long-lasting protection 56 ; heterologous bacterial vectors are safe but have not been as effective as subunit vaccines 57 ; and viral vectored vaccines are effective but may be limited by preexisting immunity. Immunology studies have shown that antibodies to F1 and LcrV provide short-term protective immunity against challenge with Y. pestis; however, F1-LcrV-specific T cell responses with a preferential Th1 polarization are also critical for protection against Y. pestis 55,57 . Our study shows that homologous prime-boost vaccination (two or three doses) with and CD8+ (d) T cells expressing IFN-γ, TNF-α, IL-2, and/or IL-17A. Shown are the frequencies of CD4+ T cells expressing IFN-γ, TNF-α, IL-2, and/or IL-17A (c) and CD8+ T cells expressing IFN-γ (d). Values in b-d are means + SEM. Differences among individual groups were evaluated by Two-way ANOVA with Tukey's correction. Values significantly different from the Sham group are marked with asterisk(s) over brackets above the comparison groups; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Results shown are representative of three similar experiments.   rLVS ΔcapB/Yp or heterologous prime-boost vaccination with rLVS ΔcapB/Yp-rLm/Yp induces significantly elevated serum antibody to F1-LcrV, balanced between IgG1 and IgG2a, elevated IFN-γ secretion by spleen and lung cells, and partial protection against pneumonic challenge with virulent Y. pestis; protection correlates with serum antibody to F1. Protection was not as great as that induced by immunization with the toxic and unlicensed EV76 vaccine, which induced much higher levels of antibody to F1 antigen; the suboptimal protective immunity induced by rLVS ΔcapB/Yp and/or rLm/Yp vaccines may have been due to the relatively poor immunogenicity of their F1-LcrV fusion protein construct. It has been shown that the mutated form of F1, F1mut, folds into a monomer, rather than polymerizing as a linear fiber, enhancing its immunogenicity 11,54 . Potentially, the efficacy of our vaccines can be substantially enhanced by similarly expressing F1mut as well as by expanding the antigen repertoire by the addition of other immunoprotective antigens.
As with plague, there are no licensed vaccines against tularemia. The most promising vaccine candidates being studied preclinically, like the unlicensed LVS vaccine, are live attenuated Francisella vaccines, including our LVS ΔcapB vectored vaccine, derived ultimately from F. tularensis subsp. holarctica 17 ; vaccines derived from the nonhuman pathogen F. novicida 58 ; and vaccines derived from the highly virulent human pathogen F. tularensis subsp. tularensis Schu S4 strain [59][60][61] . Single deletional mutants of F. tularensis Schu S4 are effective, comparable to LVS and the rLVS ΔcapB vaccines, but safety considerations dictate the need for at least one additional major attenuating deletion; thus far, such additional deletions have resulted in impotent vaccines 62 . In contrast, our rLVS ΔcapB/iglABC vaccine is highly potent; has three major attenuating deletions; is >10,000-fold less virulent than LVS in the mouse model; is unmarked, i.e. devoid of antibiotic resistance genes; and has an excellent safety profile 17 .
Our previous studies have shown that systemic (i.d.) vaccination with a single dose of rLVS ΔcapB/iglABC induces potent CD4+ and CD8+ T cell immune responses, Th1-dominated serum antibody, and partial protection against respiratory challenge with F. tularensis Schu S4 19 . Our study here shows that systemic i.d. vaccination with just two doses rLVS ΔcapB/iglABC induces substantial protection to pneumonic challenge with F. tularensis Schu S4 strain, equivalent to the LVS vaccine; mucosal (i.n.) vaccination with either two or three doses of rLVS ΔcapB/iglABC provides 100% protection.
To our knowledge, our platform is the first to demonstrate efficacy against three Tier 1 Select Agents. Vaccines have previously been described against two pathogens, including a vaccinia-based vaccine against smallpox and anthrax 32 and subunit vaccines against anthrax and plague 11,63 . Among the strategies tested in our proof of principle studies -(a) LVS ΔcapBvs. Lm-vectored vaccines; (b) homologous vs. heterologous prime-boost vaccination; (c) mucosal vs. systemic (i.d. for rLVS ΔcapB and i.m. for rLm) vaccine administration; (d) one vs. two vs. three immunization doses -we found that three homologous i.d. vaccinations with the LVS ΔcapB-vectored B. anthracis, Y. pestis, and F. tularensis vaccines is a strategy that works well for all three target pathogens. This strategy has the major advantage of being a single vector platform, induces strong antigen-specific T-cell and humoral immune responses, and provides potent efficacy against all three pathogens. In the case of F. tularensis, mucosal i.n. delivery of the rLVS ΔcapB/iglABC vaccine was somewhat superior to i.d. delivery of this vaccine; however, the i.n. route raises safety issues and administering this one vaccine by that route and the anthrax and plague vaccines by the i.d. route would preclude concurrent immunization of all three vaccines together. In the case of B. anthracis, the heterologous prime-boost vaccination strategy was also highly effective; however, this approach would require development of two vaccines based upon different vectors and likely require administration by two different routes, a significant disadvantage in terms of cost and complexity of development, manufacture, and regulatory approval, and ease of clinical administration.
With respect to safety in humans, both the LVS and the Lm ΔactA ΔinlB parental vectors have established safety profiles in humans. LVS, which retains significant virulence in animals and shows residual toxicity in humans, is the only tularemia vaccine demonstrated efficacious in humans. In comparison with its wild-type F. tularensis subsp. holarctica parent, LVS has two major attenuating deletions, FTT0918 (virulence determinate of SCHU S4) and pilA 64,65 , and as noted above, our LVS ΔcapB vector has a third major attenuating deletion, capB, rendering it >10,000 fold less virulent for mice than LVS by the i.n. route. The Lm ΔactA ΔinlB vector has been shown to be safe in humans 23 , and the additionally modified Lm ΔactA ΔinlB prfA(G155S) vector retains the attenuation of the parental Lm ΔactA ΔinlB vector in mice 25 while providing significantly enhanced antigen-specific T-cell mediated immune responses 24,25 . Our study shows that the LVS ΔcapBand Lm ΔactA ΔinlB prfA(G155S)-vectored F. tularensis, B. anthacis, and Y. pestis vaccines are safe and efficacious in BALB/c mice. Further validation of genetic stability and safety of the vaccines will be required prior to human studies.
In summary, our live attenuated single vector vaccine platform elicits robust humoral and T-cell mediated immune responses and protective immunity against three target pathogens -the agents of anthrax, plague, and tularemia -and overcomes the problems of poor immunogenicity and lack of immunological memory often associated with subunit vaccines against these pathogens. As already noted, a single vector platform vaccine has numerous advantages in terms of production cost, regulatory approval, ease of administration, and patient acceptability. This vaccine platform can be extended readily to cover other pathogens including other Tier 1 Select Agents such as Burkholderia pseudomallei and Burkholderia mallei.

Materials and Methods
Cell line, bacteria, and vaccines. Mouse macrophage-like J774A.1 cells (ATCC TIB-67) and human macrophage-like THP-1 cells (ATCC TIB-202) were negative for mycoplasma contamination and cultured as described previously 15,24 . F. tularensis Live Vaccine Strain (LVS) was obtained from the Centers for Disease Control and Prevention (Atlanta, Ga.). B. anthracis AVA (Anthrax vaccine adsorbed) vaccine was obtained from BEI Resources. Y. pestis attenuated EV76 strain was obtained from Centers for Disease Control and Prevention, Fort Collins, CO. Virulent strains of B. anthracis (Ames spores), Y. pestis (CO92), and F. tularensis (Schu S4) were originally obtained from BEI Resources, stored at −80 °C, and used in animal challenge experiments at  19 , rLVS ΔcapB/Ba, and rLVS ΔcapB/Yp, using a strategy similar to one published previously 19 . Specifically, we constructed the shuttle plasmid for expressing B. anthracis fusion protein LFnPAc, pFN-L/bfr-BaLFnPAc(GGSG), by replacing the gro-gfp expression cassette in pFNLTB6 gro-gfp (Kan R ) 66 with the bfr-BaLFnPAc(GGSG) expression cassette comprising the F. tularensis bacterioferritin promoter (upstream of FTT_1441, bfr, amplified from the genomic DNA of an clinical isolate of SCHU S4 strain), a Shine-Dalgarno sequence, a 6-nucleotide spacer, followed by the coding sequence for B. anthracis LFn 26,27 and PAc 12 separated by a GGSG linker (LFnPAc). The coding sequence for LFnPAc was amplified by two-step overlap PCRs. First, we amplified the coding sequences for LFn and PAc by using the genomic DNA of B. anthracis (BEI NR-9540) and primer pairs LF_Fw1 (TAACAATAGGAGGTACGTAATGGCGGGCGGTCATGGTGATG) and LFn_Rv (TGTTTGTTGATCGAAATTACCAGAACCACCTAGATTTATTTCTTGTT) for LFn, and primer pairs PAc_ Fw (AACAAGAAATAAATCTAGGTGGTTCTGGTAATTTCGATCAACAAACA) and PA_Rv1 (TGAAACGAG CTAGTGGATCCTTATCCTATCTCATAGCCTTTTT) for PAc, respectively. Secondly, we used the PCR products of LFn and PAc and primer pairs LF_Fw1 and PA_Rv1 to amplify the coding sequence for fusion protein LFnPAc, which was subsequently cloned into the pFNL-derived shuttle vector. We verified the inserted sequence by restriction analysis and nucleotide sequencing, electroporated the resultant shuttle plasmid into LVS ΔcapB, and selected clones that were kanamycin-resistant and stably expressed the targeted antigens shown by Western blotting. Similarly, we constructed rLVS ΔcapB/Yp carrying the shuttle plasmid pFNL/omp-YpF1V(GGSG) for expressing the Y. pestis fusion protein of F1 (170 aa) and LcrV (326 aa) separated by the GGSG linker (F1V) driven by the F. novicida outer membrane promoter (upstream of FTN_1451, omp, amplified from F. novicida genomic DNA). The coding sequence for F1 and LcrV were amplified from the genomic DNAs of Y. pestis (BEI DD-494) by using primer pair F1_Fw1 (GACTAAAAGGAGGTACGTAATGAAAAAAATCAGTTCCGTTAT) and F1_Rv (TGTTCGTAGGCTCTAATCATACCAGAACCACCTTGGTTAGATACGGTTAC) for F1 and primer pair LcrV_Fw (GTAACCGTATCTAACCAAGGTGGTTCTGGTATGATTAGAGCCTACGAACA) and LcrV_ Rv1 (CGAGCTAGTGGATCCTCATTTACCAGACGTGTCATCTA) for LcrV, respectively.

Construction and verification of attenuated recombinant Listeria monocytogenes vaccines expressing B. anthracis, Y. pestis, and F. tularensis fusion proteins. Using Lm ΔactA ΔinlB
ΔuvrAB prfA(G155S) (Lm ΔactA ΔinlB prfA, generously provided by J. Skoble of Aduro Biotech, previously Anza Therapeutics) 25,33,67 as a vector, we constructed recombinant Listeria-vectored vaccine candidates expressing immunogenic fusion proteins of B. anthracis, Y. pestis, or F. tularensis, using methodology described previously by us and others 15,22,24,68 . Briefly, we amplified the encoding sequence for the fusion bled at Week 13, and challenged i.n. with 2 CFU (2 LD 50 ) or 6 CFU (6 LD 50 ) F. tularensis Schu S4 at Week 14, and monitored for 3 weeks, as indicated. (d) Survival after vaccination and challenge and serum antibody prechallenge -Experiment VII. Upper panels -Survival. The survival curve after challenge with 2 LD 50 (left panel) or 6 LD 60 (right panel) of each vaccinated group is compared with that of either the sham-or LVS-immunized group by the log-rank test (Mantel-cox); P values that are significantly different from the control group are marked with asterisk(s) color-coded to the color of the vaccine symbol. ***P < 0.001 vs. Group A (Sham); *P < 0.05 vs. Group B (LVS). Lower panel-Serum antibody after vaccination. Sera were assayed for IgG and subtypes IgG1 and IgG2a specific to heat-inactivated LVS. Data are mean + SEM of serum antibody endpoint titer for n = 8 per group. Differences among individual groups were compared by two-way ANOVA with Tukey's correction. Values that are significantly different between two groups are marked with asterisk(s) over an open horizontal line crossing above the two groups. As indicated by the asterisks above the Sham group, its titers were significantly different from all other groups. *P < 0.05; **P < 0.01; and ****P < 0.0001.
Scientific REPORTS | (2018) 8:7009 | DOI:10.1038/s41598-018-24581-y protein of B. anthracis LFnPAc and Y. pestis F1V from the above described F. tularensis shuttle plasmids by PCR with primer pairs LF_Fw2 (AGGTGGATCCATGGCGGGCGGTCATGGTG) and PA_Rv2 (CGGTGGCGGCCGCTTATCCTATCTCATAGCCTTTTTTA) for LFnPAc and F1_Fw2 (CGAGGGATCC ATGAAAAAAATCAGTTCCGTTAT) and LcrV_Rv2 (ATATGCGGCCGCTCATTTACCAGACGTGTCATCTA) for F1V, respectively, ligated them with either the Lm hly promoter and the coding sequence for the listeriolysin O (encoded by hly) signal sequence (LLOss) or the Lm actA promoter and the coding sequence for the ActA N-terminal 100 amino acids (ActAN), and cloned into a phage-based Listeria site-specific integration vector derived from pPL2 (kindly provided by J. Skoble) 68 . We subsequently integrated the resultant plasmid into the 3′ end of tRNA arg on the bacterial chromosome of the recipient Lm ΔactA ΔinlB prfA strain from the donor SM10 strain carrying the integration plasmid through conjugation to obtain rLm vaccines. We confirmed all the molecular plasmid constructs by nucleotide sequencing and verified the final recombinant L. monocytogenes strains by colony PCR for chromosomal integration and by Western blotting for heterologous protein expression.
Heterologous protein expression by and growth kinetics of LVS ΔcapBand Listeria-vectored vaccines in broth culture and in infected macrophage-like cells. To assess protein expression by rLVS ΔcapB vaccines grown on agar, we grew each of the vaccine stocks on Chocolate agar, selected single colonies, lysed them in SDS buffer, applied the lysates to SDS-PAGE, and analyzed protein expression by Western blotting. Secreted proteins in the supernate of Brain Heart Infusion (BHI) broth culture of rLm vaccines were precipitated by the TCA-acetone method and analyzed by Western blotting. Monoclonal antibodies specific to B. anthracis PA (BEI, DD-9) and goat polyclonal antisera specific to Y. pestis LcrV (BEI, NR-31022) were used as primary antibody in Western blotting. To assay protein expression of LVS ΔcapB-vectored vaccines in macrophage-like cells, we seeded monocytic THP-1 cells at 3 × 10 5 cells/well on 24-well plates and differentiated them in the presence of PMA for 3 days. Vaccine vector (LVS ΔcapB) and rLVS ΔcapB vaccines were grown on Chocolate agar supplemented without (vector) or with (vaccines) kanamycin (7.5 μg/ml) for 3 days. The differentiated THP-1 cells were left uninfected or infected with rLVS ΔcapB/Ba or rLVS ΔcapB/Yp opsonized with human serum at a multiplicity of infection (MOI) of 10:1 (bacteria: cell) and incubated at 37 °C for 1 h. The cells were then washed with RPMI three times and incubated with complete RPMI supplemented with gentamycin (0.1 μg/ml) to inhibit extracellular bacterial growth. At 24 h post infection, medium was removed from wells; cells were lysed, and cell lysates analyzed by Western blotting using a mixture of monoclonal antibody to B. anthracis PA antigen and goat polyclonal antibody to Y. pestis LcrV antigen. Protein expression of Lm-vectored vaccines in macrophage-like cells was assayed as described previously 15,24 . Growth kinetics of LVS ΔcapBand Lm-vectored vaccines in broth and in infected macrophages were assayed as described in the Supplemental Information (legends to Figs S2 and S3, resp.) and as published by us previously 15,19,24 . Immune response analysis. Groups of 4-8 BALB/c mice were immunized as indicated in Figs 2a, 3a, 4a, 6a and 8a,c, Figs S7a and S10a. In experiments studying immunology and efficacy of B. anthracis, Y. pestis and F. tularensis vaccines, mice were bled one week prior to challenge. In experiments studying immunology only, mice were bled at one week post rLm boosting and subsequently euthanized; spleens and lungs removed; single cell suspensions of spleen and lung cells prepared and suspended in T cell medium; and cells enumerated as described previously 19 . Serum was isolated and stored at −80 °C until use. Serum antibody, in vitro stimulation and production of IFN-γ and IL-4 by murine immune lung and spleen cells, and intracellular cytokine staining of lung and spleen cells for flow cytometry analysis were assayed as published by us previously 15,19,24 .
Serum antibody. Sera were tested for IgG antibody response by enzyme-linked immunosorbent assay (ELISA) using standard procedures 15 . Briefly, ninety-six-well microtiter plates were coated with recombinant protein PA (BEI, NR-3780), LF (NR-4268), F1 (NR-44223), LcrV (NR-32875), or F1-LcrV monomer (NR-2563) at 1 µg/ml each, or heat-inactivated LVS (HI-LVS, 2 × 10 7 /ml) diluted in carbonate buffer overnight at 4 °C and afterwards processed at ambient temperature. The plates were washed three times with 0.05%Tween20-PBS, blocked in 3% BSA-PBS for 3 h, incubated with each serum sample serially diluted 2-fold twelve times at a starting dilution of 1:20 or 1:50 in 1% BSA-PBS for 90 min, and washed again. Bound antibody was detected by using alkaline phosphatase-conjugated goat anti-mouse IgG (Sigma, St. Louis, MO), subtypes IgG1 (Sigma), IgG2a (Abcam), IgG2b (sigma), or IgG3 (Sigma) diluted in 1% BSA-PBS and incubating for 90 min. Plates were developed with 100 µl of p-nitrophenylphosphate substrate (BioRad), and the A415 was read using a multiscan microplate reader (iMark, BioRad). The results are presented as the mean antibody endpoint titer and SE of the mean (SEM). Antibody endpoint titer is defined as the mean log dilution that yields an OD greater than the mean OD of Sham sera plus three standard deviations at the same serum dilution.
In vitro stimulation and production of IFN-γ and IL-4 by murine immune splenocytes. A single cell suspension of 1.0 × 10 5 splenocytes or lung cells per well was seeded in U-bottom 96-well plates and incubated with T-cell medium alone, or T-cell medium supplemented with 2 µg/mL of recombinant PA, LF, F1, LcrV, or F1-LcrV monomer for three days. After a 3-day incubation, the culture supernatant fluid was collected, cell debris removed by centrifugation, and the supernatant fluid stored in assay diluent (BD Biosciences) at −80 °C until use. The production of mouse IFN-γ and IL-4 in the culture supernatant fluid was assayed using a mouse cytokine EIA kit (BD Biosciences) 24 .
In vitro stimulation and intracellular cytokine staining for flow cytometry analysis. A single cell suspension of 7.5 × 10 5 lung cells or 1.5 × 10 6 splenocytes per well was seeded in U-bottom 96-well plates and stimulated with 2 µg/mL of recombinant PA or LF, or 2 × 10 7 /ml of HI-LVS in the presence of anti-CD28 monoclonal antibody (Clone 37.51) for a total of 6 h, and processed for Flow Cytometry analysis as described previously by us 19,24 . The frequencies of live CD4+ Scientific REPORTS | (2018) 8:7009 | DOI:10.1038/s41598-018-24581-y and CD8+ T cells producing any of the 15 possible combinations of four cytokines (IFN-γ, TNF-α, IL-2, and IL-17A) were uniquely distinguished using logic combinations of the gates for each cytokine and FACSDiva (BD) software. Background frequencies of cells producing cytokines without antigen stimulation were subtracted.
Protective efficacy. Efficacy of rLVS ΔcapBand rLm-vectored B. anthracis, Y. pestis, and F. tularensis vaccines was studied at CSU similarly to what was previously described by us [17][18][19] . Virulent strains of B. anthracis (Ames), Y. pestis (CO92), and F. tularensis (Schu S4) were originally obtained from BEI Resources, stored at −80 °C, and used to make working stocks for animal challenge experiments at CSU. Briefly, Y. pestis was grown to log phase at 37 °C in BHI broth and F. tularensis was grown on modified Mueller-Hinton agar plates and colonies collected into Mueller-Hinton broth. For both Y. pestis and F. tularensis, glycerol was added to the harvested bacterial suspensions to 15% (v/v) and the suspensions frozen in aliquots at −80 °C. B. anthracis spores were prepared using published procedures 69 , resuspended in PBS, and frozen in in aliquots at −80 °C. We pre-determined that the 50% lethal dose (LD 50 ) of B. anthracis (Ames) spores, Y. pestis CO92, and F. tularensis Schu S4 administered intranasally (i.n.) in BALB/c mice is approximately 40,000, 250, and 1 CFU, respectively. Mice were sham-immunized or immunized with 2 or 3 doses of rLVS ΔcapB or rLm vaccines (homologous prime-boost vaccination) or primed-boosted with rLVS ΔcapB -rLm vaccines (heterologous prime-boost vaccination), challenged i.n. with virulent B. anthracis (Ames) spores, Y. pestis CO92, or F. tularensis Schu S4, weighed, and monitored for illness and death for 3 weeks, as indicated. Mice that met predetermined humane endpoints for euthanasia were euthanized and counted as a death. Mean survival time was calculated by dividing the sum of the surviving days of all animals by the total number of animals examined, with animals surviving until the end of the experiment given a time of 21 days, when the experiment was terminated.
Statistical analyses. The sample sizes for assaying vaccine clearance and dissemination (4/group/time point), immune responses after vaccination (4/group), and efficacy (8/group) after challenge were estimated based on previous and pilot studies (GraphPad StatMate 2.0). Means and SE of the mean (SEM) of serum antibody endpoint titer, cytokine production, and frequencies of cytokine-producing CD4+ and CD8+ T cells were reported, and means compared across groups by ANOVA with Tukey's correction for multiple comparisons test using GraphPad Prism, 6.04 (San Diego, CA). A log-rank analysis (Mantel-Cox test) (Prism 6.04) was used to determine the significance of differences in survival curves among mice in vaccinated and control groups. Linear regression was used to obtain values for the slope and intercept and the correlation coefficient (R 2 ) between pre-challenge serum antibody endpoint titer and post-challenge mean survival time (days) at 21 days post-challenge.
Data availability. All data supporting the findings of this study are available within the article and its supplementary information files or from the corresponding author upon request.