The effect of adjuvanting cancer vaccines with herpes simplex virus glycoprotein D on melanoma-driven CD8⁺ T cell exhaustion

Abstract

Two vaccines expressing CD4⁺ and CD8⁺ T cell epitopes of MAAs by a chimpanzee-derived replication-defective AdC68 vector were compared in a mouse model of melanoma. In one vaccine, termed AdC68-gDMelapoly, the epitopes were expressed as a fusion protein within HSV-1 gD, which blocks immunoinhibitory signaling through the HVEM pathway. The other vaccine, termed AdC68-Melapoly only expressed the MAA epitopes. AdC68-gDMelapoly induced more potent MAA-specific CD8⁺ T cell responses especially to the subdominant MAA epitopes. Upon prophylactic vaccination, mice that developed CD8⁺ T cell responses to the two vaccines that were comparable in magnitude showed equal protection against tumor challenge. When mice were first challenged with tumor cells and then vaccinated results differed. In animals with comparable CD8⁺ T cell responses, the AdC68-gDMelapoly vaccine was more efficacious compared to the AdC68-Melapoly vaccine in delaying tumor growth. This effect was linked to reduced expression of 2B4, LAG-3 and PD-1 on tumor-infiltrating MAA-specific CD8⁺ T cells elicited by the gD-adjuvanted vaccine, suggesting that CD8⁺ T cells induced in presence of gD are less susceptible to tumor-driven exhaustion.

INTRODUCTION

Even cancer vaccines that are highly immunogenic in animal models commonly fail to provide benefits to patients with advanced cancers (1, 2). This has partially been linked to the highly immunosuppressive tumor microenvironment, which expresses immunoinhibitory ligands (3), recruits suppressive cell subsets such as regulatory T cells (4) and myeloid suppressor cells (5) and provides a metabolically stressed milieu (6). Biologicals that block immunoinhibitory pathways such as antibodies to PD-1 (7, 8) or CTLA-4 (9) or both (10, 11) are being tested alone or in combination with active immunotherapy in cancer patients and have yielded promising results.

Our focus has been on the herpes virus entry mediator (HVEM)² pathway. HVEM, which was first identified as a receptor for HSV-1 glycoprotein D (gD) (12), is a bimodal switch expressed on many cells including antigen presenting cells that can interact with the immunoregulatory molecules on lymphocytes (13). Binding of HVEM to LIGHT or lymphotoxin provides stimulatory signals; binding to the B and T lymphocyte attenuator (BTLA) or CD160 activates inhibitory pathways (14). Co-activators and co-inhibitors bind to different domains of HVEM and can form a trimolar complex, in which signaling through co-inhibitors dominates (14). The N-terminus of HSV-1 gD binds to a site on HVEM that is close to the BTLA/CD160 binding site and thereby blocks immunoinhibitory but not co-stimulatory HVEM signaling (15).

As we have shown previously, vaccines that express antigens fused into the C-terminus of gD elicit enhanced T cell responses, which is linked to blockade of the immunoinhibitory HVEM pathways (16). Adjuvanting vaccine antigens with gD is especially effective to augment CD8⁺ T cell responses in aging mice (17) and in mice with advanced cancers (18). Our previous cancer studies were based on human papilloma virus type 16 (HPV-16)-associated tumors, which express viral antigens that are foreign to the immune system. The current study was conducted to assess if expressing ‘self’ antigens from non-viral tumors within gD would enhance the immunogenicity and efficacy of a cancer vaccine. Experiments were conducted in a transplantable melanoma model, based on B16F10 cells that were stably transfected to express Braf_V600E (B16Braf_V600E). The vaccine antigen, termed Melapoly, was designed to express CD4⁺ and CD8⁺ T cell epitopes of melanoma-associated antigens (MAAs) including tyrosinase-related protein (Trp)-1, Trp-2, gp100 and mutated Braf linked to the _V600E universal T helper cell epitope PADRE and an endoplasmic reticulum targeting signal sequence. To test for the gD adjuvant effect, the Melapoly encoding sequence was fused into the C terminal domain of HSV-1 gD (gDMelapoly). The Melapoly and the gDMelapoly fusion proteins were expressed by a simian E1-deleted adenovirus vector of serotype 68 (AdC68).

As expected, the AdC68-gDMelapoly vector induced more potent MAA-specific CD8⁺ T cell responses, especially to subdominant epitopes, compared to the AdC68-Melapoly vector and provided superior protection if given before tumor challenge. In the same token, in a therapeutic vaccination model, the AdC68-gDMelapoly vector was superior in delaying tumor progression compared to the AdC68-Melapoly vector. To assess if the improved efficacy of the gD-adjuvanted vaccine solely reflected differences in the magnitude of MAA-specific T cell responses, we vaccinated mice with different doses of the AdC68 vectors and selected subgroups with comparable frequencies of MAA-specific CD8⁺ T cells. In a pre-challenge vaccination model, vaccine efficacy was shown to depend on frequencies of MAA-specific CD8⁺ T cells. In contrast in a post-challenge vaccination model, AdC68-gDMelapoly vaccinated mice that had MAA-specific T cell frequencies comparable to those of AdC68-Melapoly vaccinated mice survived significantly longer. This was not caused by differences in production of mediators by AdC68-gDMelapoly-induced T cells but rather by their increased resistance against differentiation towards exhaustion.

MATERIALS AND METHODS

Mice

Female C57Bl/6 mice (6–8 weeks) were purchased from the National Cancer Institute (NCI) and housed at the Wistar Institute Animal Facility. All procedures were performed under the guideline of protocols approved by the IACUC of the Wistar Institute.

Cell lines

The B16Braf_V600E cell line was derived from B16.F10 cells transduced with lentiviral vector pLU-EF1a-mCherry expressing mouse Braf_V600E (provided by Dr. M Herlyn lab, Wistar Institute, Philadelphia, PA). The mutant cell line showed the same in vivo growth characteristics as unmodified B16.F10 cells. E1-transfected HEK 293 cells were used to propagate Ad vectors. Cells were cultured with DMEM supplemented with 10% FBS.

Construction of recombinant Ad vectors

We designed the Melapoly transgene to express a number of MAA-specific CD4⁺ T cell and CD8⁺ T cell epitopes. An endoplasmic reticulum (ER) targeting signal sequence was included at the N-terminal end of the sequence (19). Three human (h)Trp-2-specific CD4⁺ T cell epitopes (20) as well as the universal T helper cell epitope PADRE (21) were incorporated into the Melapoly construct. CD8⁺ T cell epitopes were derived from human and mouse Trp-2, human gp100, mouse Trp-1 and Braf_V600E. Epitopes from Trp-1 and Braf antigens were modified to enhance their binding to MHC class I molecules (22). A Flag-tag was added to the C-terminal end. We designed the spacer sequences according to several analysis programs, including PAPROC I (http://www.paproc.de), Netchop 3.1 (http://www.cbs.dtu.dk/services/NetChop/), and IEBD Analysis Resource (http://tools.immuneepitope.org/main/). The spacers were inserted between each CD4⁺ and CD8⁺ T cell epitope to reduce interference of epitope processing (23, 24). The Melapoly gene was codon optimized and inserted into the pUC57 vector (Genescript, Piscataway, NJ). The Melapoly transgene was sequenced and cloned it into the pShuttle vector by EagI digestion. To construct the AdC68-gDMelapoly vector, the Melapoly transgene was fused into the N terminus of HSV-1 gD. The pShuttle gDMelapoly vector was generated by PCR using pShuttle Melapoly as the template, gDMelaFwAgeI CGACCGGTTAGCTAAGTTTGTGGCCGCTTG and gDMelaRvAvrII CGCCTAGGTGCTGCTGCTGCAATGCTC as primers. The PCR product was cleaved by AgeI and AvrII and cloned into the pShuttle gD-Flag vector. The inserts containing regulatory sequences of the pShuttle vector were subcloned into the E1-deleted AdC68 viral molecular clone using I-CeuI and PI-SceI sites. Recombinant AdC68 vectors were rescued, propagated on HEK 293 cells, purified by CsCl-gradient centrifugation and titrated as described (25). The concentration of each virus batch was determined by measuring virus particles (vp) by spectrophotometry at 260nm.

Real-Time PCR

We infected HEK 293 cells with 10¹⁰vp and 10¹¹vp doses of AdC68-Melapoly and AdC68-gDMelapoly vectors. Cells were harvested 24 hours later and total RNA was isolated from each sample using RNeasy Mini Kit (Qiagen, Venlo, Netherlands). cDNA samples were obtained by reverse transcription using 2ug RNA/sample with high capacity cDNA reverse transcription kit (Applied Biosystems, Foster City, CA). 20ng cDNA/sample was used for real-time PCR using Fast SYBR green mastermix (Applied Biosystems, Foster City, CA). To obtain standards for real-time PCR, we performed regular PCR for Melapoly and GAPDH (housekeeping gene as internal control) using the cDNA samples. We ran the PCR products on 1% agarose gel, purified the PCR products of the expected sizes (MiniElute gel extraction kit, Qiagen, Netherlands) and measured the DNA concentration using spectrometry. We diluted the purified Melapoly and GAPDH DNAs from 5ng to 0.156ng for real-time PCR standards. For real-time PCR, cDNA samples were replicated for both Melapoly and GAPDH using the 7500 Fast real-time PCR machine (Applied Biosystems, Foster City, CA). The primer sequences are as follows: regular PCR MelapolyFw ACAGGAAACTTCGCCGCTGC, MelapolyRv TGCCATATATCCGAGGTTGTCTG, real-time PCR MelapolyRv1 TGATCGGCTGCAGCCACGTC; regular PCR GAPDHFw GGTGAAGGTCGGTGTGAACGGATTT, GAPDHRv AATGCCAAAGTTGTCATGGATGACC and real-time PCR GAPDHFw1 TGCCCCCATGTTTGTGATGG. The final concentrations of Melapoly cDNA from both vector-infected samples were analyzed and compared after normalization to GAPDH.

Immunization and challenge of mice

Groups of C57BL/6 mice were vaccinated i.m. with the AdC68 vectors diluted in PBS into the tribialis anterior muscle of each hind limb, with doses ranging from 3 x 10⁹ to 10¹¹ vp per mouse. In tumor challenge experiments, B16Braf_V600E cells suspended in serum-free medium were inoculated s.c. into the right flank of mice. For pre-and post-tumor challenge experiments, 5x10⁴ cells were injected and mice were vaccinated 14 days before or 3 days after tumor cell challenge. We monitored tumor growth by measuring the perpendicular diameters of tumors at least three times a week. Mice were euthanized once tumors exceeded a surface area of 100 mm². Mice were monitored for a period of 60 days post tumor challenge.

Isolation of lymphocytes

PBMCs and splenocytes were harvested as described before (18). To obtain tumor-infiltrating lymphocytes (TILs), tumors were cut into small fragments and treated with 2mg/ml collagenase P and 1mg/ml DNase I (Life Technologies, Carlsbad, CA) diluted in Hank’s balanced salt solution (HBSS,1X) (Thermo Fisher Scientific, Pittsburgh, PA) with agitation for 1 hour. Tumor fragments were homogenized, filtrated and lymphocytes were purified by Percoll-gradient centrifugation.

Tetramer and intracellular cytokine staining

To detect tetramer-specific CD8⁺T cells, we stained cells with a PE-labeled tyrosinase-related protein 1 (Trp-1)-specific MHC class I (H-2D^b) tetramer carrying the TAPDNLGYM peptide and an Alexa647-labeled tyrosinase-related protein 2 MHC class I (H-2K^b) tetramer carrying the SVYDFFVWL peptide (obtained from the NIAID Tetramer Facility, Atlanta, GA). The following antibodies were also used for staining: anti-CD8-PerCPCy5.5, PacBlue or Alexa700, CD4-PercpCy5.5, CD44-PacBlue or Alexa700, LAG-3 PercpCy5.5, PD-1 PE-Cy7 (all from Biolegend, San Diego, CA), 2B4 FITC (eBioscience, San Diego, CA) and Amcyan fluorescent reactive dye (Life Technologies). For intracellular cytokine staining (ICS), CD8⁺ and CD4⁺ T cell responses to individual epitopes were measured using ~10⁶ lymphocytes per samples. Cells were cultured in 2% FBS DMEM medium with individual MHC class I or class II restricted peptides at a concentration of 5ug/ml (for peptides representing CD8⁺ T cell epitopes): mTrp-1_455–463 TAPDNLGYA, mTrp-1_481–489 IAVVAALLL, mTrp-2_522–529 YAEDYEEL, hTp-2_180–188 SVYDFFVWL, hTrp-2_343–357 STFSFRNAL, mTrp-2_363–371 SQVMNLHNL, hgp100 KVPRNQDWL and mBRaf FGLANEKSI; or 10ug/ml (for peptides 25–33594–602 representing CD4⁺ T cell epitopes): PADRE: AKFVAAWTLKAAA, hTrp-2_88–102: RKFFHRTCKCTGNFA, hTrp-2_237–256: NESFALPYWNFATGRNECDV and mTrp-2_363–376: SQVMNLHNLAHSPL (Genescript, Piscataway, NJ). To determine overall MAA-specific T cell responses, a peptide pool including all CD8⁺ and CD4⁺ T cell epitopes expressed by the Melapoly or gDMelapoly vectors was used for lymphocytes stimulation. A rabies virus glycoprotein peptide was used as a negative control. ICS was conducted as described before (14). Produced cytokines, enzymes were stained with the following antibodies: anti-IFN-γ-APC, IL-2-Alexa700 or FITC, TNF-α-PE-Cy7 (BioLegend, San Diego CA). Transcription factor T-bet was stained using Foxp3/Transcription factor staining buffer set (ebioscience) using anti-T-bet-PE-Cy7 antibody (eBioscience, San Diego CA). Cells tested by ICS were co-stained for exhaustion marker PD-1using anti-PD-1-Brilliant Violet 605 antibody (BioLegend, San Diego CA). Cells were analyzed by an LSRII (BD Biosciences, Franklin Lakes, NJ). Data were analyzed with FlowJo (TreeStar, Ashland, OR). Experiments using ICS or tetramer stains were controlled assessing responses of CD44⁻CD8⁺ cells within the same animals after preliminary experiments confirmed that results were comparable to those obtained with CD44⁺CD8⁺ cells of naïve mice or mice vaccinated with the AdC68-gD control vector.

Statistical analysis

We compared the differences of immune responses between more than two groups using one-way ANOVA followed by Holm-Sidak’s multiple comparisons. Immune responses affected by two factors (vaccine and tissue type or vaccine and cell type) were analyzed by two-way ANOVA with Holm-Sidak’s multiple comparisons. We performed Holm-Bonferroni correction for multiple comparisons within the same dataset. Survival rates between different treatment groups were compared using Gehan-Breslow-Wilcoxon test. Significance was set at p-values of or below 0.05. All statistical analyses were performed using Graphpad Prism6. Throughout the manuscript p-values adjusted for type 1 errors are shown.

RESULTS

The gD adjuvant enhances CD8⁺ T cell responses to dominant and subdominant epitopes

Levels of transgene expression can affect the immunogenicity of viral vectors. We therefore tested the levels of transgene product expression in transfected cells and showed by real-time PCR that both vectors expressed comparable levels of transgene-derived mRNA (data not shown).

We compared the magnitude of MAA-specific CD8⁺ T cell responses induced by the vaccines in young female C57BL/6 mice. Each group (n=5) of mice received either the AdC68-Melapoly or the AdC68-gDMelapoly vector given i.m. at doses of 10¹⁰ or 10¹¹ vp. Two and four weeks after vaccination, we measured CD8⁺ T cell responses to two MAA epitopes, i.e. Trp-1₄₅₅ and Trp-2₁₈₀, by staining T cells with the corresponding MHC class I tetramers (Suppl. Fig. 1). Both vaccines induced stronger CD8⁺ T cell responses to Trp-1₄₅₅ than Trp-2₁₈₀ (mean frequencies of positive CD44⁺CD8⁺ cells for Trp-1₄₅₅ vs. Trp-2₁₈₀: 10¹⁰ vp: week 2: Melapoly: 5.1% vs. 0.56%, gDMelapoly: 5.8% vs. 0.58%; week 4: Melapoly: 2.5% vs. 0.25%, gDMelapoly: 5.3% vs. 0.42%; 10¹¹ vp: week 2: Melapoly: 8.7% vs. 0.86%, gDMelapoly: 14.7% vs. 0.51%; week 4: Melapoly: 7.8% vs. 0.18%, gDMelapoly: 8.5% vs. 0.56%; naive CD44⁻ T cells: 0.043% vs. 0.056% and 0.043% vs. 0.078% for wks 2 and 4, respectively, Fig. 1 A, B). All groups mounted significant responses to the Trp-1₄₅₅ and Trp-2₁₈₀ tetramer compared to the internal CD44⁻CD8⁺ T cell controls (adjusted p-values are listed in legend to Fig. 1) or to naïve mice (data not shown). At the 10¹¹ vp dose at week 2 and the 10¹⁰ vp dose at week 4 Trp-1₄₅₅-specific CD8⁺ T cell responses were significantly higher in AdC68-gDMelapoly vaccinated animals than in AdC68-Melapoly vaccinated mice. Responses to Trp-2₁₈₀ were significantly higher in mice vaccinated with 10¹¹ vp AdC68-gDMelapoly at both weeks 2 and 4. At the 10¹⁰ vp dose at week 4 Trp-2₁₈₀ responses were significantly higher in Melapoly vaccinated mice.

FIGURE 1. Induction of MAA-specific CD8⁺ T cells by different vector doses.

Groups (n=5/group) of C57BL/6 mice were vaccinated i.m. with 10¹⁰ vp or 10¹¹ vp of either AdC68-Melapoly (open squares) or AdC68-gDMelapoly (closed squares) vector. PBMCs were collected 2 and 4 weeks later, Trp-1₄₅₅- and Trp-2₁₈₀-specific CD8⁺ T cell responses were compared. The graphs show responses in individual animals, the lines indicate means ± standard errors. * indicates significant differences between groups connected by lines (* p = 0.01–0.05, ** p = 0.001–0.01, *** p = 0.0001–0.001, ****p<0.0001). The following differences had significant p-values: all groups Trp-1₄₅₅ compared to CD44⁻CD8⁺ controls: p < 0.0001; Trp-1₄₅₅: Melapoly vs. gDMelapoly: 10¹¹ vp, wk 2: p = 0.012, 10¹⁰ vp, wk 4: p <0.0001; Trp-2₁₈₀ compared to controls: Melapoly, 10¹⁰ vp: wk 2: p < 0.0001, wk 4: p = 0.016; 10¹¹ vp: wk 2: p < 0.0001, wk 4: p = 0.034; gDMelapoly: both doses and both time points: p < 0.0001; Trp-2_180: Melapoly vs. gDMelapoly: 10¹¹ vp: wk 2: p = 0.027, 10¹⁰ vp: wk 2: p = 0.036, wk 4: p < 0.0001.

To further assess differences, we immunized mice (n=7–8) with 10¹¹ vp of the AdC68-Melapoly vector or 10¹⁰ vp of the AdC68-gDMelapoly vector. We collected blood at different time points after vaccination and measured the immune responses to individual MAA epitopes by intracellular cytokine staining (ICS) for production of IFN-γ, TNF-α and IL-2 (Fig. 2, Suppl. Fig. 2). For these experiments, blood had to be pooled to allow for testing the PBMC samples stimulated with the eight different peptides carrying CD8⁺ T cell epitopes expressed by the vaccines. Both vectors induced significantly higher tumor-antigen specific CD8⁺ T cell responses to both dominant and subdominant epitopes compared to naive controls (CD44⁻CD8⁺ T cells). In spite of the lower dose, there was a clear trend of the AdC68-gDMelapoly vaccine to induce higher responses mainly to the subdominant epitopes (Trp-1₄₅₅ epitope: mean frequencies averaging all time points: 1.9 % and 2.9 % for Melapoly and gDMelapoly, respectively, Fig. 2A; subdominant epitopes: mean frequencies of the sum of responses averaging all time points 1.3% and 3.0% for Melapoly and gDMelapoly, respectively, Fig. 2B). The preferential increase of responses to subdominant epitopes by gD was confirmed by comparing responses to the dominant Trp-1₄₅₅ epitope to those against the subdominant Trp-2₁₈₀ epitope by tetramer staining at 4 and 26 weeks after vaccination with 10¹¹ vp of AdC68-Melapoly or 10¹⁰ vp of AdC68-gDMelapoly (Fig. 2C–D). At both early and late time points the responses to Trp-1₄₅₅ epitope were comparable, while the responses to Trp-2₁₈₀ epitope were significantly higher in mice immunized with AdC68-gDMelapoly (wk 4, mean tetramer frequencies for Trp-1_455: Melapoly vs gDMelapoly: 7.8% vs 5.3%; Trp-2₁₈₀: 0.18% vs 0.42%, p = 0.0053; wk 26, Trp-1_455: Melapoly vs gDMelapoly: 4% vs 5.9%, Trp-2₁₈₀: 0.35% vs 0.66%, p = 0.033).

FIGURE 2. MAA-specific CD8⁺ T cell responses to individual epitopes.

Groups of C57BL/6 mice (n=7–8) were vaccinated i.m. with 10¹¹vp AdC68-Melapoly or 10¹⁰vp AdC68-gDMelapoly. (A–B) CD8⁺ T cell responses to individual epitopes were measured by ICS for IFN-γ, TNF-α and IL-2 upon stimulation of cells with peptides representing the T cell epitopes expressed by the vaccines over a course of 24–26 weeks. Graphs show the sum of cytokine responses. (A) shows responses to the dominant Trp-1₄₅₅ peptides. (B) shows the sum of responses over time for the other peptide; responses to individual peptides (but for Trp-1₄₅₅) were totaled. (C–D) shows Trp-1-teramer⁺ and Trp-2 tetramer⁺ CD8⁺ T cell responses 4 weeks and 26 weeks after vaccination in tumors of vaccinated mice. Differences between the two vaccines were not significant for Trp-1-specific CD8⁺ T cells (p = 0.12 and 0.18, respectively) but significant for Trp-2-specific CD8⁺ T cells (p = 0.0053 and 0.035, respectively). (E–J) The patterns of the sum of cytokine responses to the 8 CD8⁺ T cell epitopes at 3 time points, i.e., week 6 (E,F), 12 (G,H) and 24 (I,J) after vaccination.

We selected three time points (weeks 6, 12 and 24) to compare the overall response pattern (Fig. 2E–J). Responses to the subdominant epitopes developed with a delay as compared to Trp-1₄₅₅ responses. At all three time points the gD-adjuvanted vector induced a more prominent CD8⁺ T cell response to the subdominant epitopes compared to the AdC68-Melapoly vector (sum of frequencies to Melapoly and gD-Melapoly, respectively: wk 6: 3.8%, 4.4%, wk 12: 2%, 6.7%, wk 24: 4.5 %, 6.6%).

In a pre-tumor challenge vaccination model, gD adjuvant efficacy is solely linked to magnitude of the MAA-specific CD8⁺ T cell responses

To determine if vaccination with the AdC68-gDMelapoly vector resulted in superior protection against tumor growth than vaccination with the AdC68-Melapoly vector, we immunized groups of mice (n=5–18) with 10¹⁰ vp of AdC68-gDMelapoly, 10¹¹ vp of AdC68-Melapoly or 10¹¹ vp of AdC68-gD, the latter as a control for non-specific effects of the vector or gD. Fourteen days later mice were challenged with B16Braf_V600E cells given s.c.. Total cytokine⁺CD8⁺ T cell responses to the eight CD8⁺ T cell epitopes were measured before tumor challenge. As shown in Fig. 3A, both vaccines induced significantly higher responses when compared to background stains of naïve T cells (mean frequencies 4.5% and 9.3% for Melapoly and gD-Melapoly, 0.02% for control cells). The AdC68-gDMelapoly vector induced significantly higher CD8⁺ T cell responses compared to the AdC68-Melapoly vector (p < 0.0001) and this was linked to enhanced survival rates upon tumor challenge (Fig. 3C, p = 0.0018 by Gehan-Breslow-Wilcoxon test). Specifically > 60% of AdC68gD-Melapoly vaccinated mice as compared to < 20% of AdC68-Melapoly vaccinated mice remained tumor-free. When we compared subgroups with similar frequencies of MAA-specific CD8⁺ T cells by removing the low responders from the Melapoly group and the high responders from the gD-Melapoly group (Fig. 3B, mean frequencies of the Melapoly subcohort: 5.7% and the gD-Melapoly subcohort: 6.5% p = 0.1), survival rates of the two vaccine groups ceased to be significantly different (p = 0.08, Fig. 3D). This suggests that in the pre-tumor challenge model vaccine efficacy is significantly influenced by the magnitude of MAA-specific CD8⁺ T cell responses.

FIGURE 3. Vaccine efficacy in mice vaccinated before tumor cell challenge.

Groups of C57BL/6 mice (n=5–18) were vaccinated with 10¹¹vp AdC68-gD, 10¹¹vp AdC68-Melapoly or 10¹⁰vp AdC68-gDMelapoly vectors. Fourteen days later PBMC samples were tested for CD8⁺ T cell responses to all of the MAA epitopes present in the vaccine. (A) The sum of tumor antigen-specific CD8⁺ T cell percentages for individual mice. (*) indicates significant differences as detailed in legend to Fig. 1. Melapoly vs. gDMelapoly: p < 0.0001, Melapoly vs. naïve: p = 0.0014, gDMelapoly vs. naïve: p < 0.0001. (B) The sum of tumor antigen-specific CD8⁺ T cell percentages in the two subcohorts that showed comparable CD8⁺ T cell responses. Subcohorts were selected by excluding animals with frequencies of MAA-specific CD8⁺ T cells below the mean of the Melapoly group or above the mean of the gDMelapoly group. Melapoly vs. gDMelapoly, p = 0.1, Melapoly vs. naïve: p < 0.0001, gDMelapoly vs. naïve: p < 0.0001. (C) Survival rates of the entire group, measured for 60 days. Melapoly vs. naïve: p < 0.0001, gD-Melapoly vs. naïve: p < 0.0001, Melapoly vs. gDMelapoly: p = 0.0018 (D) Survival rates of the of the Melapoly subcohort in comparison to the gDMelapoly and the control group. Melapoly vs. controls: p < 0.0001, gD-Melapoly vs. controls: p = 0.0003, Melapoly vs. gDMelapoly: p = 0.08. In (C) and (D) control animals are shown as X, AdC68-Melapoly-vaccinated mice are shown as open squares, AdC68-gDMelapoly-vaccinated mice are shown as closed squares.

In a post-tumor challenge vaccination model, gD adjuvant efficacy is not solely linked to the magnitude of MAA-specific CD8⁺ T cell responses

We further compared the efficacy of the two vectors in a post-challenge vaccination model. B16Braf_V600E cells grow very rapidly in vivo; after tumor cell inoculation most control mice require euthanasia within 14–18 days. AdC68-induced T cell responses on the other hand do not peak till 12–14 days after vaccination. Therefore, we chose a fairly short interval of 3 days between tumor cell inoculation and vaccination for our therapeutic vaccine studies. Accordingly, groups of mice (n=5–10) were challenged with B16Braf_V600E tumor cells on day 0 and vaccinated three days later with either 10¹¹vp of the AdC68-Melapoly vector (n=9) or 3x10⁹vp and 10¹⁰vp of the AdC68-gDMelapoly vector (n=8 and 10 for 3x10⁹vp and 10¹⁰vp, repectively). A control group (n=28) was vaccinated with 10¹¹vp of the AdC68-gD vector. CD8⁺ T cell responses were measured two weeks later from blood by ICS for production of IFN-γ, TNF-α, granzyme B and perforin. In addition, we assessed CD8⁺ T cell responses by staining with the Trp-1₄₅₅- and Trp-2₁₈₀-specific tetramers. MAA-specific CD8⁺ T cell responses tested by ICS were significantly higher compared to background stains for both vaccine groups, and higher in gDMelapoly than Melapoly vaccinated mice (means: Melapoly: 3.6%, gD-Melapoly: 5.2%, controls :0.27%, adjusted p-values are shown in the figure legend, Fig. 4Aa). The AdC68-gDMelapoly and Melapoly vector induced significantly higher frequencies of Trp-1₄₅₅ (means: Melapoly: 5.2%, gD-Melapoly: 6.3%, controls: 0.037%, Fig. 4Ba) and Trp-2₁₈₀ tetramer⁺CD8⁺ T cells (means: Melapoly: 0.19% and gD-Melapoly: 0.14% control cells: 0.042%, Fig. 4Ca) compared to controls. Both vaccines significantly delayed tumor growth and prolonged overall survival compared to control mice (p<0.0001, Fig. 4Da). Comparing the two vaccine groups showed that AdC68-gDMelapoly vaccinated mice survived significantly longer than AdC68-Melapoly vaccinated mice (p = 0.0039, Fig. 4Ad).

FIGURE 4. Vaccine efficacy in mice vaccinated after tumor cell challenge.

Groups of C57BL/6 mice (n=9–30) were challenged with B16Braf_V600E cells given s.c. 3 days before vaccination with 10¹¹vp AdC68-gD, 10¹¹vp AdC68-Melapoly or 3 x10⁹ vp and 10¹⁰ vp AdC68-gDMelapoly vector. Subcohorts were selected by excluding animals that had frequencies of MAA-specific CD8⁺ T cells below the mean of the Melapoly group divided by the fold difference between the gDMelapoly over Melapoly group (1.34) or above the mean of the gDMelapoly group multiplied by the fold difference.

(a, b) PBMCs were tested 12 days after vaccination. (c) Cells from blood, spleens and tumors were tested at necropsy. (a) shows responses and survival rates of all mice, (b,c) show responses and survival rates of the subcohorts.

(*) indicates significant difference between two groups as described in legend to Fig. 1 with the following p-values: (Aa) both vaccine groups vs. controls: p < 0.0001; Melapoly vs. gDMelapoly: p = 0.01. (Ab) both vaccine groups vs. to controls: p < 0.0001; Melapoly vs. gDMelapoly: p = 0.99. (Ac) Blood: Melapoly vs. controls: p < 0.0001, gDMelapoly vs. controls: p = 0.0003, Melapoly vs. gDMelapoly: p = 0.011, Spleen: Melapoly vs. controls: p = 0.0003, gDMelapoly vs. controls: p < 0.0001, Tumors: all comparisons: p < 0.0001.

(Ba) both vaccines vs. controls p < 0.0001, Melapoly vs. gDMelapoly p = 0.28. (Bb) both vaccine groups vs. controls: p < 0.0001, Melapoly vs. gDMelapoly: p = 0.07. (Bc) Blood: Melapoly vs. controls: p < 0.0001, gDMelapoly vs. controls: p = 0.0029, Melapoly vs. gDMelapoly p =0.047, Spleen: Melapoly vs. controls: p < 0.0001, gDMelapoly vs. controls: p = 0.04, Tumors: all comparisons: p < 0.0001. (Ca) both vaccines vs. controls: p < 0.0001. (Cb) Melapoly vs. controls: p < 0.0001, gDMelapoly vs. controls: 0.0002. (Cc) Spleen: Melapoly vs. controls: p = 0.02, gDMelapoly vs. controls: p < 0.0001, Tumors: both vaccine groups vs. controls: p < 0.0001, Melapoly vs. gDMelapoly p = 0.0187.

As MAA-specific CD8⁺ T cell responses were higher in the AdC68-gDMelapoly group than in the Melapoly group, we selected 8 mice from the AdC68-Melapoly vaccine group and 14 mice from the AdC68-gDMelapoly vaccine group with comparable frequencies of total cytokine⁺CD8⁺ T cell responses (means: Melapoly: 3.9%, gDMelapoly: 4.0%, Fig. 4Ab), Trp-1₄₅₅ tetramer⁺ CD8⁺ T cells (means: Melapoly: 3.0% and gDMelapoly: 3.1%, Fig. 4Bb) and Trp-2₁₈₀ tetramer⁺ CD8⁺ T cell responses (means: Melapoly: 0.19% and gD-Melapoly 0.14%, Fig. 4Cb). Again in the subcohorts both vaccines significantly delayed tumor growth and prolonged survival compared to control mice (p<0.0001). Although mice had been selected based on comparable frequencies of MAA-specific CD8⁺ T cell responses, the AdC68-gDMelapoly vaccinated subcohort showed significantly prolonged survival rates compared to the AdC68-Melapoly vaccinated subcohort (p=0.015, Fig. 4Db).

To assess if vaccine-induced T cell responses differed in tissues, we analyzed MAA-specific CD8⁺ T cell responses to all epitopes by ICS in blood, spleens and tumors at the time of necropsy, when the size of tumor exceeded a surface area of ~ 100 mm². Interestingly, total cytokine⁺CD8⁺ T cell responses in the blood and tumors of AdC68-Melapoly vaccinated mice were significantly higher than those of AdC68-gDMelapoly vaccinated mice (mean frequencies: blood, spleens, tumors: Melapoly: 2.2%, 1.5%, 3.7%; gD-Melapoly: 1.2%, 1.4%, 2.1%; controls: 0.049%, 0.12%, 0.027%; Fig. 4Ac, p-values are shown in the figure legend). For Trp-1₄₅₅-specific CD8⁺ T cell responses, both vaccines induced significant responses in the different compartments compared to naïve cells. The AdC68-Melapoly group again showed significantly higher responses in tumors compared to mice immunized with AdC68-gDMelapoly (mean frequencies: blood, spleens, tumors: Melapoly: 2.7%, 2.0%, 7.8%; gD-Melapoly: 1.6%, 1.0%, 2.7%; controls: 0.085%, 0.043%, 0.14%; Fig. 4Bc; gating strategies Suppl. Figure 3). Trp-2₁₈₀- specific responses were insignificant in blood but reached significance for both vaccine groups for cells from spleens and tumors compared to naive samples (mean frequencies: blood, spleens, tumors: Melapoly: 0.18%, 0.25%, 0.66%; gDMelapoly: 0.14%, 0.33%, 0.89%; controls: 0.016, 0.02, 0.13%; Fig. 4Bc). MAA-specific CD4⁺ T cell responses were analyzed in parallel; they showed no differences before challenge but after challenge were higher in spleens of Melapoly-vaccinated mice (mean frequencies in spleens 1.5% and 0.59% for Melapoly and gD-Melapoly, respectively, p = 0.012, data not shown). Overall these results indicated that in the post-challenge vaccine model better protection achieved with the gD adjuvanted vaccine could not solely be explained by induction of higher frequencies of MAA-specific CD8⁺ T cells.

MAA-specific CD8⁺ T cell phenotypes

Chronic exposure to antigen can lead to T cell exhaustion, which is characterized by progressive loss of T cell functions and eventual cell death (26, 27). To determine if gD affected the differentiation pathways of T cells, we assessed the expression of several exhaustion markers, i.e., 2B4, LAG-3 and PD-1 on Trp-1₄₅₅-and Trp-2₁₈₀-specific CD8⁺ T cells in spleens and tumors at the time of necropsy, using the two subcohorts with comparable CD8⁺ T cell responses. Expression of exhaustion markers was also assessed on CD44⁻CD8⁺ (naïve) T cells and CD44⁺tet⁻CD8⁺ (memory) T cells. The latter population, negative for Trp-1₄₅₅ and Trp-2₁₈₀ tetramer-specific CD8⁺ T cells, presumably contained CD8⁺ T cells to other MAAs in the vaccine and to unrelated antigens encountered previously. Fig. 5A shows expression levels of the three exhaustion markers as mean fluorescence intensity (MFI) of the different dyes attached to the marker-specific antibodies. Significant differences were found between CD44⁻CD8⁺ and CD44⁺CD8⁺ T cells in spleens and tumors only for 2B4 (Fig. 5A). Differences between CD44⁻CD8⁺ or CD44⁺CD8⁺ T cells and vaccine-induced tetramer-positive CD8⁺ T cells were seen for all markers with the exceptions of 2B4 on Trp-1₄₅₅⁺CD8⁺ cells from spleens and 2B4 and PD-1 on Trp-2₁₈₀⁺CD8⁺TILs. We also observed significant differences between the two vaccine subcohorts as well as between CD8⁺ T cells to the two epitopes. In particular, 2B4 showed higher expression on Trp-2₁₈₀- than Trp-1₄₅₅- specific CD8⁺ T cells induced by either vaccine in spleens (mean MFI Melapoly: Trp-1₄₅₅ vs. Trp-2₁₈₀: 158 vs. 269, gDMelapoly: Trp-1₄₅₅ vs. Trp-2₁₈₀: 166 vs. 284, adjusted p-values are shown in the legend to Fig. 5A). In both vaccine groups LAG-3 expression in spleens and in tumors was higher on Trp-1₄₅₅- than Trp-2₁₈₀-specific CD8⁺ T cells. In tumors, Trp-1₄₅₅- specific CD8⁺ T cells showed significantly lower level of LAG-3 expression in AdC68-gDMelapoly vaccinated animals compared to those from AdC68-Melapoly vaccinated mice (mean MFI: spleen: Melapoly: Trp-1₄₅₅ vs. Trp-2₁₈₀: 436 vs. 224, gDMelapoly: Trp-1₄₅₅ vs. Trp-2₁₈₀: 401 vs. 313; tumors: Melapoly: Trp-1₄₅₅ vs. Trp-2₁₈₀: 875 vs. 504, gDMelapoly: Trp-1₄₅₅ vs. Trp-2₁₈₀: 728 vs. 609). In spleens, PD-1 was expressed at higher levels on Trp-2₁₈₀- than Trp-1₄₅₅-specific CD8⁺ T cells and this reached significance for those induced by AdC68gD-Melapoly. Within tumors again Trp-1₄₅₅-specific CD8⁺ T cells that had been elicited by the AdC68-gDMelapoly vaccine showed lower PD-1 expression compared to those elicited by the AdC68-Melapoly vaccine. In both vaccine groups, Trp-1₄₅₅-specific CD8⁺ T cells expressed higher levels of PD-1 compared to those specific for Trp-2₁₈₀ (mean MFI: spleen: Melapoly: Trp-1₄₅₅ vs. Trp-2₁₈₀: 97 vs. 126, gDMelapoly: Trp-1₄₅₅ vs. Trp-2₁₈₀: 89 vs. 151; tumors: Melapoly: Trp-1₄₅₅ vs. Trp-2₁₈₀: 1190 vs. 400, gDMelapoly: Trp-1₄₅₅ vs. Trp-2₁₈₀: 721 vs. 297).

FIGURE 5. MAA-specific CD8⁺ T cell phenotypes.

Phenotypes of CD8⁺ T cells isolated from spleens and tumors of the two subcohorts shown in Fig. 4 were tested for expression of 2B4, LAG-3 and PD-1.

5A Mean fluorescence intensity (MFI) of 2B4 (A), LAG-3 (B) and PD-1 (C) fluorochrome-labeled antibodies on CD44⁻CD8⁺, tetramer⁻CD44⁺CD8⁺ and tetramer⁺CD44⁺CD8⁺ T cells from spleens (a) and tumors (b). Open boxes show results for CD8⁺ T cells from AdC68-Melapoly vaccinated mice, grey boxes indicated CD8⁺ T cells from AdC68-gDMelapoly vaccinated mice.

(*) indicates significant differences as detailed in legend to Fig. 1 between groups connected by lines (dashed lines for differences between control cells or between control cells and antigen-specific CD8⁺ T cells, solid lines for differences between specific CD8⁺ T cells). The following p-values were obtained for Trp-1₄₅₅ and Trp-2₁₈₀ comparisons: (Aa): Trp-1₄₅₅ vs. Trp-2₁₈₀: Melapoly: p < 0.0001, gDMelapoly: p < 0.0001; (Ab): none; (Ba) Trp-1₄₅₅ vs. Trp-2₁₈₀: Melapoly: p < 0.0001, gDMelapoly: p = 0.0056; (Bb) Trp-1₄₅₅ vs. Trp-2₁₈₀: Melapoly: p < 0.0001; (Ca) Trp-1₄₅₅ vs. Trp-2₁₈₀: gDMelapoly: p < 0.0001; (Cb) Trp-1₄₅₅ vs. Trp-2₁₈₀: Melapoly: p < 0.0001, gDMelapoly: p = 0.028, Melapoly vs. gDMelapoly: Trp-1₄₅₅: p = 0.0008.

5B Percentages of tumor-derived naïve CD44⁻CD8⁺, tetramer⁻CD44⁺CD8⁺, and tetramer⁺CD44⁺CD8⁺ T cells that were positive for a given marker or combinations of exhaustion markers. Cells from AdC68-Melapoly vaccinated mice are shown as open boxes, cells from AdC68-gDMelapoly vaccinated mice are shown as grey boxes. Whiskers show 5–95 percentile, lines show 50 percentile, + within boxes indicates means.

(*) indicates significant differences between groups as described in legends to Fig. 1. (Aa) Trp-1₄₅₅ vs. Trp-2₁₈₀: Melapoly: p = 0.031. (Ab) Melapoly vs. gDMelapoly: Trp-1₄₅₅: p = 0.035; Trp-1₄₅₅ vs. Trp-2₁₈₀: Melapoly: 0.019. (Ba): Trp-1₄₅₅ vs. Trp-2₁₈₀: Melapoly: p < 0.0001, gDMelapoly: p < 0.0001. (Bb) Melapoly vs. gDMelapoly: Trp-1₄₅₅: p = 0.0036; Trp-1₄₅₅ to Trp-2₁₈₀: Melapoly: p < 0.0001, gDMelapoly: p < 0.0001. (Ca) Melapoly vs. gDMelapoly: Trp-1₄₅₅: p = 0.0006, Trp-1₄₅₅ vs. Trp-2₁₈₀: Melapoly p = 0.016. (Cb) Melapoly vs. gDMelapoly: Trp-1₄₅₅: p = 0.039; Trp-1₄₅₅ vs. Trp-2₁₈₀: Melapoly: p = 0.0004, gDMelapoly: p = 0.021.

In addition, we determined the percentages of Trp-1₄₅₅- and Trp-2₁₈₀-specific CD8⁺TILs that were positive for either of the exhaustion markers and compared the results with those on naïve (CD44⁻) or memory tetramer⁻ CD8⁺ T cells (Fig. 5B, Suppl. Fig. 4). Only very low percentages of CD44⁻CD8⁺ T cells expressed any of the three exhaustion markers (< 2%). Percentages of marker⁺CD44⁺tet⁻CD8⁺ T cells were intermediate (2B4: 19%, 15%; LAG-3: 17%, 12%; PD-1: 23%, 41% for Melapoly and gDMelapoly vaccinated mice). Expression of individual markers was significantly higher on CD44⁺CD8⁺ cells than on naïve cells. Percentages of Trp-1₄₅₅-or Trp-2₁₈₀-specific CD8⁺ T cells positive for any of the three exhaustion markers were again significantly higher when compared to CD44⁻CD8⁺ or tetramer⁻CD44⁺CD8⁺ cells. Comparing the two vaccine subcohorts or Trp-1₄₅₅- and Trp-2₁₈₀-specific CD8⁺ T cells within each vaccine group showed no significant differences for percentages of 2B4⁺ cells (means: Melapoly: Trp-1₄₅₅ vs. Trp-2₁₈₀: 52% vs. 39%, gDMelapoly: Trp-1₄₅₅ vs. Trp-2₁₈₀: 44% vs. 33%). LAG-3⁺ cells were significantly more common within Trp-1₄₅₅- specific CD8⁺ T cells in both vaccine groups (means: Melapoly: Trp-1₄₅₅ vs. Trp-2₁₈₀: 87% vs. 48%, gDMelapoly: Trp-1₄₅₅ vs. Trp-2₁₈₀: 81% vs. 58%) as were PD-1⁺CD8⁺ cells in Melapoly-vaccinated mice. Interestingly percentages of PD-1⁺ Trp-1₄₅₅-specific CD8⁺ cells were significantly lower in the AdC68gD-Melapoly group than in mice immunized with Melapoly (means: Melapoly: Trp-1₄₅₅ vs. Trp-2₁₈₀: 84% vs. 67%, gDMelapoly: Trp-1₄₅₅ vs. Trp-2₁₈₀: 63% vs. 57%).

Exhaustion leads to sequential expression of several markers (27). Accordingly, we further analyzed tumor-derived T cells for co-expression of PD-1, LAG-3 and 2B4 (Fig. 5B). Again we found that higher percentages of Trp-1₄₅₅- and Trp-2₁₈₀-specific CD8⁺ T cells co-expressed two or three markers compared to CD44⁻CD8⁺ or tetramer⁻CD44⁺CD8⁺ cells. Trp-1₄₅₅-specific CD8⁺ T cells had in general higher frequencies of PD-1⁺LAG-3⁺ and PD-1⁺LAG-3⁺2B4⁺ cells compared to Trp-2₁₈₀-specific CD8⁺ T cells. Trp-1₄₅₅-specific CD8⁺ cells expressing two or three of the markers were present at significantly lower frequencies in gDMelapoly than Melapoly-immunized mice (means: Melapoly vs. gDMelapoly: PD-1⁺2B4⁺: Trp-1₄₅₅: 47% vs. 35%, Trp-2₁₈₀: 32% vs. 24%; PD-1⁺LAG-3⁺: Trp-1_455: 69% vs. 53%, Trp-2₁₈₀: 29% vs. 24%; PD-1⁺2B4⁺LAG-3⁺: Trp-1_455: 45% vs. 34%, Trp-2₁₈₀: 25% vs. 20%).

Markers that are indicative for exhaustion are also upregulated upon activation. To ensure that AdC68-gDMelapoly-induced TILs showed better preservation of function as would be expected of less exhausted CD8⁺ T cells we compared the proportions of Trp-1 specific CD8⁺ T cells that produced only IFN-γ or IFN-γ together with TNF-α between the two vaccine groups. For these comparisons cells were isolated from tumors at days 30–35 after vaccination. AdC68-gDMelapoly induced Trp-1-specific CD8⁺ T cells had significantly higher proportions of Trp-1 specific CD8⁺ TILs that produced both cytokines compared to AdC68-Melapoly induced TILs (mean AdC68-gDMelapoly: 0.47, AdC68-Melapoly: 0.3, p = 0.04 by two-tailed t-test). To further ensure that Trp-1-specific TILs with increased expression of PD-1 or PD-1 and LAG-3 were indeed exhausted rather than recently activated, we tested PD-1⁺ and PD-1⁻ TILs for production of IFN-γ and TNF-α upon stimulation with the Trp-1₄₅₅ peptide. As shown in Fig. 6A, slightly higher frequencies of Trp-1 specific CD8⁺ T cells from blood produced IFN-γ together with TNF-α than IFN-γ only. In tumors this proportion shifted and nearly 70% of Trp-1 specific TILs produced only IFN-γ suggesting loss of function (mean frequencies of Trp-1-specific CD8⁺ T cells: blood: IFN-γ⁺TNF-α⁻ vs. IFN-γ⁺TNF-α⁺: 0.32% vs. 0.57%; tumor: 1.85% vs. 0.84%). Co-stains for PD-1 showed significantly higher PD-1 expression on monofunctional Trp-1-specific CD8⁺ TILs as compared to those that produced both cytokines (mean PD-1 MFI values: blood: IFN-γ⁺TNF-α⁻ vs. IFN-γ⁺TNF-α⁺: 174 vs. 143,. tumor: 1236 vs. 774, Fig. 6B). An analysis for expression of T-bet, which controls transcription of effector molecules and inhibits expression of PD-1 showed that T-bet levels were low on naïve circulating CD8⁺ T cells and markedly increased on blood-derived Trp-1-specific CD8⁺ T cells. Tumor-infiltrating Trp-1-specific CD8⁺ T cells showed a marked reduction in T-bet expression (Fig. 6C), which corresponded to expression levels of exhaustion/activation markers (mean T-bet MFI values: CD44⁻CD8⁺ T cells: 6.2; Trp-1 specific CD8⁺ T cells blood vs. TILs: 464.2 vs. 204.6); Trp-1-specific CD8⁺ TILs that expressed high levels of PD-1 or PD-1 and LAG-3 had reduced levels of T-bet as compared to those with low PD-1 or PD-1 and LAG-3 expression (Fig. 6D–F) (mean T-bet MFI values: PD-1⁺ vs. PD-1⁻: 191.4 vs. 286.2; PD-1⁺LAG-3⁺ vs. PD-1⁻LAG-3⁻ : 181.3 vs. 256). Overall these results show that increased expression of PD-1 was associated with loss of polyfunctionality and downregulation of T-bet, confirming that MAA-specific CD8⁺ T cells induced by the vaccines were differentiation towards exhaustion within the TME.

FIGURE 6. MAA-specific T cell functions in relation to phenotypes.

Mice with 3-day old B16Braf_V600E tumors were vaccinated with 10¹⁰ vp of AdC68-gDMelapoly. Lymphocytes from blood and tumors were isolated 4 weeks later. (A) Frequencies of CD8⁺ T cells producing IFN-γ only or IFN-γ together with TNF-α were determined by ICS. By t-tests frequencies of Trp-1-specific IFN-γ⁺TNF-α⁺CD8⁺ T cells were higher in blood than those of IFN-γ⁺TNF-α⁻⁻ CD8⁺ T cells (p = 0.036) while in TILs frequencies of Trp-1-specific CD8⁺ T cells producing only IFN-γ were significantly higher (p = 0.0066). (B) shows levels of PD-1 expression on Trp-1-specific CD8⁺ T cells that produced only IFN-γ or IFN-γ together with TNF-α. Expression levels were comparable in blood-derived cells but significantly higher on TILs that were only positive for IFN-γ than on those that produced both cytokines (p = 0.024). (C) shows levels of T-bet expression on naïve CD44⁻CD8⁺ lymphocytes from blood and on Trp-1-specific CD8⁺ T cells from blood and tumors. Trp-1-specific CD8⁺ T cells expressed significantly higher levels of T-bet as compared to naïve CD8⁺ T cells (p < 0.0001). T-bet expression was higher on blood than on tumor-derived Trp-1-specific CD8⁺ T cells (p < 0.0001). (D) shows levels of T-bet expression on Trp-1-specific TILs separated into subgroups according to levels of PD-1 or PD-1 and LAG-3 expression. Levels of T-bet expression were significantly higher on PD-1⁻ as compared to PD-1⁺ cells (p = 0.0001) and on PD-1⁻LAG-3⁻ cells than on PD-1⁺LAG-3⁺ cells (p = 0.0002). (E) Gating scheme onto PD-1⁺ and PD-1⁻ cells following initial gating on live lymphoid cells, CD8⁺ cells, CD44^hi cells and Trp-1 tetramer⁺ cells as shown in Suppl. Fig. 1. (F) Expression levels of T-bet on PD-1⁻ (dashed line) and PD-1⁺ Trp-1-specific CD8⁺ T cells.

Taken together these results demonstrate complex patterns of up-regulation of exhaustion markers during tumor progression that were affected by the vaccines, the TcR specificity as well as the anatomic sites from which T cell had been isolated. Within tumors MAA-specific CD8⁺ T cells induced by the AdC68-gDMelapoly vector were less exhausted compared to those induced by the AdC68-Melapoly vector, fewer cells expressed exhaustion markers and they were able to delay tumor progression for longer periods of time. In addition, tumor-infiltrating CD8⁺ T cells specific to the dominant Trp-1₄₅₅ epitopes showed more evidence of exhaustion compared to those specific to the subdominant Trp-2₁₈₀ epitope.

DISCUSSION

T cell activation is finely tuned upon binding of the TcR to its cognate antigen-MHC class I molecule complex through additional interactions with co-stimulatory and co-inhibitory receptors (28, 29). Co-inhibitory receptors include CTLA-4, PD-1, BTLA and others. BTLA, unlike PD-1 and CTLA-4, is expressed on both naïve and activated T cells (30) and thereby presumably regulates T cell activation at several stages. PD-1 and CTLA-4 inhibit downstream signals transmitted upon TcR ligation such as induction of the protein kinase B pathway (PKB), albeit through distinct pathways (31, 32). PKB in turn regulates cell survival as well as glucose metabolism and is therefore crucial for successful activation and expansion of T cells (33). The exact signaling pathway downstream of BTLA remains unknown. Similar to PD-1 and CTLA-4, the cytoplasmic domain of BTLA contains immune-receptor tyrosine based inhibitory motifs (34) that may dampen TcR signaling.

Although modulators of HVEM signaling have not yet been tested in humans, the TNFRSF14 locus, which encodes HVEM, has been linked to increased risks for ulcerative colitis (35), systemic lupus erythomatodus (36) and rheumatoid arthritis (37), stressing the importance of this regulatory pathway. Similarly, genetic deletions of BTLA or HVEM cause exacerbated inflammatory reactions (38) and accelerated rejection of partially mismatched transplants in mice (39).

It has been shown that HVEM is expressed on tumor cells such as B cell lymphomas (40) or metastatic melanomas (41). BTLA on the other hand is highly expressed on melanoma-specific CD8⁺ T cells and it has been argued that the impaired proliferative capacity of MAA-specific T cell in melanoma patients is linked to immunoinhibitory signals from HVEM⁺ tumor cells (41). In another study NY-ESA-1-specific CD8⁺ T cells from end-stage melanoma patients were shown to express elevated levels of PD-1 and BTLA (42). T cells were defective in their ability to proliferate or secret factors in response to antigenic stimulation, which could in part be restored by blocking PD-1 and/or BTLA signaling. This suggests that inhibitors of BTLA should be explored as additives to active immunotherapy of melanoma.

As we showed previously, blockade of the BTLA-HVEM pathway through HSV-1 gD during antigen-driven T cell stimulation numerically augments CD8⁺ T cell responses (17, 18). This was confirmed in the current study using vaccines expressing multiple epitopes derived from MAAs. In a prophylactic melanoma model, the superior protection achieved upon gD-mediated blockade of the HVEM pathways at the time of T cell induction was correlated with the magnitude of MAA-specific CD8⁺ T cell responses. Using different doses of the two vaccines, i.e., AdC68-Melapoly and AdC68-gDMelapoly, mice that developed approximately equal frequencies of MAA-specific T cells showed comparable levels of protection against tumor protection. In contrast in mice that were inoculated first with melanoma cells and then vaccinated with either of the two vectors, the magnitude of the vaccine-induced MAA-specific CD8⁺ T cell response was not the sole factor that determined vaccine efficacy; in mice with approximately comparable frequencies of MAA-specific CD8⁺ T cells following vaccination with the AdC68-Melapoly or AdC68-gDMelapoly vectors, those that received the latter survived significantly longer. Cytokine profiles of antigen-specific CD8⁺ T cells were largely comparable between the two vaccine groups before and after challenge with responses being dominated by CD8⁺ T cells producing IFN-γ alone or in combination with TNF-α (not shown) as is typical for T cells induced by Ad vectors.

Paradoxically, at the time of necropsy AdC68-Melapoly vaccinated animals had significantly higher frequencies of MAA-specific CD8⁺ T cells in some tissues compared to those vaccinated with AdC68-gDMelapoly. This again confirms that protection did not solely correlate with frequencies of vaccine-induced T cells. Further analysis of MAA-specific CD8⁺ T cells for expression of activation/exhaustion markers showed that within tumors lower frequencies of AdC68-gDMelapoly-induced CD8⁺ T cells expressed LAG-3, PD-1, PD-1 together with LAG-3 or 2B4 or all three markers as compared to AdC68-Melapoly induced CD8⁺ T cells. These markers are not only indicative for exhaustion but they are also increased upon recent CD8⁺ T cells activation. Antigens derived from the tumor may have provided further activation signals to the MAA-specific CD8⁺ T cells, which then contributed to their elevated levels on tumor-infiltrating MAA-specific CD8⁺ T cells. Nevertheless, as these activation signals should have been provided in both vaccine cohorts, we assume that the differences in expression levels on tumor-infiltrating MAA-specific CD8⁺ T cells in Melapoly and gDMelapoly-vaccinated reflected differences in their stage of exhaustion. This was further confirmed by loss of polyfunctionality and T-bet expression on cells with increased expression of immunoinhibitory markers. In turn, our finding suggests that the relative resistance to tumor-driven exhaustion of MAA-specific CD8⁺ T cells generated in presence of gD resulted in prolonged control of tumor progression.

T cell exhaustion, a consequence of continued antigen-driven-stimulation of T cells, was initially described in chronic viral diseases such as those caused by lymphocytic choriomeningitis virus infection of mice (26) or human immunodeficiency virus (43) or hepatitis C virus infection (44) of humans and subsequently in cancer patients (45, 46). Differentiation towards exhaustion is initially characterized by increased expression of PD-1 on the T cell surface, which over time is joined by other co-inhibitors (47). Expression of PD-1 is regulated by TcR ligation (48) and at the transcriptional level by NFAT1c (49). T cells with high affinity TcRs appear to be more sensitive to antigen-driven exhaustion compared to those with lower affinity TcRs (50). This is supported by our finding that within tumors CD8⁺ T cells specific to the subdominant Trp-2₁₈₀ epitope appeared less exhausted; vaccine-induced Trp-2₁₈₀-specific CD8⁺T cells expressed lower levels of PD-1 or LAG-3 as compared to CD8⁺ T cells specific to the more dominant Trp-1₄₅₅ epitope. The same pattern was seen for MAA-specific CD8⁺ T cells that were double or triple positive for PD-1, 2B4 and/or LAG-3. Notwithstanding, it should be pointed out that the finding that higher TcR affinity promotes exhaustion remains debatable, although it is compatible with the highly reproducible finding that exhaustion is primarily driven by continued TcR ligation. (51) Other pathways that prevent antigen-driven exhaustion of CD8⁺ T cells used by the gD-adjuvanted vaccine may have contributed the results, but cannot be explored without additional knowledge of the BTLA/CD160 signaling pathways.

Active cancer immunotherapy with traditional vaccines that boost tumor-specific T cell responses has performed poorly in patients with advanced cancer. Additional strategies that block the immunoinhibitory pathways may improve the efficacy of cancer vaccines. Here we show that adjuvanting a tumor vaccine with HSV-1 gD could induce tumor antigen-specific CD8⁺ T cells with increased resistance to exhaustion. This approach, unlike other reagents that block immunological checkpoints such as antibodies to PD-1 or CTLA-4, selectively affects tumor antigen-specific T cells and presumably will not globally perturb the exquisitely fine-tuned balance of the immune system.