In Vivo Synthesis of Cyclic-di-GMP Using a Recombinant Adenovirus Preferentially Improves Adaptive Immune Responses against Extracellular Antigens

With a limited number of adjuvants approved for human administration, there is a pressing need for the development and testing of vaccine adjuvants that can improve the efficacy and maintain the safety profile of vaccines against resilient infectious diseases and cancers (1). The addition of adjuvants to vaccine formulations can serve to significantly improve vaccine efficacy when using less immunogenic Ags (2), to decrease vaccine toxicity by diminishing the need for higher vaccine dosages or reduce the need for repeated boosting (3).

Many significant cellular functions in bacteria, including regulation of motile/sessile phenotypes, virulence capabilities, and global gene expression are mediated by the second messenger bis-(3′–5′)-cyclic-dimeric-guanosine monophosphate (c-di-GMP) (4). C-di-GMP is generated by diguanylate cyclase (DGC) enzymes combining two GTP molecules (5). In the mammalian cytosol, the presence of c-di-GMP molecules can be detected by nucleotide sensors, including absent in melanoma 2 (AIM2) (6), the DEAD box-containing helicase (DDX41) (7), and stimulator of IFN genes (STING), each of which directly binds to c-di-GMP, resulting in the increased expression of type I IFNs and other innate immune responses (8, 9).

The direct administration of c-di-GMP has been shown to induce innate immune responses that can enhance protection of mice against challenges with Klebsiella pneumoniae (10), Staphylococcus aureus (11), methicillin-resistant S. aureus (12), Bordetella pertussis (13), Streptococcus pneumoniae (14), and avian influenza A/H5N1 (15, 16). Specifically, after intranasal challenge with B. pertussis in BALB/c mice, c-di-GMP induced production of cytokines such as IFN-γ, TNF-α, IL-6, and the chemokine MCP-1 in lung tissue (13). Recently, the ability of c-di-GMP to cause robust induction of IFN-β has been shown to attenuate experimental autoimmune encephalitis progression and onset through the induction of T regulatory cells, which suppress Th/effector T cell responses (17, 18).

The ability of c-di-GMP to trigger mammalian inflammatory responses has recently been harnessed for potential use as a promising vaccine adjuvant (19). Several studies suggest that inclusion of c-di-GMP in vaccine formulations can improve vaccine efficacy to provide immune protection against various bacterial infections (13, 20) and cancers (21–23). Local coadministration (intranasal and sublingual) of H5N1 virosomes and c-di-GMP to BALB/c mice resulted in strong H5N1-specific B cell and T cell adaptive immunity, but the i.m. route of vaccination resulted in significantly less protection (15). A liposome-based delivery system that improved c-di-GMP cell uptake in vivo resulted in IFN-β induction and enhanced tumor-specific cytotoxic T cell activity associated with regression of tumor growth in mice (21). However, this study suggests that pure extracellular c-di-GMP does not efficiently enter target cells.

V. cholerae encodes upward of 40 unique DGCs, many of which have been shown to synthesize c-di-GMP in this bacterium (24–26). These DGCs have highly divergent synthesis activities (27). Approximately half of these DGCs are thought to be integral inner membrane proteins, whereas the other half are cytoplasmic. Each contains a unique N-terminal sensory domain that is predicted to be regulated by environmental or host-derived cues (28). Tens of thousands of DGCs have been identified across bacterial genomes (29). Thus, these genes offer a wide range of unique enzymes possessing different properties that can be transduced by vectors to potentially modulate immune responses.

We have recently developed a novel strategy to synthesize high concentrations of c-di-GMP in vivo by transducing a Vibrio cholerae–encoded DGC, VCA0956, via a replication-deficient adenovirus vector in mammalian cells. Transduction of VCA0956 and subsequent production of c-di-GMP stimulated several innate immune responses, including significant induction of IFN-β and other cytokines, as well as the expression of several IFN-responsive genes. When coadministered with an adenovirus-based vector carrying the Clostridium difficile–derived toxin A (TA) Ag, the adenovirus serotype 5 (Ad5) vaccine expressing VCA0956 also enhanced, although moderately, the induction of adaptive immune responses against the TA Ag in mice (30).

In this study, we examined a second DGC encoded by the gene VCA0848 that synthesizes higher levels of c-di-GMP in V. cholerae (unpublished data). This DGC was recombined into the nonreplicative Ad5 vector to create the AdVCA0848 transducing particle. AdVCA0848 significantly enhances c-di-GMP production compared with our previous platform, AdVCA0956, by producing ∼400-fold more c-di-GMP in vivo. Delivery of AdVCA0848 improved the induction of innate immune responses including dramatic inductions of IFN-β. Importantly, coinjection of AdVCA0848 with OVA protein resulted in significant enhancement of OVA-specific T cell and B cell adaptive immune responses. However, we also found that inclusion of AdVCA0848 with a second adenovirus-based vaccine transducing either the HIV-1–derived Gag Ag (AdGag) or the C. difficile–derived toxin B (ToxB) Ag (AdToxB) resulted in the decreased induction of Gag or ToxB-specific adaptive immune responses relative to animals vaccinated with only AdGag or AdToxB. Therefore, in vivo–produced c-di-GMP can both enhance and inhibit adaptive immune responses depending on the nature of the coadministered Ag.

Adenovirus-based vectors used in this study were all replication deficient. An adenovirus vector that does not express a transgene (AdNull) and AdGag were constructed as previously described (31, 32). AdVCA0848 was constructed similarly to AdVCA0956 as previously described (30). In brief, the V. cholerae gene VCA0848 gene (GenBank sequence: CP007635.1) was subcloned into pShuttle-CMV as previously described (33). Primers used for AdVCA0848 construction were as follows: forward: 5′-ATAGGTACCCCACCATGAATGACAAAGTGCT-3′ and reverse: 5′- ATACTCGAGTTAGAAAAGTTCAACGTCATCAGAA-3′. The mutant version of AdVCA0848, AdVCA0848^mut, carrying the following amino acid changes, GGEEF > AAEEF in the GGDEF domain of VCA0848 allele, was mutated using the QuikChange Lightning site-directed mutagenesis kit (Agilent) with the primer 5′-GTCTTCTCAACTATTTCGCTTTGCTGCTGAAGAGTTCGTGATTATTTTTT-3′. AdToxB was constructed as previously described (34). In brief, a synthetic gene was designed based on the C. difficile ToxB sequence data from previous studies (35, 36) and ordered from GENEART (Regensburg, Germany). The synthetic gene representing the C-terminal portion of ToxB, including 617 aa (residues 1750–2366), was subcloned into pShuttle-CMV as previously described (33). Primers used for AdToxB construction were as follows: forward: 5′-GCTACTACGAGGACGGCCTG-3′ and reverse: 5′-CTCATCGATGATCAGCTTGCC-3′. The C-terminal region of the new synthetic gene did not contain the enzymatic domain, and recombination and viral propagation were carried out as previously described (30, 33, 37). Constructs were confirmed to be replication-competent adenovirus negative using replication-competent adenovirus PCR and direct sequencing methods as previously described (32, 38). All procedures with recombinant adenovirus constructs were performed under biosafety level 2 conditions.

The Michigan State University Institutional Animal Care and Use Committee approved the animal procedures conducted in this study. Care was provided to mice in this study in accordance with Public Health Service and American Association for the Accreditation of Laboratory Animal Care standards. Mice were purchased from Taconic Biosciences (Germantown, NY).

To determine the amount of c-di-GMP produced by the AdVCA0848 vector, male 6- to 8-wk-old BALB/c mice were i.v. injected (retro-orbitally) with AdNull (n = 3), AdVCA0956 (n = 4), or AdVCA0848 (n = 4) in 200 μl of a PBS solution (pH 7. 4) containing 2 × 10¹¹ viral particles (vps)/mouse, or they were not injected (naives) (n = 3) as previously described (30). The same viral dose was also used for additional experiments in which mice were injected with AdVCA0848 or AdVCA0848^mut, or were not injected (naives). At 24 h postinjection (hpi), mice were sacrificed and liver samples were collected, immediately snap frozen, and used later for c-di-GMP quantification as described later.

For innate immunity studies, 6- to 10-wk-old male C57BL/6 mice (n = 4) were i.v. injected (retro-orbitally) with AdNull or AdVCA0848 in 100 μl of a PBS solution (pH 7.4) containing 1 × 10¹⁰ vps/mouse, or they were not injected (naive). The same viral dose was also used for additional experiments in which mice were injected with AdVCA0848 or AdVCA0848^mut, or were not injected (naives). At 6 hpi, mice were sacrificed. Blood samples were collected and used for ELISA analysis, and splenocytes were harvested, counted, and used for immune cell-surface staining. Liver samples were immediately stored at −80°C for c-di-GMP quantification.

To determine the effect of AdVCA0848 on adaptive immune responses against OVA, we coinjected male 8- to 10-wk-old C57BL/6 mice (n = 4) with AdVCA0848 or AdNull in 30 μl of a PBS solution (pH 7. 4) containing 1 × 10¹⁰ vps/mouse via i.m. injection and 100 μg/mouse OVA via i.p. injection, with an additional group of mice that were not injected (naives). At 6 d postinjection (dpi), retro-orbital bleeding was used to collect blood samples for ELISA analysis. At 14 dpi, mice were sacrificed, peripheral blood samples were collected, and spleen was harvested in 2% FBS RPMI 1640 media.

To determine the effect of AdVCA0848 on the adaptive immune response against the AdGag, we initially conducted a dose-dependent study to determine the optimum AdVCA0848 dose that would significantly modulate adaptive immunity specific to the coinjected 5 × 10⁶ vps/mouse dose of AdGag. Six- to eight-week-old male BALB/c mice (n = 4) were i.m. coinjected in the tibialis anterior with vps in a PBS solution in 30 μl (pH 7.4) containing a dose of 5 × 10⁶ vps AdGag along with three different doses of 5 × 10⁷, 5 × 10⁸, or 5 × 10⁹ vps/mouse of either AdNull or AdVCA0848. An additional group of mice was not injected (naive). Additional experiments were conducted in which mice were coinjected with AdGag at 5 × 10⁶ and 5 × 10⁹ vps/mouse AdVCA0848 or AdVCA0848^mut, or were not injected (naives). At 14 dpi, mice were sacrificed, peripheral blood samples were collected, and spleen was harvested in 2% FBS media. To determine the effect of AdVCA0848 on the adaptive immune response against AdToxB, female 6- to 8-wk-old C57BL/6 mice (n = 4) were i.m. coimmunized in the tibialis anterior with vps of AdToxB (5 × 10⁸ vps/mouse) along with 5 × 10⁸ vps/mouse of either Ad5 vector expressing GFP or AdVCA0848. At 21 dpi, mice were terminally sacrificed, and blood samples were collected for B cell analysis with ELISA. To verify the expression of Gag protein in the injected mice, we i.v. injected 6- to 8-wk-old male BALB/c mice with 1 × 10¹¹ vps/mouse of AdGag only (n = 3) or coinjected them with 1 × 10¹¹ vps/mouse of AdGag along with 1 × 10¹¹ vps/mouse of either AdNull or AdVCA0848. At nearly 24 hpi, mice were humanely sacrificed and liver samples were obtained and frozen at −80°C until analysis by Western blot for Gag protein levels.

Liver samples were harvested from mice injected with 2 × 10⁹ vps/mouse AdVCA0848, mice injected with 2 × 10¹¹ vps/mouse AdVCA0848, AdVCA0848^mut, AdVCA0956, or AdNull, or mice that were not injected (naives), as described in the animal procedures. A total of 20 mg from each liver sample was placed in 500 μl PBS and homogenized using an Omni Tissue Homogenizer (Omni International). A total of 300 μl homogenate was added to an equal volume of equilibrated phenol solution (Sigma-Aldrich, St. Louis, MO). The homogenate-phenol solution was then vortexed and centrifuged at 15,000 rpm for 10 min. The aqueous phase was removed and added to 500 μl chloroform. The mixture was vortexed and then centrifuged at 15,000 rpm for 10 min. The aqueous phase was removed and stored at −80°C until analysis. Quantification of c-di-GMP was conducted by liquid chromatography coupled with tandem mass spectrometry at Michigan State University spectrometry and metabolomics core facility as previously described (39).

Liver samples from mice injected with AdGag alone or coinjected with AdGag and AdNull or AdVCA0848 as described earlier were harvested and later were homogenized in ice-cold lysis buffer containing 1% Triton and complete protease inhibitor. Supernatant was collected and analyzed for protein concentration (BCA protein kit; Sigma-Aldrich). Total protein of 15 μg was heated at 100°C for 5 min with Laemmli sample buffer (Sigma Aldrich, St. Louis, MO), and samples were loaded on 1-mm-thick 10% gel Mini-Protean TGX Precast Gels (Bio-Rad, Hercules, CA). Transfer was completed overnight at 4°C using a 0.2-μm nitrocellulose membrane (Millipore, Billerica, MA). The membrane was blocked for 1 h in Odyssey Blocking Buffer (Licor Biosciences, Lincoln, NE), then incubated for 1 h at room temperature with primary monoclonal mouse anti-Gag (1:10,000) Ab (183-H12-5C) obtained from the National Institutes of Health AIDS research and reference reagent program (gift from Dr. Y-H Zheng, Michigan State University) and mouse anti–β-actin (1:3000; 8224; Abcam, Cambridge, MA) diluted in Odyssey Blocking Buffer (927-40000; Licor Biosciences). The blot was washed with TBS-T three times and then incubated with labeled anti-mouse secondary Ab (926-32210; Licor) diluted in blocking buffer (1:10,000) for 1 h at room temperature. The blotted membrane was washed and developed on the Licor Odyssey (Licor).

Effects of AdVCA0848 on IFN-β induction were determined by quantifying IFN-β using the VeriKine mouse IFN-β ELISA kit (PBL Assay Science, Piscataway, NJ) according to the manufacturer’s instructions. To determine the effect of AdVCA0848 on B cell adaptive immune responses specific to Ags delivered by the coadministered AdGag or AdToxB, or the extracellular Ag OVA with the use of AdNull or AdVCA0848^mut as a negative control, we conducted ELISA-based titering experiments as previously described (40). In brief, 5 × 10⁸ vps/well inactivated Ad5 particles, 0.2 mg/well Gag protein, 50 μg/well OVA, or 100 ng/well ToxB (each diluted in PBS) was used to coat wells of a 96-well plate overnight at 4°C. Plates were washed with PBS-Tween 20 (0.05%) solution, and blocking buffer (3% BSA in PBS) was added to each well and incubated for 1–3 h at room temperature. For measuring total IgG Abs, plasma from injected mice was serially diluted in PBS buffer. Following dilution, plasma was added to the wells and incubated at room temperature for 1 h. Wells were washed using PBS-Tween 20 (0.05%), and HRP-conjugated rabbit anti-mouse Ab (Bio-Rad, Hercules, CA) was added at a 1:5000 dilution in PBS-Tween 20. Tetramethylbenzidine (Sigma-Aldrich) substrate was added to each well, and the reaction was stopped with 2 N sulfuric acid. OD was then obtained by reading the plates at 450 nm in a microplate spectrophotometer.

Splenocytes were harvested from individual mice, and RBCs were lysed using ACK lysis buffer (Invitrogen, Grand Island, NY). Ninety-six-well Multi-Screen high protein binding Immobilon-P membrane plates (Millipore, Billerica, MA) were wetted with 70% ethanol, coated with mouse anti–IFN-γ or IL-2 capture Abs, incubated overnight, and blocked before the addition of 5 × 10⁵ (AdGag studies) or 1 × 10⁶ (OVA studies) splenocytes/well. Additional studies were conducted using AdVCA0848^mut as a control (AdGag studies) with the use of 1 × 10⁶ splenocytes/well. Ex vivo stimulation included incubation of splenocytes in 100 μl media alone (unstimulated) or media containing 4 μg/ml Gag-specific AMQMLKETI (AMQ) peptide (GenScript, Piscataway, NJ) for the AdVCA0848 and AdGag studies, or 10 μg/ml OVA or SIINFEKL [MHC class I–restricted OVA-derived peptide (41)] for AdVCA0848 and OVA studies, overnight in a 37°C, 5% CO₂ incubator. Staining of plates was completed per the manufacturer’s protocol. Spots were counted and photographed by an automated ELISPOT reader system (Cellular Technology, Cleveland, OH). Ready-SET-Go! IFN-γ and IL-2 mouse ELISPOT kits were purchased from eBioscience (San Diego, CA).

To investigate innate immune responses after AdVCA0848 vaccination, we injected mice with 1 × 10¹⁰ vps/mouse AdVCA0848 vector, and activation of innate immune cells was evaluated 6 h after i.v. injection. Splenocytes were stained with various combinations of the following Abs: PE-CD69 (clone: H1.2F3), allophycocyanin-Cy7-CD3 (clone: 145-2C11), PerCP-Cy5.5-CD19 (clone: 1D3), Pacific blue-CD8α (clone: 53-6.7), and PE-Cy7-NK1.1 (clone: PK136) (4 μg/ml). To assess the effect of AdVCA0848 on dendritic cells (DCs), splenocytes were stained with combinations of the following Abs: PE-Cy7-CD11c (clone: HL3), allophycocyanin-Cy7-CD11b (clone: M1/70), Alexa Fluor 700-CD8a (clone: 53-6.7), FITC-CD40 (clone: HM40-3), PerCP-Cy5.5-CD80 (clone: 16-10A1), and V450-CD86 (clone: GL1) (4 μg/ml). All Abs were obtained from BD Biosciences. To determine the intracellular cytokine levels 14 dpi of AdVCA0848 and AdGag coinjections, we performed intracellular staining as previously described (37). In brief, splenocytes (2.5 × 10⁶/well) were stimulated with Gag-specific AMQ peptide for 6 h with brefeldin A (Sigma-Aldrich) for 30 min and stored at 4°C overnight. Cells were washed twice with FACS buffer and surface stained with allophycocyanin-CD3, Alexa Fluor 700-CD8a, and CD16/32 Fc-block Abs, fixed with 2% formaldehyde (Polysciences, Warrington, PA), permeabilized with 0.2% saponin (Sigma-Aldrich), and stained for intracellular cytokines with PE-Cy7-TNF-α and Alexa Fluor 488-IFN-γ (4 μg/ml) (all obtained from BD Biosciences, San Diego, CA). We included a violet fluorescent reactive dye (ViViD; Invitrogen) as a viability marker to exclude dead cells from the analysis. Tetramer staining of splenocytes at 1 × 10⁶ cells/well was performed using PE-labeled MHC class I tetramer folded with the AMQ peptide (generated at the National Institutes of Health Tetramer Core Facility, Atlanta, GA) for 30 min at room temperature, and for memory T cell staining, a mixture of the following Abs (at 2 μg/ml) was used: allophycocyanin-CD3, Alexa Fluor 700-CD8a, PerCP-Cy5.5-CD127, FITC-CD62L, and CD16/32 Fc-block Abs. All Abs were purchased from BD Biosciences (San Diego, CA). After washing with FACS buffer, data for stained cells were collected with the use of BD LSR II instrument and analyzed using FlowJo software (Tree Star, San Carlos, CA). Gating strategy was based on negative control results (naives) that were applied consistently across all samples examined. Representative examples from this gating approach are presented in this article for activation of innate immunity cells and for the frequency of cytokine-producing CD8⁺ T cells.

Statistically significant differences in innate immune responses were determined using a one-way ANOVA with a Student–Newman–Keuls post hoc test (p < 0.05 was deemed statistically significant). The ELISPOT and ELISA studies were all analyzed using one-way ANOVA with a Student–Newman–Keuls post hoc test (p < 0.05 was deemed statistically significant). For flow cytometry, a one-way ANOVA with a Student–Newman–Keuls post hoc test was used (p < 0.05 was deemed statistically significant). Statistical analyses were performed using GraphPad Prism (GraphPad Software).

We previously demonstrated the feasibility of in vitro and in vivo production of c-di-GMP in mammalian cells by using Ad5 vectors to transduce DGCs (30). Our prior unpublished studies suggested that use of an alternative DGC, VCA0848, which has greater enzymatic activities, might generate a significantly elevated amount of c-di-GMP in vivo. We constructed an Ad5 vector with a CMV enhancer/promoter element to drive VCA0848 expression in mammalian cells. The use of the AdVCA0848 platform resulted in a significant in vivo c-di-GMP production measured in the liver of injected mice. Injecting with increasing viral loads of 2 × 10⁹ and 2 × 10¹¹ vps/mouse AdVCA0848 resulted in ∼130 and 3000 μmol/g c-di-GMP in the liver, respectively. This confirms that the in vivo c-di-GMP production is entirely due to the enzymatic activity of the delivered VCA0848 because AdVCA0848^mut vectors and naive mice failed to produce detectable levels of c-di-GMP (Fig. 1). In addition, when compared with an earlier DGC-expressing platform that was constructed using the exact same adenovirus vector backbone, the AdVCA0848 platform produces significantly higher levels of c-di-GMP in the mouse liver (∼400-fold increase) than that produced by an equal viral dose of the AdVCA0956 platform per gram of mouse liver (p < 0.05). As expected, similar to AdVCA0848^mut control, the AdNull vectors, which lack the DGC gene, did not produce detectable levels of c-di-GMP (Supplemental Fig. 1). These results confirm the feasibility of transducing the bacterial DGC VCA0848 using Ad5 to synthesize in vivo larger amounts of c-di-GMP in vivo.

It is thought that activation of beneficial innate immune responses by adjuvants is the underlying mechanism that is critical for achieving effective and long-lived, Ag-specific, adaptive immune responses. i.v. administration of AdVCA0848 dramatically induced plasma levels of IFN-β (p < 0.05) nearly 1000-fold compared with the level produced by the AdNull control (Fig. 2A). Importantly, administration of AdVCA0848^mut control produced similar levels of IFN-β, as compared with AdNull, suggesting the increased IFN-β levels after AdVCA0848 is due to the enzymatic activity of the transduced VCA08484 (Supplemental Fig. 2A). Also, administration of AdVCA0848 significantly induced DC maturation and NK activation as compared with an identical cell population derived from AdNull controls (p < 0.05) (Fig. 2B, 2C). Furthermore, administration of AdVCA0848 resulted in increased numbers of CD69-expressing B cells, CD3⁺CD8⁻ T cells, and CD3⁺CD8⁺ T cells, as compared with the use of the AdNull vector in this experiment (p < 0.05) (Fig. 2D–F). Utilization of AdVCA0848^mut control suggested that the activation of immune cells is largely due to the enzymatic activity of the transduced VCA0848 (Supplemental Fig. 2B–F). Our results also confirmed previous findings that the Ad5 vector itself results in increased activation of NK cells, macrophages, CD3⁺CD8⁻ T cells, CD3⁺CD8⁺ T cells, and B cells as indicated by the significant expression of the activation marker CD69 (37). Together, these data suggest a significant induction of innate immune responses by AdVCA0848 in the mouse model, surpassing that caused by the adenovirus itself.

Direct administration of the OVA protein is a model Ag frequently used to study Ag-specific adaptive immune responses (42, 43). C57BL/6 mice were vaccinated with 100 μg/ml OVA alone or simultaneously with AdNull or AdVCA0848; a fourth untreated group served as a naive control. At 14 dpi, IFN-γ ELISPOT results from the experimental and control animals indicated that OVA-specific T cell responses from mice coadministered with AdVCA0848 and OVA were significantly higher (upon ex vivo stimulation with the entire OVA protein or the OVA-derived MHC class I–restricted peptide SIINFEKL) as compared with splenocytes derived from mice receiving only OVA, or OVA concomitant with the AdNull control vector (p < 0.05) (Fig. 3A). The simultaneous use of AdVCA0848 with OVA vaccination also increased the number of SIINFEKL and the intact OVA protein–specific, IL-2–secreting T cells present in the splenocytes of OVA-treated mice as compared with mice injected with OVA alone or concomitant with AdNull control (p < 0.05) (Fig. 3C). The noticeable variability of T cell responses resulted from the ex vivo stimulation with whole OVA protein and the MHC class I–restricted SIINFEKL peptide likely suggest a CD8⁺ T cell–driven response indicated by higher SIINFEKL-specific, IFN-γ–producing T cells and smaller SIINFEKL-specific, IL-2–producing T cells. Interestingly, splenocytes harvested from mice coinjected with AdVCA0848 and OVA also had dramatically increased numbers of Ad5 capsid-specific IFN-γ–secreting T cells and IL-2–secreting T cells, as compared with mice injected with OVA alone or concomitant with AdNull control (p < 0.05) (Fig. 3B, 3D). These results indicate that AdVCA0848 provides enhancement of OVA-specific adaptive T cell immune responses when coinjected with the extracellular Ag OVA.

Coadministering AdVCA0848 and OVA also resulted in enhancement of OVA-specific (Fig. 4A) and Ad5-specific (Fig. 4B) B cell responses 6 dpi. At 14 dpi, OVA-specific B cell response was enhanced compared with mice coinjected with the AdNull control vector (Fig. 4C) or when injected with OVA alone (p < 0.05) (Supplemental Fig. 3). Ad5-specific IgG Ab B cell responses were also detected in those mice that received either of the Ad5 vectors. Although the presence of AdVCA0848 significantly increased the Ad5-specific B cell response compared with that exerted by the AdNull control (p < 0.05) when measured at 6 dpi, this effect was observed to be minimal when measured at 14 dpi (Fig. 4D). Despite the transient enhancement of humoral response against the delivering vector, these results demonstrate the beneficial effects of AdVCA0848 on the OVA-specific adaptive B cell response from a single administration of OVA.

Our previous results indicated a modest, although significant, enhancement of adaptive immune responses specific against Ags expressed from Ad5-based vaccines coinjected with AdVCA0956, a vector expressing a less active DGC (30). Therefore, we assessed whether the enhanced ability of AdVCA0848 to produce c-di-GMP in vivo would also improve adaptive immune responses specific for adenovirus-expressed Ags. We have previously used an adenovirus-based vector to express the Gag protein, an HIV-1–derived Ag, and demonstrated the platform’s ability to induce Gag-specific humoral and cellular immune responses (31, 33, 40, 44). Based on our previous work, we administered the AdGag vaccine at the dose of 5 × 10⁶ vps/mouse along with escalating doses (5 × 10⁷, 5 × 10⁸, or 5 × 10⁹ vps/mouse) of AdVCA0848 or the AdNull control. After 14 d, Gag-specific memory T cell immune responses were evaluated by IFN-γ ELISPOT assay. The results demonstrated that concurrent administration of AdVCA0848 along with the AdGag vaccine inhibited T cell responses to the Gag Ag, which were especially significant at the highest AdVCA0848 dose of 5 × 10⁹ vps/mouse compared with that seen from the concurrent administration of AdNull control along with AdGag vaccine (p < 0.05) (Fig. 5A). Similar to our previous observations (45), as the viral load of AdNull coinjected with AdGag increased, the Gag-specific T cell response measured by IFN-γ ELISPOT decreased in a dose-dependent manner (p < 0.05). In contrast, ELISPOT assays demonstrated a dramatic enhancement of Ad5-specific, IFN-γ–producing T cells at 5 × 10⁹ vps/mouse AdVCA0848 compared with the AdNull control group (p < 0.05), whereas the first two doses of 5 × 10⁷ and 5 × 10⁸ vps/mouse showed minimal Ad5-specific T cell response (Fig. 5B). We confirmed that the inhibitory effects on IFN-γ–secreting T cells were lost in a VCA0848 mutant that cannot synthesize c-di-GMP (Supplemental Fig. 4A).

A multiparameter tetramer-binding assay showed a significantly decreased number of Gag-specific Tet⁺CD8⁺ T cells present in mice coinjected with three different doses of AdVCA0848 along with AdGag as compared with mice coinjected with AdGag and the AdNull control vector (p < 0.05) (Fig. 6A), confirming the negative impact of AdVCA0848 on the induction of Gag-specific CD8⁺ T cells. We also performed intracellular staining and FACS analysis to evaluate the impact of AdVCA0848 on the numbers of Gag-specific CD8⁺ T cells upon ex vivo stimulation with the Gag-specific peptide, AMQ. The number of IFN-γ– and TNF-α–producing CD8⁺ T cells specific for this potent Gag peptide were significantly inhibited in mice coinjected with AdVCA0848 as compared with equal viral loads of AdNull (p < 0.05), with the highest dose of AdVCA0848 of 5 × 10⁹ vps/mouse showing the strongest inhibitory effects (Fig. 6B, 6C). We also looked at the effect of AdVCA0848 on Gag-specific IFN-γ–, TNF-α–, and IL-2–producing CD4⁺ T cells and observed no significant effect (data not shown). Together, these data strongly suggested that despite a strong induction of innate immunity and improved induction of adaptive immune responses to extracellular proteins such as the OVA protein and the Ad5 capsid, expressing high levels of c-di-GMP using VCA0848 from an Ad5 vector significantly inhibited induction of Ag-specific CD8⁺ T cell responses to Ags expressed intracellularly by another Ad5 vector.

We next evaluated humoral B cell responses after AdVCA0848 coadministration with AdGag. Similar to its effect on T cell responses, the presence of AdVCA0848 resulted in significant inhibition of HIV-1/Gag-specific B cell responses as compared with those mice administered with equal amounts of the AdNull control vector (p < 0.05) (Fig. 7A). The inhibition of Gag-specific B cell responses by AdVCA0848 was very potent at the doses of 5 × 10⁷ and 5 × 10⁸ vps/mouse (compared with AdNull, p < 0.05). AdNull exhibited inhibition similar to AdVCA0848 at the highest dose of 5 × 10⁹ vps/mouse (Fig. 7A). Alternatively, increasing doses of both the AdNull and the AdVCA0848 increased B cell responses against the Ad5 vector in a dose-dependent manner (Fig. 7B). The inhibitory effects on Gag-specific B cell responses were lost using the AdVCA0848^mut that cannot synthesize c-di-GMP (Supplemental Fig. 4B). We confirmed the ability of AdVCA0848 to enhance Ad5-specific B cell response compared with that shown by AdVCA0848^mut (Supplemental Fig. 4C).

To confirm this interesting observation using a different Ag expressed by an Ad5-based vaccine, we coadministered AdVCA0848 along with an Ad5 vector expressing the truncated form of the AdToxB protein. The presence of AdVCA0848 with AdToxB also resulted in significantly reduced ToxB-specific B cell responses as compared with control vaccinations (p < 0.001) (Fig. 7C). Importantly, we again observed significantly (p < 0.01) increased Ad5-specific IgG titers in mice vaccinated with AdVCA0848 and AdToxB, as compared with controls (Fig. 7D). These results further confirm the inhibitory effects of the strong c-di-GMP producer, AdVCA0848, on another Ag intracellularly expressed from an adenovirus vector (AdToxB).

One possible explanation for the inhibition of response to Ad-expressed Ags is that the presence of the AdVCA0848 vector inhibits in trans the in vivo expression of the Ad expressed Ags. However, mice coinjected with AdVCA0848 and AdGag demonstrated the presence of the HIV-1–derived Gag protein whether delivered by the AdGag platform alone or when coinjected with the AdNull control or with AdVCA0848 (Fig. 8). These results suggest that inhibitory effects exerted by AdVCA0848 on B cell and T cell adaptive immune responses against Gag are not due to lack of Gag expression and translation in vivo.

Understanding the molecular mechanisms underlying how a putative adjuvant acts to enhance the efficacy of a specific vaccine will help to guide the formulation of newer-generation vaccines that efficiently generate specific long-term immunity against difficult Ags derived from pathogens or cancer cells (46). The use of pure c-di-GMP has been demonstrated to be an immunomodulatory molecule with potential therapeutic and prophylactic properties (19). Whereas the presence of nucleic acids can be sensed by AIM2, and signals the activation of caspase-1 (47, 48), the presence of cytosolic c-di-GMP can be sensed by other sensors including the STING and helicase DDX41 pathways, and subsequently lead to the release of IFN-β, primarily from CD11b⁺ DCs (17). In addition, c-di-GMP has been shown to stimulate the MYPS/STING-dependent induction of TNF-α and IL-22, not type I IFN, when used as a nasal mucosal adjuvant, suggesting c-di-GMP may have different effects on different innate immunity pathways (49, 50).

In this study, we demonstrated the ability of a potent, bacterial-derived DGC to be delivered by an Ad5 vector (AdVCA0848) that produced >400-fold more c-di-GMP than our previous Ad5 DGC vector (30), resulting in a robust induction of several innate immune responses, including IFN-β induction. By using a mutant version of VCA0848 delivered by AdVCA0848^mut, our data suggest that these significant levels of c-di-GMP are products of the enzymatic activity of the transduced VCA0848. These strong innate immune responses allowed the induction of enhanced adaptive immune responses to an extracellular Ag, that is, OVA, coadministered with the AdVCA0848, but also suppressed adaptive immune responses to virally expressed Ags. The recent characterization of mammalian endogenous cyclic GMP-AMP (2′3′-cGAMP) synthase (51–53) provided the rationale for testing cGAMP as a vaccine adjuvant, and initial studies demonstrated its usefulness in stimulating innate immune responses and improving Ag-specific adaptive immune responses (54–56). When compared with the bacterial c-di-GMP, cGAMP had higher binding affinity to STING. However, it has also been shown that c-di-GMP results in higher IFN-β induction than that induced by 2′3′-cGAMP or its isomers, suggesting that higher binding affinity to STING does not correlate with IFN-β induction. These results may be attributable to possible differences in biological stability between c-di-GMP and the mammalian cGAMP (53).

The adenovirus-based platforms we used in these studies are also expected to activate multiple innate immune responses. The vector is known to activate innate immune responses via interactions with extracellular and intracellular TLRs, and can simultaneously trigger early proinflammatory responses such as the induction of IP-10 (57) and the activation of the PI3K signaling cascade (58). We and others have also demonstrated that upon penetrating host cells and escaping the endosomal compartment, adenoviral vectors have the ability to ignite the MAPK and NF-κB signaling pathways through TLR-dependent (TLR2, TLR3, TLR4, and TLR9) and non-TLR–dependent mechanisms (59–61) leading to the induction of several chemokines and cytokines, fostering its utility as a vaccine platform in and of itself. In addition, the adenoviral dsDNA genome can be sensed by cytoplasmic sensors such as DAI (leading to type I IFN induction) (62) and AIM-2, resulting in activating the inflammasome and the induction of caspase-1–dependent IL-1β (47). Recent data also suggest that STING is central and acts as a major PRR after vaccination with Ad5-based platforms including Ad5 vectors (63). With these facts in mind, it is clear that our results confirm that the additional production of c-di-GMP from an already immunogenic platform such as Ad is significant enough to further promote the induction of proinflammatory immune responses beyond that provided by the Ad vector platform itself. Whether expression of DGCs from other vaccine platforms will yield similar results awaits future studies beyond the scope of this article.

The broad impact of the AdVCA0848 platform on innate immune responses clearly demonstrates its promising potential for use as a vaccine adjuvant to enhance adaptive immune responses. For example, relative to enhancing adaptive immune responses to extracellular Ags, plasmacytoid DC precursors are thought to be the major source of IFN-β (64). In agreement with previous reports that demonstrated the stimulatory effects of c-di-GMP on murine and human DCs (13, 19), AdVCA0848 improved the induction of CD11c⁺CD11b⁻ CD86⁺ DCs. Ultimately, plasmacytoid DC precursors can differentiate into typical DCs capable of stimulating naive T cells in an Ag-specific manner (65). IFN-β has also been shown to enhance DC maturation and the efficiency of DCs to activate the cross-priming of CD8⁺ T cells, and increase induction of CD4⁺ Th I differentiation (66). In addition to increasing the number of CD86⁺CD11c⁺CD11b⁻ DCs and activating CD69⁺NK1.1⁺ NK cells that are involved in regulating innate immune responses, AdVCA0848 activated cells directly involved in adaptive immune responses such as B cells and CD4⁺ and CD8⁺ T cells.

AdVCA0848 also enhanced induction of OVA-specific B cell and T cell adaptive responses. These results parallel recent studies evaluating the beneficial effects of direct administration of c-di-GMP as an adjuvant during vaccination with OVA (49, 50) and 4-hydroxy-3-nitrophenylacetyl-chicken γ-globulin, in which c-di-GMP was shown to have the capacity to enhance germinal center development (67). In addition, the presence of c-di-GMP in an adjuvant formulation containing chitosan improved adaptive immune responses to H5N1 Ags (16) and (along with a conventional aluminum salt-based adjuvant) improved adaptive immune responses specific to the hepatitis B surface Ag (67). Recently, it was demonstrated that nasal administration of c-di-GMP significantly increases the MYPS-mediated uptake of OVA Ag via endocytosis and pinocytosis in vivo. This generates mucosal adjuvant activities that are mediated by type II and type III IFN, but not type I IFN, suggesting variable c-di-GMP pleiotropic effects on innate immune responses against extracellular Ags. The in vivo production of c-di-GMP by i.m. administration of our AdVCA0848 platform potentially enhanced the OVA uptake and processing by DCs, and subsequently resulted in improved OVA-specific adaptive immune responses (50). As a proof of principle, our results suggest that adenovirus-based platforms expressing DGCs may also be used to promote improved immunity against other disease-specific Ags, such as those found in current cholera, diphtheria, and tetanus vaccines, because each is an example of a protein-based vaccine. In addition, because our approach also enhances activation of APCs and induction of Ag CD8⁺ CTLs, future studies using tumor Ag-specific peptides may also enhance the induction of antitumor cellular immune responses (21, 22, 68, 69).

Our results also revealed the potential for inhibitory effects on adaptive immune responses to Ags expressed intracellularly, simultaneous with provision of high levels of c-di-GMP. Although the dose of 5 × 10⁸ vps/mouse AdVCA0848 did not show significant inhibition of IFN-γ–secreting splenocytes compared with that shown by the AdNull control, this dose caused significant inhibition of Gag-specific IFN-γ– and TNF-α–secreting CD8⁺ T cells, suggesting that CD8⁺ T cells may be the specific targets for these inhibitory effects. Furthermore, increasing the AdVCA0848 dose to 5 × 10⁹ vps/mouse further inhibited Gag-specific T cell responses. Notably, the use of higher doses of the AdNull control vector also resulted in decreased induction of Gag-specific CD8⁺ T cell responses. Despite this, the provision of elevated c-di-GMP levels resulted in additional inhibitory effects on Gag-specific adaptive immune responses.

We have previously reported that increasing the dose of AdVCA0956 to 5 × 10⁹ vps/mouse did not improve B cell responses specific for an Ag delivered by an Ad5 vector in mice (30). Specifically, AdVCA0956 moderately suppressed B cell responses against the C. difficile–derived TA Ag expressed from the coinjected Ad5 vector at the dose of 5 × 10⁹ vps/mouse. In this article, our results suggest that those trends were likely real. Even stronger inhibitory effects were noted after administration of the more potent AdVCA0848 on B cell and T cell adaptive immune responses against the intracellularly expressed Gag and ToxB Ags. These results suggest that in mice the magnitude of inhibitory effects on adaptive immune responses to intracellularly expressed Ags is likely to increase with excessive amounts of c-di-GMP production.

There is also the possibility that the transduced DGC, and ultimately the synthesized c-di-GMP, interferes with the expression of these Ags when using the CMV expression cassette (used in constructing the vectors). We explored this possibility in vitro and found enhanced GFP expression in HEK293 cells coinfected with AdVCA0848 and an Ad5 vector expressing GFP from the same CMV enhancer/promoter elements used in these studies (data not shown). Our data also suggest that coadministration of the AdGag vaccine along with the strong c-di-GMP–producing AdVCA0848 did not prevent Gag translation. It remains unclear how the significant induction of c-di-GMP and subsequently high levels of type I IFN can inhibit the T cell and B cell responses of an intracellularly expressed Ag (63), and the impact of strong type I IFN induction on the availability of intracellular Ag-loaded APCs requires further investigation. We do note that the production of another bacterial second messenger, c-di-AMP, by the intracellular pathogen Listeria monocytogenes was shown to induce IFN-β in a STING-dependent manner leading to the inhibition of T cell–mediated immunity, similar to our results with excessive production of c-di-GMP (70).

In summary, we demonstrated the feasibility of in vivo synthesis of extremely large amounts of c-di-GMP via an Ad5-based platform expressing a highly potent DGC. Although high amounts of c-di-GMP production can inhibit adaptive immune responses to Ags expressed simultaneously with significant increasing c-di-GMP levels, this unique platform appears to preferentially improve Ag-specific B cell and T cell adaptive immune responses specific for coadministered extracellular Ags. This approach can be used to develop and improve protein-based prophylactic and therapeutic vaccines targeting infectious diseases and cancers.

We are grateful to Dr. Louis King at the Michigan State University Flow Cytometry facility for assistance in conducting our FACS experiments and to the Michigan State University Laboratory Animal Support Facility for help in the humane care and maintenance of the animals used in this study. We also thank the Michigan State University RTSF Mass Spectrometry and Metabolomics Core for help in c-di-GMP quantification.

The authors have no financial conflicts of interest.