Herpes Simplex Virus Glycoprotein D Targets a Specific Dendritic Cell Subset and Improves the Performance of Vaccines to Human Papillomavirus-Associated Tumors

Since the mid-1970s, many studies have shown that human papillomavirus (HPV) is present in virtually all clinical cases, being considered the etiologic agent of cervical cancer (1–3). Despite the high efficacy in preventing HPV infection, prophylactic vaccines do not benefit women with precursor lesions or cancer (4). Therapeutic vaccines aim to control the growth of HPV-induced lesions/tumors by activation of cell-based immune responses, particularly tumor-infiltrating cytotoxic CD8⁺ T cells (5, 6). A few other antitumor vaccines evaluated under clinical conditions demonstrated encouraging results regarding induction of immune responses, such as generation of specific antibodies and T-cell responses to HPV antigens (7–10). However, the complete regression of HPV-induced lesions/tumors by therapeutic antitumor vaccines, particularly those at more advanced stages, is still an unmet goal. Thus, the discovery of effective and safe therapeutic approaches may radically change the way cervical cancer is currently treated.

In this scenario, our group has developed two therapeutic vaccine strategies to control HPV-16–associated tumors based either on naked DNA (11) or on purified recombinant protein (12). Both strategies are based on the expression of a hybrid protein generated after the genetic fusion of HSV-1 gD with HPV-16 E7 oncoprotein (gDE7). Regarding the vaccine formulation based on the purified recombinant protein, our previous observations demonstrated that repeated immunizations with the non-adjuvanted gDE7 protein conferred partial therapeutic antitumor protection to mice transplanted with tumors cells expressing the HPV-16 E6 and E7 oncoproteins (TC-1; ref. 12). The therapeutic antitumor protection mediated by vaccination was linked to the adjuvant properties of the gD protein.

In the present study, we showed that the therapeutic antitumor efficacy of the gDE7-based vaccine could be further enhanced by co-administration of the poly (I:C) adjuvant, inducing a complete and long-lasting therapeutic antitumor protection. Furthermore, immunization with gDE7 and poly (I:C) induced infiltration of E7-specific CD8⁺ T cells into the tumor microenvironment and prevented the expansion of systemic immunosuppressive cell populations [regulatory T cells (Treg) and myeloid-derived suppressor cells (MDSC)]. Remarkably, the adjuvant role of gD protein was linked with the activation of a dendritic cell (DC) subset specialized in antigen cross-presentation to CD8⁺ T cells, both in mice and in humans. The present observations shed light on the cellular mechanisms of the gD-mediated adjuvant effects involved in the induction of a potent antitumor response to HPV-16–associated tumors.

The animal facility unit of the Department of Parasitology at the University of São Paulo (São Paulo, Brazil) provided female C57BL/6 wild-type (WT), C57BL/6 TLR4 knockout (KO), and C57BL/6 CD4⁺ T-cell KO mice. The animal facility unit of the Department of Immunology at the University of São Paulo provided female C57BL/6 CD8⁺ T-cell KO mice. All animal handling, immunization, and euthanasia procedures were approved by the ethics committee for animal experimentation (CEUA 147/2011) and followed the standard rules approved by the National Council for Control of Animal Experimentation (CONCEA). TC-1 tumor cell line was kindly provided by Dr. T.C. Wu at Johns Hopkins University (Baltimore, MD) in 2002 (13). Groups of mice were challenged by subcutaneous injection of 0.1 mL of serum-free media containing 7.5 × 10⁴ TC-1 cells in the right rear flank of mice. Tumor growth was monitored by visual inspection and palpation twice a week after challenge. Mice were considered as tumor-bearing when tumors became solid and were euthanized when tumors exceeded 10 mm in diameter. For experiments of tumor microenvironment analysis, TC-1 cells were diluted in Matrigel (BD Biosciences) before inoculation.

The gD, gDE7, and E7 recombinant proteins were expressed and purified as previously described (12). The concentration and the integrity of the antigens were routinely checked by PAGE and the molecular weights corresponded to those predicted in silico. Lipopolysaccharide (LPS) quantification was performed with the Limulus Amebocyte Lysate Kit (Lonza) according to the instructions of supplier. LPS amounts found in gD, gDE7, and E7 protein solutions corresponded to 0.9, 3.2, and 1.2 ng of LPS per μg of protein, respectively. Vaccine formulations contained 30 μg of gDE7 or 10 μg of E7 each dose, suspended in 200 μL of apyrogenic saline. Poly (I:C) (InVivogen) was admixed with gDE7 or E7 recombinant proteins (50 μg/dose). In most experiments, mice were subcutaneously vaccinated with two doses of vaccine formulations in weekly intervals, starting one day after TC-1 cell transplantation. The therapeutic vaccination protocol was also performed with two doses of gDE7 admixed with poly (I:C) vaccine formulation starting at the following different time points: 1, 3, 5, 7, or 10 days after challenge. To evaluate the impact of gD-pre-existing immunity in the antitumor response, groups of mice were primed with a single dose of non-adjuvanted gD recombinant protein (22.5 μg/mice), challenged with TC-1 cells, and immunized with two doses of gDE7 admixed with poly (I:C).

To evaluate the kinetics of the CD8⁺ T-cell response, blood samples were collected weekly and cultured with the CD8-specific E7 peptide (⁴⁹RAHYNIVTF⁵⁷; GeneScript; 1.5 μg/mL; ref. 14), brefeldin A (10 μg/mL; Sigma), and IL2 (5 ng/mL; Sigma). After incubation, cells were stained with anti-CD8a-FITC (cat: 553031) and, after fixation and permeabilization, with anti-IFNγ-PE (cat: 554412) monoclonal antibodies (mAb). The percentages of CD8⁺ IFNγ⁺ cells over all CD8⁺ T cells were determined. To study the functional profile of CD8⁺ T cells, spleen cells from immunized mice were re-stimulated with an anti-CD28 antibody (2 μg/mL; cat: 553295), the CD8-specific E7 peptide (1.5 μg/mL), and brefeldin A-GolgiPlug (10 μg/mL; BD Biosciences) for 12 hours. After the incubation period, cells were stained with anti-CD8a-Pacific Blue (cat: 558106) mAb, fixed and permeabilized using the Cytofix/Cytoperm Kit, and stained with anti-IL2-PE (cat: 561061), anti-TNFα–PE-Cy7 (cat: 557644), and anti-IFNγ-APC (cat: 554413) mAbs. Boolean platform was used to create all possible combinations of cytokines over the CD8⁺ population. For memory phenotyping assays, spleen cells were stained with the APC-labeled H-2Db E7–specific dextramer (Immudex) and, subsequently, stained with anti-CD8a-Pacific Blue (cat: 558106), anti-CD44-PE (cat: 553134), and anti-CD62L PE-Cy5 (cat: 555545) mAbs. The number of effector memory (T_EM; CD44⁺CD62L^low) and central memory (T_CM; CD44⁺CD62L^high) E7-specific CD8⁺ T cells were determined. For Treg cell analysis, spleen cells were cultured with 10 μg/mL of E7 protein for 5 days (10, 15). After incubation, cells were stained with anti-CD4-FITC (cat:553651), anti-CD25-APC (eBiosciences; cat: 17-0251), and anti-Foxp3-PE (eBiosciences; cat: 12-4771-80) mAbs using a Foxp3 staining buffer kit (eBioscience). For MDSC analysis, spleen cells were stained with anti-CD11b-FITC (cat: 553310), anti-Gr-1-PE (cat: 551084), and anti-CD45-PercP-Cy5.5 (eBiosciences; cat: 45-0451) mAbs. The frequencies of CD25⁺ FoxP3⁺ cells over all CD4⁺ T cells and CD11b⁺ Gr-1⁺ over all CD45⁺ cells were determined. Unless stated differently, all antibodies were from BD Biosciences. Samples were acquired on a LSR Fortessa flow cytometer (BD Biosciences) and analyzed using the FlowJo software (Tree Star).

Cells from naïve mice were stained with 0.5 or 5 μmol/L of carboxyfluorescein diacetate succinimidyl ester (CFSE; Invitrogen). The 5 μmol/L CFSE-labeled population was also pulsed with 2.5 μg/mL of the CD8-specific E7 peptide. Both peptide-pulsed (target) and non-pulsed (control) cells were injected intravenously (10⁷ cells each) into mice previously immunized. One day after the transfer of CFSE-labeled cells, splenocytes were harvested and analyzed for CFSE staining by flow cytometry. The percentages of target-specific cell killing were calculated as described elsewhere (16).

CD11c⁺ cells were positively selected with magnetic beads according to instructions of the kit supplier (Miltenyi Biotec). The DC yield represented more than 80% of the total cells recovery at the end of the procedure. The protein–DC binding assays were performed with isolated DCs from C57BL/6 WT mice (4 × 10⁵ cells/well). Recombinant proteins were used in equimolar amounts [gDE7 (3 μg/mL), gD (2.25 μg/mL), E7 (0.75 μg/mL)]. DCs Fcγ receptors were blocked with anti-CD16/CD32 antibodies (cat: 553140). Cells were incubated with the recombinant proteins and, subsequently, treated with purified anti-His-tag IgG mAb (Thermo Fischer; cat: MA1-21315). After washes, cells were stained with anti-IgG-PE antibodies (Jackson ImmunoResearch; cat: 115-116-146) and, subsequently, incubated with anti-CD3e-biotin (cat: 553060), anti-CD19-biotin (cat: 553784), and anti-CD49b-biotin (cat: 553856) mAbs. Finally, after additional washes, cells were stained with anti-CD11c-APC (cat: 550261), anti-MHC-II-BV421 (cat: 562928), anti-CD8a-BB515 (cat: 564422) mAbs, and streptavidin-PerCP-Cy5.5 (BD Biosciences; cat: 551419). The CD11c⁺ MHC-II⁺ population was gated, and the binding of the recombinant proteins was evaluated through the median of fluorescence intensity (MFI) of PE for the CD11c⁺ CD8α⁺ and CD11c⁺ CD8α⁻ DCs.

For DC activation assays, polymixin B (200 μg/mL) was previously incubated with LPS (2 μg/mL), or gDE7 (60 μg/mL), gD (45 μg/mL), or E7 (15 μg/mL) proteins for 2 hours at 37°C (17). Isolated DCs (4 × 10⁵ cells/well) were incubated for 48 hours with the following combinations: LPS (100 ng/mL); polymixin B (10 μg/mL); LPS + polymixin B (100 ng/mL + 10 μg/mL); gDE7 + polymixin B (3 μg/mL + 10 μg/mL); gD + polymixin B (2.25 μg/mL + 10 μg/mL); and E7 + polymixin B (0.75 μg/mL + 10 μg/mL). Cells were stained with anti-CD11c-PE (cat: 553802), anti-MHC-II-FITC (cat: 553605), anti-CD8α-Alexa Fluor 700 (cat: 557959), and anti-CD40-APC (cat: 558695), anti-CD80-APC (cat: 560016), or anti-CD86-APC (cat: 565479) mAbs. For intracellular cytokine evaluation, DCs were incubated with the Golgi Stop Reagent (BD Biosciences) for the last 12 hours. DCs were fixed, permeabilized, and stained with anti-IL12-APC (cat: 554480) mAb. CD40 and CD86 MFI values are shown from gated CD11c⁺ CD8α⁺ and CD11c⁺ CD8α⁻ DCs. The percentages of IL12⁺ cells over all CD11c⁺ CD8α⁺ or CD11c⁺ CD8α⁻ DCs were determined. Unless stated differently, all antibodies were from BD Biosciences. Samples were acquired on a LSR Fortessa flow cytometer (BD Biosciences) and analyzed using the FlowJo software (Tree Star).

Blood DCs from healthy donors were separated by double centrifugation procedures with Ficoll-Hypaque and Percoll 51% (GE Healthcare). After washes, blood DCs and previously differentiated Mo-DCs were incubated with LPS (100 ng/mL); polymixin B (10 μg/mL); LPS + polymixin B (100 ng/mL + 10 μg/mL); gDE7 + polymixin B (3 μg/mL + 10 μg/mL); gD + polymixin B (2.25 μg/mL + 10 μg/mL); or E7 + polymixin B (0.75 μg/mL + 10 μg/mL). The cells were incubated with the Golgi Plug (10 μg/mL; BD Biosciences) reagent for 6 hours and stained with anti-HLA-DR V500 (cat: 561224), anti-CD11c V450 (cat: 560369), anti-CD14 APC (cat: 555399), anti-BDCA1 PE (Miltenyi Biotec; cat: 130-090-508), and anti-BDCA3 PE-Vio770 (Miltenyi Biotec; cat: 130-100-217) mAbs. Cells were fixed/permeabilized and stained with the anti-TNFα-FITC (cat: 554512) mAb. The percentages of TNFα⁺ cells over all Mo-DCs, BDCA3⁺, or BDCA1⁺ DCs were determined. Unless stated differently, all antibodies were from BD Biosciences. Samples were acquired on a LSR Fortessa flow cytometer (BD Biosciences) and analyzed using the FlowJo software (Tree Star).

One-way or two-way ANOVA followed by Bonferroni post-test was performed when individual data were compared. Log-rank tests were employed whenever survival curves were compared. P < 0.05 was considered statistically significant. Statistical analysis and graphical representation of data were performed using GraphPad Prism 5.0.

We have previously shown that repeated immunizations with non-adjuvanted gDE7 enhanced the E7-specific CD8⁺ T-cell responses and conferred partial therapeutic antitumor protection in mice transplanted with TC-1 cells (12). In this article, we sought to improve the therapeutic antitumor effect of the gDE7-based vaccine by co-administration of poly (I:C), an extensively used vaccine adjuvant with clinical trial records (18). Administration of four subcutaneous doses of gDE7 admixed with poly (I:C) resulted in complete antitumor protection to challenges with transplanted TC-1 cells (Supplementary Fig. S1A). The same result was achieved in mice immunized with two subcutaneous doses of this vaccine formulation (Fig. 1A). Under the same conditions, administration of two doses of non-adjuvanted gDE7 or purified E7 admixed with poly (I:C) conferred only 20% of therapeutic antitumor protection, whereas no protection was observed in mice immunized with the E7 antigen alone (Fig. 1A). On the basis of these results, subsequent immunization experiments were performed with the regimen of two subcutaneous doses of the vaccine formulations. The therapeutic antitumor effects of gDE7 admixed with poly (I:C) were also monitored in mice primed up to 10 days after the TC-1 cell transplantation, when tumors were already palpable. As shown in Fig. 1B, all mice immunized with the first dose of the adjuvanted gDE7 up to 7 days after challenge were tumor-free at the end of the follow-up period (60 days). On the other hand, when the interval between tumor cell transplantation and the vaccine priming was delayed by 10 days, the therapeutic antitumor protection declined to 40%.

We monitored the kinetics of CD8⁺ T-cell responses until 28 days after challenge. As shown in Fig. 1C, two subcutaneous doses of adjuvanted gDE7 induced the highest frequency of IFNγ-producing E7-specific CD8⁺ T cells in peripheral blood mononuclear cells (PBMC). The response peak was achieved 14 days after the administration of the second dose of the vaccine formulation. Despite the lower frequency, no statistical differences were found between mice immunized with the gDE7 alone and the adjuvanted gDE7. No E7-specific CD8⁺ T-cell response was detected in mice immunized with either E7 or E7 admixed with poly (I:C). Similar results were detected in mice immunized with four doses of the tested vaccine formulations (Supplementary Fig. S1B).

To further understand the role of the E7-specific CD8⁺ T cells on the antitumor protective immunity, we measured the in vivo cytotoxic activity of these lymphocytes induced by vaccination. The observed results indicated that, despite the similar performance in activation of E7-specific CD8⁺ T-cell responses, mice immunized with gDE7 or gDE7 admixed with poly (I:C) differed in the in vivo cytotoxic activities. While mice immunized with adjuvanted gDE7 exhibited a significant lysis of cells pulsed with the CD8⁺ T-cell–specific E7-derived peptide (∼60%), mice immunized with non-adjuvanted gDE7 showed a lower activation of E7-specific cytotoxic CD8⁺ T cells (∼25%; Fig. 1D). Residual or no in vivo cytotoxic activity was detected in mice immunized with E7 alone or E7 admixed with poly (I:C).

To test whether the residual LPS content of gDE7 or E7 recombinant proteins could affect the antitumor immunity conferred on mice transplanted with TC-1 cells, C57BL/6 TLR4 KO mice were immunized with the tested vaccine formulations. As previously observed with WT C57BL/6 mice, two subcutaneous doses of adjuvanted gDE7 conferred complete antitumor protection to challenged TLR4 KO mice (Supplementary Fig. S1C). Similarly, immunizations with adjuvanted gDE7 induced the highest frequency of IFNγ-producing E7-specific CD8⁺ T cells in TLR4 KO mice (Supplementary Fig. S1D). These results indicated that the presence of residual LPS did not affect the protective immunity conferred by the vaccine formulations.

To further understand the contribution of T-cell subsets in the therapeutic antitumor protection, CD4⁺ T-cell KO and CD8⁺ T-cell KO C57BL/6 mice were challenged with TC-1 cells and subsequently immunized with gDE7 admixed with poly (I:C). CD4⁺ T-cell KO mice immunized with adjuvanted gDE7 showed the same antitumor protection detected in WT isogenic mice, whereas, in contrast, CD8⁺ T-cell KO mice showed no protective immunity (Supplementary Fig. S1E). This observation is in accordance with previous observations reporting that protective immunity induced in animals immunized with E7-based antitumor vaccine formulation was CD8⁺ T-cell–dependent but CD4⁺ T-cell–independent (19—21). In addition, we also demonstrated that pre-existing immunity to gD, as generally observed in those previously exposed to HSV, did not hamper the therapeutic antitumor effects induced by gDE7 admixed with poly (I:C) (Supplementary Fig. S1F).

Collectively, these results demonstrated that the genetic fusion of E7 to gD and the combination of poly (I:C) synergistically enhanced the therapeutic antitumor protection to TC-1 cells challenge by activation of cytotoxic E7–specific CD8⁺ T cells. To further advance on the understanding of cellular mechanisms that support the antitumor effects induced by gDE7 admixed with poly (I:C), we characterized the specific features of the CD8⁺ T-cell populations specifically induced by the vaccine formulation.

Recently, it has been suggested that multifunctional CD8⁺ T cells are a predictive biomarker of therapeutic vaccines aiming the control of high-grade cervical lesions under clinical conditions (9). Multifunctional CD8⁺ T cells were evaluated by multiparametric flow cytometry 14 days after the second dose of the tested vaccine formulations (Fig. 2A). Immunizations with adjuvanted gDE7 induced higher frequencies of IFNγ-, TNFα-, or IL2-producing CD8⁺ T cells compared with the other tested vaccine formulations (Fig. 2B–D). Besides that, co-administration of poly (I:C) increased the frequencies of IFNγ-, TNFα-, or IL2—producing CD8⁺ T cells by approximately 4-, 28-, and 140-fold, respectively, regarding mice immunized with non-adjuvanted gDE7. Similarly, administration of gDE7 admixed with poly (I:C) increased the frequencies of IFNγ-, TNFα-, or IL2-producing CD8⁺ T cells by approximately 4-, 4-, and 9-fold, respectively, with regard to mice immunized with E7 admixed with poly (I:C) (Fig. 2B–D). Boolean combinations of cytokine-producing CD8⁺ T cells showed that immunizations with gDE7 admixed with poly (I:C) induced high frequency of multifunctional CD8⁺ T cells capable to produce IFNγ and TNFα simultaneously (Fig. 2E). We also observed a significant proportion of IL2⁺ CD8⁺ T cells from mice immunized with gDE7 admixed with poly (I:C) regarding the other tested vaccine formulations. Multifunctional CD8⁺ T cells capable to produce all three cytokines were also mainly detected in mice immunized with gDE7 admixed with poly (I:C) (Fig. 2E).

To further understand the differentiation of the E7-specific CD8⁺ T-cell populations involved in the control of tumor development, we determined the phenotype of these T-cell populations. Mice immunized with adjuvanted gDE7 elicited an E7-specific CD8⁺ T-cell response with a prevailing cell population with the effector memory phenotype (T_EM; CD44⁺CD62L^low; Fig. 2F). In addition, the frequency of tumor-infiltrating CD8⁺ T cells was significantly higher in mice immunized with adjuvanted gDE7 than in mice immunized only with gDE7 or the other tested immunization groups (Fig. 2G). Noticeably, the vast majority of tumor-infiltrating CD8⁺ T cells activated by immunization with adjuvanted gDE7 was E7-specific CD8⁺ T cells (∼80%; Fig. 2H). In contrast, immunization with E7 admixed with poly (I:C) promoted activation of E7-specific CD8⁺ T cells characterized by a central memory phenotype (T_CM; CD44⁺CD62L^high; Fig. 2F), with limited effector function and reduced ability to migrate into the tumor microenvironment (Fig. 2G and H). Thus, the present data demonstrate that the genetic fusion of E7 to gD protein and co-administration of poly (I:C) efficiently activated multifunctional E7-specific CD8⁺ T cells with an effector memory phenotype, compatible with the ability to promote tumor growth control.

Frequently, Treg cells expand in blood and spleen during tumor development and may impair antitumor immune responses at both induction and effector levels (22). We wondered whether immunization with the gDE7-based vaccines would affect the immune suppressive mechanisms driven by the tumor cells. Regarding the analysis of splenic Treg cells, sham-treated mice as well as mice treated with poly (I:C), E7 or E7 admixed with poly (I:C) displayed higher frequencies of splenic Treg cells (∼12%, 8%, 12%, and 11%, respectively). In contrast, mice treated with non-adjuvanted gDE7 or gDE7 admixed with poly (I:C) showed much lower frequencies of Treg cells (∼4% and 2.5%, respectively; Fig. 3A and B).

MDSCs also proliferate in lymphoid organs and tumor tissues in response to cytokines and other factors secreted by tumor cells (23). Under our experimental conditions, sham-treated mice as well as mice immunized with E7 or E7 admixed with poly (I:C) showed similar frequencies of splenic MDSCs (∼6.7%, 6.2%, and 5%, respectively). On the other hand, mice immunized with gDE7 or gDE7 admixed with poly (I:C) presented significantly lower frequencies of splenic MDSCs (∼2.5% for both immunization groups; Fig. 3C and D). Administration of poly (I:C) also caused a significant reduction of splenic MDSCs with regard to sham-treated mice. Collectively, these results demonstrated that the control of tumor development mediated particularly by gDE7 admixed with poly (I:C) vaccine formulation led to the impairment of splenic Treg cells and MDSC expansion.

To further understand the mechanisms involved with the antitumor effects of gDE7-based vaccines, we evaluated whether the recombinant protein could specifically activate DCs, particularly those directly involved with the cross-presentation of antigens to CD8⁺ T cells. For that purpose, we measured the interaction of the different recombinant proteins with CD11c⁺ CD8α⁺ and CD11c⁺ CD8α⁻ DCs (Fig. 4A). As shown in Fig. 4B and C, both gDE7 and gD bound with greater affinity to the CD11c⁺ CD8α⁺ DCs, which are particularly relevant for the activation of CD8⁺ T-cell populations. In contrast, the purified E7 antigen showed low specific binding to both DC subsets.

The activation of DCs mediated by recombinant proteins promoted upregulation of CD40 and CD86 expression and IL12 production (Fig. 4D–F). gDE7 and gD activated CD11c⁺ CD8α⁺ DCs more efficiently than CD11c⁺ CD8α⁻ cells. As expected, induction of CD40 and CD86 expression on CD11c⁺ CD8α⁺ DCs did not occur after exposure to E7 (Fig. 4D and E). The frequency of IL12-producing CD11c⁺ CD8α⁺ DCs increased after incubation with gDE7 and gD (Fig. 4F). To rule out any effect associated with residual LPS present in the protein preparations, we repeated the activation experiment in the presence of excess endotoxin and polymixin B at the same concentration used with the recombinant proteins. The results demonstrated that the amount of polymixin B present in the assays was sufficient to reduce any effect associated with residual LPS (Supplementary Fig. S2A–S2C). Collectively, these results indicate that the gD component specifically targeted the E7 antigen to the CD11c⁺ CD8α⁺ DCs and promoted differential activation of a murine DC subset specialized in antigen cross-presentation.

As an additional step toward the clinical testing of the gDE7-based vaccines, we measured the activation of human DCs under in vitro conditions. To demonstrate the direct effects of the recombinant proteins on human DCs, BDCA1⁺ and BDCA3⁺ DCs were obtained from healthy human donors (Fig. 5A), promptly stimulated with equimolar amounts of E7, gDE7, or gD and analyzed for TNFα production. In vitro differentiated Mo-DCs were also tested in these experiments. As shown in Fig. 5B, activation of BDCA3⁺ DCs mediated by gDE7 and gD was significantly higher than BDCA1⁺ DCs. In addition, gDE7 induced higher frequencies of TNFα-producing BDCA3⁺ DCs regarding incubation with E7 (Fig. 5B). Considering that the frequency of BDCA3⁺ DCs in human PBMCs is approximately 10-fold lower than BDCA1⁺ DCs (Fig. 5A; ref. 24), our results demonstrate that both gDE7 and gD promote the specific activation of the DCs directly involved with antigen cross-presentation to CD8⁺ T cells also in humans. Similar to the results observed in mice, the presence of polymixin B efficiently counteracted the effects associated with contaminant LPS but did not interfere with the effects associated with the recombinant proteins (Supplementary Fig. S2D). Taken together, these results demonstrated that the antitumor activity of gDE7 could be ascribed to the adjuvant effects mediated by the gD component, which activates a specific DC population involved with antigen cross-presentation to CD8⁺ T cells both in mice and in human samples.

In the present study, we evaluated the adjuvant effects conferred by the HSV-1 gD as an antigen delivery platform for the development of therapeutic vaccines against HPV-16–associated tumors. The approach involves the parenteral administration of gDE7 admixed with poly (I:C) to mice previously transplanted with TC-1 cells. The complete therapeutic antitumor effect was achieved after two doses of the vaccine formulation, and the protective immunity was ascribed to the activation of a robust cytotoxic E7–specific response mediated by multifunctional and effector memory CD8⁺ T cells. In addition, the control of tumor development mediated by the gDE7-based vaccine halted the expansion of systemic immunosuppressive cells, such as Treg cells and MSDCs. Of particular relevance was the finding that gDE7 activated a specific DC subset involved with antigen cross-presentation, both in mice and in humans, to CD8⁺ T cells. Altogether, the present study expands the knowledge about the cellular mechanisms of the immune responses induced by gD-based vaccines and further supports the use of this antigen delivery/adjuvant platform in the development of antitumor immunotherapeutic approaches.

Vaccine-induced antitumor protection in the TC-1 tumor cell model strongly correlates with the activation of cytotoxic CD8⁺ T-cell responses (20, 25). However, purified E7 protein is poorly immunogenic and usually does not induce an effective CD8⁺ T-cell response capable to control tumor growth in TC-1 transplanted mice (12, 19). The genetic fusion of E7 to non-related proteins is frequently applied to enhance the immunogenicity by increasing presentation via class-I/II MHC molecules (26, 27). Previous studies performed by our group showed that the genetic fusion of different antigens nearby the C-terminal region of HSV-1 gD improved the T- or B-cell responses to these antigens (11, 12, 28–30). The adjuvant role of gD has been correlated to the binding to the herpesvirus entry mediator (HVEM) receptor and the blockage of inhibitory signals mediated by B and T lymphocyte attenuator (BTLA) and CD160, which are natural HVEM ligands (31). In addition, the gD/HVEM interaction promotes NF-κB activation and cell survival by itself (32, 33), modulating a complex multidirectional signaling pathway that result in the activation of T-cell responses (34). Our data now indicate that the gD-mediated adjuvant effects may also involve a targeting mechanism to a specific DC subset specialized in antigen cross-presentation to CD8⁺ T cells.

The murine CD11c⁺ CD8α⁺ DCs efficiently cross-present exogenous antigens on MHC class-I molecules to CD8⁺ T cells (35). The present study demonstrated that gDE7 bound and activated the CD11c⁺ CD8α⁺ DCs more efficiently than the CD11c⁺ CD8α⁻ DCs. When activated, CD11c⁺ CD8α⁺ DCs are committed with IL12 production and activation of cytotoxic CD8⁺ T cells (36). Indeed, we demonstrated an enhancement of IL12 production by CD11c⁺ CD8α⁺ cells after incubation with gDE7. In addition, in vivo evidences showed that immunization with the gDE7 admixed with poly (I:C) induced a strong antitumor CD8⁺ T-cell response. Collectively, our results demonstrate that gD adjuvant effects involve targeting and activation of a specific mouse DC subset specialized in antigen cross-presentation, a key step on the mounting of antitumor CD8⁺ T-cell response.

Several studies have demonstrated that TLR engagement also improves antigen cross-presentation to CD8⁺ T cells (37). Shulz and colleagues showed that TLR3 signaling helps the recognition of double-stranded RNA [poly (I:C)] by CD11c⁺ CD8α⁺ DCs and enhances antigen cross-presentation to CD8⁺ T cells (38). As described elsewhere, co-delivery of an antigen and a TLR ligand to the same antigen-presenting cell improves the activation of T cells (39). In our study, the gD-mediated antigen delivery, leading to activation of CD11c⁺ CD8α⁺ DCs, and co-administration of poly (I:C) support a similar explanation for the synergistic antitumor effect induced by the tested vaccine formulation.

As discussed above, both gDE7 and poly (I:C) activate DCs specialized in antigen cross-presentation, providing all signals required for activation of an effective CD8⁺ T-cell response. Our study demonstrated that immunization with adjuvanted gDE7 induced a strong cytotoxic, multifunctional and effector memory CD8⁺ T-cell response. Activation of such specific CD8⁺ T-cell population has been correlated with the control of HPV-associated tumors based on the capability to migrate to peripheral tissues and to respond to antigen contact by immediate effector functions (40). Indeed, we found a significant amount of tumor-infiltrating E7-specific CD8⁺ T cells in mice immunized with gDE7 admixed with poly (I:C). This finding explain, at least in part, the strong vaccine-mediated therapeutic antitumor protection and corroborates clinical data showing that CD8⁺ T-cell infiltration positively correlates with lesion regression (5, 6).

The presence of HPV-16 oncoproteins–specific Treg cells in lymph nodes and tumor tissues from patients with cervical cancer raised the possibility that oncoprotein-based immunotherapies would result in Treg cell expansion and activation (22). Our study showed that the frequency of splenic Treg cells in mice immunized with adjuvanted gDE7 was similar to the frequency found in naïve mice (∼2%; ref. 41). This result indicates that immunization with adjuvanted gDE7 had no direct effect on Treg cell expansion but contributed to the induction of an effective antitumor CD8⁺ T-cell response. In addition to Treg cells, MDSCs also have a direct role on tumor-mediated immunosuppression (20). Similarly, immunizations with gDE7, regardless of co-administration of poly (I:C), impaired the expansion of splenic MDSCs. Altogether, these results indicated that the gD-mediated immunomodulatory effects influence different cell populations affecting both the effector cytotoxic responses and immunosuppressive mechanisms.

Human BDCA3⁺ DCs, the counterpart of the mouse CD11c⁺ CD8α⁺ DCs (24), produced elevated TNFα levels compared with Mo-DCs or BDCA1⁺ subsets when exposed to gDE7. The ability of human BDCA1⁺ and BDCA3⁺ DCs in antigen cross-presentation is linked to several factors including the nature of the antigen, the antigen-processing pathway, and the DC activation status (42). Regarding the antigen-processing pathways, antigens delivered to late endosomes are better cross-presented by BDCA3⁺ DCs, whereas antigens delivered in early endosomes are similarly cross-presented by both BDCA1⁺ and BDCA3⁺ DCs (43). Although not explored in the present study, the identification of the specific intracellular pathways ascribed to gDE7 processing/presentation in combination to TLR ligands represents a quite interesting topic and deserve further evaluation.

The use of DCs has been critical for the development of effective antitumor strategies (44, 45). As demonstrated here, immunization with adjuvanted gDE7 induced strong antitumor protection and promoted the activation of a DC population specifically involved with antigen cross-presentation and activation of CD8⁺ T cells. Recent studies report that the priming of specific T-cell responses occurs in different HPV-induced cancer patients, including cervical cancer (46), head and neck (47), and oropharyngeal squamous cell carcinoma (48). Nonetheless, an effective tumor control requires adequate immunologic stimuli that could afford the concomitant activation of cytotoxic responses and control of tumor-induced immunosuppressive mechanisms. Collectively, our data provide basis for an immunotherapeutic approach by using a DC targeting strategy that can impact the way HPV-induced cancer is currently treated.

No potential conflicts of interest were disclosed.

Conception and design: B.F.M.M. Porchia, A.C.R. Moreno, M.O. Diniz, J.A.M. Barbuto, L.C.S. Ferreira

Development of methodology: B.F.M.M. Porchia, A.C.R. Moreno, M.O. Diniz, L.H.T.M. de Andrade, J.A.M. Barbuto

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): B.F.M.M. Porchia, A.C.R. Moreno, M.O. Diniz, L.H.T.M. de Andrade, D.S. Rosa, J.A.M. Barbuto, S.B. Boscardin

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): B.F.M.M. Porchia, A.C.R. Moreno, R.N. Ramos, D.S. Rosa, J.A.M. Barbuto, S.B. Boscardin, L.C.S. Ferreira

Writing, review, and/or revision of the manuscript: B.F.M.M. Porchia, A.C.R. Moreno, M.O. Diniz, D.S. Rosa, J.A.M. Barbuto, S.B. Boscardin, L.C.S. Ferreira

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A.C.R. Moreno

Study supervision: L.C.S. Ferreira

Other (Financial support): A.C.R. Moreno

Other (Performed experiments using the human samples): R.N. Ramos

The authors thankfully acknowledge the technical support of E.G. Martins and C. Bertelli of Vaccine Development Laboratory, University of São Paulo, São Paulo, Brazil.

B.F.M.M. Porchia was a fellow from Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP), grant 2011/20917-0. A.C.R. Moreno was a fellow from FAPESP, grants 2015/16505-0 and 2016/00708-1 and from Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), grant 560713. R.N. Ramos was a fellow from FAPESP, grant 2011/08905-7. M.O. Diniz was a fellow from FAPESP, grant 2011/51218-0. L.H.T.M. de Andrade was a fellow from Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), grant 560486/2010-3. D.S. Rosa was a fellow from FAPESP, grant 2014/15061-8. SB Boscardin was a fellow from FAPESP, grant 2013/11442. J.A.M. Barbuto was a fellow from FAPESP, grant 2009/54599-5. L.C.S. Ferreira was a fellow from CNPq, grant 520931/1996-3.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.