High Therapeutic Efficacy of a New Survivin LSP-Cancer Vaccine Containing CD4+ and CD8+ T-Cell Epitopes

The efficacy of an antitumoral vaccine relies both on the choice of the antigen targeted and on its design. The tumor antigen survivin is an attractive target to develop therapeutic cancer vaccines because of its restricted over-expression and vital functions in most human tumors. Accordingly, several clinical trials targeting survivin in various cancer indications have been conducted. Most of them relied on short peptide-based vaccines and showed promising, but limited clinical results. In this study, we investigated the immunogenicity and therapeutic efficacy of a new long synthetic peptide (LSP)-based cancer vaccine targeting the tumor antigen survivin (SVX). This SVX vaccine is composed of three long synthetic peptides containing several CD4+ and CD8+ T-cell epitopes, which bind to various HLA class II and class I molecules. Studies in healthy individuals showed CD4+ and CD8+ T-cell immunogenicity of SVX peptides in human, irrespective of the individual's HLA types. Importantly, high frequencies of spontaneous T-cell precursors specific to SVX peptides were also detected in the blood of various cancer patients, demonstrating the absence of tolerance against these peptides. We then demonstrated SVX vaccine's high therapeutic efficacy against four different established murine tumor models, associated with its capacity to generate both specific cytotoxic CD8+ and multifunctional Th1 CD4+ T-cell responses. When tumors were eradicated, generated memory T-cell responses protected against rechallenge allowing long-term protection against relapses. Treatment with SVX vaccine was also found to reshape the tumor microenvironment by increasing the tumor infiltration of both CD4+ and CD8+ T cells but not Treg cells therefore tipping the balance toward a highly efficient immune response. These results highlight that this LSP-based SVX vaccine appears as a promising cancer vaccine and warrants its further clinical development.

The efficacy of an antitumoral vaccine relies both on the choice of the antigen targeted and on its design. The tumor antigen survivin is an attractive target to develop therapeutic cancer vaccines because of its restricted over-expression and vital functions in most human tumors. Accordingly, several clinical trials targeting survivin in various cancer indications have been conducted. Most of them relied on short peptide-based vaccines and showed promising, but limited clinical results. In this study, we investigated the immunogenicity and therapeutic efficacy of a new long synthetic peptide (LSP)-based cancer vaccine targeting the tumor antigen survivin (SVX). This SVX vaccine is composed of three long synthetic peptides containing several CD4 + and CD8 + T-cell epitopes, which bind to various HLA class II and class I molecules. Studies in healthy individuals showed CD4 + and CD8 + T-cell immunogenicity of SVX peptides in human, irrespective of the individual's HLA types. Importantly, high frequencies of spontaneous T-cell precursors specific to SVX peptides were also detected in the blood of various cancer patients, demonstrating the absence of tolerance against these peptides. We then demonstrated SVX vaccine's high therapeutic efficacy against four different established murine tumor models, associated with its capacity to generate both specific cytotoxic CD8 + and multifunctional Th1 CD4 + T-cell responses. When tumors were eradicated, generated memory T-cell responses protected against rechallenge allowing long-term protection against relapses. Treatment with SVX vaccine was also found to reshape the tumor microenvironment by increasing the tumor infiltration of both CD4 + and CD8 + T cells but not Treg cells therefore tipping the balance toward a highly efficient immune response. These results highlight that this LSP-based SVX vaccine appears as a promising cancer vaccine and warrants its further clinical development.
Keywords: cancer vaccine, immunotherapy, long synthetic peptide, T-cell responses, tumor associated survivin antigen INTRODUCTION Therapeutic vaccination aiming to stimulate tumor specific T-cell responses is a promising tool in antitumoral strategies. Hundreds of early phase clinical trials have already been performed so far but they showed only modest clinical effects (1)(2)(3). The main hurdles limiting efficacy of cancer vaccines have been identified: their weak immunogenicity, targeted tumor antigen often being self-antigens (4) or dispensable antigens, thus increasing the risk of immune tumor evasion by the loss of target antigen expression (5); their restricted indication to certain cancer types or patients, as they do not target broadly expressed tumor antigens or do not use promiscuous T-cell epitopes restricted to various HLA molecules; the difficulty in achieving the right balance of CD4 + and CD8 + T-cell responses required to induce effective anti-tumor immune responses and long-term protection against relapses since CD4 + T cells are required for CD8 + memory Tcell induction and maintenance (6)(7)(8). Finally, a major obstacle is the presence of immunosuppressive mechanisms in the tumor microenvironment (3,9). Specifically, therapeutic vaccination can induce regulatory CD4 + T cells (Treg) limiting the vaccine efficacy (10)(11)(12).
To address these hurdles, personalized cancer vaccine based on mutant tumor antigens (neoantigens) (13), and vaccine targeting universal tumor antigens are very attractive approaches. As such, the inhibitor of apoptosis (IAP) protein survivin is an attractive target for anticancer immunotherapy, since it is a near universally over-expressed tumor antigen in human cancers, whereas its expression is low or undetectable in most differentiated adult tissues (14). Human transcriptome analysis reveals that survivin is the fourth most highly expressed transcript in human cancer cells when compared to normal cells (15). In multiple cancers, survivin expression is a marker of poor prognosis (16).
In addition, survivin exerts critical functions in tumor cells by contributing notably to apoptosis evasion, cell division, resistance to therapy (chemotherapy and radiotherapy) (17,18) and angiogenesis (19). Therefore, a potential down-regulation or loss of survivin expression as a strategy of immune escape would severely impair tumor cell's survival capacity.
Accordingly, several clinical trials have been conducted with different vaccine candidates targeting survivin in multiple cancer indications. These studies showed promising but limited clinical results (20)(21)(22). We hypothesized that all these defects relied on the inability of these vaccines to induce both robust and balanced CD4 + and CD8 + T-cell responses as most of them are based on short peptides mainly focusing on CD8 + cytotoxic T lymphocytes (CTL) responses. Long synthetic peptides (LSPs) are a vaccine modality using 25-to 100mer peptides that can contain both CD4 + and CD8 + T-cell epitopes. Several studies have demonstrated the relevance of LSPbased vaccines over short peptides or recombinant proteins to generate robust and long-term CD4 + and CD8 + T-cell responses, while limiting the induction of immune tolerance (7,(23)(24)(25). LSP-based vaccines were found to be safe, well-tolerated, and showed promising clinical efficacy in patients with pre-neoplastic lesions (3).
In this study, we thus designed a new survivin-based vaccine composed of three LSPs (SVX) encompassing multiple CD4 + and CD8 + T-cell epitopes. We evaluated the immunogenicity of the SVX peptides by conducting in vitro assays on human blood samples (from both healthy and cancer cell donors) and the therapeutic efficacy of the SVX vaccine in vivo in murine models using various established tumor cell lines.

Generation and Specificity of SVX-Specific T-Cell Lines Isolated From Healthy Donors
Monocyte-derived dendritic cells (DCs) generated as previously described (30) were loaded with the pool of survivin peptides (S1, S2, and S3, each at 5 µMol) and incubated for 4 h at 37 • C.
For the induction of CD8 + T-cell lines, CD8 + T cells were isolated by positive selection using anti-CD8 mAbs coupled to magnetic microbeads (Miltenyi Biotec). Pulsed mDCs were washed and 2 × 10 4 were added to each round-bottom microwell already containing 2 × 10 5 autologous CD8 + T lymphocytes  in 200 µL IMDM medium (Invitrogen) supplemented by 10% Human AB serum (Lonza), and 30 ng/mL IL-21 (R&D systems). After 3 days of culture, 5 ng/mL IL-7 and 5 ng/mL IL-15 (R&D systems) were added to the cultures. CD8 + T-cell lines were restimulated once on day 7 with 2 × 10 4 autologous peptideloaded DC and were grown in complete IMDM supplemented with IL-7 and IL-15 (5 ng/mL, each). The specificity of the CD8 + T-cell lines was assessed by IFN-γ ELISpot assay on day 14 using C1R-A2 cells loaded with the pool or the individual SVX peptides.
A response was considered positive if the number of spots per well obtained in peptide(s) stimulated conditions was two-fold higher than the number of spots counted without peptide(s), with a cut-off at 10 spot-forming cells after subtracting background.

Assessment of Survivin-Specific T-Cell Responses in Cancer Patients
PBMCs were cultured for 7 days at 2 × 10 6 cells/ml with the pool of SVX peptides (each peptide at a concentration of 10 µg/mL in complete RPMI 1640 medium supplemented with 5% serum AB). Cultured T cells were then added with SVX peptides to triplicate wells at 1 × 10 5 cells/well in AIM V medium for 24 h at 37 • C in 5% CO 2 . Plates were revealed using human IFN-γ ELISpot assays (Diaclone). A response was considered positive if the number of spots per well obtained in peptide(s) stimulated conditions was two-fold higher than the number of spots counted without peptide(s), with a cut-off at 10 spot-forming cells after subtracting background.
For CD8 + T-cell depletion studies, 100 µg of anti-CD8 antibody (clone 2.43; BioXcell) or isotype control antibody (rat IgG2a) was administered i.p to tumor bearing mice the day before vaccination and each subsequent week. CD8 depletion was verified by flow cytometry.

IFN-γ Elispot
Spots were counted using an ImmunoSpot analyzer (C.T.L) and enumerated as number of spot-forming cells per well. Cells incubated with medium alone or 100 ng/mL of phorbolmyristate acetate (PMA) and 500 ng/mL of ionomycin (Sigma Aldrich) were used as negative and positive controls, respectively. The number of specific T cells was calculated after subtracting negative control values. A response was considered positive if the number of spots per well-obtained in peptide(s) stimulated conditions was two-fold higher than the number of spots counted without peptide(s), with a cut-off at 10 spot-forming cells.

Tumor Dissociation
Tumors were extracted from mice upon termination and chopped into small pieces using a scalpel. Pieces were mechanically ground using a gentleMACS TM Dissociator (Miltenyi Biotec), filtered (40 µM) and treated with ACK solution. Cells were then washed in PBS and used for flow cytometry.

Statistical Analysis
Results are expressed as means ± SEM. The Mann-Whitney test was used to compare two groups. The Kruskal-Wallis test was used to compare three or more groups. Comparison between tumor growth curves has been performed using a two-way ANOVA test and multiple comparisons have been corrected with the Bonferroni coefficient. Statistical significance was determined with Prism software (GraphPad Software). Significance was assumed at P < 0.05.

Induction of CD4 + and CD8 + T-Cell Responses Specific for SVX in Healthy Donors
The SVX vaccine is composed of three long synthetic peptides (LSPs) derived from the native sequence of the human tumor antigen survivin (namely S1, S2, and S3) ( Table 1). Collectively, these new LSPs contain 8 promiscuous CD4 + T-cell epitopes (30,34) overlapping with 6 main described CD8 + T-cell epitopes restricted to HLA-A2, A3, A1, and A24 for which spontaneous CD8 + T-cell responses have been identified in numerous cancer patients (26-29, 35, 36) (Table 1). In order to anticipate the T-cell responses in vaccines as shown for a HIV LSP vaccine (37), the SVX peptides were introduced in longterm in vitro T-cell amplification assays using cells collected from healthy donors. We first quantified the CD4 + SVXspecific T-cell responses in 12 healthy donors with diverse HLA-DRB1 genotypes (Supplementary Table 1). Purified CD4 + T lymphocytes were seeded in multiple wells and stimulated weekly by autologous DCs previously loaded with the pool of SVX peptides. After three rounds of stimulation, the specificity of the T-cell lines (each well containing the expanded T cells) was tested by IFN-γ ELISpot assays. All donors generated a T-cell response specific for at least two of the three SVX peptides ( Figure 1A and Supplementary Table 1). The SVX-specific CD4 + T-cell responses were highly directed against the S1 and S3 peptides, generating specific T-cell response in 91.7 and 100% of the donors respectively, while S2 peptide generated specific T-cell response in only 75% of the donors. We also determined the intensity of such responses by defining the percentage of positive T-cell lines for each patient (Figure 1B and Supplementary Table 1). An average of 31.7% ± 3.9 of T-cell lines contained specific CD4 + T lymphocytes against the pool of SVX peptides and 17.8% ± 3.5 and 16.4% ± 2.9 of T-cell lines contained S1 or S3 specific CD4 + T lymphocytes, respectively. S2 peptide was found to be slightly but not significantly less immunogenic than S1 and S3 as 10.3% ± 4.4 of T-cell lines contained S2-specific CD4 + T lymphocytes ( Figure 1B).
The CD8 + T-cell response to SVX peptides was also evaluated in long-term in vitro T-cell amplification assays using CD8 + T cells harvested from HLA-A * 02:01 positive healthy donors. All donors generated a specific T-cell response to the S1 and S2 peptides, which contain HLA-A2 restricted CD8 + T-cell epitopes ( Table 1) (Figure 1C and Supplementary Table 2) but not to the S3 peptide which does not incorporate any HLA-A2 restricted CD8 + T-cell epitope. In addition, an average of 85% ± 7.6 of the seeded T-cell lines were found to contain S1 and S2 specific CD8 + T-cell lymphocytes, corresponding to a high level of immunogenicity (Figure 1D and Supplementary Table 2).
Collectively, these data demonstrated the high CD4 + and CD8 + T-cell immunogenicity of the SVX peptides in healthy individuals, regardless of the individual's HLA type and confirmed their potential in vaccination to generate survivin specific T-cell responses.

Spontaneous T-Cell Responses Against SVX Peptides in Selected Cancer Patients
In cancer patients, immune tolerance may preclude efficient Tcell activation. To evaluate the presence of pre-existing SVXspecific T-cell precursors in cancer patients, PBMCs (including both CD4 + and CD8 + T cells) of cancer patients (and healthy donors as controls) were cultured with the three SVX peptides for 1 week (short-term in vitro culture). Spontaneous T-cell responses to SVX peptides, assessed by IFN-γ ELISpot assays, were found in cancer patients [9/11 head and neck (82%), 6/10 lung (60%) and 11/14 (79%) renal cancers], but not in healthy donors ( Figure 1E). These results demonstrate the presence of SVX specific T-cell precursors and the absence of immune tolerance against SVX peptides in most cancer patients studied. For the mRCC responder patients, specific spontaneous T-cell responses against each individual peptide have been monitored ( Figure 1F). Interestingly the responses were heterogeneous between each patient, with some patients eliciting a specific T-cell response against one individual peptide (either S1, S2, or S3), some against two peptides and one patient against the three peptides. These results highlight that pooled together, the three peptides will elicit specific-SVX responses in a large fraction of cancer patients irrespective of their HLA's types. The demonstration that specific spontaneous responses against SVX peptides were detected in cancer patients was a prerequisite to develop a preclinical model to evaluate their potential as a vaccine.

Survivin Vaccine Is Immunogenic in Mice
Based on these results, we next performed preclinical vaccination in mouse models benefiting from the fact that the human and the mouse survivin proteins are highly conserved (84% identical and 91% homologous) (38). Importantly, the human SVX peptides contained highly conserved immunodominant epitopes described in mice, notably in H2 d background (33,39,40) (Supplementary Table 3) allowing evaluating the immune response induced by this new SVX vaccine in BALB/c mice. We first validated the immunogenicity of SVX vaccine following immunization in tumor-free BALB/c mice. Three weeks after vaccination, we performed an in vitro restimulation of splenocytes with either the mix of three LSPs or each individual peptide (Figure 2A, left graph). In all conditions, specific Tcell responses of similar intensity were observed as assessed by IFN-γ secretion, confirming the immunogenicity of SVX vaccine in mice. Similar results were obtained with murine tumor-cell lines expressing the human survivin (Figure 2A, middle graph) demonstrating that naturally processed epitopes were also recognized. Next, we studied the impact of different adjuvants or adjuvant combinations on the immunogenicity of SVX vaccine. T-cell immunogenicity of SVX vaccine was found to be significantly higher when formulated with CpG (TLR9), Poly ICLC (TLR3), or the combination of CpG with GM-CSF, compared to IFA, IC31, or MPLA (Figure 2A, right graph). The combination of CpG with IFA provided the highest immunogenicity of SVX vaccine and was thus selected as the optimal adjuvant combination.

High Therapeutic Efficacy of SVX Vaccine Against Various Established Tumor Models
The therapeutic efficacy of the formulated SVX vaccine candidate was then evaluated in BALB/c mice engrafted with various syngeneic tumor models expressing the human survivin. Treatment with SVX vaccine was found to significantly inhibit the growth of established colorectal carcinoma (hCT26) (Figure 2B, left) and B-lymphoma (hA20) (Figure 2C, left) tumor cells compared to non-immunized mice or mice treated only with adjuvant combination (Figure 2D). To fully optimize vaccination's protocol, different strategies of boosting vaccination were investigated. Performing SVX boost + adjuvants (CpG/IFA) at day 7 or at day 14 provided similar efficacy as a boost without adjuvants at day 7 ( Figure 2E). Inhibition of tumor growth was also observed with hRenca tumor cells in BALB/c mice or hSarc-A2 tumor model in HLA-A2/DR1 transgenic mice (Figures 2F,G). SVX vaccine also led to increased survival as 37.5% of mice were still alive at day 40 in the hCT26 model and 50% of mice still alive at day 50 in the hA20 model (Figures 2B,C, middle graphs).
This was associated with an induction of survivin-specific T-cell responses secreting high amounts of IFN-γ of similar intensity in vaccinated tumor-bearing (TB) and tumor-free (TF) mice (Figures 2B,C, right graphs).
Remarkably, over 5 experiments (representing 40 mice) performed in both hCT26 and hA20 models, treatment with SVX vaccine resulted in complete and sustained eradication of tumor in 27.5 and 42.5% of mice, respectively ( Figure 3A). The capacity of SVX vaccine to induce anti-tumor memory immune responses was evaluated in all SVX-vaccinated mice that eradicated hA20 tumor cells in primary response. After an initial but small tumor growth, all animals completely eradicated tumor cells after a secondary challenge, compared to naive mice grafted with tumor cells (Figure 3B), leading to 100% of mice survival for more than 60 days ( Figure 3C).
Altogether, these data demonstrate the high therapeutic efficacy of SVX vaccine against various established tumor cells, associated with its capacity to induce robust and specific Tcell responses but also effective memory T-cell responses for long-term protection against relapses.

The Therapeutic Efficacy of SVX Vaccine Is Mainly Mediated by CD8 + T Cells
To evaluate the capacity of SVX vaccine to generate specific CD8 + T-cell responses and their role in tumor eradication, similar tumor rejection assays were performed in CD8-depleted Tumor-free BALB/c (H2 d ) mice were subcutaneously (s.c) vaccinated as followed: priming with the three survivin LSPs (SVX) and adjuvants and boost 2 weeks later with SVX without adjuvant. One week after the boost, the induction of survivin-specific T-cell responses was analyzed by IFN-γ ELISpot assay on total splenocytes (2 (Continued)  mice. In absence of CD8 + T cells, the therapeutic efficacy of SVX vaccine was totally abolished against established hCT26 tumors (Figures 4A,C) while it was only partially impaired against hA20 tumors (Figures 4B,D). As a control, CD8 depleting treatment was also performed in non-vaccinated tumor-bearing (TB) animals. Similar tumor growths were observed in CD8depleted and isotype control treated tumor-bearing animals, in both hCT26 ( Figure 4C) and hA20 models ( Figure 4D).
Finally, we monitored by flow cytometry, the expression of Granzyme B (GrB) by CD8 + T cells isolated from the tumor of SVX-vaccinated or not vaccinated mice. The percentage of GrB + CD8 + T cells was found to be substantially higher among tumor infiltrating lymphocytes (TILs) isolated from vaccinated compared to non-vaccinated mice ( Figure 4F).
Altogether, these results highlight the SVX vaccine's capacity to generate robust and effective anti-tumoral CTL responses and demonstrate their crucial role in its therapeutic efficacy against different MHC class I + tumor cell types.

SVX Vaccine Induces Robust and Specific Th1 Responses
To evaluate the impact of CD4 + T-cell responses, we initially performed CD4 depletion in both hCT26 and hA20 models. CD4 depletion did not affect SVX vaccine efficacy in both vaccinated groups compared to non-depleted groups (Supplementary Figure 1). However, as CD4 + T cells are a heterogeneous population containing both T helper (Th) and Treg cells, the experimental protocol using total CD4 depletion is most likely inappropriate. To more accurately evaluate the role of CD4 T-cell help, we analyzed the cytokine profile of CD4 + T-cell responses induced with SVX vaccine.
After 24 h of in vitro restimulation, we predominantly detected CD4 + T cells with a Th1 profile, in both hCT26 tumor-bearing (TB) and tumor-free (TF) vaccinated groups, secreting IFN-γ, IL-2, TNF-α, and GM-CSF. CD4 + T cells isolated from non-vaccinated TB mice secreted limited levels of these cytokines (Figure 5A). The production of IFN-γ evaluated FIGURE 4 | SVX therapeutic efficacy against tumor cells in CD8-depleted mice. BALB/c mice (8 mice per group) were engrafted s.c with hCT26 (A,C) or hA20 tumor cells (B,D). When tumors reached 10 mm 2 , mice were s.c injected with PBS, or were immunized with SVX + CpG/IFA and received a boost 1 week later without adjuvant (SVX). (A,B) Groups of vaccinated mice engrafted with hCT26 (A) or hA20 tumor cells (B) were depleted of CD8 + T cells, using anti-CD8 mAbs (100 µg) injected intra-peritoneally (i.p) once a week, starting 1 day before SVX immunization (SVX + αCD8). Data are presented as mean tumor size (mm 2 ) ± SEM from cohorts of 8 mice with *P < 0.05, **P < 0.01 and ***P < 0.001. Experiments have been done twice. (C,D) An additional group of mice for each experiment is shown and represents BALB/c mice (8 mice per group) engrafted with hCT26 (C) or hA20 tumors (D) and depleted of CD8 + T cells using anti-CD8 mAbs (100 µg) injected i.p once a week during 3 weeks. Data represent the mean of tumor size (mm 2 ) ± SEM at day 28 (C) or day 32 (D). **P < 0.01 and ***P < 0.001. (E) Intensity of survivin CD8 + specific T-cell responses in hCT26 TB mice vaccinated (SVX) or not with SVX (PBS). Evaluation of functional responses was performed 2 weeks after the last vaccination using IFN-γ ELISpot assays on total splenocytes restimulated overnight in vitro with the CD8 + T-cell epitope surv85-93. Data are presented as means of IFN-γ spots ± SEM of 16 mice per group from two independent experiments. **P < 0.01. (F) Same mice as in (E). Two weeks after the last immunization, the tumors were harvested, and the percentage of CD8 + GrB + in TIL was evaluated by flow cytometry. Data are presented as means of percentage of cells ± SEM of 16 mice per group from two independent experiments. ***P < 0.001. by ELISpot also showed similar level of IFN-γ secretion between vaccinated TB and TF mice (Supplementary Figure 2A). A same profile of cytokine secretion was observed in the B-lymphoma model (Supplementary Figures 2B,C). Cytokine production was also evaluated at 48 h in the hCT26 model. The level of cytokine secretion remained very low in nonvaccinated TB mice, while it was maintained and even increased in vaccinated groups. The production of IFN-γ, IL-2, and TNF-α was similarly increased in TB and TF vaccinated mice ( Figure 5B). Considering Th2 responses, low level of cytokine production was observed at 24 h (Figure 5A). At 48 h, CD4 + T cells isolated from vaccinated groups secreted significantly higher levels of IL-4 and IL-13 than those FIGURE 5 | Cytokine profile of the CD4 + T-cell responses induced with SVX vaccine. Tumor-bearing (TB) mice engrafted s.c with hCT26 (n = 8 per group) were s.c injected with PBS or vaccinated with SVX vaccine (SVX). Tumor-free (TF) were immunized with SVX vaccine (SVX). Two weeks after the last immunization, splenic CD4 + T cells were cell sorted by magnetic beads. CD4 + T cells (2 × 10 5 ) were then co-cultured with BM-DC (5 × 10 4 ) pulsed with medium or the pool of SVX peptides. Cytokine productions were measured by Luminex assay performed on the supernatant after 24 h (A) or 48 h (B) of culture. Data are mean ± SEM of 8 mice per group with *P < 0.05, **P < 0.01, and ***P < 0.001.
in non-vaccinated TB mice. However, low levels of IL-5 were detected in all groups (Figure 5B). SVX vaccine also induced some inflammatory cytokines such as IL-6 and IL-18 (data not shown), which could be detected in the supernatant of CD4 + T cells from vaccinated groups (Figures 5A,B).
Altogether these results demonstrate that SVX vaccine induces in vivo specific and multifunctional CD4 + T-cell responses with a predominant Th1 profile, not impacted by the presence of established tumors. This strong activation of CD4 + T cells could greatly participate to the high efficacy of the SVX vaccine by helping to induce and sustain CD8 + T-cell responses.
SVX Vaccine Highly Increase the Ratio of Both Conventional CD4 + and CD8 + T Cells Over Treg Cells but Has No Significant Impact on PD-1 Expression Among CD4 + T cells, the generation of immunosuppressive cells such as Treg cells may limit the efficacy of vaccine strategy. Using flow cytometry analysis, we thus evaluated the impact of SVX treatment on the frequencies and phenotypes of CD8 + T cells, CD4 + FoxP3 − Tconv, and CD4 + FoxP3 + Treg cells both in the spleen and among tumor infiltrating lymphocytes (TILs) (Figures 6A,B) of TB mice vaccinated or not with SVX but also in TF mice.
In SVX-vaccinated mice bearing hCT26 tumor, we observed significant increased frequencies of both splenic ( Figure 6C) and intra-tumoral ( Figure 6D) CD4 + Tconv and CD8 + T cells, and decreased frequencies of Treg cells. The effect was more pronounced within the tumor: two-fold increased in the vaccinated group compared to the non-vaccinated one. Similar results were observed in the spleen of TF vaccinated mice ( Figure 6C). SVX vaccine thus favored high CD4 + Tconv/Treg and CD8/Treg ratios in both vaccinated groups compared to non-vaccinated TB animals (Figures 6C,D, right graphs).
Intra-tumoral T-cell populations generated upon SVX vaccination were also evaluated by assessing the expression of markers associated with central vs. effector memory differentiation (CD44, CD62L); KLRG1, a marker associated with better CD8 + anti-tumoral responses upon peptide vaccination in some studies (41) and PD-1, as a marker of T-cell activation and/or exhaustion. TILs expressed high level of CD44 but low level of CD62L (Figure 6E), identifying them as effector cells. KLRG1 was highly expressed on Treg, but mildly expressed on CD4 + Tconv and CD8 + T cells. Finally, low frequencies of PD-1 expressing CD4 + T cells and Treg, and high frequencies of PD-1 expressing CD8 + T cells (more than 70%) were detected among TILs of TB mice. Importantly, the phenotype of each cell subset was not different in vaccinated and non-vaccinated mice ( Figure 6E). Similar observations were made in the hA20 model: a strong increase of intratumoral CD4 + Tconv/Treg and CD8/Treg ratios in vaccinated mice and no difference in PD-1 expression between vaccinated and non-vaccinated mice (Supplementary Figure 3).
Altogether, these results demonstrate the capacity of SVX vaccine to reshape the tumor microenvironment by strongly increasing the tumor infiltration of both CD4 + Tconv and CD8 + T cells over Treg cells therefore tipping the balance toward an efficient immune response.

DISCUSSION
A number of survivin-targeting immunotherapies have been developed so far including short peptide-based vaccines (20)(21)(22). These vaccines contained one or several HLA class I restricted survivin peptides and showed prolonged overall survival and induction of peptide specific CD8 + T-cell responses in various advanced cancers, with no adverse effects (42)(43)(44)(45)(46). However, CD8 + T-cell responses alone have not led to consistent clinical responses. Numerous studies have highlighted the fundamental role of CD4 + T cells in enhancing CD8 + T-cell responses and long term protection against relapses (20,34,47,48). Combining CTL and CD4 + antigen-specific T-cell activation during vaccination may provide a theoretical advantage to improve survivin-targeted vaccination approaches. One study using such a vaccine strategy has been recently tested in a phase I trial in patients with malignant glioma (49). The survivin vaccine relies on the modified peptide SVN53-67/M57-KLH (SurVaxM) that contains antigen-binding motifs for multiple HLA class I and one potential HLA class II (50). This study showed that six of eight patients developed both cellular and humoral responses and had an overall survival of 86 vs. 25.7 weeks in an indirect comparative chemotherapy clinical trial (51), suggesting that such strategy may be highly beneficial.
In our study, we developed a more promiscuous survivin (SVX) vaccine, composed of three LSPs containing numerous CD4 + and CD8 + T-cell epitopes, presented to a large spectrum of HLA class II and I molecules. We first demonstrated the immunogenicity of the SVX peptides after long-term in vitro amplification of human PBMCs from healthy donors displaying either various HLA-DRB1 genotypes or HLA-A2 molecules for CD4 + and CD8 + T-cell responses, respectively: all tested donors responded to SVX peptides. These results unequivocally demonstrate that SVX vaccine includes immunogenic CD4 + and CD8 + T-cell epitopes. Because immune tolerance is a major hurdle in therapeutic vaccine strategies, we evaluated the efficacy of SVX peptides to stimulate T cells from the peripheral blood of cancer patients in short-term culture. We observed that 60-82% of cancer patients were responsive to SVX stimulation. These data suggested that SVX vaccine might boost specific T-cell responses in various cancer types and in a high proportion of patients. The presence of SVX T-cell precursors in cancer patients, that can be activated or re-activated, is crucial. Indeed, evaluating the efficacy of the vaccine only makes sense if targeted specific T cells are available. These results obtained in human were an important pre-requisite before going further and assessing the efficacy of the SVX vaccine in preclinical model. In BALB/c mice, we validated the therapeutic efficacy of SVX vaccine against different established tumor models. In all models tested, the reduction of tumor growth was associated with the induction of survivin specific T-cell responses. These results emphasize the broad potential of our therapeutic survivin-based vaccine approach.
The use of human survivin vaccine in murine models can be considered as a xeno-vaccination model that has been shown in some studies to generate higher immunogenicity that syngeneic vaccination (39,52). Although the potential benefit of xeno-vaccination is debated (53,54), it may be a limitation that needs to be taken into account in our model. However, we also performed studies in the humanized HLA-A2/DR1 transgenic mice, which more closely resemble human peptide presentation. In these mice, the SVX vaccine allowed a delay in the hSarc-A2 tumor growth,  confirming the efficiency of the SVX vaccine in a human HLA context. In BALB/c mice, the protection upon secondary challenge with tumor cells demonstrated that SVX vaccine was able to generate effective anti-tumor memory responses resulting in long-term protection against relapses. We next evaluated the impact of SVX vaccine on CD4 + and CD8 + T-cell responses. As expected, tumor rejection assays performed in CD8-depleted mice highlighted the capacity of SVX vaccine to induce strong anti-tumoral CD8 + T-cell responses. Interestingly, the impact of SVX vaccine on tumor growth was less strictly mediated by CD8 + T cells in the hA20 than in hCT26 tumor model, suggesting that other immune cells responsive to the survivin vaccine may participate to tumor control. In contrast, CD4 + T-cell depletion failed to demonstrate any impact on tumor growth, presumably due to the opposite effects provided by Th and Treg CD4 + T cells, respectively. When addressing directly CD4 + T-cell functions, CD4 + T cells exhibited multifunctional cytokine responses to SVX peptides, with a predominant Th1 profile in vaccinated tumor-bearing animals. These strong CD4 + T-cell responses may greatly participate to the efficacy of the survivin vaccine, in particular, in helping effective CD8 + memory T-cell generation. Because CD4 + T-cell responses are heterogeneous, we also examined Treg proportions following SVX vaccination in spleen and tumors. We demonstrated that SVX vaccine highly increased tumor infiltration by both CD4 + Tconv and CD8 + T cells. Importantly there is no increase in the percentage of Treg cells as previously described in some therapeutic vaccine designs (10)(11)(12). Consequently, the SVX vaccine appears to favor the balance toward strong effector immune responses. In addition to the analysis of T-cell compartment, the evaluation of the myeloid-derived suppressor cells (MDSCs) and tumor associated macrophages (TAMs) populations in the tumor microenvironment would also be an important issue to address, as they play a major role in tumor-related immunosuppression and can hamper successful immunotherapy approaches (55,56).
Whereas, therapeutic vaccines may increase PD-1 expression on TILs (57, 58), we did not observed such increase in our study. Similar percentage of PD-1 expressing CD8 + TILs was observed between SVX-vaccinated and nonvaccinated animals. SVX vaccine might thus not be able to overcome all the immunosuppressive mechanisms developing during tumor development. These observations suggest that the combination of SVX vaccine with immune checkpoint blockade antibodies targeting PD-1 may potentially further increases its therapeutic efficacy against such established tumor cells, a hypothesis that we are currently addressing.
In summary, the SVX vaccine allowed to generate both CD4 + and CD8 + specific T-cell responses, leading in mice to high therapeutic efficacy in various tumor models. The SVX vaccine was also able to generate anti-tumor memory T-cell responses. Additionally, limited Treg expansion was observed, thus minoring Treg immunosuppressive effects. In humans, our assays on blood samples demonstrated that the SVX peptides were able to elicit specific T-cell responses in various cancer types and in a high proportion of patients. This highlights the absence of immune tolerance against the SVX peptides. Based on this set of experiments, SVX vaccine may constitute a very promising new vaccine candidate for clinical trials that could be used in a large spectrum of patients suffering from different cancers, and irrespectively of their HLA types.