Tuning Subunit Vaccines with Novel TLR Triagonist Adjuvants to Generate Protective Immune Responses against Coxiella burnetii

Vaccines are a powerful prophylactic tool to prevent disease. Inactivated whole-organism vaccines stimulate both innate and adaptive immunity and do not require additional immunostimulants (adjuvants) to induce a robust and long-lasting immune memory. However, inactivated vaccines for some diseases can cause adverse events in previously exposed or immunized individuals (1–3). Conversely, subunit formulations based on recombinant proteins are a safer alternative for use in both immunocompetent and immunocompromised patients (4–7). The Ags used in subunit vaccines do not induce adverse inflammatory responses, but they do not stimulate robust immune responses. The low immunogenicity of these Ags necessitates the use of adjuvants in subunit vaccine formulations. Selecting the appropriate Ags and adjuvants to induce a protective response remains a challenge. Thus, we hereby present a primary example of using novel adjuvants to tailor immune responses against Coxiella burnetii.

Adjuvants enhance Ag recognition, uptake, and processing by APC (8–10). The addition of adjuvants to subunit vaccine formulations can also reduce the amount of Ag needed per immunization, reduce the number of doses required for protection, enhance immunogenicity in individuals at the extremes of age, and induce broad spectrum protection against related pathogenic strains (8, 10, 11). Tailoring the immune response to the Ag can be achieved through modulation with specific adjuvants, which enhance the immunogenicity and protective efficacy of subunit vaccines. A large area of adjuvant research uses agonists of TLRs, a class of pattern recognition receptors, which activate NF-κB–induced proinflammatory cytokines. TLR agonists are naturally present in heat-killed or attenuated whole-cell vaccines and are responsible for stimulating the innate immune system’s response to the immunizing pathogen (12). In fact, pathogens contain distinct TLR agonist combinations that stimulate tailored responses, which can be exploited in vaccine design to develop adjuvants that are best suited for combating infection caused by a pathogen (13). Including adjuvants in subunit vaccine formulations is necessary to generate immune responses against the Ag; TLR agonists have been heavily explored for this role (14). In addition, different TLR agonists activate immune responses synergistically when administered simultaneously, improving the efficacy of the vaccine (15). However, small-molecule TLR agonists may cause systemic toxicity because of rapid diffusion from the site of injection, but this effect can be overcome through conjugation to other chemical moieties or polymers (16, 17).

To circumvent these challenges, we linked different combinations of three TLR agonists to mimic their spatial organization on a pathogen to generate unique and distinct innate immune responses in vitro and in vivo (18). In this study, we evaluate the ability of these TLR triagonists to stimulate protective Ag-specific immune responses in subunit vaccine formulations in vivo. We chose C. burnetii as the target of our subunit vaccine efforts because of previous challenges in generating protective subunit vaccines against this bacterium in addition to the lack of U.S. Food and Drug Administration (FDA)–approved vaccines for C. burnetii.

C. burnetii, the etiologic agent of Q fever, is a category B bioterrorism agent that is easily aerosolized and has a single bacterium infectious dose (19, 20). The only licensed human vaccine against C. burnetii is Q-Vax, a formalin-inactivated whole-bacteria vaccine approved for use in Australia and some European countries (see the visual abstract for this article in the online version). Individuals previously exposed or vaccinated against C. burnetii can have severe hypersensitivity reactions to Q-Vax (21–24), necessitating prescreening by serology and a skin test for previous sensitization to C. burnetii proteins (22–24). The potential adverse response to immunization, general impracticalities presented by Q-Vax immunization, and lack of FDA approval are major concerns, given the potential of C. burnetii’s use as a bioterrorism agent (20, 25). Thus, there is a critical need to develop a protective but nonreactogenic vaccine against this bacterium.

In these studies, we evaluated the adjuvant activity of a library of TLR triagonists in subunit vaccines formulated with recombinant C. burnetii Ags (Visual Abstract). We characterized the ability of our (to our knowledge) novel adjuvant platform to stimulate robust, Ag-specific adaptive immune responses in in vivo immunogenicity studies and evaluated the functional efficacy of these responses in a live C. burnetii aerosol challenge model in mice. Given the reactogenicity, safety, and implementation issues associated with Q-Vax, we chose to take a subunit vaccine approach to develop our vaccine candidates against C. burnetii. From these studies, we identified multiple vaccine candidates, formulated with specific TLR triagonist combinations, that conferred indications of protection against a live C. burnetii challenge.

The design, synthesis, and in vitro evaluation of the TLR triagonist compounds are described in detail in our previous work (18). Briefly, TLR triagonists were constructed through bioconjugation reactions (amide bond formation, maleimide-thiol Michael addition, and azide-alkyne click chemistry) of individual TLR triagonists, functionalized with chemical handles, to a central triazine core. The individual agonists included Pam₃CSK₄ (TLR1/2a), Pam₂CSK₄ (TLR2/6a), a pyrimido-indole derivative (TLR4a) (26), an imidazoquinoline derivative (TLR7a) (17), and a single-stranded CpG 1826 class B DNA (TLR9a). The five combinations synthesized were TLR1/2_4_7a, TLR2/6_4_7a, TLR 1/2_4_9a, TLR2/6_4_9a, and TLR4_7_9a, which were generated by conjugation of the indicated agonists in a 1:1:1 ratio to the triazine core. See Albin et al. (18) for chemical structures.

Five- to six-week-old female C57BL/6 mice were purchased from Charles River Laboratories International. Mice were housed in a specific pathogen-free facility with a 12-h dark/light cycle with autoclaved bedding and irradiated food. All handling of mice at both the University of California, Irvine and U.S. Army Medical Research Institute of Infectious Diseases was performed under an approved Institutional Animal Care and Use Committee protocol. Groups of five or eight mice were given a priming dose (day 0) and boosted 2 wk later (day 14). Mice were administered 1 nmol of each TLR agonist linked to the inert core molecule or mixed in liquid formulations with 0.5 nmol of C. burnetii Ags. C. burnetii Ags were synthesized and purified by GenScript (Piscataway, NJ). In relevant vaccine formulations, AddaVax (InvivoGen, San Diego, CA) was added at 50% of total vaccine volume. Fifty microliters of vaccine formulations was delivered i.m. in the thigh using a 31-gauge needle (3/10cc, Insulin Syringes; Becton Dickinson, Laguna Hills, CA). Animals were periodically weighed throughout the in vivo experiment protocol.

Animals were anesthetized under 2–2.5% isoflurane and 2 l/min of oxygen flow under standard atmospheric pressure for these experiments. Plasma was collected on days 0, 1, 14, and 15 from isoflurane-anesthetized mice via cheek bleed using a 25-gauge needle (Becton Dickinson PrecisionGlide Needle). Blood was collected in Microvette CB 300 LH lithium heparin tubes (Sarstedt; Aktiengesellschaft and Co., Sparks, NV), and plasma was separated at 2000 × g for 5 min at 4°C. Serum was collected on termination day (day 17 or 21) via cardiac puncture of carbon dioxide–euthanized mice using a 29-gauge needle (1 ml, Insulin Syringe; Exelint International, Redondo Beach, CA) and separated at 2000 × g for 10 min at 4°C. Plasma and serum samples were stored at −20°C for further analysis.

Spleen and draining inguinal lymph node (LN) samples were collected from mice on day 17 or 21. Whole spleens and LNs were ground up and passed through 40-μm nylon mesh strainers (Thermo Fisher Scientific, Waltham, MA) with PBS to make single-cell suspensions. Splenic single-cell suspensions were treated with ACK lysing buffer (Thermo Fisher Scientific) to remove RBCs. A total of 1 × 10⁶ cells from each spleen/LN suspension were then incubated with anti-CD16/32 before subsequent staining with allophycocyanin/Cy7–anti-B220, FITC–anti-CD4, and PE/Cy7–anti-CD8α in cell staining buffer (BioLegend, San Diego, CA). All Abs were purchased from BioLegend. Samples were analyzed on a NovoCyte 3000 flow cytometer (ACEA, San Diego, CA) using the NovoExpress software. Total numbers of spleen/LN lymphocytes were back-calculated from the number of marker-positive cells read and the total volume of sample processed by the NovoCyte 3000 flow cytometer.

A total of 1 × 10⁶ splenocytes per ml were incubated in CytoOne flat-bottom tissue culture plates (USA Scientific, Ocala, FL). Cells were stimulated with 10 μg/ml C. burnetii Ag in RPMI 1640 supplemented with 10% FBS, 2% penicillin and streptomycin, and 0.2% 2-ME (complete RPMI 1640) and incubated for 48 h at 37°C at 5% CO₂. IFN-γ or IL-4 levels were analyzed in undiluted supernatants by ELISA (BioLegend) according to the manufacturer’s instructions. For ELISpot Analysis, 2 × 10⁵ splenocytes per well were incubated with 10 μg/ml C. burnetii Ag in complete RPMI 1640 in double-color IL-4/IFN-γ ELISpot plates (ImmunoSpot; Cellular Technology Ltd, Shaker Heights, OH) for 48 h at 37°C at 5% CO₂. Plates were processed using manufacturer’s instructions and analyzed using Cellular Technology Ltd ImmunoSpot scanning services.

Groups of 10 mice were given a priming dose (day 0) via i.m. injection with either three recombinant protein vaccines (CBU_1910 alone, CBU_1910 and TLR triagonist 2/6_4_7a, or CBU_1910 and TLR agonists 2/6 + 4 + 7a), positive control Q-Vax, or saline control. Blood was collected from vaccinated mice on day 0 (baseline) and day 14 (just before boost, except Q-Vax–vaccinated mice) and 70 (at the end of study).

Exposure to aerosolized bacteria was accomplished as previously described (27, 28). All mice were exposed to a target dose of 10¹¹ genome per milliliter of aerosolized C. burnetii Nine Mile strain 6 wk following last vaccination. Mice were transferred to wire mesh cages, and wire mesh cages were placed in a whole-body aerosol chamber within a class three biological safety cabinet located inside the Biosafety Level 3 Laboratory. Aerosols were created by a three-jet collision nebulizer for 10 min at a constant flow rate of 19 l/min followed by a 5-min wash cycle. Following the wash cycle, mice were removed from aerosol chamber and transported back to their housing room. The aerosolization was performed at ambient temperature and humidity. The generated aerosol was sampled with an all-glass impinger sampling at a rate of 6 l/min. All-glass impinger samples were analyzed by real-time PCR assay. The inhaled dose was calculated using the following formula: dose = [aerosol (μg/ml) × minute volume (ml) × exposure time (min)]. The minute volume of mice was estimated using the mean weight of all mice on the day of exposure and Guyton formula (29).

All of the mice, including negative saline control, survived the aerosol exposure with C. burnetii. Mice were weighed once a day. The weight change was calculated as the percent difference between starting weight on the day of challenge (day 42 of the study) and weight on each subsequent day following challenge. All surviving mice were euthanized at the end of study on day 70 ± 2 d postexposure via i.p. barbiturate overdose, followed by cervical dislocation.

Custom-purified C. burnetii proteins (Genscript) were diluted to 0.1 mg/ml in PBS-0.001% Tween 20 and then printed in triplicate onto nitrocellulose-coated glass AVID slides (Grace Bio-Labs, Bend, OR) using an Omni Grid 100 microarray printer (Genomic Solutions). Mouse serum samples were diluted 1:100 in protein array blocking buffer (GVS, Sanford, ME) and then incubated on microarrays overnight at 4°C with gentle agitation. Arrays were washed with TBS-0.05% Tween 20 before incubation with biotinylated anti-mouse total IgG, IgG1, or IgG2c Abs (Jackson ImmunoResearch Laboratories, West Grove, PA) diluted 1:200 in array blocking buffer. Bound anti-mouse Abs were detected by incubation with streptavidin-conjugated Qdot800 or streptavidin-Qdot 585 (Thermo Fisher Scientific, Eugene, OR) diluted 1:200 in array blocking buffer. Slides were washed and then air dried by brief centrifugation. Images were acquired using an ArrayCAM Imaging System (Grace Bio-Labs). Signal intensities were corrected for spot-specific background before further analysis.

For Q-Vax immunogenicity studies, groups of five female C57BL/6 mice received 50 μl of Q-Vax (CSL, Melbourne, VIC, Australia) administered either i.m. or s.c. Mice were bled on day 0, 21, 28, and 67, and plasma was used to probe a C. burnetii proteome microarray (30, 31) to identify reactive Ags.

All animal protocols were approved by the Institutional Animal Care and Use Committee of the University of California, Irvine, and was conducted under an Institutional Animal Care and Use Committee–approved protocol. Animal research at the U.S. Army Medical Research Institute of Infectious Diseases was conducted under an Institutional Animal Care and Use Committee–approved protocol in compliance with the Animal Welfare Act, Public Health Service Policy, and other federal statutes and regulations relating to animals and experiments involving animals. The facility where this research was conducted is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care, International, and adheres to principles stated in the Guide for the Care and Use of Laboratory Animals, National Research Council, 2011.

All data were analyzed for statistical significance using one- or two-way ANOVA on GraphPad Prism 7 (GraphPad Software, La Jolla, CA).

The first goal of our study was to identify immunogenic bacterial Ags that could be included in our subunit vaccine formulations. Several C. burnetii Ags have been previously reported to stimulate robust Ag-specific humoral immune responses in a naturally infected human patient population (30–32). The C. burnetii outer membrane protein CBU_1910, or COM1, was the dominant Ag identified in many of these studies, including ours (30–33). We also wanted to identify the Ag-specific humoral response stimulated by immunization with Q-Vax because this vaccine is known to confer protection to immunized individuals and in mice. To accomplish this, we immunized C57BL/6 mice with Q-Vax and analyzed the Ag-specific Ab response elicited 28 d postimmunization (Fig. 1A) using a C. burnetii protein microarray (30–32). Despite immunization with a whole-cell vaccine containing over 2000 potential Ags, the Ab response was observed to target very few Ags (6 Ags >5000 signal intensity) from the C. burnetii proteome (Fig. 1B, data not shown). Furthermore, the only bacterial protein that generated robust and consistent Ab responses across all five immunized animals was CBU_1910, with an average signal intensity at least 3.4-fold higher than any other Ag (Fig. 1B, 1C). In addition, the Abs generated to CBU_1910 were all IgG2c biased (Supplemental Fig. 1). Because of its relevance in human C. burnetii infections and the protective vaccine, we selected CBU_1910 as the Ag for immunogenicity testing in formulations with our novel TLR triagonist adjuvants.

In our first vaccine candidate, we formulated CBU_1910 with TLR2/6_4_7 triagonist (TLR2/6_4_7a) in both the linked and unlinked forms. Our preliminary in vitro and in vivo studies indicated that this TLR triagonist activates dendritic cells without stimulating long-lasting serum cytokines following immunization in the linked form (18), an indication of low systemic adjuvant-induced toxicity. We immunized cohorts of C57BL/6 mice i.m. on day 0 followed by a boost of the identical formulation on day 14. We then collected sera and draining LNs at the completion of the experiment on day 21 postimmunization (Fig. 2A). Animals received either CBU_1910 with the TLR triagonist in either the linked (blue bars) or unlinked (orange bars) form to evaluate whether the physical linking of the three TLR agonists affected their immunomodulatory activity.

Compared to PBS alone, the linked and unlinked TLR2/6_4_7a stimulated a significant expansion of B220⁺ cells (2.9- and 3.5-fold increase, respectively) in the draining LNs (Fig. 2B). This expansion of B cells in the draining LNs correlated with the increased production of IgG, which we detected in the sera on day 21 (Fig. 2C). We analyzed the CBU_1910-specific Ab response to immunization and detected both IgG1 and IgG2c Ab subtypes specific for the immunizing Ag in animals that received either the linked or unlinked forms of TLR2/6_4_7a, with no significant difference between these groups (Fig. 2C).

These two IgG subtypes provide insight into the type of T cell responses generated in response to immunization; IgG2c suggests a Th1 type, or inflammatory, T cell response, whereas IgG1 production is consistent with a Th2, or humoral biasing, T cell response (34). CBU_1910 alone stimulated a low Ag-specific IgG1 response and no IgG2c CBU_1910-specific Abs following immunization. This suggests that this Ag is not inherently immunogenic and requires adjuvants to stimulate immune responses. Linked and unlinked TLR2/6_4_7a generated similar levels of CBU_1910-specific IgG1 Ab. However, the unlinked triagonist generated 2.1-fold higher IgG2c than the linked TLR triagonist, although the difference was not statistically significant (Fig. 3C). Previous studies suggest that T cell responses (Th1), in addition to B cell responses, are important for the protection against and clearance of C. burnetii (35, 36). Although the Ab response in this case skewed toward IgG1, vaccines adjuvanted with linked or unlinked TLR2/6_4_7a resulted in a significant increase in CD4⁺ (2.1- and 2.2-fold increase, respectively) and CD8⁺ (2.4- and 2.6-fold increase, respectively) T cells in the draining LN (Fig. 2B).

We next evaluated the CBU_1910 vaccine candidates adjuvanted with linked or unlinked TLR2/6_4_7a in a live C. burnetii aerosol challenge model (27, 28). For this experiment, we followed the same vaccination schedule used in our immunogenicity studies (Fig. 2E): cohorts of C57BL/6 mice received a prime immunization on day 0 followed by a boost on day 14. Blood was collected on day 28 for serum Ab analysis. On day 42, sera were collected before each cohort was challenged with aerosolized C. burnetii. In this model, mice should exhibit symptoms of infection between days 51 and 54 (9–12 d postchallenge). The experiment was terminated on day 70 (28 d postchallenge), and sera and tissue were collected to analyze the efficacy of our vaccine candidates. A cohort of animals received Q-Vax as a positive control for vaccine efficacy.

Mice from each cohort were weighed and their temperature was measured daily, as weight loss and fever indicate an animal’s clinical response to a C. burnetii challenge (Supplemental Fig. 2A, 2B) (37–39). As expected, the animals that received PBS or Ag only immunizations lost weight in response to the C. burnetii challenge, especially on day 10 (9.9 and 10.6% loss, respectively) postchallenge (Fig. 2F, 2G). Q-Vax–immunized mice continued to gain weight throughout the study, as expected and previously reported (40, 41). However, it was observed that mice immunized with the linked TLR2/6_4_7a had minimal weight loss (day 10: 7.9% loss) in comparison with the cohort that received the unlinked TLR agonists (day 10: 14.6% loss) (Fig. 2F) in response to C. burnetii challenge. Animals in both our TLR triagonist cohorts generated robust anti–CBU_1910 Ab responses to immunization prior to challenge, which were both significantly higher than the anti–CBU_1910 response generated by Q-Vax immunization (Fig. 2H). This potentially indicated a better efficacy of the linked agonists compared with the unlinked agonists in the challenge model. However, there was no significant difference in the levels of C. burnetii in the organs of any of the challenged cohorts when bacterial genomic equivalents were quantified by PCR (Supplemental Fig. 2C). This is likely because the bacteria had been cleared at the time the tissues were collected 28 d postchallenge (37, 42). In addition, the only group that showed significantly lower organ weight following challenge was Q-Vax (Supplemental Fig. 2D), although in the lung and spleen, the Ag-alone group and the linked TLR2/6_4_7a group, respectively, showed a modest reduction in organ weight compared with PBS control. These data indicate that the linked TLR2/6_4_7a triagonist generates Ag-specific Abs in response to immunization but does not confer protection in response to C. burnetii challenge. These data suggest the linked TLR2/6_4_7a is not the ideal adjuvant for use in vaccine candidates against C. burnetii when compared with Q-Vax–immunized mice that did not lose weight in response to challenge.

We hypothesized that to achieve better protection, we would need to emulate the immune response type generated by Q-Vax. One potential difference, noted above, was that mice immunized with TLR2/6_4_7a adjuvanted CBU_1910 vaccines generate predominately Th2-biased Ab responses, in contrast to Q-Vax–induced responses, which are predominantly Th1 biased (Supplemental Fig. 1). Therefore, we began characterizing the immunogenic activity of additional TLR triagonists with the goal of identifying a formulation that stimulated a more Th1-biased immune response.

As reported by Albin et al. (18), we synthesized four additional unique TLR triagonists to be evaluated for their immunogenic activity, specifically focusing on Ag-specific IgG subtypes and T cell responses elicited in response to immunization. CBU_1910 was used as the model immunizing Ag for all of these studies, and we followed the same vaccination protocol as our prior experiments, with each cohort receiving a prime and boost immunization. The resulting immune responses were then evaluated at the termination of the experiment on day 21.

All of the TLR triagonists generated CBU_1910–specific B and T cell responses to immunization (Fig. 3). As we observed in the TLR 2/6_4_7a immunogenicity studies, some of the linked and unlinked TLR triagonists generated distinct CBU_1910–specific humoral responses (Fig. 3A). Linked TLR1/2_4_7a generated relatively weak anti–CBU_1910 IgG1 (72% less) and IgG2c (79% less) Abs compared with the unlinked form (Fig. 3A). Similar to TLR2/6_4_7a, the Ab profile was skewed toward Th2 IgG1 Abs (26% IgG2c linked, 32% IgG2c unlinked) (Fig. 3B). Linked and unlinked TLR2/6_4_9a elicited the most IgG1 Abs compared with the other combinations by at least a factor of 1.7. However, linked TLR2/6_4_9a elicited a 2.0-fold increase in IgG2c Abs over the unlinked combination (Fig. 3A). This difference resulted in 46% IgG2c-skewed profile with linked TLR2/6_4_9a compared with 31% IgG2c-skewed profile with the unlinked combination (Fig. 3B). Although lower in signal intensity, linked and unlinked TLR1/2_4_9a behaved similarly to TLR2/6_4_9a (52% IgG2c linked, 43% IgG2c unlinked). The linked and unlinked forms of the TLR4_7_9a generated some of the least-robust Ag-specific Ab responses out of all five TLR triagonist combinations we analyzed in vivo (Fig. 3A). However, both linked and unlinked TLR4_7_9a skewed the IgG1/IgG2c ratio the furthest toward IgG2c (60 and 64% IgG2c, respectively), suggesting this TLR triagonist preferentially stimulates Th1 responses to immunization (Fig. 3B). This was the only TLR triagonist of the five in our library that exhibited this property.

To directly analyze Ag-specific T cell responses to immunization with these TLR triagonist adjuvanted vaccines, we evaluated IL-4 and IFN-y production by splenocytes restimulated with CBU_1910 using ELISpot. IL-4 is a classical Th2 cytokine that stimulates humoral responses (34, 43). IFN-y is produced by inflammatory T cells in response to stimulation with their cognate Ag (43, 44) and is indicative of Th1 CD4 T cell and CD8 T cell responses to immunization (43, 44). In line with the robust Ab responses observed in Fig. 3A, we found that T cells from cohorts immunized with each of the TLR triagonist combinations produced IL-4 in response to restimulation with CBU_1910 (Fig. 3C). However, only cohorts immunized with TLR4_7_9a generated IFN-y–producing T cells (Fig. 3C). In addition, the linked TLR4_7_9a elicited significantly more IFN-y–producing T cells (1.9-fold higher) compared with the unlinked combination (Fig. 3C). Similar to our initial TLR2/6_4_7a study, Triagonists increased the number of B220⁺, CD4⁺, and CD8⁺ in the draining LN, with the exception of TLR1/2_4_7a, which was similar to PBS control (Supplemental Fig. 3).

Based on the in vivo immunogenicity analysis of our expanded TLR triagonist library, we focused on optimizing the TLR triagonist TLR4_7_9a because of its Th1-skewing activity (Fig. 3). We adjusted two of our TLR triagonist formulations to include AddaVax to determine whether this could overcome the Th2-skewing activity of our adjuvants. We specifically chose TLR2/6_4_7a and TLR4_7_9a because of their opposing T cell–skewing activity we observed in our previous experiments. AddaVax is an MF59-like squalene oil-in-water emulsion adjuvant with known Th1- and Th2-skewing properties (45, 46). We hypothesized that the inclusion of AddaVax in these two TLR triagonist adjuvanted vaccine formulations would further skew the resulting immune responses toward Th1 responses while maintaining the Ag-specific Ab responses.

Formulation of TLR2/6_4_7a with AddaVax increased the CBU_1910–specific IgG1 response by 38% to match that of AddaVax alone (Fig. 4A). However, the Ab response was further skewed toward IgG2c when TLR2/6_4_7a was included with AddaVax compared with formulations with only AddaVax and CBU_1910 (Fig. 4B). The inclusion of AddaVax in TLR4_7_9a adjuvanted vaccines also resulted in IgG1 production similar to CBU_1910 formulated with only AddaVax (Fig. 4C), much higher than CBU_1910 adjuvanted with only TLR4_7_9a (Fig. 3A). However, including linked TLR4_7_9a with AddaVax resulted in significantly higher production (92–114% increase) of IgG2c (Fig. 4C). As a result of the increased IgG2c, the ratio of IgG1/IgG2 became more balanced (57% IgG2c with linked TLR4_7_9a, Fig. 4D) compared with CBU_1910 formulated with only AddaVax.

As we observed that high, IgG1-biased Ab titers against CBU_1910 were insufficient for protection (Fig. 2H), we sought to include the other C. burnetii Ags that were immunogenic in Q-Vax. We generated vaccine formulations including multiple C. burnetii Ags, an approach that has achieved better vaccine protection in other disease models compared with individual Ags alone (47–50) (Fig. 2F–H). We added three additional C. burnetii Ags to our CBU_1910 vaccine formulations (CBU_0545, CBU_0630, and CBU_0370). These Ags were identified from our Q-Vax–immunized mice serum (Fig. 1) and from human patients that were either acutely or chronically infected with C. burnetii at the time their serum was collected (30–32). We considered Ags that generated robust humoral immune responses to either vaccination or infection as candidates to be included in our vaccine formulations. In formulations adjuvanted with TLR2/6_4_7a and AddaVax, IgG1 and IgG2c Ag-specific responses were detected against all four immunizing Ags (Fig. 4E). Importantly, the Ab responses detected in animals immunized with a single Ag were not significantly different when animals were immunized with all four Ags concurrently (Fig. 4E). This result suggests that the inclusion of multiple C. burnetii Ags in our vaccines should have an additive immune benefit and that a single Ag does not dominate or interfere with the immune response to the other Ags in the vaccine formulation. With the exception of CBU_0307, all of the Ag-specific Ab responses generate a balanced IgG1/IgG2 response (Fig. 4F). These optimization studies indicated that a polyvalent vaccine would result in a broader immune response, potentially leading to a more-efficacious vaccine formulation in challenge studies.

Our immunogenicity studies (Fig. 3) and previous in vitro studies (18) indicated that the linked forms of TLR4_7_9a and TLR1/2_4_9a would elicit the strongest Th1-biased immune responses to immunization while maintaining Ag-specific Ab responses. We combined each of these TLR triagonists with AddaVax and six C. burnetii Ags (CBU_1910, CBU_0545, CBU_0630, CBU_0370, CBU_0612, and CBU_0891) and evaluated the Ag-specific responses elicited in response to immunization to determine whether the responses were additive.

Similar to what we had observed in our initial (4-plex) multiantigen vaccine formulation (Fig. 4E), these 6-plex vaccines stimulated Ag-specific Ab responses to all six Ags (Fig. 5A), although CBU_0307 and CBU_0612 did not elicit robust Ab responses. We observed that the magnitude of the Ab response and the specific IgG subtype profile elicited in response to immunization were dependent on the TLR triagonist adjuvant in the formulation. The Ab response generated against CBU_0307 provided the clearest measure of the change mediated by our TLR triagonist adjuvants (Fig. 5A, purple column). With TLR4_7_9a, we observed a statistically significant Ag-specific IgG2c humoral response against CBU_0307, but this response is significantly blunted in the vaccine formulation adjuvanted with TLR1/2_4_9a. When these Ags were formulated with TLR4_7_9a, the Ab response was skewed more strongly toward an IgG2c response (94%), compared with TLR1/2_4_9a (78%) (Fig. 5A, 5B). The humoral response to CBU_1910 and CBU_0891 was not influenced by our TLR triagonist adjuvant system (Fig. 5A, 5B). In addition, TLR1/2_4_9a led to a significant expansion of B220⁺ B cells (7.4-fold increase) and CD4⁺ (4.0-fold increase) and CD8⁺ (4.4-fold increase) T cells in the draining LNs in response to immunization (Fig. 5C). TLR4_7_9a only showed significant increases in B220⁺ B cells (5.7-fold increase). The increases in CD4⁺ (2.7-fold increase) and CD8⁺ (3.0-fold increase) were not significant; however, this was likely due to relatively high error. In summary, these two vaccine candidates elicit robust, Th1-biased immune responses toward several Ags and were identified as our lead candidates. We hypothesized that immunizing animals with these vaccine candidates would result in a more-efficacious immune response in a live C. burnetii aerosol challenge model than single-Ag vaccines. We adjusted the Ag mixture, however, as CBU_0612 was nonimmunogenic and CBU_0630 could not be produced on sufficient scale for all further experiments. We substituted CBU_1398 as it addressed both issues (data not shown).

Following the optimization of our two lead vaccine candidates (TLR 1/2_4_9a and TLR 4_7_9a) (Fig. 5), we tested their protective efficacy in a live C. burnetii aerosol challenge model. We followed the same immunization protocol as our initial challenge model (Fig. 2E) and immunogenicity studies (Fig. 2A). For this experiment, we split the cohort of 10 animals into two groups and terminated half the cohort on day 9 postchallenge, when the peak of infection should occur, in an attempt to evaluate changes in bacterial load due to immunization using PCR (Fig. 6A). The other half of each cohort was terminated 28 d postchallenge and similarly evaluated for changes in bacterial load (Fig. 6A). Following the challenge on day 42, animals from each cohort were monitored daily for changes in weight and body temperature. Similar to our previous challenge study (Fig. 2E–H), a Q-Vax–immunized cohort was the positive control for protective efficacy. On day 9 postchallenge (day 51 of the experiment), only the Q-Vax–immunized mice were fully protected from weight loss in response to the live C. burnetii challenge (Fig. 6B). However, the two triagonist-vaccinated groups appeared to gain weight back faster than the PBS group (Supplemental Fig. 4A). In addition, animals immunized with the five-Ag mix and TLR4_7_9a adjuvanted formulations exhibited significantly lower fever compared with the PBS immunization cohort (Fig. 6C).

PCR analysis revealed promising signs of protection with the TLR triagonist adjuvanted vaccine candidates. The day 9 bacterial load in spleens and lungs in animals that received our vaccine candidates was significantly reduced. However, TLR4_7_9a adjuvanted formulation conferred better protection (spleen: 8.3-fold less and lung: 4.3-fold less than PBS) as evidenced by the significant reduction in bacterial genomes detected compared with TLR1/2_4_9a adjuvanted formulation (spleen: 4.7-fold less and lung: 3.5-fold less than PBS) (Fig. 6D, 6E, blue diamonds versus red inverted triangles). Although Q-Vax results in more profound decreases in bacterial burden compared with PBS (spleen: 93.2-fold less and lung: 7.2-fold less), importantly, we could not detect significant differences between the bacterial burden in our TLR triagonist adjuvanted formulations compared with Q-Vax (Fig. 6D, 6E, blue diamonds and red inverted triangles versus green triangles). By day 28, none of the vaccinated group showed statistically different levels of C. burnetii by PCR of organs (Supplemental Fig. 4D), suggesting that this vaccine formulation has efficacy similar to Q-Vax in its ability to reduce bacterial load in tissues.

The formulations containing TLR triagonist generated robust Ag-specific Ab responses to each Ag in the vaccine (Fig. 6F) and were either IgG1/IgG2c balanced or IgG2c biased (Fig. 6G). In line with our hypothesis that IgG2c responses are more protective in C. burnetii challenge, we observed that Q-Vax immunization generated a strongly IgG2c-biased (95% IgG2c) Ab response to CBU_1910 and almost no IgG1 (Fig. 6F). The Q-Vax–vaccinated group showed statistically lower spleen to total body weight percentage on day 9 and 28 following challenge; for our experimental groups, only the TLR 1/2_4_9a group showed a reduction in spleen to body weight percentage on day 28 following challenge (Supplemental Fig. 4E). Taken together, these data strongly support our findings that TLR4_7_9a and TLR1/2_4_9a adjuvanted formulations are capable of generating IgG2c-skewing immune responses following immunization and indicate that these responses confer partial protection against a live C. burnetii challenge.

The current inactivated whole-cell vaccine for Q fever, Q-Vax, although protective, causes reactogenicity in previously exposed individuals, limiting its practical application. This is a concern given C. burnetii’s low infectious dose, durability in the environment, and its potential use as a biological weapon (20, 25). It is necessary to develop an approved prophylactic vaccine against this bacterium. Because of reactogenicity limitations of the whole-cell vaccine, we took the approach of formulating recombinant protein subunit vaccine candidates, composed of our (to our knowledge) novel TLR triagonists and immunogenic C. burnetii Ags, to generate vaccines that are as protective as and do not have the safety issues shown with the whole-cell vaccine.

Our initial C. burnetii vaccine candidate was composed of the Ag CBU_1910, identified as an ideal target through proteome microarray analysis of Q-Vax–vaccinated mice (Fig. 1A) and in the literature (51–53), formulated with either linked or unlinked TLR2/6_4_7a. The cause of the reactogenic response generated by Q-Vax is unknown, but we hypothesized that a recombinant protein-based subunit vaccine would be unlikely to produce a reactogenic response as it lacks the potentially reactogenic components found in Q-Vax, specifically LPS (54). Although our subunit vaccine candidate did not generate reactogenic responses, the initial candidate did not induce a protective response compared with Q-Vax. We showed that this was due to the Th2-biased response generated toward CBU_1910, whereas the Q-Vax–induced humoral response was heavily Th1 biased, which is a critical component in Q-Vax–induced protection (21, 35, 36). In other efforts to develop protective C. burnetii vaccines, adding multiple MHC class I epitopes from various C. burnetii Ags improved protection in other attempts to develop Q fever vaccines (55–57), consistent with previously published studies suggesting that a cell-mediated response is important for protection against C. burnetii (35) (58). Finally, emulsions are commonly added to subunit vaccine formulations (45, 46) but were not included in our initial vaccine candidate. Thus, we investigated optimizing our vaccine formulation by evaluating the adjuvant, including an emulsion (AddaVax), and increasing the number of Ags included in the subunit vaccine.

In evaluating our adjuvants, we found that a more Th1-biased response could be achieved using different TLR triagonist combinations (Fig. 2A). TLR triagonists containing TLR2/6a showed to be more Th2 biasing (Figs. 1, 3). This result mirrored our in vitro studies (18) and previous studies showing that TLR2 agonists are potent Th2-inducing adjuvants. When TLR2/6a was replaced with TLR1/2a, we observed a slight shift toward a more Th1-biased response, perhaps due to the lower potency of TLR1/2a compared with TLR2/6a, despite similar signaling pathways (Fig. 3). When TLR7a was exchanged for TLR9a, we observed a further shift in the immune response toward Th1 biasing (Fig. 3). We speculate that this is due to the higher potency of TLR9a when linked to the core than the TLR7a, which results in a stronger ability to overcome the Th2 biasing of TLR1/2a and TLR2/6a. Although these TLR9a-containing triagonists were more Th1 biased, they still generated high Ab titers when combined with a TLR2 agonist. When TLR2/6a was exchanged for TLR7a (TLR4_7_9a), we observed the most Th1-biased immune response of the triagonist panel. However, the Ab response was significantly diminished. We speculate that the TLR2/6 agonist was essential for generating potent Ab responses from our compounds. Overall, we found that we could generate unique immune response profiles by changing the combination of TLR triagonist coadministered as the adjuvant. We envision this analysis will be impactful in vaccine development as particular TLR triagonist combinations, and their resulting immune response profiles, may be well suited for specific vaccine targets. For our Q fever vaccine, we selected TLR1/2_4_9a and TLR4_7_9a as adjuvants in our C. vburnetii vaccine candidates because they were the two most Th1-biasing combinations of the TLR triagonists evaluated in our studies.

Subunit vaccines often require an agent that increases Ag availability for presentation by APC (8–10) to improve vaccine efficacy. AddaVax is a commercially available vaccine adjuvant, similar to a squalene emulsion that has been licensed for use in influenza vaccines in Europe (59–61). AddaVax has been shown to recruit and activate APC (46, 62). Squalene oil-in-water emulsions, including AddaVax, elicit both B cell and T cell responses to immunization (46). When AddaVax was added to our initial formulation (CBU_1910 and TLR2/6_4_7a), we observed a more Th1-biased response compared with either AddaVax alone or TLR2/6_4_7a alone. This result suggests there is some synergistic Th1-stimulating activity when the emulsion and TLR triagonist are combined. The same was also true when our most Th1-biasing triagonist, TLR4_7_9a, was formulated with AddaVax. In this formulation, however, AddaVax significantly improved the quantity of total Ag-specific IgG generated compared with the triagonist alone, resulting in a Th1-biased response, without compromising Ab generation.

In addition to optimizing the adjuvant and emulsion, we included additional Ags to our formulation to increase the breadth of the immune response to immunization. We were concerned that the addition of multiple Ags to the formulation would increase the likelihood of our candidates to induce a reactogenic response, similar to Q-Vax. However, other attempts at generating subunit vaccines for Q fever have shown that including multiple T cell epitopes of C. burnetii Ags improved the efficacy of the vaccine (35, 58). Thus, we felt it was ideal to include additional C. burnetii Ags in the formulation. When adjuvanted with TLR2/6_4_7a and AddaVax, Ab responses were generated to four Ags in the same vaccine, and the responses were additive. This differs from Q-Vax, in which CBU_1910 appears to be the immunodominant Ag. Thus, multiple Ags could be incorporated into the vaccine with our adjuvant system, without compromising the individual Ag-specific immune responses.

In our second challenge study, we developed two Th1-skewing candidate vaccines based on the optimization experiment results: 1) TLR4_7_9a and TLR1/2_4_9a were more Th1-biasing adjuvants, 2) combining an oil-in-water emulsion (AddaVax) with our TLR triagonists generates a more Th1-biased immune response and significantly increases Ab titers, and 3) multiple Ags can be included in the formulation to expand the immune response breadth without generating any observed reactogenic responses. The two formulations with these design principles were shown to elicit robust, Th1-biased Ab responses to four of the six Ags in the formulation. This result confirmed that we could use our formulation design principles to generate a vaccine with the desired type of immune response to selected Ags. When these formulations were tested in the challenge study, both were observed to reduce the bacterial burden in mice following a live C. burnetii aerosol challenge. In addition, compared with Q-Vax, the vaccine candidate adjuvanted with TLR4_7_9a resulted in a lower fever and no weight loss in mice challenged with C. burnetii. Furthermore, this formulation resulted in significantly less C. burnetii in the spleen and lungs of vaccinated and challenged mice compared with sham vaccinated mice, although it was unable to reduce the organ weight to total body weight percentage in any of the tissues analyzed day 9 and 28 postchallenge. Although the subunit vaccines were not as protective as Q-Vax (Fig. 6), it is likely that with further tuning of the vaccine components, similar levels of protection could be achieved. For example, the Q-Vax sample resulted in near-complete IgG2c biasing of the Ab response, significantly more so than the subunit vaccine groups, leaving room for further refinement of the Th1-biased response. Conversely, the response generated in this experiment could account for partial protection, whereas Q-Vax elicits an additional, uncharacterized response that provides further protection, which is not present in our candidate vaccines. Thus, further evaluation of the protective nature of Q-Vax could guide further refinement of subunit vaccine alternatives.

In conclusion, we developed two promising Q fever subunit vaccine candidates. The vaccines were developed through rational design using a (to our knowledge) novel TLR adjuvant platform that had not been previously evaluated for protective efficacy in an in vivo challenge model. By tuning the formulations’ adjuvant, emulsion, and Ag compositions, we generated an immune response that is specified for protection against C. burnetii challenge. Further efficacy experiments in more-relevant animal models (nonhuman primates) and evaluations of reactogenicity in guinea pig models are underway and will further elucidate the potential use of these adjuvant systems in FDA-approved subunit vaccine formulations. Compared to traditional empirical approaches that use a single adjuvant system, we envision this adjuvant platform and rational vaccine design approaches could be used to generate subunit vaccine candidates against other human pathogens, resulting in more-effective and safer vaccines specifically tailored to each human pathogen.

We acknowledge help from Laser Spectroscopy Labs at University of California, Irvine, for light scattering experiments. We thank John Stenos (Barwon Health, Geelong, VIC, Australia) and Stephen Graves (University Hospital Geelong, Geelong, VIC, Australia) for gifts of Q-Vax. The views expressed in this article are those of the authors and do not reflect the official policy or position of the U.S. Department of Defense or the U.S. Army.

The authors have no financial conflicts of interest.