Optimization of Prime-Boost Vaccination Strategies Against Mouse-Adapted Ebolavirus in a Short-Term Protection Study

Abstract

In nonhuman primates, complete protection against an Ebola virus (EBOV) challenge has previously been achieved after a single injection with several vaccine platforms. However, long-term protection against EBOV after a single immunization has not been demonstrated to this date. Interestingly, prime-boost regimens have demonstrated longer protection against EBOV challenge, compared with single immunizations. Since prime-boost regimens have the potential to achieve long-term protection, determining optimal vector combinations is crucial. However, testing prime-boost efficiency in long-term protection studies is time consuming and resource demanding. Here, we investigated the optimal prime-boost combination, using DNA, porcine-derived adeno-associated virus serotype 6 (AAV-po6), and human adenovirus serotype 5 (Ad5) vector, in a short-term protection study in the mouse model of EBOV infection. In addition, we also investigated which immune parameters were indicative of a strong boost. Each vaccine platform was titrated in mice to identify which dose (single immunization) induced approximately 20% protection after challenge with a mouse-adapted EBOV. These doses were then used to determine the protection efficacy of various prime-boost combinations, using the same mouse model. In addition, humoral and cellular immune responses against EBOV glycoprotein were analyzed by an enzyme-linked immunosorbent assay, a neutralizing antibody assay, and an interferon γ–specific enzyme-linked immunospot assay. When DNA was used as a prime, Ad5 boost induced the best protection, which correlated with a higher cellular response. In contrast, when AAV-po6 or Ad5 were injected first, better protection was achieved after DNA boost, and this correlated with a higher total glycoprotein-specific immunoglobulin G titer. Prime-boost regimens using independent vaccine platforms may provide a useful strategy to induce long-term immune protection against filoviruses.

Ebolavirus (EBOV) causes the most severe form of hemorrhagic fever known to date. Current management of EBOV-infected individuals mainly consists of supportive care, as no approved treatments or vaccines against EBOV are commercially available [1, 2]. The lack of approved therapeutics, high mortality rate, and ease of transmission have prompted research on the development of both treatments and vaccines against EBOV.

Protection against EBOV challenge has been demonstrated in nonhuman primates (NHPs) after immunization with various vaccine platforms based on adenovirus (Ad) [3–5], vesicular stomatitis virus (VSV) [6], virus-like particles [7], human parainfluenza virus type 3 (HPIV3) [8], Newcastle disease virus [9], and replication-competent rabies virus vector [10] encoding EBOV glycoprotein (GP). The efficacy of these platforms was mainly demonstrated 28 days after immunization. However, lack of long-term protection by these vaccine platforms would limit their clinical use. Although protection against Marburg virus challenge has been reported in NHPs approximately 14 months after a single shot of VSV [11], to date, long-term protective immunity against EBOV has only been reported after a single dose of replication-competent rabies virus vector or after prime-boost regimens. After DNA/Ad5 or ChAd3/MVA prime-boost, NHPs were fully protected against EBOV challenge 3 or 8 months, respectively, after the last injection [12, 13]. Although single immunization regimens could probably protect NHPs against EBOV challenge for >28 days after vaccination, the studies cited above highlight the potential for prime-boost regimens to induce long-term protective immunity against EBOV. Another advantage of prime-boost regimens over single-injection vaccines is their ability to overcome preexisting immunity against viral vectors. The detrimental effects of preexisting vector immunity against the immunogenicity and efficacy of Ad-based vectors have been extensively reported both in animal models [14, 15] and clinical trials [16, 17]. Although less documented, preexisting immunity is also expected to reduce the efficacy of other virus-derived vaccine platforms, such as HPIV3 [9]. To circumvent preexisting vector immunity, various strategies, such as developing vectors based on low-seroprevalence viruses [14, 18], varying the immunization route [19, 20], and, notably, prime-boost regimens, have been developed. Regarding the latter, both in rodents [21] and in a clinical trial [22, 23] DNA prime followed by Ad or modified vaccinia virus Ankara (MVA) boost were not significantly affected by preexisting Ad or MVA immunity. Although in the studies described above, DNA was used as a prime, prime-boost regimens with other platforms may also have the potential to overcome preexisting immunity against viral vectors.

Because of the potential of prime-boost regimens to induce durable protective responses against EBOV and their ability to overcome the host's preexisting immunity against viral vectors, determining the optimal vector combination for prime-boost regimens against EBOV is of great interest. However, this task is complicated by several factors. First, there are an increasing number of platforms under development for immunization against EBOV. Second, challenge studies are required to determine vaccine efficacy against EBOV infection. Although high humoral responses are associated with survival against EBOV challenge in Ad5-GP–vaccinated [24, 25] or VSV-GP–immunized NHPs, antibody (Ab) titers are not sufficient to predict survival in all animals. Furthermore, in some vaccine platforms, strong humoral responses were not predictive of survival against EBOV [10], indicating that, depending on the vaccine platform, protection might correlate with different immune parameters. Supporting this notion, depletion studies identified CD8⁺ T cells or antibodies as crucial for protection in NHPs immunized with Ad5 [26] or VSV [27], respectively. Finally, as short-term protection against EBOV challenge was demonstrated with the platforms of interest, the efficacy of prime-boost regimens is usually evaluated in long-term challenge studies.

As a proof of concept, we compared the protective efficacy of prime-boost regimens in a short-term protection study, using the mouse model of EBOV infection. For this study, we tested 2 well-characterized vaccine platforms, Ad5 and DNA, which were previously demonstrated to protect mice against mouse-adapted EBOV (MA-EBOV), as well as a newly developed vector, porcine serotype 6 adeno-associated virus (AAV-po6), a vector of swine origin. Mice were immunized and boosted 28 days apart before challenge on day 56. Furthermore, the protective efficacy of each vector boost was correlated with either cellular or humoral responses against EBOV GP.

MATERIALS AND METHODS

Mouse, Vectors, and Viruses

B10.Br mice (MHC-2K) were purchased from The Jackson Laboratory (Bar Harbor, Maine). DNA and Ad5, both expressing a human codon optimized EBOV GP from the Mayinga strain, were previously described [28, 29]. AAV-po6 was isolated as previously reported [30], and the codon-optimized GP sequence was cloned into the AAV-po6 as previously described [31].

MA-EBOV [32] and EBOV–enhanced green fluorescent protein (EGFP) [33], which were previously described, were also derived from the Mayinga strain of EBOV.

Immunization and Challenge

The viral platforms were administered intramuscularly into the left posterior hind limb. In contrast, DNA was delivered by a shallow intramuscular injection into the quadriceps muscle, followed by electroporation, as previously described [34].

For vector titration, mice were immunized with varying doses of each individual platform, as follows: 0.1, 1, and 10 µg of DNA; 1 × 10⁸, 1 × 10⁹, and 1 × 10¹⁰ gene copies of AAV-po6; and 1 × 10³ and 5 × 10³ plaque-forming units (PFU) of Ad5. On day 28, vaccinated mice were transferred into the biosafety level 4 laboratory before intraperitoneal challenge with 10 focus forming units (FFU) of MA-EBOV 1000 TCID50.

For prime-boost immunization, mice were vaccinated with the various platforms and boosted 28 days later with the denoted vectors. On day 56 (28 days after the last injection), vaccinated mice were transferred into our biosafety level 4 laboratory and challenged intraperitoneally with 10 FFU of MA-EBOV.

Following MA-EBOV challenge, the animals were monitored daily for disease progression for 12–16 days. All procedures were executed according the guidelines outlined by the Institutional Animal Care Committee at the National Microbiology Laboratory of the Public Health Agency of Canada according to the guidelines of the Canadian Council on Animal Care. All infectious work was performed in the biosafety level 4 facility at the National Microbiology Laboratory of the Public Health Agency of Canada.

Immunoglobulin G (IgG)–Specific Enzyme-Linked Immunosorbent Assay (ELISA)

To determine Ab titers, 96-well plates were coated overnight with 30 ng of purified EBOV GP lacking a transmembrane domain (from EBOV Zaire), which was purchased from IBT bioservice (Gaithesburg, Maryland). After extensive washes, plates were incubated with the different dilution of tested sera. Samples were washed before incubation with 0.33 µg of goat anti-mouse IgG-conjugated horseradish peroxidase Ab (Cedarlane, Ontario, Canada). After washing, ABST (2,2′-azino-bis[3-ethylbenzothiazoline-6-sulphonic acid]) and peroxidase substrate (Cedarlane) were added to visualize Ab binding. Absorbance was read at 405 nm. Samples with a background-subtracted OD that was >3 times above background were considered positive.

Neutralizing Ab Assay

The EBOV neutralization assay was performed as previously described by Richardson et al [29]. Briefly, sera collected from immunized mice were inactivated by heating before undergoing serial dilution. Diluted sera were incubated with approximately 100 infectious units (based on EGFP expression) of EBOV expressing the EGFP reporter gene (EBOV-EGFP) at 37°C for 90 minutes. The mixture was then incubated with Vero E6 cells for 1 hour at room temperature. After addition of Dulbecco's modified Eagle's medium supplemented with 20% fetal bovine serum, samples were incubated for an additional 48 hours. Cells incubated with EBOV-EGFP alone were used as positive controls. GFP-positive cells were counted in each well, and sample dilutions showing >50% reduction in GFP-positive cells, compared with controls, were scored as positive for neutralizing Abs. All infectious work was performed in the biosafety level 4 facility at the National Microbiology Laboratory of the Public Health Agency of Canada.

Enzyme-Linked Immunospot (ELISPOT) Assay

ELISPOT assays were conducted according to the manufacturer's instructions (BD Bioscience, San Jose, California). Briefly, 96-well ELISPOT plates (Millipore, Billerica, Massachusetts) were coated overnight with anti-mouse interferon γ (IFN-γ) Ab (R&D Systems, Minneapolis, Minnesota), washed with phosphate-buffered saline, and blocked with 10% fetal bovine serum in Roswell Park Memorial Institute medium. On day 38, splenocytes were harvested from 3 mice and pooled together to assess T-cell responses. A total of 5 × 10⁵ splenocytes were plated per well and stimulated for 18–24 hours with 2.5 µg/mL of 36 overlapping peptide pools spanning the entire gene encoding EBOV GP. Roswell Park Memorial Institute medium alone and staphylococcal enterotoxin B were used as negative and positive controls, respectively. The following day, samples were extensively washed before incubation with biotinylated anti-mouse IFN-γ Ab (R&D Systems). After incubation with streptavidin–alkaline phosphatase (MabTech, Cincinnati, Ohio), IFN-γ–secreting cells were detected using streptavidin–alkaline phosphatase (MabTech), followed by AEC Chromagen (BD biosciences). Finally, spots were counted with an automated ELISPOT reader (Cellular Technology, Shaker Heights, Ohio).

Statistical Analysis

Survival data was analyzed using the log rank test and the Gehan–Wilcoxon test, while IgG-specific ELISA and neutralizing Ab titer assay data were analyzed by 1-way analysis of variance followed by a Tukey test. All statistical analysis was performed using GraphPad software, version 5.03.

RESULTS

Vector Titrations

Our goal was to test various prime-boost combination involving Ad5, DNA, and AAV-po6, with all encoding a consensus codon-optimized EBOV GP (hereafter denoted Ad5-GP, DNA-GP, and AAV-po6–GP, respectively), to determine which vector combination would generate the best protective immunity against EBOV challenge. First, to test the various prime-boost combinations in a short-term challenge study using the mouse model of EBOV infection, each vector platform was titrated. To do so, B10.Br mice were immunized with varying doses of each individual platform. Of note, a single injection with 10⁴ PFU of Ad5 encoding the codon-optimized EBOV GP was previously demonstrated to induce full protection in immunized mice [29]. In contrast, protection in mice against MA-EBOV has only been reported after 2–4 DNA injections with doses varying from 1.5 to 5 µg [35, 36]. As a result, titration was started at 5 × 10³ PFU and 10 µg for Ad5-GP and DNA-GP, respectively.

Twenty-eight days after immunization, mice were challenged with 10 FFU of MA-EBOV; mock-treated, challenged mice were used as negative control. Infected mice were scored for signs of disease (data not shown) and monitored for weight loss and survival (Figure 1). In mice immunized with DNA-GP, 60%, 20%, and 0% survival was observed in mice receiving 10, 1, and 0.1 µg, respectively (Figure 1A). Similarly, mice vaccinated with 10¹⁰, 10⁹, or 10⁸ gene copies of AAV-po6–GP displayed 100%, 20%, and 0% protection, respectively (Figure 1B); this is the first report of full protection against EBOV challenge that used an AAV vector. Finally, 87.5% and 12.5% protection were observed in mice immunized with 5 × 10³ or 1 × 10³ PFU Ad5-GP (Figure 1C). All surviving mice showed no signs of the disease and minimal weight loss, while all control mice succumbed to the disease around days 6–8 after infection (Figure 1).

For experiments involving different prime-boost combinations, vector doses conferring approximately 20% protection were selected. Using these doses, any increase in protection efficacy after boost immunization could be detected without having to significantly increase the period between the boost and the EBOV challenge. Indeed, using the protective dose of each platform for prime-boost regimens, animals would have to be challenged several months after the boost to see differences in protection between the prime-boost regimens. Therefore, 1 µg of DNA-GP, 10⁹ gene copies of AAV-po6–GP, and 10³ PFU of Ad5-GP, which conferred 20%, 20%, and 12.5% protection, respectively, were selected for prime-boost immunization. In these experiments, only prime-boost regimens with identical prime or boost could be compared. Therefore, the results were grouped on the basis of the vector used for priming.

Prime-Boost Regimens After a DNA Prime

Groups of 8 B10.Br mice were vaccinated with 1 µg of DNA and then boosted 4 weeks later with DNA-GP, Ad5-GP, or AAV-po6–GP. Immunized mice were then challenged with MA-EBOV 28 days after the boost. Weight loss (data not shown) and survival (Figure 2A) were monitored using challenged naive mice as controls. Of the 3 tested platforms, Ad5-GP boost induced the most protective immunity in mice primed with DNA-GP, as evidenced by survival rates. Indeed, 87.5%, 70%, and 25% protection was observed in DNA-GP–primed mice boosted with Ad5-GP, DNA-GP, and AAV-po6–GP, respectively (Figure 2A).

Immune responses after prime-boost were assessed to determine whether survival correlated with humoral or cellular immunity. Sera were collected before and after the boost from mice immunized with the prime-boost regimens described above, and the EBOV-specific neutralizing Ab titer and GP-specific IgG titer were determined (Figure 2B and 2C). These parameters did not correlate with the boosting efficacy of the various platforms after a DNA prime. Indeed, mice immunized with DNA-GP/Ad5-GP and those immunized with DNA-GP/AAV-po6–GP had similar neutralization titers, which were higher than those in mice vaccinated with DNA-GP/DNA-GP. However, in all 3 groups, neutralizing titers were relatively low (Figure 2B). In contrast, GP-specific IgG titers were relatively high in all 3 groups. However, no significant difference in IgG titers was observed in mice before and after the boost, independently of the platform used for boosting (Figure 2A and 2C). In addition to humoral responses, cellular responses against EBOV GP were also assessed by IFN-γ–specific ELISPOT, using splenocytes obtained from immunized mice 26 days after the prime or 10 days after the boost. Interestingly, the magnitude of the IFN-γ–specific ELISPOT responses correlated with the boosting ability of the different vectors, as higher survival rates and cellular responses were observed in mice immunized with DNA-GP/Ad5-GP, DNA-GP/DNA-GP, and DNA-GP/AAV-po6–GP (Figure 2D). Taken together, the above experiments indicate that, of the DNA-GP, AAV-po6–GP, and Ad5-GP vectors, boosting with the latter platform generated the best immune response for protection against MA-EBOV challenge and that this superior boosting correlated with a stronger cellular response.

Prime-Boost Regimens After an AAV-po6–GP Prime

Similar experiments were performed in mice primed with AAV-po6–GP followed by boosting with DNA-GP, AAV-po6–GP, or Ad5-GP 4 weeks later. In mice primed with AAV-po6–GP, the highest survival rate was observed in the ones boosted with DNA-GP, followed by those boosted with AAV-po6–GP and those boosted with Ad5-GP (Figure 3A). Next, the protective efficacy of the various boost platforms was compared to the generated humoral and cellular responses. Neutralizing Ab titers were low in all 3 groups, with titers averaging a dilution of 1:25 in mice primed with AAV-po6–GP and boosted with AAV-po6–GP or Ad5-GP (Figure 3B). Unlike neutralizing Ab titers, GP-specific IgG titers correlated with the boosting ability of the different vectors. Indeed, higher IgG titers and survival rates were obtained in mice primed with AAV-po6–GP and then boosted with DNA-GP, AAV-po6–GP, and Ad5-GP (Figure 3C). Finally, in mice primed with AAV-po6–GP, no correlation was observed between protective efficacy and the IFN-γ–specific ELISPOT response. Indeed, mice immunized with AAV-po6–GP/Ad5-GP were not protected against EBOV challenge but had the highest cellular response (Figure 3D). Taken together, the experiments described above suggest that DNA was a superior boost in mice primed with AAV-po6 and that this superior protective immunity correlated with higher GP-specific IgG responses. In contrast, although it induced a strong cellular response, Ad5 boost did not complement AAV-po6 prime, as illustrated by the lack of survivors after MA-EBOV challenge and by weak humoral responses.

Prime-Boost Regimens After an Ad5-GP Prime

Finally, mice were primed with Ad5-GP and then boosted with DNA-GP, AAV-po6–GP, or Ad5-GP before MA-EBOV challenge. On the basis of survival data after MA-EBOV challenge, DNA-GP provided the best boost for the immune response generated after Ad5-GP immunization (Figure 4A). To correlate protection with immune parameters, neutralizing and total IgG Ab titers, in addition to GP-specific T-cell responses, were measured in mice vaccinated with the prime-boost regimens asserted above. When Ad5-GP was used as a prime, a clear difference in neutralization titers (Figure 4B) or the number of IFN-γ–secreting cells (Figure 4D) was not detectable between the Ad5-GP/DNA-GP or Ad5-GP/AAV-po6–GP groups. However, these groups demonstrated 100% and 75% protection, respectively (Figure 4A). Although both parameters were lower in mice immunized with Ad5-GP/Ad5-GP, which only demonstrated 37.5% survival, they could not reliably predict the boosting ability of the vector tested in Ad5-GP–primed mice. In contrast, on the basis of GP-specific total IgG titers, a clear hierarchy was observed between the different groups, with a higher GP-specific total IgG titer correlating with enhanced protective efficacy against MA-EBOV challenge (Figure 4C). Ad5-GP/DNA-GP was the best prime-boost regimen in mice that received an Ad5-GP prime, and boosting efficacy correlated with GP-specific IgG titers.

DISCUSSION

In this study, we used Ad5, AAV-po6, and DNA encoding GP to determine which vectors better complement each other, to generate the most-efficient immune responses against MA-EBOV challenge. To evaluate the protective efficacy of the prime-boost combination in a short-term protection study that used the mouse model of EBOV infection, each individual platform was first titrated to obtain approximately 20% protection. This suboptimal dose of each vector was then used for evaluating prime-boost combinations. Although titration is required to test prime-boost regimens in our short-term challenge study, it restricts the comparisons that can be made between these vaccine regimens. By use of this approach, the boosting or priming ability of the different vectors can only be compared in animals that receive a common prime or boost. Alternatively, if the optimal dose (100% protection) of each platform is used, the efficacy of each prime-boost combination would have to be determined after an extended period (6–12 months) following vaccination.

In DNA-GP–primed mice, better protection was observed after Ad5-GP boost. The superior boosting ability of Ad5 in these mice correlated with increased GP-specific cellular responses but not with GP-specific humoral responses. Similarly, in mice first immunized with Ad5-GP, superior protection was obtained after DNA-GP boost, illustrating the high degree of complementarity between these 2 platforms. Interestingly, in Ad5-GP–primed mice, superior boosting was associated with stronger GP-specific Ab responses but did not correlate with GP-specific cellular responses. Finally, for animals primed with AAV-po6–GP, DNA-GP boost provided the best protection efficacy. The superior DNA-GP boosting ability correlated with higher mean GP-specific titers than those observed in animals boosted by means of other platforms.

Although this work was restricted to 3 vaccine platforms as a proof of concept for our short-term challenge method, valuable lessons can be learned. We observed that, depending on the vector used as a prime, boosting efficacy correlates with either cellular immunity or total IgG titer against EBOV GP, indicating that the immune parameter that correlates with protection may be dependent on the platform used for priming. Of note, neutralizing Ab titers were relatively low and did not correlate with boosting ability in any of the tested vaccine platforms. Similarly, an earlier study demonstrated a lack of correlation between neutralizing Ab titers and survival in Ad5-and VSV-immunized NHPs [25]. In contrast, total GP-specific IgG titers were predictive of a strong boost in mice primed with either AAV-po6–GP or Ad5-GP. Similarly, strong humoral responses against GP in NHPs immunized with Ad5-GP or VSV-GP [24, 25, 27] has been previously correlated with survival after EBOV challenge. In mice primed with DNA-GP, a strong boost was associated with higher IFN-γ–specific ELISPOT responses. To date, a clear correlation between cellular responses and survival after EBOV challenge has not yet been reported. For example, clear differences in cellular response were not observed in NHPs immunized with a fully protective dose of Ad5-GP and in NHPs immunized with partially protective dose of Ad35-GP or Ad26-GP [37]. The lack of correlation between cellular response and protection against EBOV challenge may be due to the high variability of assays used to measure cellular responses, such as intracellular cytokine staining assays [38]. It would be interesting to determine whether ELISPOT assays are more reproducible than intracellular cytokine staining. However, one major drawback of ELISPOT assays performed on total splenocytes or peripheral blood mononuclear cells is the inability to differentiate between CD4⁺ T-cell, CD8⁺ T-cell, dendritic cell, or macrophage responses.

Our work also illustrated the importance of the order of the vectors used in prime-boost vaccine strategies. This was striking in prime-boost regimens involving AAV-po6–GP. Although 75% of the mice immunized with Ad5-GP/AAV-po6–GP or AAV-po6–GP/DNA-GP were protected against MA-EBOV challenge, only 0% and 25%, respectively, of mice immunized with the same vectors in the reverse order survived after MA-EBOV infection. The decrease in protection observed in mice primed with AAV-po6–GP and boosted with Ad5-GP indicated, that as a prime, AAV-po6 vector induced immune tolerance toward its transgene, as previously reported with AAV serotype 2 and 8 [39, 40]. Therefore, subsequent boosting of AAV-po6–primed animals may result in suboptimal cellular and humoral responses. In support of this hypothesis, Ad5 and, to a less extent, DNA boost of AAV-po6–primed mice resulted in lower humoral responses than a single immunization with either vector. Although AAV-po6, DNA, and Ad5 boost of AAV-po6–primed mice all resulted in higher cellular responses, a high proportion of these IFN-γ–producing cells may be regulatory T cells. Several studies have demonstrated that regulatory T cells are able to secrete IFN-γ upon inflammatory stimulation [41, 42].

Three vaccine platforms were tested in this pilot study, with the potential of expanding platform selection to those that previously demonstrated complete short-term protection in the NHP model of EBOV infection. Evaluation of prime-boost combinations in a short-term challenge study has several advantages. First, the efficacy of the numerous prime-boosts regimens can be compared at limited cost. The optimal prime-boost regimens could later be tested for sustained, durable protection. This approach reduces the number of vaccine regimens to be tested in long-term protection studies, which are both time consuming and resource demanding. Prime-boost combinations can be optimized in mice before further studies are performed in larger animal models, such as guinea pigs and primates.

Notes

Financial support. This work was supported by the Center for Research in Technology & Innovation (CRTI grant 453TD) and the Public Health Agency of Canada.

Potential conflicts of interest. All authors: No reported conflicts.

All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

References

Author notes