Local HPV Recombinant Vaccinia Boost Following Priming with an HPV DNA Vaccine Enhances Local HPV-Specific CD8⁺ T-cell–Mediated Tumor Control in the Genital Tract

Persistent infection with an oncogenic human papillomavirus (HPV) is a necessary, but insufficient cause of cervical cancer (1), the third most common cancer in women worldwide (2). Despite the availability of prophylactic vaccines, to date, uptake has been uneven, and so HPV disease remains common. Although most infections are cleared without intervention or clinical sequelae, mechanisms of immune-mediated clearance in humans are not well understood. Emerging data from human cohorts demonstrate that not all high-grade dysplastic lesions in the cervix, cervical intraepithelial neoplasia 2/3 (CIN2/3), progress to invasive disease (3). Lesions associated with HPV16, the genotype most commonly associated with cancer, undergo complete regression in 20% to 25% of immune-competent women (3). Because expression of two viral proteins, E6 and E7, is functionally required for initiation and persistence of disease, they both represent rational, non-“self” antigenic targets for immune therapies (4).

Recent clinical data from a trial testing a heterologous DNA-prime, recombinant vaccinia vector-based boost (TA-HPV) demonstrate tissue localization of effector immune responses following peripheral, intramuscular vaccination in the deltoid muscles prior to standard therapeutic resection (5). Because memory T cells display pronounced tropism for the tissue in which they first encounter their cognate antigen, ongoing clinical trials are also testing the feasibility and immunogenicity of direct, intralesional vaccination in HPV16⁺ CIN2/3 (6). We developed a candidate therapeutic HPV vaccine, pNGVL4a-sig/E7(detox)/HSP70, based upon a naked DNA expressing a chimeric protein consisting of a signal peptide (sig) linked to HPV-16 E7 antigen and also heat shock protein 70 (HSP70), described previously (7). Intramuscular administration of pNGVL4a-sig/E7(detox)/HSP70 DNA vaccine has been well tolerated by patients with HPV16⁺ CIN2/3. However unlike the preclinical murine models, vaccination with this construct in humans elicited weak systemic E7-specific CD8⁺ T-cell responses that did not correlate with lesion regression (8). TA-HPV is a recombinant vaccinia virus vaccine that encodes HPV-16/18 E6 and E7 proteins. In humans, TA-HPV has elicited limited detectable systemic HPV-specific cellular immune responses (9–13). However, peripheral vaccination with this construct has elicited striking changes in the target lesions, suggesting that vaccine-induced immune responses are capable of trafficking to the site of antigen (5).

Many investigators have demonstrated that in preclinical murine models, heterologous vaccination regimens consisting of DNA vaccine priming, followed by boosting with viral vector constructs, elicit effector responses that are far greater in magnitude than vaccination with either DNA alone or viral vectors alone (14–19). In our TC-1 model, vaccination with an E7-expressing DNA vaccine followed by boosting with E7 recombinant vaccinia virus also elicited greater immune responses compared with repeat immunization with either vaccine (14). In humans, Maldonado and colleagues demonstrated that intramuscular vaccination with two doses of pNGVL4a-sig/E7(detox)/HSP70 DNA, followed by TA-HPV, was well tolerated as well as immunogenic in patients with HPV16⁺ CIN2/3 (NCT00788164; refs. 5, 20). This vaccination regimen also elicited effector memory CD8 T-cell infiltrates in and around the CIN lesions, a phenotype associated with lesion clearance, suggesting that systemic immunization can elicit local immunity.

Tissue-resident memory T cells (Trm) have been shown to play a central role in the local control of infection, including in the genital tract (for reviews see refs. 21, 22). Site-specific vaccination in the setting of established disease may present a strategy for induction of therapeutic immunity for HPV⁺ mucosal tumors (for review see ref. 23). Indeed, in a preclinical murine model using an orthotopic HPV⁺ tumor, intranasal mucosal administration of a candidate head and neck cancer vaccine targeting E7 generated significantly enhanced therapeutic effects and mucosal-targeted antigen-specific CD8⁺ T-cell responses compared with intramuscular administration of the vaccine (24).

Here, we examine whether intramuscular vaccination with E7-expressing DNA followed by intratumoral vaccination with TA-HPV would elicit enhanced clearance of vaginal HPV16⁺ tumors, as compared with peripheral boosting with intramuscular injection of TA-HPV, and dissect the signals attracting the HPV-specific CD8⁺ T cells to the genital tract in an orthotopic model of HPV-associated cervical cancer.

Six- to 8-week-old female C57BL/6 mice were purchased from the NCI (Frederick, MD). β7 integrin–deficient C57BL/6 mice (C57BL/6-Itgb7^tm1cgn/J) were purchased from The Jackson Laboratory. Rag1 knockout OT-1 transgenic C57BL/6 mice were purchased from Taconic. All animal procedures were performed according to approved protocols and in accordance with recommendations for the proper use and care of laboratory animals.

TC-1 cells expressing the HPV16 E6-E7 proteins (25) and the TC-1 cells expressing the firefly luciferase gene (TC-1 luc) were developed in our laboratory and have been described previously (26).

FITC, phycoerythrin (PE), and APC-conjugated anti-mouse CD8a (clone 53.6.7), FITC-conjugated anti-mouse IFNγ (clone XMG1.2), PerCP-Cy5.5–conjugated anti-mouse CD103 (clone M290) antibodies were purchased from BD Pharmingen. APC-conjugated anti-mouse α4β7 (clone DATK32) was purchased from eBiosciences. PE-conjugated anti-mouse S1P1/EDG-1 (clone 713412) was purchased from R&D Systems. Purified anti-mouse CD4 antibody (clone GK1.5) was purchased from Bio X Cell. FITC-conjugated anti-mouse CCR9 (clone CW-1.2) and purified rat anti-mouse MAdCAM-1 antibody (clone MECA-367) were purchased from BioLegend. Target retrieval solution, peroxidase block, horseradish peroxidase (HRP)–conjugated rabbit anti-rat secondary antibody, anti–rabbit-labeled polymer, and diaminobenzidine (DAB) were all purchased from DAKO. HPV16 E7aa49-57 and HPV16 E7aa43-62 peptides were synthesized by GenScript at purity over 80%. CpG-ODN 1826 was synthesized by Invitrogen. PE or APC-conjugated, HPV16 E7aa49-57 peptide-loaded H-2D^b tetramer was provided by National Institute of Allergy and Infectious Diseases tetramer core facility. Retinoic acid receptor beta (RAR-β) antagonist, LE540, was purchased from STEMCELL Technologies. All trans retinoic acid (RA) was purchased from Sigma. A CD11c-positive selection kit was purchased from Miltenyi Biotech. Carboxyfluorescein succinimidyl ester (CFSE) was purchased from BD Pharmingen. Ammonium chloride solution (ACK) was purchased from Quality Biological INC.

The DNA vaccine, pNGVL4a- sig/E7(detox)/HSP70, used in this study has been described previously (7). A recombinant vaccinia virus expressing HPV16 and HPV18 E6 and E7, TA-HPV, has also been described previously (9). Mice were immunized by cervicovaginal vaccination (lateral wall of the cervicovaginal tract) or intramuscular (biceps femoris muscle of hind leg) injection on day 0 with 50 μg of pNGVL4a-sig/E7(detox)/HSP70 DNA vaccine and boosted with 1 × 10⁷ PFU of TA-HPV 7 days later. For peptide vaccination, 10 μg of HPV16 E7aa43-62 peptide was mixed with 1 μg of CpG1826 before the injection. For subcutaneous vaccination, the vaccines were injected subcutaneously at the base of tail. The total volume injected was 50 μL for both routes. Mice were anesthetized before immunization.

To deplete CD4⁺ T cells, mice were injected with 100 μg of anti-mouse CD4 antibody (Clone GK1.5) for 3 days via intraperitoneal injection. The depletion was maintained through the experiment by injecting the antibody once per week. The success of depletion was confirmed by staining peripheral blood cells with an anti-mouse CD4 antibody that recognizes a different epitope than the depleting antibody.

Peripheral blood mononuclear cell (PBMC) were obtained by bleeding from the tail vein of mice and after obtained red blood cells were lysed with ACK lysis buffer. Vaginal (cervical and vaginal tissues) cell suspensions were obtained by enzymatic dispersion in RPMI 1640 digestion buffer for 1 hour at 37°C while shaking. Cervicovaginal cells were passed through a 70-μm cell strainer (BD). Iliac lymph node and spleen cell suspensions were mechanically disrupted and filtered through a 70-μm cell strainer. PBMC, cervicovaginal, spleen, and lymph node cell suspensions were washed in RPMI/FBS 2% and freed from erythrocytes by treatment with ACK.

For cell surface staining, single cells from spleen, lymph node, or cervicovaginal tissue were resuspended in 50 μL of FACS buffer (PBS + 0.5% BSA). To avoid nonspecific antibody binding through surface Fc receptor, cells were preincubated with CD16/32 mouse BD Fc Block (BD pharmingen). Fluorochrome-conjugated anti-mouse antibodies or PE or APC-conjugated HPV16 E7aa49-57 peptide-loaded H-2 D^b tetramer were added to the cells and incubated at 4°C for 1 hour. Cells were acquired with FACSCalibure flow cytometer, and the data were analyzed with CellQuest (Becton Dickinson Immunocytometry System) or FlowJo (FlowJo) software.

To detect HPV16 E7–specific CD8⁺ T-cell responses by IFNγ intracellular staining, splenocytes or single vaginal tissue cells were isolated using methods described previously (27), then stimulated with HPV16 E7aa49-57 peptide (1 μg/mL) in the presence of GolgiPlug (BD Pharmingen) at 37°C overnight. The stimulated cells were stained with PE-conjugated rat anti-mouse CD8a monoclonal antibody. Cells were subjected to intracellular cytokine staining using the Cytofix/Cytoperm Kit according to the manufacturer's instructions. Intracellular IFNγ was stained with FITC-conjugated rat anti-mouse IFNγ monoclonal antibody. Flow cytometry analysis was performed using FACSCalibur and analyzed with either CellQuest or FlowJo software.

To establish orthotopic cervicovaginal cancers, 2 × 10⁴ of TC-1 luc cells were injected into the vaginal wall of female wild-type or β7 integrin–deficient C57BL/6 mice. Three days later, mice were vaccinated with 50 μg of pNGVL4a-sig/E7(detox)/HSP70 DNA in 50-μL volume by either cervicovaginal vaccination (lateral wall of the cervicovaginal tract) or intramuscular injection. Seven days later, the mice were boosted with 1 × 10⁷ PFU of TA-HPV in 50-μL volume by either cervicovaginal vaccination or intramuscular injection. Genital tumor growth was monitored once a week by bioluminescence using a Xenogen imaging system. Briefly, D-Luciferin was dissolved to 7.8 mg/mL in PBS, filter sterilized, and stored at −80°C. Mice were given D-luciferin by i.p. injection (200 μL/mouse, 75 mg/kg) and anesthetized with isoflurane. In vivo bioluminescence imaging for luciferase expression was conducted on a cryogenically cooled IVIS system using Living Image acquisition and analysis software (Xenogen). Mice were placed onto the warmed stage inside the light-tight camera box with continuous exposure to 1% to 2% isoflurane. Images were acquired 10 minutes after D-luciferin administration and imaged for 2 minutes. The levels of light from the bioluminescent cells were detected by the IVIS imager, integrated, and digitized. The region of interest from displayed images was designated around the vagina and quantified as total photon counts using Living Image 2.50 software (Xenogen).

Cervicovaginal tissues were harvested en bloc from vaccinated mice 7 days after the last immunization. Paraffin-embedded tissue sections were prepared by the Department of Pathology at Johns Hopkins University. The sections were deparaffinized in xylene, and heat-based antigen retrieval was performed for 30 minutes using target retrieval solution followed by peroxidase block. The sections were incubated with rat anti-mouse MAdCAM-1 antibody at 4°C overnight, followed by HRP-conjugated secondary rabbit anti-rat antibody at room temperature for 1 hour. The sections were then incubated with anti–rabbit-labeled polymer at room temperature for 30 minutes. DAB served as a chromogen, and Gill's hematoxylin was used as a counterstain.

Splenocytes from intramuscular pNGVL4a-sig/E7(detox)/HSP70 DNA primed and cervicovaginal TA-HPV vaccinia virus boosted C57BL/6 mice, either wild-type or β7 integrin–deficient, were labeled with 0.5 μmol/L of CFSE. CFSE-labeled splenocytes (4 × 10⁷) were then adoptively transferred into wild-type C57BL/6 mice through retro-orbital injection. One day after transfer, splenocytes and cervicovaginal cells were prepared from recipient mice. The cells were stained with PE-conjugated anti-mouse CD8a antibody and acquired with FACSCalibur. CD8⁺CFSE⁺ cells were analyzed with FlowJo software.

Single cells from spleen and vaginal tissue were prepared as described above. CD11c⁺ dendritic cells were positively isolated from single cells of spleen and vagina tissue using CD11c magnetic beads (Miltenyi Biotec) according to the manufacturer's protocol. CD11c⁺ cells were resuspended in complete RPMI medium and counted.

Naïve OVA-specific OT-1 CD8⁺ T cells were isolated from Rag1 knockout OT-1 TCR transgenic mice and labeled with 0.5 μmol/L of CFSE. 1 × 10⁵ of these CFSE-labeled OT-1 T cells were plated into each well of 48-well plate and mixed with purified CD11c⁺ DCs (described above) at a DC: T-cell ratio of 1:5 in the presence of OVA protein (25 μg/mL) and IL2 (20 IU/mL). As a positive control, retinoic acid was added to the cultures at the concentration of 50 ng/mL. After 4 days, the expression of α4β7 on proliferating OT-1 cells was evaluated by flow cytometry analysis. For the retinoic acid inhibition assays, the DC/T cell cocultures were incubated in the presence of 1 μmol/L of RAR-β antagonist, LE540.

All data are expressed as mean ± SD and are representative of at least two separate experiments. Comparisons between individual data points were made using the Student t test. The nonparametric Mann–Whitney test was used for comparing two different groups. All P values <0.05 were considered significant.

Chen and colleagues demonstrated the synergistic induction of HPV16 E7–specific IFNγ⁺ CD8⁺ T-cell responses by heterologous prime-boost vaccination in C57BL/6 mice administered naked pcDNA3-sig/E7/LAMP-1 DNA intramuscularly and a recombinant vaccinia virus expressing sig/E7/LAMP-1 delivered intraperitoneally. We sought to confirm that a heterologous prime-boost strategy utilizing vaccines previously utilized in the clinic (pNGVL4a-sig/E7(detox)/HSP70 DNA and TA-HPV vaccinia virus) is also effective therapeutic vaccine approach. Three regimens were evaluated and compared with unvaccinated controls: (i) naïve C57BL/6 mice (five per group) were vaccinated intramuscularly with pNGVL4a-sig/E7(detox)/HSP70 DNA at weeks 0 and 1, followed by a single TA-HPV vaccination at week 2, or (ii) received homologous prime-boost vaccination comprising repeat vaccination with either pNGVL4a-sig/E7(detox)/HSP70 DNA alone, or (iii) TA-HPV alone. The spleens were harvested from each group of immunized mice at week 3, and the fraction of splenocytes derived from each mouse comprising CD8⁺ T cells producing IFNγ in response to E7 peptide stimulation for 24 hours was measured by intracellular cytokine staining (Fig. 1). The heterologous prime-boost regimen generated a stronger E7-specific IFNγ+ CD8⁺ T-cell response compared with repeat vaccination with either pNGVL4a-sig/E7(detox)/HSP70 DNA alone or TA-HPV alone.

HPV-specific T cells must traffic to the genital tract to clear intraepithelial neoplasia of the vagina or cervix. Therefore, we sought to elicit E7-specific CD8⁺ T cells in cervicovaginal tissues of mice via local vaccination. After a single priming vaccination with pNGVL4a-sig/E7(detox)/HSP70 DNA, C57BL/6 mice were boosted 7 days later by TA-HPV administered either by cervicovaginal or intramuscular vaccination. One week after TA-HPV vaccination, systemic and local E7-specific CD8⁺ T-cell responses were analyzed by flow cytometry after staining with E7 peptide-loaded H-2D^b tetramer of splenocytes or cells recovered from the cervicovaginal tissues, respectively. As shown in Fig. 2A and B, mice receiving either cervicovaginal or intramuscular heterologous prime-boost vaccination generated robust systemic E7-specific CD8⁺ T-cell responses. Intramuscular delivery generated a significantly higher systemic E7-specific CD8⁺ T-cell responses than cervicovaginal delivery. In contrast, mice that received the cervicovaginal heterologous prime-boost regimen generated significantly higher number of total number of CD8⁺ T cells as well as E7-specific CD8⁺ T cells in the cervicovaginal tract compared with mice vaccinated intramuscularly (Fig. 2C–E). The total number of CD8⁺ T cells in the cervicovaginal tissues of mice that received intramuscular vaccination was not significantly different from naïve mice, indicating that intramuscular administration did not lead to increase of CD8⁺ T cells in the cervicovaginal tract, regardless of E7 specificity.

We also characterized the presence of E7-specific CD8⁺ T cells in the spleen and cervicovaginal tract by intracellular cytokine staining for IFNy followed by flow cytometry analysis. In contrast with the measurement of total E7-specific CD8⁺ T-cell responses by staining with E7 peptide-loaded H-2D^b tetramer of splenocytes (Fig. 2A and B), no significant difference in the systemic E7-specific IFNγ⁺ CD8⁺ T-cell response was observed upon assay by intracellular cytokine staining of peptide-stimulated splenocytes from mice vaccinated with either the intramuscular or cervicovaginal heterologous prime-boost regimen (Fig. 2F and G). However, cervicovaginal heterologous prime-boost vaccination induced a significantly higher (8-fold) E7-specific IFNγ⁺ CD8⁺ T-cell response in murine cervicovaginal tissue as compared with intramuscular administration (Fig. 2H and I). Taken together, our data generated from two different immunologic assays indicate that compared with intramuscular administration, cervicovaginal vaccination with pNGVL4a-sig/E7(detox)/HSP70 DNA and TA-HPV generates significantly more E7-specific CD8⁺ T-cell infiltrates in the cervicovaginal tissues, without compromising the generation of E7-specific IFNγ⁺ CD8⁺ T-cell responses in the spleen.

We next examined whether it is required to administer both the pNGVL4a-sig/E7(detox)/HSP70 DNA priming vaccination and the TA-HPV boost ICV to induce both a robust systemic E7-specific CD8⁺ T-cell response and its recruitment in to cervicovaginal tissues. For the induction of a systemic E7-specific CD8⁺ T-cell response, intramuscular administration of the pNGVL4a-sig/E7(detox)/HSP70 DNA priming vaccination was more effective than cervicovaginal delivery before the cervicovaginal TA-HPV booster vaccination, and not significantly different than with intramuscular administration of the TA-HPV booster vaccination (Supplementary Fig. S1A and S1B). This may reflect an improved DNA uptake and E7 presentation by muscle cells as compared with cervicovaginal tissues. Indeed, cervicovaginal vaccination with the pNGVL4a-sig/E7(detox)/HSP70 DNA did not effectively prime the E7-specific CD8⁺ T-cell response to an intramuscular booster vaccination with TA-HPV (Supplementary Fig. S1A and S1B).

When examining the numbers of E7-specific CD8⁺ T cells in the genital tract of the immunized mice (Supplementary Fig. S1C and S1D), significantly higher number of E7-specific CD8⁺ T cells was observed in the cervicovaginal tract in the groups that received the cervicovaginal TA-HPV booster vaccination as compared with intramuscular vaccination, regardless of the route by which the DNA priming immunization was administered. Interestingly, after cervicovaginal vaccination boosting with TA-HPV, a higher number of E7-specific CD8⁺ T cells was detected in the genital tract of the intramuscular primed mice as compared with those primed with the pNGVL4a-sig/E7(detox)/HSP70 DNA by ICV administration (Supplementary Fig. S1C and S1D). We speculate that this may reflect a higher overall response because of improved DNA vaccine delivery in muscle as compared with cervicovaginal tissues.

The growth of luciferase-expressing TC-1 (TC-1 luc) tumor in the genital tract can be readily monitored by imaging of bioluminescence, and it provides a model of HPV16⁺ high-grade CIN/microinvasive cervical cancer. Therefore, we assessed the impact of route of administration of the prime-boost vaccination regimen upon control of luciferase-expressing TC-1 tumor established in the genital tract. TC-1 luc cells were injected submucosally into the vaginal wall of C57BL/6 mice. One day later, groups (n = 5) of these mice were immunized via either intramuscular or cervicovaginal administration with pNGVL4a-sig/E7(detox)/HSP70 DNA and 7 days later with TA-HPV by the same route. Mice were monitored for tumor growth by luminescence imaging on day 3 and day 17 after tumor challenge. The imaging on day 3 confirmed that, as shown in Fig. 3A and B, cervicovaginal vaccination with pNGVL4a-sig/E7(detox)/HSP70 DNA and TA-HPV exhibited significantly better control of the TC-1 luc tumor, as measured by decreased luminescence in the genital tract, on day 17 compared with mice receiving intramuscular vaccination or no vaccination. Furthermore, most mice which had cervicovaginal vaccination survived past 80 days, whereas all control mice and those vaccinated intramuscularly died before 30 days (Fig. 3C). These data indicate that cervicovaginal vaccination better controls HPV⁺ tumor in the genital tract and prolongs survival compared with vaccination through intramuscular injection.

In order to determine whether the E7-specific CD8⁺ T cells induced by the prime-boost regimen were Trms attracted to the genital tract, we evaluated them for the expression of tissue-specific molecules α4β7 and CCR9. α4β7 is a mucosa-associated homing integrin, is known to be expressed by T cells in the cervicovaginal tract of humans (28), and functions by interacting with Mucosal Addressin Cell Adhesion Molecule-1 (MAdCAM-1; for review, see ref. 29). Also involved in the homing and retention of lymphocytes in mucosal tissue is the chemokine receptor CCR9 whose ligand is CCL25, which is commonly expressed in the epithelium of respiratory, gastrointestinal, and urogenital tissues (for review, see ref. 23). As shown in Fig. 4A–D, naïve mice which were vaccinated in the cervicovaginal tissues with pNGVL4a-sig/E7(detox)/HSP70 DNA and TA-HPV 1 week later had a significantly higher percentage of E7-specific CD8⁺ T cells expressing α4β7 or CCR9 among all E7 tetramer-positive cells in the cervicovaginal tract as compared with mice vaccinated intramuscularly with this regimen. Likewise, a greater fraction of E7-specific CD8⁺ T cells expressing α4β7 or CCR9 were detected in the regional iliac lymph node of mice vaccinated in cervicovaginal tissue as compared with those vaccinated intramuscularly (Fig. 4E and F). These data indicate that intravaginal boost vaccination is more effective than intramuscular vaccination in generating antigen-specific CD8⁺ T cells that express α4β7 or CCR9 in the cervicovaginal tract.

Because MAdCAM-1 is an important addressin for nonlymphoid tissue-specific T-cell homing programs (23), we sought to determine whether cervicovaginal vaccination can induce local expression of MAdCAM-1 on the cervicovaginal tissue. The genital tracts were excised from mice vaccinated either within cervicovaginal tissues or intramuscularly with pNGVL4a-sig/E7(detox)/HSP70 DNA followed by TA-HPV, and their tissues were fixed and analyzed by immunohistochemistry staining. As shown in Supplementary Fig. S2, mice receiving cervicovaginal vaccination were associated with significantly higher expression as indicated by higher amount of positive brown staining of MAdCAM-1 within plump high endothelial venule cells in the cervicovaginal tissue compared with naïve mice or intramuscular-vaccinated mice. These data suggest that cervicovaginal vaccination with this heterologous prime-boost regimen induces the expression of addressin molecules in high endothelial venules of the cervicovaginal tract, which may further assist in homing of antigen-specific CD8⁺ T cells by interacting with integrin α4β7.

Since our ICV heterologous prime-boost regimen generated E7-specific CD8⁺ T cells that express α4β7 and induced increased expression of MAdCAM-1 in high endothelial venules of cervicovaginal tissue, we reasoned that α4β7 played a significant role in the recruitment of E7-specific CD8⁺ T cells to the genital tract. To test this, we administered the cervicovaginal heterologous prime-boost regimen in wild-type mice and Itgb7^−/− mice. As shown in Fig. 5A and B, cervicovaginal vaccination generated similar number of E7-specific CD8⁺ T cells in the spleen in wild-type mice compared with mice deficient of α4β7. However, cervicovaginal vaccination induced a significantly higher number of E7-specific CD8⁺ T cells in the cervicovaginal tissue of wild-type mice compared with mice lacking α4β7 (Fig. 5C and D). Furthermore, in experiments to treat genital TC-1 luc tumor, cervicovaginal heterologous prime-boost vaccinations effectively control tumor growth in wild-type mice, whereas α4β7-deficient mice showed significant tumor growth 17 days after tumor challenge (Fig. 5E and F). In addition, as shown in Fig. 5G, 80% of wild-type mice receiving cervicovaginal vaccination survived past 60 days, whereas all the α4β7-deficient mice died within 30 days after tumor challenge. These results suggest that α4β7 is responsible for the increase in local antigen-specific CD8⁺ T cells in the cervicovaginal tract. Furthermore, the findings suggest that it is not a direct killing effect by the vaccinia virus in the genital tract that is main mechanism responsible for local tumor control, but rather the recruitment of antigen-specific CD8⁺ T cells and immunologic control.

We sought to further assess the role of α4β7 in the traffic of CD8⁺ T cells into the cervicovaginal tract. Wild-type and Itgb7^−/− mice were vaccinated within cervicovaginal tissues with the heterologous prime-boost regimen, and 7 days after the last immunization, lymphocytes were isolated from the spleens and labeled with CFSE. The CFSE-labeled T cells were adoptively transferred retro-orbitally into WT naïve mice, and one day after transfer, T cells in spleen and cervicovaginal tract were isolated and analyzed by flow cytometry. Expression of α4β7 on the isolated T cells was detected through anti-α4β7 antibody staining (Supplementary Fig. S3A). As shown in Supplementary Fig. S3B and S3C, there is no significant difference in the numbers of adoptively transferred wild-type CD8⁺ T cells compared with the number of transferred α4β7-deficient CD8⁺ T cells in the spleen of the recipient mice. In contrast, a significantly higher number of adoptively transferred wild-type CD8⁺ T cells compared with number of transferred α4β7-deficient CD8⁺ T cells was detected in the cervicovaginal tissues of the recipient mice (Supplementary Fig. S3C and S3D). These data indicate that the expression of α4β7 is responsible for the trafficking of CD8⁺ T cells into the cervicovaginal tract, and that the increased number of E7-specific CD8⁺ T cells in the cervicovaginal tract observed is mainly contributed by T-cell homing mediated by α4β7.

Previously, it has been shown that retinoic acid can induce expression of α4β7 on CD8⁺ T cells (30–33). It has also been reported that dendritic cells in the gut but not in the spleen express retinal dehydrogenase (RALDH), a class I aldehyde dehydrogenase (ALDH1), which is responsible for the production of retinoic acid (34). We hypothesized that dendritic cells in the cervicovaginal tract produce retinoic acid, which induces expression of α4β7 in CD8⁺ T cells that have differentiated to a tissue-resident phenotype. We cocultured CD11c⁺ dendritic cells isolated from the spleen or the cervicovaginal tract with CFSE-labeled OT-1 cells at a ratio of 1:5 and stimulated with ovalbumin (OVA) protein and IL2. Unstimulated OT-1 cells served as negative control, whereas stimulated OT-1 cells cocultured with retinoic acid served as positive control. After 4 days, α4β7 expression on OT-1 cells was analyzed by flow cytometry. As shown in Fig. 6A and B, a significantly higher number of OT-1 cells expressed α4β7 when cocultured with dendritic cells isolated from the cervicovaginal tract compared with coculturing with dendritic cells isolated from spleens. These data suggest that, under these conditions, dendritic cells from the cervicovaginal tract are capable of inducing α4β7 expression in CD8⁺ T cells, whereas dendritic cells from the spleen are not. We then assessed the role of retinoic acid in the induction of α4β7 expression. LE540, a retinoic acid receptor inhibitor, was added to the OT-1 cell culture to block the effects of retinoic acid. Blockade of retinoic acid stimulation resulted in a reduction in the expression of α4β7 in activated OT-1 cells (Fig. 6C and D). This result suggests that retinoic acid is responsible for triggering expression of α4β7 in CD8⁺ T cells, consistent with previous studies (30–33). Taken together, our data indicate that dendritic cells in the cervicovaginal tract can induce expression of α4β7 in CD8⁺ T cells in a retinoic acid–dependent manner.

In the current study, we identified a heterologous prime-boost regimen that induces vaccine antigen-specific CD8⁺ T-cell immune responses and therapeutic effect against an HPV16 E6/E7⁺ orthotopic tumor model in the cervicovaginal tract. Both the pNGVL4a-sig/E7(detox)/HSP70 naked DNA vaccine and TA-HPV have been used safely, either alone or together, in clinical trials (5, 8–13). Indeed, they are being tested in sequence in an ongoing clinical trial, in women with HPV16⁺ CIN2/3 (20). In our murine model, cervicovaginal delivery elicits greater intensity of E7-specific CD8⁺ T cells than intramuscular administration of TA-HPV induction in cervicovaginal tissues. Furthermore, the enhanced local E7-specific CD8⁺ T-cell responses generated by cervicovaginal route of vaccine boost extend to the E7 peptide vaccine system (Supplementary Fig. S4). Furthermore, in mice bearing established E6/E7-expressing cervicovaginal tumors, cervicovaginal vaccination, presumably adjacent to the tumor, produces significantly better tumor control. Indeed, subcutaneous vaccination of the DNA prime followed by vaccinia vaccine boost also elicited a weaker E7-specific CD8⁺ T-cell response compared with cervicovaginal vaccination (Supplementary Fig. S8). This result further supports the necessity of local administration of the vaccines in the cervicovaginal tract. Importantly, DNA prime is necessary to elicit the enhanced antitumor responses, as homologous prime-boost with only TA-HPV results in significantly weaker tumor control in mice compared with heterologous DNA prime followed by vaccinia boost (Supplementary Fig. S5). Of note, cervicovaginal administration of the DNA vaccine was not required; in fact, intramuscular priming generated a better response than ICV.

To assess whether CD4⁺ cells play a role in the observed enhancement of local E7-specific CD8⁺ T-cell responses by our DNA prime followed by vaccinia boost regimen, we vaccinated CD4 T cells depleted mice with our vaccines. As shown in Supplementary Fig. S6, CD4 depletion almost completely abolish the generation of E7-specific CD8⁺ T cells by the DNA prime followed by vaccinia boost vaccination in the spleen, draining lymph nodes, and the cervicovaginal tissues. These results indicate that CD4 cells play a critical role in eliciting E7-specific CD8⁺ T-cell responses both locally and systemically.

The pNGVL4a-sig/E7(detox)/HSP70 naked DNA vaccine has previously been administered intramuscularly in clinical trials (8). Although TA-HPV has previously been administered either intramuscularly (5), or intradermally (9, 10), the cervicovaginal route of administration has not been tested in humans. However, cervical administration of a recombinant E2 modified vaccinia Ankara (MVA) was well tolerated and was associated with lesion regression in a phase II trial with women with HPV disease (35). Taken together, the cervicovaginal route of administration of therapeutic HPV vaccinia virus vaccine such as TA-HPV may potentially be applicable to HPV-associated lesions, particularly in advanced refractory cervical and head and neck cancer. However, recombinant vaccinia virus vaccine may potentially be risky for immunocompromised patients. Thus, for future clinical translation, such population should be excluded to minimize risk of general vaccinia infection.

The environment in the cervicovaginal tract is subjected to effects of the variation in hormones during cell cycle. In a previous study, the transfection efficiency of DNA vaccine administered through vaginal tract (lumen) using a micropipette followed by electroporation was indeed affected by hormonal differences (36). However, here we are directly injecting our DNA and vaccinia vaccines into the cervicovaginal tissue, thus the complication in absorption due to hormonal variation may not be applicable to our method of administration. Nevertheless, the full effect of hormonal variation on the immune responses generated by the cervicovaginal boost with TA-HPV vaccinia vaccine should be further investigated in future studies.

The E7-specific CD8⁺ T cells induced in the cervicovaginal tissues by cervicovaginal boost with TA-HPV express the mucosa-associated homing integrins α4β7 and CCR9, suggesting that these CD8 T cells are Trms. Using CD103 as a marker for Trm, we observed that there are slightly more CD103⁺ E7-specific Trm in the draining lymph node (DLN) and in the cervicovaginal tract compared with in the spleen (Supplementary Fig. S7). Furthermore, CD103⁺ cells from cervicovaginal tract express higher amount of S1P1 proteins (Supplementary Fig. S7). It has been shown that S1P1 can help with tissue retention of T cells (47). Thus, the increased expression of S1P1 may contribute to the retention of the CD103⁺ T cells in the cervicovaginal tract, which may contribute to the antitumor effect against local cervicovaginal tumor. However, previously Skon and colleagues have reported that S1P1 transcriptional downregulation is required for establishment of tissue resident memory T cells in the spleen, salivary glands, and small intestine (37). In comparison, here we observed an increase in S1P1 expression in CD103⁺ E7-specific CD8⁺ T cells in the cervicovaginal tract. The difference in the expression level of S1P1 may be explained by the difference in the local tissue environment. Another potential explanation for the discrepancy of our data may be related to the timing of evaluating S1P1 expression on T cells following vaccination. Unlike the experimental method in Skon and colleagues which evaluates S1P1 mRNA expression through qPCR at least 4 weeks after treatment, we assess S1P1 protein level in just 1 week after the last vaccination by flow cytometry. The expression of S1P1 changes based on different location and phases of T-cell activation and differentiation (38). S1P1 expression is high while the T cells are still in circulation, and as they traffic into the local tissue environment, cytokines milieu triggers the downregulation of S1P1 in T cells, leading to their retention in the local environment and gradual differentiation into S1P1-low Trms. We only assessed S1P1 protein level just 1 week after treatment and the downregulation of S1P1 may have not been fully initiated. It will be of interest to further characterize the expression kinetics of S1P1 in CD103⁺ T cell in the cervicovaginal tract following vaccination in future studies in order to generate more insight into the molecular mechanism of the generation of Trms in the cervicovaginal tract.

In addition, the expression of α4β7 is necessary for the observed enhancement of antigen-specific CD8⁺ T-cell antitumor immune responses in cervicovaginal tissues elicited by our prime-boost regimen. Of note, ltgb7^−/− mice also lack CD103. However, CD103 mainly plays a role for the retention of T cells in the epithelia, whereas α4β7 is responsible for trafficking (39). Our results with ltgb7^−/− mice show that the enhanced local E7-specific CD8⁺ T-cell response by our vaccination approach is mainly mediated by α4β7 (Fig. 5). However, Itgb7 deficiency may also lead to impairment of the generation of Trms due to lacking CD103, and consequently a decrease in the retention of CD8⁺ T cells in the local environment. Finally, consistent with previous findings by Zeng and colleagues (33), we demonstrated that the expression of α4β7 on CD8⁺ T cells can be mediated by stimulation with retinoic acid that is secreted by dendritic cells obtained from the cervicovaginal tissues. Taken together, these data suggest that, after a DNA priming vaccination, the administration of this recombinant HPV vaccinia vaccine via the cervicovaginal route may be more effective than intramuscular or subcutaneous injection in generating local cell-mediated immune responses for the control of HPV-associated diseases in cervicovaginal tract.

We showed that dendritic cells (DCs) from the cervicovaginal tract can induce the expression of α4β7 on CD8⁺ T cells and that this effect is mediated by retinoic acid production. The current treatment regimen of heterologous prime followed by cervicovaginal boost may well have generated enhanced production of retinoic acid in the cervicovaginal DCs that induced expression of α4β7 by CD8⁺ T cells. This regimen was associated with the accumulation of both E7-specific and total CD8⁺ T-cell populations in the cervicovaginal tissues, raising the possibility that it is nonspecific inflammation associated with the vaccinia expression rather than the presence of the antigen that is driving this homing response. Other cell types, including stromal cells, epithelial cells, and macrophages, have also been shown to produce retinoic acid (40–43). Whether these cells play a role in the generation of Trms in the cervicovaginal tissues following the heterologous prime-cervicovaginal boost vaccination regimen should be investigated in future studies. Furthermore, α4β7 and CCR9 have also been shown to interact with MAdCAM-1 and TECK, the CCR9 ligand, expressed in the small intestine (34).

Our current data are consistent with previous observations in terms of the efficacy of DNA-vaccinia prime-boost vaccination regimens in controlling HPV-associated disease in this preclinical model (14). Previously, we employed Vac-Sig/E7/LAMP-1 and Sig/E7/LAMP-1 DNA in various homologous and heterologous combinations in a preclinical model and found that priming with Sig/E7/LAMP-1 DNA and boosting with Vac-Sig/E7/LAMP-1 generated the strongest E7-specific CD8⁺ T-cell responses (14). The encouraging preclinical data led to an ongoing clinical trial assessing the efficacy of pNGVL4a-sig/E7(detox)/HSP70 DNA prime followed by TA-HPV boost in patients with grade 3 cervical intraepithelial neoplasia using intramuscular administration (20). Patients who were vaccinated peripherally, who had residual dysplasia after vaccination, nonetheless had striking histologic and molecular changes in the lesions, including markedly increased intensity of CD8⁺ T-cell infiltrates in both the stromal and epithelial compartments, compared with prevaccination. These infiltrates were organized in tertiary lymphoid structures, showed evidence of proliferation induced by cognate antigen, and expression of genes associated with immune activation and effector function. These findings suggest that, in humans, peripheral therapeutic vaccination can induce immune responses that localize in target lesions and that analyses of immune responses at sites of antigen are likely to be much more informative than analyses of systemic responses.

In this model, cervicovaginal boost vaccination with TA-HPV is more effective in generating E7-specific CD8⁺ T cells in the cervicovaginal tract than intramuscular administration (Supplementary Fig. S1), suggesting that future clinical trials should consider using cervicovaginal administration for TA-HPV. In our study, we demonstrated that cervicovaginal boost with TA-HPV generated CD8 T cells with a Trm phenotype in the cervicovaginal tract. Previously, it has been shown that cervicovaginal vaccination with heterotypic HPV pseudovirions induced antigen-specific T cells expressing CD103 (44). The induction or enhancement of existing Trm-mediated immunity may be advantageous in the context of HPV-associated disease. In other peripheral tissues, including skin and lung, tissue-resident memory T cells have been shown to provide a “first response” to antigen, secreting cytokines that recruit circulating T cells (22, 45). In murine models, Trms in normal mucosa have been shown to be able to protect against mucosal challenge with HIV, whereas systemic CD8⁺ T cells were unable to provide protection (46). It is likely that analyses of tissue immune responses will provide insights that are not obvious in cells in circulation; clinical trials testing therapeutic HPV vaccination regimens should include monitoring immune responses whenever possible.

In summary, we showed that cervicovaginal administration of therapeutic HPV vaccines in mice with established tumors generates enhanced antigen-specific CD8⁺ T cells with the mucosal Trm phenotype as well as therapeutic effects, compared with intramuscular vaccination.

R.B.S. Roden reports receiving commercial research grants from, holds ownership interest (including patents) in, and is a consultant/advisory board member for Papivax Biotech Inc. T.-C. Wu holds ownership interest (including patents) in, and is a consultant/advisory board member for Papivax Biotech Inc. No potential conflicts of interest were disclosed by the other authors.

Conception and design: Y.-Y. Sun, S. Peng, L. Han, R.B.S. Roden, C.L. Trimble, C.-F. Hung, T.-C. Wu

Development of methodology: Y.-Y. Sun, S. Peng, L. Han, J. Qiu, Y. Tsai, C.-F. Hung, T.-C. Wu

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y.-Y. Sun, S. Peng, L. Han, J. Qiu, L. Song, B. Yang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y.-Y. Sun, S. Peng, L. Han, J. Qiu, L. Song, B. Yang, T.-C. Wu

Writing, review, and/or revision of the manuscript: S. Peng, B. Yang, R.B.S. Roden, C.L. Trimble, T.-C. Wu

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y.-Y. Sun

Study supervision: L. Han, C.-F. Hung, T.-C. Wu

This work was supported by the NIH/National Cancer Institute Cervical Cancer SPORE P50CA098252, Head and Neck Cancer SPORE P50 CA-DE019032, 2R01CA114425-06, and R01CA142691 grants.

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.