AdrA as a Potential Immunomodulatory Candidate for STING-Mediated Antiviral Therapy That Required Both Type I IFN and TNF-α Production Free

The innate immune system provides a primary line of defense against pathogens in higher organisms, particularly in eukaryotes. Detection of pathogens—bacteria, viruses, and microbes—occurs via pattern-recognition receptors, which are sensors positioned strategically to detect a multitude of pathogen-associated molecular patterns in the extra- and intracellular environments. Pathogen-associated molecular patterns include sugars, lipids, and nucleic acids, and their binding to pattern-recognition receptors triggers a signaling cascade that culminates in the production of inflammatory cytokines, chemokines, and type I IFNs (1, 2).

Detection of viruses occurs via TLRs as well as DNA-sensing molecules such as cyclic GMP–AMP (c-GAMP) synthase or IFN-γ–inducible protein 16 (IFI16) and RNA sensors including retinoic acid–inducible gene I (RIG-I)/melanoma differentiation–associated protein 5 (MDA5) or the NOD-like receptor family, which are sensors of the inflammasome pathway (3).

c-GAMP synthase (cGAS) and other DNA sensors activate the stimulator of IFN genes (STING; also known as MITA, MPYS, ERIS, and transmembrane protein 173 [TMEM173]), which is located at the endoplasmic reticulum and functions as an essential signaling adaptor, linking the cytosolic detection of DNA to the type I IFN response. STING can also induce the activation of NF-κB, promoting the induction of NF-κB–dependent cytokines/chemokines like TNF-α (1, 4–7). The involvement of cGAS and several other dinucleotide cyclases in STING-mediated immunity and the potential therapeutic use of naturally occurring and synthetic cyclic dinucleotides (CDNs) have been extensively studied, especially for cancer (8–10). Over the last decade, various studies successfully used STING agonists as inducers of antitumor immunity (11–15). Another study reported the attenuation of experimentally acquired encephalomyelitis by STING activation using DNA nanoparticles (16). However, studies addressing the potential of CDNs in antiviral therapy are far and few between. Existing ones aimed at overcoming the failure to mount an immune response in chronic hepatitis B virus (HBV) infection (17) induce protection against genital herpes (18) and enhance responses in a macaque model of latent infection with SIV (19).

Salmonella enterica serovar Enteritidis (S. Enteritidis), a Gram-negative member of the genus Salmonella, contains 12 genes encoding putative diguanylate cyclases (20, 21). These signal transduction proteins are involved in regulation of cellular processes via the effector molecule cyclic-di-GMP (c-di-GMP). This second messenger produced in bacteria, but not in mammals, is involved in the regulation of complex biological bacterial processes, such as biofilm formation, virulence, cell cycle progression, or differentiation. The cyclase activity, which converts two molecules of GTP to c-di-GMP, is encoded in the GGDEF protein domain (22, 23). Besides its role as an intracellular and intercellular signaling molecule in prokaryotes, c-di-GMP also affects eukaryotes. Because of its similarity in the CDN structure with c-GAMP, c-di-GMP can also be recognized by STING, promoting the induction of IFNs and proinflammatory cytokines (7, 24). Over the past couple of decades, there has been an increasing interest in c-di-GMP as a vaccine adjuvant. Pure c-di-GMP was demonstrated to stimulate innate immune responses and to have an immunomodulatory and thus therapeutic and prophylactic function in protection against bacterial infection (25). Furthermore, Koestler (26) and Alyaqoub (27) reported that the use of diguanylate cyclases from Vibrio cholerae delivered by a replication-deficient adenovirus (Ad) serotype 5 could activate innate immunity and improved immune responses against extracellular Ags.

In the current study, a replication-deficient Ad vector expressing the S. Enteritidis–derived diguanylate cyclase AdrA (Ad AdrA) was developed and employed to address its antiviral potential in a mouse model of chronic HBV infection. Upon in vitro infection of dendritic cells (DC), they turned into highly activated APCs. Most importantly, AdrA induced a STING-dependent antiviral effect, which required both type I IFNs and TNF-α for maximum effectiveness. Our data indicate a potential usefulness of Ad AdrA as an antiviral agent.

Male C57BL/6J–Tmem173 goldenticket (gt)/J–transgenic (also referred to as gt or STING^−/−) mice were obtained from The Jackson Laboratory. STING^−/− are chemically induced mutant (mut) mice that carry a missense mutation in exon 6 of the TMEM173 (or STING) gene, which results in an isoleucine-to-asparagine change in aa 199 in the C-terminal of the protein that functions as a null allele and fails to produce detectable protein (28). STING^−/− mice, deficient for the IFN-αR (IFNAR^−/−), and HBV-transgenic (HBVtg) mice (kindly provided by F.V. Chisari) were bred and maintained under pathogen-free conditions and genotyped at 3 wk. Age-matched C57BL/6 wild-type (wt) (B6) males were purchased from Harlan Laboratories (Barcelona, Spain). For HBVtg experiments, animals were matched for age, sex, and levels of HBV DNA and HBV surface Ag in serum.

Six- to eight-week-old male mice were used in all experiments. Mice were kept under controlled temperature and light, with water and food ad libitum. Mice were injected i.v. with Ad (1 × 10⁷ or 1 × 10⁸ infectious units [infectious units per mouse]) and adenovirus-associated virus (AAV) vector (5 × 10¹⁰ viral genomes [vg] per mouse) in a volume of 100 μl. In the case of treatment with Etanercept (Enbrel; Pfizer), mice were injected i.p. every other day with a dose of 9 mg/kg in a volume of 100 μl. For all procedures, animals were anesthetized by i.p. injection of a mixture of xylazine (Rompun 2%; Bayer) and ketamine (Imalgene 50; Merial) 1:9 v/v. Blood collection was performed by submandibular bleeding, and serum samples were obtained after centrifugation of total blood. The experimental design was approved by the Ethics Committee for Animal Testing of the University of Navarra.

HEK293T cells were cultured in DMEM (Invitrogen) supplemented with 10% FBS, 1% Na-pyruvate, 1% l-glutamine 200 mM, and 1% penicillin/streptomycin (Invitrogen).

RAW-Lucia IFN-stimulated gene (ISG) cells and RAW-Lucia STING^−/− ISG cells were obtained from InvivoGen. They were generated from the murine RAW 264.7 macrophage cell line by stable integration of an expression cassette containing an ISG54 minimal promoter in conjunction with five IFN-stimulated response elements that control the expression of the secretable luciferase reporter gene. The levels of IRF-induced luciferase in the culture supernatant were monitored using a Quanti-Luc detection reagent (InvivoGen). RAW cells were grown in DMEM (Invitrogen) supplemented with 10% FBS, 100 μg/ml Normocin, 1% l-glutamine 200 mM, and 200 μg/ml Zeocin (InvivoGen). All cells were cultured at 37°C and 5% CO₂.

Bone marrow–derived DC (BMDC) were isolated from B6 or STING^−/− mice as previously described (29). In brief, bone marrow was isolated from femurs and tibias and RBCs depleted using ammonium–chloride–potassium lysing buffer. Cells were resuspended at 1 × 10⁶ cells/ml in complete RPMI 1640 (10% FBS, 2-ME 5 × 10⁻⁵ M, and 1% penicillin/streptomycin; all purchased from Life Technologies) supplemented with GM-CSF (10 ng/ml; Thermo Fisher Scientific) and seeded into tissue-treated plates (664160; Greiner). The next day, nonadherent cells were collected, washed, and seeded in petri dishes. Cells were incubated at 37°C with 5% CO₂ and media changed after 4 d. On day 7, adherent cells were harvested, washed, and resuspended in RPMI 1640 and 10% FBS. BMDC were cultured at 37°C and 5% CO₂.

Genes encoding 3xFLAG-tagged diguanylate cyclases from S. Enteritidis (AdrA, AdrA mut, SEN1023, SEN4316, yeaJ, and yfiN proteins) were PCR amplified from derivative strains of the clinical isolate S. Enteritidis 3934 (21). They were then subcloned into pcDNA3.1 plasmid (Invitrogen) using the TOPO-TA cloning system under the transcriptional control of the ubiquitous CMV promoter. The plasmids expressing cGAS and the inactivated version of cGAS (cGAS mut) were kindly provided by Dr. Nistal-Villan (Universidad San Pablo Centro de Estudios Universitarios, Madrid, Spain).

AdrA wt– and mut–encoding genes, together with the elongation factor 1 α (EF1α) promoter upstream and the bovine growth factor polyadenylation signal downstream, were cloned in the pSAdBST vector (kindly provided by Dr. R. Hernandez, Centro de Investigación Médica Aplicada), which is based on the commercial pAdEasy-1 vector and contains most of the human Ad serotype 5 (Ad5) genome necessary to produce a first-generation Ad that lacks genes E1 and E3. In addition, it contains a kanamycin resistance gene and a unique BstBI site for cloning expression cassettes into the deleted E1A gene.

AAV-HBV plasmid was constructed by insertion of 1.3× copies of the HBV genome as previously described (30).

Ad production and purification was performed as described previously (31). In short, AdrA wt and mutated AdrA pSAsBST plasmids were digested with Pac-1 and transfected into HEK293 cells using Lipofectamine 2000. Virus was collected 7–10 d after transfection, and limiting dilution was performed to obtain recombinant Ad clones. Viral DNA of each clone was extracted following the specifications of High Pure Viral Nucleic Acid Kit (Roche Diagnostics) and verified by PCR. One clone each was selected for amplification in HEK293 cells. For purification, cells were lysed, and virus was purified by ultracentrifugation in two consecutive cesium chloride gradients. Remaining cesium chloride was removed by size exclusion column of (Sephadex G-50; Sigma-Aldrich). The virus was eluted in 0.1 M Tris (pH 8.1), titrated, and stored at −80°C in elution buffer/glycerol 10%.

AAV-HBV production has been described elsewhere (32).

To determine the induction of IFN-β and ISG54 expression, HEK293T cells (5 × 10⁵/well in 12-well plate) were transiently transfected with two plasmids: an IFN-β or an ISG54 reporter construct (gift from Dr. A. García Sastre, Ichan School of Medicine, Mount Sinai, NY), each driving the expression of the firefly luciferase. In addition, cells were transfected with a control plasmid (pRL-TK), which drives the expression of Renilla luciferase under the control of a constitutive thymidine kinase promoter to allow normalization. Transfections were performed using Lipofectamine 2000 (Invitrogen). Luminescence was detected with the Dual-Luciferase Reporter Assay System (Promega) according to instructions. The activation of the IFN-β and ISG54 promoter expression was calculated as firefly luminescence relative to Renilla luminescence.

In the case of RAW-Lucia ISG cells, the coelenterazine-based luminescence assay reagent Quanti-Luc (InvivoGen) was used to detect luciferase activity in culture medium.

Detection of flag-tagged diguanylate cyclases was performed by lysing transfected HEK293T cells (5 × 10⁵) with 100 μl of a buffer containing 50 mM Tris-HCl (pH 7.5), 200 mM NaCl, 0.1% Triton X-100, 1 mM EDTA, 1 mM 2-ME, and 5% glycerol supplemented with a protease inhibitor (P8340; Sigma-Aldrich). Cell extracts were incubated at 99°C for 10 min, and 30 μl of extract was used to perform an electrophoresis in a 12% acrylamide gel at 100 V for 2 h. Proteins were transferred to a polyvinylidene difluoride membrane (Thermo Fisher Scientific) by tank transfer (85 V, 385 mA, 2 h), and subsequently, the membranes were blocked for 2 h with 0.01% PBS–Tween 20/10% reconstituted skim milk. The membrane was incubated overnight at 4°C with either mouse anti-flag M2 Ab (1/1000; Sigma-Aldrich) or rabbit anti-mouse β-actin Ab (1/2500; Sigma-Aldrich) in 0.01% PBS–Tween 20/10% milk. After three washes of 10 min each in 0.01% PBS–Tween 20, membranes were incubated with anti-mouse or anti-rabbit IgG-HRP Ab (both 1/5000 in 0.01% PBS–Tween 20; GE Healthcare). Membranes were washed three more times, and the HRP substrate was added (Super Signal West Pico Chemiluminescent Substrate; Thermo Fisher Scientific). Chemiluminescence was detected using a ChemiDoc Gel Imaging System (Bio-Rad Laboratories).

HBV DNA was isolated from 50 μl serum samples using the High Pure Viral Nucleic Acid Kit (Roche Diagnostics) according to the manufacturer’s instructions. Real-time quantitative PCR assays were performed using iQ SYBR Green Supermix (Bio-Rad Laboratories) in a CFX96 Real-Time Detection System (Bio-Rad Laboratories) and primers specific for hepatitis B core Ag (HBcAg) encoding sequence forward (5′-TTCGCACTCCTCCAGCTTAT-3′) and reverse (5′-GGCGAGGGAGTTCTTCTTCTA-3′) (Sigma-Aldrich).

Concentration of murine IFN-α and IFN-β was determined with VeriKine Mouse IFN α or β ELISA Kit (both PBL Assay Science). Serum concentrations of murine IL-12 p70 and IL-12 p40/p70 were measured using OptEIA mouse IL-12 p70 and IL-12 p40/p70 ELISA Kits (BD Biosciences/Pharmingen, San Diego, CA) according to instructions.

Concentrations of mouse GCSF, GM-CSF, IL-2, IL-3, IL-4, IL-5, IL-6, IL-9, IL-10, IL-12 p40/p70, IL-12 p70, IL-13, IL-17, IFN-γ, MCP-1, MCP-5, RANTES, SCF, soluble TNF receptor I (sTNFRI), TNF-α, thrombopoietin, and VEGF were determined by Mouse Cytokine Ab Array (ab133993; Abcam).

Expression of Ag-presenting and costimulatory molecules on the surface of DC was analyzed by flow cytometry using the following Abs and dilutions: MHC class I H-2Kb-FITC (1:200; BD), CD11c-PE (1:200; BioLegend), SIINFEKL/class I-PE (1:200; BioLegend), CD40–PE-Cy5 (1:100; eBioscience), MHC class II PE-Cy5 (1:200; eBioscience), CD86-APC (1:200; BioLegend), and CD80-BV421 (1:500; BioLegend). In short, BMDC were harvested with a cell scraper, washed, and counted, and 5 × 10⁵ cells were incubated with the respective Ab mixes for 15–20 min at 4°C in the dark. After washing twice, cells were analyzed with an eight-color BD FACSCanto II equipped with a 488-, 633-, and 405-nm laser or a four-color FACSCalibur with two lasers (argon 488 and 635 nm diode). Cells were gated as live (based on forward scatter/side scatter properties) and all expressed CD11c. Data were processed with FlowJo (V10; Tristar) software.

Liver sections were fixed in 4% paraformaldehyde (Panreac), embedded in paraffin and sectioned (5 μm). HBcAg (B0586; Dako, Glostrup, Denmark) was detected with the EnVision system (Dako) as previously described (33).

Statistical analysis was performed using Prism (GraphPad). Data are presented as mean ± SD. Comparisons between two groups were made using a two-tailed unpaired t test. Multiple groups were compared using ANOVA, followed by Bonferroni posttest. Statistical significance was assigned to p values <0.01.

The potential of S. Enteritidis diguanylate cyclases to activate the STING response in eukaryotic cells was investigated. To this end, five of its diguanylate cyclases, AdrA, SEN1023, SEN4316, yeaJ, and yfiN, were subcloned into mammalian expression plasmids under the transcriptional control of the CMV promoter. An inactive form of AdrA wt, in which the GGDEF motif inside the GGDEF domain had been mutated to GGGSF, which was termed AdrA mut, was included. Expression was tested by transfection of HEK293T cells and subsequent Western blot analysis (Fig. 1A). All cyclases were expressed at similar amounts, with the exception of SEN4316.

Next, the capacity of the cyclases to induce the type I IFN pathway was tested in HEK293T cells, making use of an IFN-β reporter system. None of the diguanylate cyclases were able to induce IFN-β on their own. However, when they were cotransfected with a plasmid carrying human STING (hSTING), induction of the IFN-β as well as ISG54 reporter systems (Fig. 1B) was achieved for all cyclases except for yeaJ. Interestingly, even though SEN4316 expression measured by Western blot was very low, its amount was sufficient to induce IFN-β in the presence of STING. Being the strongest inducer of IFN-β, AdrA wt was chosen for further characterization. AdrA wt expression had a dose-dependent effect on IFN-β levels but, again, only in the presence of cotransfected hSTING (Fig. 1C). AdrA mut, in contrast, could not induce expression of IFN-β, either in the absence or in the presence of hSTING. Furthermore, AdrA wt, and not AdrA mut, was able to induce the IFN pathway in the presence of murine STING (Fig. 1D). Comparing the ability of AdrA wt to induce IFN-β with that of the well-described cGAS, it turned out that although cGAS had the tendency to induce higher levels of IFN-β, no differences in the induction of ISG54 expression could be observed (Fig. 1E, 1F). Cyclase activity (i.e., c-di-GMP production by AdrA wt, but not AdrA mut) was further confirmed by HPLC in cell extract from transfected HEK293T cells (data not shown).

Having established that AdrA wt can serve as a strong activator of the type I IFN pathway in the presence of STING, its potential use as immunomodulatory agent was then addressed. An Ad-based delivery system was developed. Both AdrA wt and AdrA mut were each placed under the control of the EF1α promoter. The ability of both vectors to induce the type I IFN pathway and its dependency on STING was tested using RAW-Luc-ISG54 cells or its STING-deficient variant RAW-Luc-STING^−/−-ISG54. RAW-ISG54 cells were derived from the murine macrophage cell line Raw 264.7 and contain a stably expressed IFN-inducible ISG54 promoter enhanced by a multimeric IFN-stimulated response elements (ISRE). When infecting the two RAW reporter cell lines with either Ad AdrA wt or Ad vector containing AdrA mut (Ad AdrA mut), luciferase expression was only induced in RAW-Luc-ISG54 cells, but not in the STING-deficient cells (Fig. 2A). Induction by Ad AdrA wt was 4-fold stronger than by Ad AdrA mut. Given that plasmid carrying AdrA mut was not able to induce IFN in HEK293T in the presence of STING, it is likely that the induction of the type I IFN pathway was the result of recognition of the Ad vector as has been previously described (34).

With a functional vector for delivery at hand, the immunomodulatory potential of AdrA wt and its mut version was investigated in DC. DC were generated from the bone marrow of B6 or STING^−/− mice. STING^−/− mice fail to produce detectable levels of STING protein because of a single nucleotide mutation (T596A) in the STING gene (28). Susceptibility of DC from either strain to infection with Ad did not differ (data not shown). wt and STING^−/− DC were infected with Ad AdrA wt or Ad AdrA mut, and 24 h postinfection, a selection of cytokines was measured in culture supernatants by either cytokine array or ELISA (Fig. 2B–D). B6 DC infected with Ad AdrA wt produced large amounts of IL-6, IL-12, MCP-1, and type I IFN and, to a lesser degree, Rantes, sTNFRI, IL-15R, and several others, including TNF-α. B6 DC infected with mutated AdrA only produced IL-12, Rantes, and sTNFRI, and they did so at much lower levels than the ones infected with Ad AdrA wt, indicating that the activation of immune cells is largely due to the enzymatic activity of AdrA. For DC derived from STING^−/− mice, none of the above cytokines could be detected in cell supernatants. In addition, different markers of activation were analyzed (Fig. 2E). Only B6 DC infected with Ad AdrA wt or Ad AdrA mut showed increased upregulation of MHC class I and class II, and levels were much higher in cells infected with the AdrA wt variant. In line with this finding, pulsing the cells with the OVA-derived peptide SIINFEKL postinfection and measuring its presentation by flow cytometry led to strongly increased peptide presentation (Fig. 2F). High expression of classical activation markers such as CD80, CD86, and PD-L1 was also only observed in B6 DC infected with AdrA wt.

As described, CDNs can be potent inducers of the expression of a number of antiviral molecules. To test the potential of AdrA in modulating the antiviral response, an HBVtg mouse model was used (35). This model has been widely used to test immunostimulatory therapies aimed at clearing HBV infection (33, 36, 37).

HBVtg mice were injected systemically with a dose of 1 × 10⁸ infectious units per mouse of Ad AdrA wt or Ad AdrA mut, and the viremia in blood (HBV DNA copies/ml blood) was followed for 10 d (Fig. 3A). A significant reduction of viremia in serum of Ad AdrA–treated mice compared with PBS controls was observed; however, while in the Ad AdrA wt group, viremia dropped until day 7 and remained low by day 10, and in the group treated with Ad AdrA mut, the effect was more transient: by day 3, viremia had dropped to levels equivalent to those observed in the Ad AdrA wt group, but although still being lower than in the PBS-injected control mice on day 10, it started increasing again much faster than in the Ad AdrA wt–treated group. When comparing presence of HBcAg expression in livers by immunohistochemistry (IHC) on day 10, the proportion of HBcAg-stained area was lower in Ad AdrA wt–treated animals than in the ones that had been treated with Ad AdrA mut or the PBS control group, but the difference was nonsignificant (data not shown). Interestingly though, in livers from the Ad AdrA wt group, HBcAg was generally localized at a higher rate in the nucleus than in the cytoplasm, and the staining of the cytoplasm was weak. The presence of cytoplasmic HBcAg is an indicator of active HBV replication, whereas nuclear HBcAg is highly stable, persisting in the absence of HBV replication (38, 39). Indeed, when quantifying the ratio of stained area/nuclei in IHC images, it was found that it was significantly lower for the Ad AdrA wt group than in the Ad AdrA mut or PBS groups (Fig. 3B). For Ad AdrA mut, partial (upper right) or no cytoplasmic reduction of HBcAg (lower left) was observed.

Given that the initial drop in blood viremia was observed in all groups injected with Ad and that it was likely to be an effect of the virus as described previously by Cavanaugh et al. (40), the experiment was repeated using a 10-fold lower dose of virus. This should allow for a better distinction between effect of the Ad and AdrA wt. The results obtained were in line with those obtained for the higher dose, but this time, albeit NS, there was already a difference in blood viremia between Ad AdrA wt and Ad AdrA mut recipients on day 3 (Fig. 3C). Mice were followed until day 14, and even though the HBV viremia in the Ad AdrA wt group started increasing again after day 7, significant differences in viremia between the two groups were maintained for the duration of the experiment.

Possible modes of action of AdrA wt include a direct induction of an intracellular antivirus mechanism and an indirect recruitment of immune cells. Direct action of AdrA wt would depend on STING, which has been reported to be expressed at only very low levels in hepatocytes (41) but is present in immune cells such as macrophages and DC, liver resident or not. Although the recruitment of immune cells could be excluded, as no signs of infiltration could be detected (not shown), the involvement of liver-resident leukocytes is a likely possibility. To elucidate the molecular mechanism of AdrA in HBV clearance, an AAV-HBV infection model was used in B6, STING^−/−, and IFNAR^−/− mice. HBV replication in these animals was established by using an AAV-HBV vector, as described previously (42). Mice received an i.v. injection of AAV-HBV, and once HBV steady-state levels were reached in circulation (day 21), recombinant Ad was injected (1 × 10⁷ infectious units). To establish the effect of Ad AdrA wt and Ad AdrA mut in the new model, a first experiment was performed with B6 animal only, and groups of mice were sacrificed on days 1, 3, 7, and 10 after injection of Ad. In the Ad AdrA wt group, serum viremia had dropped significantly on day 1 compared with the PBS control group (Fig. 4A). On day 3, levels were at the detection limit and started recovering thereafter. In Ad AdrA mut–treated animals, an initial drop in viremia could also be observed, albeit less pronounced than in the Ad AdrA wt groups. Measurement of relative HBV DNA and RNA levels were in line with these findings: only in Ad AdrA wt–injected groups, HBV DNA and RNA levels were significantly decreased compared with both PBS controls and Ad AdrA mut groups (Fig. 4B, 4C). Detection of HBcAg expression by IHC revealed that only in Ad AdrA wt recipients, the cytoplasmic stain typically observed (especially around blood vessels) was greatly reduced on days 1 and 3 after injection of Ad (Fig. 4D). Although quantification of the area staining positive for HBcAg did reflect this finding (Fig. 4E), it was NS between groups.

With day 3 established as the day on which differences in blood viremia were most prominent, in the following experiments, mice were treated as before, and the blood viremia was analyzed for each group on days 0, 3, and 7 and at sacrifice on day 10 (Fig. 5A). As before, in B6 animals administered with Ad AdrA wt or Ad AdrA mut (1 × 10⁷ infectious units), by day 3, the blood viremia had dropped near 100-fold in the Ad AdrA wt group only, and by day 7, the difference to the mut and control groups had already almost disappeared (Fig. 5B).

Administering Ad AdrA wt or its mutated version to STING^−/− mice after establishing HBV infection by means of AAV-HBV did not induce any changes in blood viremia (Fig. 5C). This implicates that the observed reduction in viremia is STING dependent indeed. Interestingly, in AAV-HBV, IFNAR^−/−-treated mice, injection of Ad AdrA wt led to a less pronounced reduction (10-fold) of serum viremia than in the AAV-HBV B6 animals (Fig. 5D).

TNF-α has long been implicated in the downregulation of HBV in animal models (35, 43), and upregulation of TNF-α as part of the proinflammatory cytokine repertoire induced by STING signaling (44) has also been described. Hence, to investigate the role of TNF-α as a possible candidate that may be involved in the control of viremia in the absence of type I IFNs, B6 and IFNAR^−/− mice were treated with Enbrel, which blocks signaling via the TNFR. Anti–TNF-α–treated B6 mice showed an ∼10-fold reduction in viremia on day 3 after adenoviral infection with AdrA wt (Fig. 5E). This drop was lower than the one observed in B6 mice infected with AdrA wt, and interestingly, it was more transient than the effect observed in IFNAR^−/− mice: by day 7, viremia had returned to levels found in the Ad AdrA mut– and PBS-treated groups. Taken together, these data suggested that both type I IFNs and TNF-α are key players in the mechanism of inhibition of HBV replication and that their contribution may be similar. Indeed, the synergistic action of type I IFNs and TNF-α was confirmed by treating IFNAR^−/− mice with Enbrel, in which the antiviral effect of AdrA wt was completely lost (Fig. 5F).

c-di-GMP was first described in the bacterium Gluconacetobacter xylinus as a regulator of cellulose production (45). Following its initial discovery, it was shown to occur in a wide range of bacteria, in which it acts a second messenger and is involved in many physiological processes important for prokaryote survival, such as cell cycle progression, motility, formation of biofilms, and virulence (46). To date, there is only one report on its existence in eukaryotes, namely in all major groups of the amoeba genus Dictyostelium, in which it plays a role in stalk formation in conditions of famine (47). Like other CDNs, c-di-GMP has the ability of activating the innate immune system by binding to STING (48), and its immunostimulatory potential has drawn attention to its use as vaccine adjuvant (25, 49–51). However, although so far most research has focused on the direct use of CDNs or the use of diguanylate cyclases in a vaccine setting (26, 27), in this study, the potential of diguanylate cyclases from S. Enteritidis to elicit antiviral responses was investigated.

Among the five diguanylate cyclases tested (AdrA, yeaJ, SEN1023, SEN4316, and yfiN), AdrA was found to have the strongest capacity to induce STING-dependent IFN-β expression for both murine and hSTING. The observed differences in the magnitude of IFN-β induction by the tested bacterial enzymes may be due to differences in their enzymatic activity. AdrA was reported to be a highly active enzyme that can produce up to 60% of total c-di-GMP in bacteria (52)—whether it has the same level of activity in eukaryotic cells remains currently unknown. Comparing the induction of IFN-β and downstream ISG54 by AdrA and the eukaryotic CDN synthase cGAS in HEK293T cells transiently transfected with a reporter system, induction by cGAS was found to be similar. This finding is in line with a study by Yi et al. (12), which using the same transient transfection approach but treating the cells directly with CDNs, reported a higher induction of IFN-β by c-GAMP compared with c-di-GMP.

Importantly, and in contrast to studies suggesting that the recognition of protozoan c-di-GMP and metazoan c-GAMP might be differential for human and murine STING (53), AdrA wt was able to induce IFN-β responses that were very similar for human and mouse STING.

Using a replication-deficient Ad as delivery vehicle, Ad AdrA, but not its AdrA inactive version, was found to be able to activate DC in vitro, leading to increased expression of MHC class I and II as well as costimulatory molecules. In line with the upregulation of MHC class I molecules on the DC surface, an increased presentation of the OVA-derived CD8 peptide SIINFEKL was observed. As previously described (54), part of the observed upregulation of MHC and several costimulatory molecules could be attributed to the adenoviral vector itself as DC infected with Ad AdrA mut also showed a degree of activation. However, upregulation was clearly STING dependent because DC derived from STING^−/− mice did not respond. Importantly, only Ad AdrA wt–, but not Ad AdrA mut–infected B6 DC displayed a proinflammatory cytokine profile, with IFN-α and -β, IL-6, IL-12 p70/p40, and MCP-1 being the predominant cytokines produced. This observation indicates that the presence of the adenoviral vector was not the sole driver of DC activation. Interestingly, IFN-γ production was absent, and in contrast to other publications, only a low induction of TFN-α was observed (55).

A potential antiviral function of AdrA was addressed in an HBVtg mouse model. This model resembles a chronic infection in many ways; however, unlike in patients in whom HBV-specific T cells are rendered unresponsive by exhaustion, in HBVtg mice, they undergo deletion via central tolerance. Administration of both Ad AdrA wt and Ad AdrA mut led to a transient drop of viremia in serum—an effect that in the initial phase could be partially attributed to Ad-mediated activation of the innate immune response and one that could be diminished by reducing the dose of virus used. The drop in blood viremia and the disappearance of HBcAg from centrilobular hepatocytes upon Ad infection was previously described by Cavanaugh et al. (40). In agreement with their results, detection of intranuclear capsid Ag was not affected. However, although the viremia started increasing again after 3 d in the Ad AdrA mut group, it stayed low in the Ad AdrA wt group until 10 d and was still significantly lower on day 14 when a lower dose of Ad was used. This finding was confirmed by HBcAg IHC. Together, these findings support that the adenoviral vector is only involved during the early phase of the antiviral response and that the prolonged effects are conferred by AdrA.

Furthermore, using AAV-HBV to establish an HBV chronic replication model in B6 and STING^−/− mice, the reduction of HBV viremia was shown to be clearly STING dependent. In both HBV mouse models used in the current study, administration of Ad AdrA wt did neither result in a hepatic immune infiltrate nor liver damage; thus, a likely mechanism is a direct antiviral effect mediated by proinflammatory cytokines such as type I IFNs and TNF-α. However, as murine hepatocytes express STING only at very low levels (41), it is generally considered that STING does not partake in the initiation of innate antiviral responses in the murine liver. In contrast, in 2017, Guo et al. (41) used an in vitro system comprising immortalized mouse hepatocytes with TET-inducible HBV replication and HBV-infected, NTCP-expressing human hepatoma cells to show that direct treatment of these cells with STING agonists c-GAMP or DMXAA led to an inhibition of HBV replication, most likely conferred by the induction of an intracellular antiviral cytokine response. Whether this is the case in our models or whether the intracellular response is triggered by the activation of liver-resident macrophages (Kupffer cells) and DC and their subsequent chemokine/cytokine production or a combination of both cannot be concluded from the data obtained to date.

Although type I IFNs are widely associated with STING-dependent innate immune responses, the systemic administration of Ad was shown to lead to an acute cytokine response, especially the production of TNF-α and IL-6, from DC and Kupffer cells within the first 6 h (40, 56). Further elucidating the role of IFN-α/β and TNF-α in the Ad AdrA wt–mediated antiviral response, we found that the observed reduction in viremia was a combined effect of type I IFNs and TNF-α. Various papers have shown the effects of type I IFNs and TNF-α on HBV viremia in vivo combining HBVtg with cytokine injections, knockout models, Abs, coinfection with other viruses, and transfer of cytotoxic lymphocytes (35, 40, 43, 57, 58). No study identified a single molecule as the sole mediator; however, the contribution of each varied with the study setup: TNF-α [e.g., was proposed as a possible main driver by Heise et al. (59) but discarded as such by McClary et al. (43) a year later in a study that investigated the effect of an Ad on HBV clearance in HBVtg mice]. In our hands, both type I and TNF-α contributed about equally; in both IFNAR^−/− and anti–TNF-α mice, HBV viremia was partially reduced (∼10-fold each) upon treatment with Ad AdrA wt and only blocking the binding of TNF-α to its receptor in IFNAR^−/− animals led to the complete abrogation of the reduction of blood viremia. Thus, although type I IFN may be more important during the later response (beyond day 3), to achieve a maximum effect, both mediators have to act in an additive manner or even synergistically.

Although c-di-GMP may be considered a molecule less useful as immunostimulatory adjuvant because of its inability to induce activation of TBK-1/IRF (60), we, in this study, described the bacterial di-guanylate cyclase AdrA as a potent inducer of IFN-β and the downstream ISG54 through both murine and hSTING. When delivered by an adenoviral vector in vitro, AdrA had the potential to turn BMDC into mature APCs capable of T cell activation. Furthermore, using in vivo HBV mouse models, Ad AdrA was not only found to have an antiviral effect, but importantly, this effect was mediated by a synergistic interaction of TNF-α and type I IFNs. Taken together, AdrA may offer an alternative to existing options as an immunomodulatory candidate for STING-mediated antiviral therapy. Furthermore, with its potential to activate DC, AdrA could also be useful as an alternative adjuvant for vaccines, a field in which various other activators of STING are readily being tested.

We thank Cristina Olagüe, Africa Vales, and Niklas Rieck for excellent technical assistance.

The authors have no financial conflicts of interest.