INTRODUCTION

Breast cancer accounts for 28.2% of all cancer cases in European females and 16.2% of deaths in European women.1 Triple-negative breast cancers (TNBCs) constitute 10–20% of all breast cancers, and more frequently affect younger patients and women of African heritage. TNBCs are generally larger in size, are of higher grade, have lymph node involvement at diagnosis and are biologically more aggressive. The management of TNBC remains a significant clinical challenge because of the lack of targeted therapies, with patients having a poorer prognosis and relapsing more frequently than patients with hormone receptor-positive subtypes of the disease. Less than 30% of women with metastatic TNBC survive 5 years, and almost all die from their disease despite adjuvant chemotherapy and the initial higher rates of clinical response that can be achieved with neoadjuvant chemotherapy.2 The majority of TNBCs metastasize to the lungs, liver, bone and the brain, with brain metastases being particularly difficult to treat and associated with the poorest survival.3 More advanced, effective treatments for TNBC are, therefore, urgently needed.

The aim of cancer vaccine-based immunotherapy is to induce protective, robust and sustained antitumor immunity against relevant target antigens. Success is critically dependent on identifying appropriate tumor antigens based on their specificity and frequency of expression in cancer tissues.

Cancer testis antigens (CTAs) are almost exclusively expressed by the testes and/or placenta in healthy adult tissues, with their expression in other settings being primarily restricted to malignant tumors. These tumor-restricted expression and immunogenic features make CTAs ideal candidates as targets for immunotherapy. The helicase antigen (HAGE) (DEAD-box protein DDX43) has previously been shown to be expressed in 14.8% of early primary TNBC tumors and 43% of locally advanced primary TNBC.4 The role of HAGE as a potential prognostic marker for patients with breast cancer and a predictor of response to adjuvant chemotherapy has also been demonstrated.4 Herein, we have investigated HAGE as a potential therapeutic target for CTA-targeted immunotherapy for TNBC expressing HAGE.

Vaccination using a peptide sequence within a native protein which is predicted to be immunogenic can maximize the probability of inducing both helper and cytotoxic T-cell responses. Among several algorithms, the SYFPEITHI (http://www.syfpeithi.de) database enables the identification of immunogenic peptides within a protein sequence based on the presence and position of anchor residues that influence their binding affinities to different human leukocyte antigen (HLA) alleles.

The ease of synthesis and generation of large quantities of peptide or DNA vaccines and their relatively low cost have made these attractive formats for anticancer vaccines. Moreover, the use of long peptides incorporating epitopes for CD8 rather than using only CD8-specific epitopes, or the whole protein has been shown to be superior at inducing CD8+ T cells.5, 6 The use of long peptides also enables the possible presentation of CD4-specific T-cell epitopes. The use of adjuvants such as incomplete Freund’s adjuvant (IFA) and CpG-ODN (TLR9 agonist) promotes the recognition of these peptides by professional antigen-presenting cells (APCs), thereby increasing antigen cross-presentation and promoting type 1 T helper-type immunity.7 However, although many of these peptide/adjuvant vaccines have been reported to generate CD8+ T-cell responses, significant clinical efficacy has yet to be observed.8, 9

Delivery of antigen or antigenic fragments using viral or plasmid DNA vectors has received much attention in recent years. The ImmunoBody vaccine is a plasmid DNA designed to encode a human antibody molecule with complementarity-determining regions engineered to express cytotoxic and helper T-cell epitopes derived from the antigen of interest.10, 11 In the case of the study presented herein, these epitopes were derived from the HAGE sequence. ImmunoBody-DNA vaccines have been shown to maximize T-cell activation and the avidity of reactive cells by two distinct mechanisms: “direct” and “cross-presentation.” Direct presentation is achieved by the direct transfection of APCs and cross-presentation by secretion of antibody which efficiently stimulates helper and cytotoxic T-lymphocyte responses by effective targeting of dendritic cells via the high-affinity Fc receptors. The first developed ImmunoBody, called SCIB1, is a DNA plasmid that encodes a human antibody molecule engineered with T-cell epitopes (both helper and cytotoxic) derived from melanoma antigens TRP-2 and gp-100 within the complementarity-determining regions. The SCIB1 ImmunoBody is currently being evaluated as a therapeutic for melanoma in Phase I/II clinical trials and has been shown to significantly prolong survival rates, particularly in patients following the resection of disease.12, 13

The release of interferon-γ (IFNγ) by vaccine-induced T cells in response to tumor cells can induce or increase the expression of programmed death ligand 1 (PD-L1) by the tumor cells which, in turn, will downregulate the effector functions of activated T cells expressing programmed cell death protein 1 (PD-1) via PD-L1/PD-1 binding. We and others have previously shown that TNBCs exhibit a high level of tumor-infiltrating lymphocytes (TILs) and a higher expression of the immunoregulatory checkpoint pathway ligand PD-L1 than other breast cancer subtypes,4, 14, 15 and that the presence of TILs is associated with a favorable outcome. Antibody-mediated inhibition of PD-L1–PD-1 interactions (so-called checkpoint inhibition) combined with cytotoxic chemotherapies that induce immunogenic tumor cell death has been shown to induce an objective response in 39.4% of patients with metastatic TNBC.16 The checkpoint receptor PD-1 is primarily expressed by activated T cells, the effector functions of which are downregulated upon PD-L1 engagement. Consequently, the efficacy of PD-1 and/or PD-L1 checkpoint inhibitors relies on the expression of PD-1 on immune effector cells and PD-L1 on tumor cells. However, if the patients’ immune system has not been sufficiently stimulated, their T cells will not express PD-1 and their tumor is also unlikely to have been induced to express PD-L1, except for tumors expressing a mutated PTEN which has been shown to induce PD-L1 expression.17 Antigen/specific vaccination is, therefore, an approach that is likely to activate T cells and initiate the development of protective antitumor immunity.

The main aim of this study was to identify regions within the HAGE protein that harbor immunogenic epitopes having the potential to trigger CD4+ and CD8+ T-cell responses that can kill tumor cells expressing HAGE in vitro and tumors expressing HAGE in vivo. Herein, we have identified a 30-amino-acid-long immunogenic HAGE-derived sequence. Immune responses induced by a HAGE-30mer peptide-based vaccine and a HAGE ImmunoBody-DNA vaccine (ImmunoBody-HAGE) were then compared. ImmunoBody vaccines are designed to generate potent high-avidity T-cell responses capable of a broad antitumor effect. We have shown that immunizing animals with ImmunoBody-HAGE287–316 induced higher frequencies of HAGE peptide-specific T cells in the spleen than the peptide-adjuvant vaccine. Splenocytes from ImmunoBody-HAGE-vaccinated mice stimulated in vitro could recognize HAGE+ tumor cells (HHDII+DR1+HAGE+ B16) and the human TNBC cell line MDA-MB-231. More importantly, the growth of implanted HHDII+DR1+HAGE+ B16 cells was significantly delayed by the ImmunoBody-HAGE287–316 vaccine in both prophylactic and experimental metastasis settings. Overall, we demonstrate the potential of HAGE-derived vaccines for treating HAGE-expressing cancers and that such vaccines could be considered as therapeutic options for patients with TNBC expressing HAGE after conventional treatment to prevent disease recurrence.

RESULTS Immunogenicity of the HAGE-derived 30mer peptide

The SYFPEITHI database searching tool identified a 30mer HAGE-derived sequence encompassing several peptides that were predicted to bind to different HLA haplotypes including HLA-A2, A1, A3, DR1 and DR4 with high affinity. The binding scores for these sequences are shown in Table 1.

Table 1. Peptides derived from HAGE 30mer QTGTGKTLCYLMPGFIHLVLQPSLKGQRNR Peptide number Peptide position Sequence SYFPEITHI score Length HLA haplotype P3 287–296 QTGTGKTLCY 25 10mer HLA-A1 P4 297–305 LMPGFIHLV 28 9mer HLA-A2 P5 296–304 YLMPGFIHL 27 9mer HLA-A2 P6 296–305 YLMPGFIHLV 30 10mer HLA-A2 — 303–311 HLVLQPSLK 25 9mer HLA-A3 P7 291–305 GKTLCYLMPGFIHLV 30 15mer HLA-DR1 P8 298–312 MPGFIHLVLQPSLKG 26 15mer HLA-DR1 — 305–309 CYLMPGFIHLVLQPS 26 15mer HLA-DR4 — 309–313 PGFIHLVLQPSLKGQ 26 15mer HLA-DR4 HLA, human leukocyte antigen.

In order to determine the immunogenicity of the HAGE-derived 30mer sequence, as well as to compare the efficacy of two most commonly used adjuvants, namely, IFA18-21 and CpG ODN 1826,22-24 these adjuvants were tested individually and together with the HAGE vaccine. HHDII+DR1+ mice (described in Methods) were immunized with the 30mer peptide sequence on day 0 and a cocktail of MHC class I peptides (peptides 4, 5 and 6 from Table 1) on day 15 with IFA or CpG ODN 1826, either individually or in combination. A cocktail of MHC class I peptides was used as a boost in order to promote the expansion of CD8+ T cells specific for these peptides. Seven days after the last immunization, mice were culled and the responsiveness of splenocytes against all the predicted HLA-A2 and HLA-DR1 peptides (listed in Table 1) from within the HAGE 30mer sequence was assessed using an IFNγ ELISpot assay. The results of the ex vivo assays (Figure 1) show that the HLA-A*0201-restricted peptides 4 (HAGE297–305) and 6 (HAGE296–305) and the HLA-DRB*0101-restricted peptide 7 (HAGE291–305) induced the greatest peptide-specific IFNγ responses among all of the HLA-A2 and DR1 epitopes tested. Interestingly, despite the high similarity between the sequences of peptides 4, 5 and 6, only peptides 4 and 6 induced higher numbers of IFNγ-secreting cells. Combining IFA and CpG significantly enhanced the peptide-specific IFNγ response (Figure 1).

Ex vivo interferon-γ (IFNγ) ELISpot assay using splenocytes from mice immunized with HAGE 30mer peptide vaccine with different adjuvants. IFNγ cytokine release evaluated by IFNγ ELISpot assay using freshly isolated splenocytes derived from mice immunized with HAGE 30mer peptide formulation containing IFA, CpG and IFA+CpG. Graph shows the number of IFNγ spots released by splenocytes upon stimulation with MHC class I (1 µg mL−1) and MHC class II (10 µg mL−1) peptides. Data show that IFA in combination with CpG induces significantly higher number of IFNγ-releasing cells compared with cells induced by IFA or CpG individually. All data are presented as means ± SEM. (n = 3 mice per group in each independent experiment). **P-value < 0.01 and ****P-value < 0.0001.

Peptide versus DNA vaccine (ImmunoBody)

The ImmunoBody-HAGE construct was designed to carry the HAGE 30mer (HAGE287–316) peptide and this was inserted in the heavy variable region.

HHDII+DR1+ mice were immunized with either ImmunoBody-HAGE or HAGE 30mer peptide/IFA+CpG vaccine. Seven days after the last immunization, splenocytes from immunized animals were harvested and responsiveness to the HLA-A2 and DR1 peptides derived from the HAGE 30mer sequence listed in Table 1 was assessed in vitro using an IFNγ ELISpot assay.

Figure 2a demonstrates the superiority of the ImmunoBody-HAGE vaccine in generating a specific immune response against the assessed peptides, as shown by the significantly higher response (in the context of the number of cells secreting IFNγ) against peptides 5, 6 and 7. Interestingly, immunization of mice with the HAGE 30mer peptide followed by a boost using a cocktail of MHC class I peptides generated a significant number of IFNγ-secreting cells in response to peptide 4, a response which was not observed when animals were immunized using the ImmunoBody-HAGE. By contrast, strong immune responses against peptide 5 were observed (Figure 2). Immune responsiveness against peptide 6 was detected in both immunization strategies. These findings indicate that peptide 4 was not naturally endogenously processed, despite being predicted to have a higher binding affinity for HLA-A2 molecules.

Ex vivo ELISpot assay: comparison between the ImmunoBody-HAGE vaccine and 30mer peptide vaccine (IFA+CpG). IFNγ cytokine release by (a) IFNγ ELISpot assay using freshly isolated splenocytes derived from mice immunized with HAGE-ImmunoBody or HAGE-30mer peptide + IFA/CpG or no immunization plated with major histocompatibility complex (MHC) class I (1µg mL−1) and class II peptide at 1 and 10 µg mL−1 final concentration, respectively, at 37°C to measure the immune response induced by individual HAGE-derived short peptides between different immunization groups compared with naïve. Functional avidity of (b) peptide 5 and (c) peptide 6. All data are represented as mean ± SEM. (n = 3, three mice in each independent experiment). EC50, half maximal effective concentration; IFN, interferon. **P-value < 0.01, ***P-value < 0.001 and ****P-value < 0.0001.

The ImmunoBody-HAGE vaccine was also better at generating high-avidity T-cell responses than the peptide/IFA+CpG vaccination strategy (Figure 2b, c).

Induction of HAGE-specific cytotoxicity

The ability of immunization strategies to trigger cytotoxic T cells having the capacity to recognize and kill peptide-pulsed target cells and TNBC cells naturally expressing HAGE was then assessed.

As a model system, the murine melanoma cell line B16F1 (C57BL/6 background) was knocked out for both murine endogenous MHC class I and II by zinc finger nuclease technology and stably transfected with plasmids to express both HHDII and HLA-DR1 molecules. Thereafter, the resultant cells were transduced with HAGE-encoding viral construct and transfected with the luciferase (Luc2)-encoding plasmid vector. HAGE expression was confirmed at both messenger RNA (Figure 3a) and protein levels (Figure 3b, c). HHDII, DR1 expression (Figure 3d) and luciferase reporter gene expression (Figure 3e) of the HHDII+DR1+HAGE+Luc+ B16, Luc2 cells were assessed by flow cytometry and luciferase reporter assays, respectively.

Establishment of stable B16 cell line (knockout for murine β2-microglobulin and then transfected with chimeric HHDII and HLA-DR1) to coexpress HAGE and luciferase. B16 cells transduced with lentiviral particles from vector with and without HAGE were analyzed for HAGE expression. (a) HAGE messenger RNA (mRNA) levels in B16 transfectants showing three technical replicates. (b) Representative western blotting for HAGE expressions. The bands shown are from the same blot but probed with HAGE protein on top and β actin at the bottom, with PCI13 cell lysate used as a positive control for HAGE protein. (c) Immunofluorescence staining of HHDII+DR1+Luc+ B16 cells transfected with empty vector and HAGE. Cells were stained with DAPI (blue, nuclear staining) and primary DDX43antibody (red); magnification: 4×. (d) Flow cytometry analysis of HLA-DR1 and β2 microglobulin expression. (e) Luciferase expression measured as luminescence (total flux) with D-luciferin (15 µg mL−1) expressed activity in HHDII+DR1+HAGE+Luc+ B16 cells. HAGE expression assessed in TNBC cells (f) as mRNA level and (g) protein levels. Ab, antibody; DAPI, 4′,6-diamidino-2-phenylindole; HLA, human leukocyte antigen.

Splenocytes from mice immunized with either HAGE 30mer peptide vaccine or ImmunoBody-HAGE DNA vaccine were cultured for 1 week with the MHC class I peptide cocktail (peptide 4-LMPGFIHLV, 5-YLMPGFIHL, 6-YLMPGFIHLV), after which the ability of these cells to recognize and kill peptide-pulsed or HAGE-expressing target cells was assessed using IFNγ ELISpot and 51chromium release cytotoxicity assays. Splenocytes cocultured with target cells (T2 ± peptide, HHDII+DR1+HAGE+/–B16 cells MDA-MB-231 (HAGE+HLA-A2+) and MDA-MB-468 (HAGE+HLA-A−) in an IFNγ ELISpot plate showed HAGE-specific cytokine release in a peptide- and HLA-A2-restricted manner (Figure 4a). These cells were also able to specifically kill peptide-pulsed T2 (Figure 4b) cells and B16 cells expressing HAGE (Figure 4c). The ImmunoBody-HAGE DNA vaccine induced significantly higher cytotoxicity against HHDII+DR1+HAGE+ B16 tumor cells than the HAGE peptide vaccine. Moreover, the cytotoxicity of ImmunoBody-derived T cells against human MDA-MB-231 and MDA-MB-468 TNBC cells was assessed as they also express HAGE (Figure 3f, g). Figure 4d demonstrates the capacity of the ImmunoBody-HAGE vaccine to generate immune cells with the ability to specifically target HAGE-expressing and HLA-A2+ TNBC cells. The cytotoxicity against MDA-MB-231 (HLA.A2+) cells was significantly higher than the cytotoxicity against MDA-MB-468 (HLA.A2–) cells.

HAGE-specific CD8+ T-cell responses after in vitro stimulation: ELISpot assay. (a) HAGE-specific target induced interferon-γ (IFNγ) cytokine release by pure CD8+ T cells after 48 h of coculture with unpulsed and peptide-pulsed (100 µg mL−1) T2 cells, B16 +/− HAGE, or human triple-negative breast cancer (TNBC; MDA-MB-231, MDA-MB-468) cell lines to measure the IFNγ induced as a result of the presence of target cells. HAGE-specific target-induced IFNγ cytokine release was observed following incubation with HLA-A2+ and HAGE+ cell lines. Splenocytes from immunized mice were cocultured in vitro with LPS-activated cells and plated with 51chromium-labeled target cells. 51Chromium release assay to determine HAGE-specific cytotoxicity against (b) T2 cells pulsed overnight with a cocktail of MHC class I peptides, (c) HHDII+DR+Luc+ B16 cells+/– HAGE and (d) ImmunoBody-HAGE vaccine-induced T-cell cytotoxicity against MDA-MB-231 and MDA-MB-468 cells. Groups were compared to obtain statistical P-values by two-way ANOVA. All data presented are mean ± SEM, (n = 2, 3 mice per group) tested in triplicates in each independent experiments. HLA, human leukocyte antigen; LPS, lipopolysaccharide; MHC, major histocompatibility complex. *P-value < 0.05, **P-value < 0.01, ***P-value < 0.001 or ****P-value < 0.0001.

Preimmunization with ImmunoBody-HAGE vaccine delays the growth of implanted HHDII+DR1+HAGE+Luc+ B16 cells in both subcutaneous and metastasis settings

Having demonstrated the superiority of the ImmunoBody-HAGE over HAGE-30mer peptide, only the HAGE-ImmunoBody was used in tumor model experiments. In these experiments, the ability of the ImmunoBody-HAGE vaccine to prevent/delay the growth of HAGE-expressing HHDII+DR1+HAGE+Luc+ B16 tumors after either subcutaneous implantation or intravenous injection (as an experimental model of metastasis) into HHDII+DR1+ mice was tested in both prophylactic and therapeutic settings.

In prophylactic studies, HHDII+DR1+ mice were immunized on days 0, 7 and 15, and 7 days after the last immunization (Day 22) animals were challenged with HHDII+DR1+HAGE+Luc+ B16 tumor cells. Control mice received only tumor cells as shown in Supplementary figure 2. Figure 5a shows representative images of tumor-bearing mice from different experimental groups. It was observed that by day 34 after tumor cell injection, 3/11 mice from the prophylactic group of the subcutaneous model remained alive (Figure 5e), whereas none of the mice from the control and therapeutic groups were alive at that point. Figure 5b–d shows that immunization with the ImmunoBody-HAGE vaccine significantly delayed tumor uptake and growth. The ability of the vaccine to slow/eradicate established tumor growth was assessed by first implanting HHDII+DR1+HAGE+Luc+ B16 cells and then starting vaccination 2 days later. Figure 5e shows that the vaccine was unable to delay the growth of established tumors in this setting. The antitumor efficacy of the ImmunoBody-HAGE vaccine was also tested in an experimental metastatic model. A similar antitumor protection with significant delay in the onset of tumor growth (Figure 6b–d) was observed in immunized mice compared with nonimmunized mice groups. Mice administered with vaccine before tumor injections showed prolonged survival compared with nonimmunized mice and mice that received therapy (Figure 6e). While such a model of experimental metastasis does not, however, adequately mimic human metastatic breast cancer because the cells injected systemically directly go to the metastasis site without first establishing primary disease elsewhere, it is the only model available to assess the efficacy of a vaccine in HHDII+DR1+ mice.

Efficacy of ImmunoBody-HAGE vaccine against HHDII+DR1+HAGE+Luc+ B16 cells subcutaneous tumors. Tumor-bearing animals were monitored for tumor growth by measuring luciferase activity. (a) Representative images of tumor-bearing mice observed over time, (b, c and d) comparison of total flux between control and treatment groups and (e) survival curves showing growth of subcutaneous tumors (n = 1, 11 mice per group). One experiment was performed. Three representative images are shown for each group. **P-value < 0.01.

Efficacy of ImmunoBody-HAGE vaccine in a metastasis model. Mice were intravenously injected with HHDII+DR1+HAGE+Luc+ B16 cells. (a) Representative images of tumor-bearing mice observed over time, (b) comparison of total flux between control and treatment groups and (c) survival curves indicating the proportions of surviving mice in each group (n = 1, 10/group). Survival curve analysis was performed using the Gehan–Breslow–Wilcoxon test to obtain significant differences (P-value) between control and treatment groups. One experiment was performed. Three representative images are shown for each group. *P-value < 0.05.

Prophylactic vaccinations were then combined with therapeutic vaccinations to assess whether the continued vaccination, even after the appearance of tumor, could improve efficacy and outcomes.

Tumor growth was significantly further delayed when vaccination was continued after the administration of a lethal dose of HHDII+DR1+HAGE+Luc+ B16 tumor cells (Figure 7a–c). However, the addition of anti-PD-1 antibody did not further improve this outcome (Figure 7d, e). Interestingly, we have found that more than 60% of all tumor-infiltrating T cells were PD-1+, whereas less than 20% of the T cells located in the spleen of the same animals express PD-1.24 Hence, to understand the association between TILs and additional mechanisms of T-cell suppression, HHDII+DR1+HAGE+Luc+ B16 tumor-derived TILs were profiled using multiparameter flow cytometry. For these studies, mice were grouped according to their tumor size/weight (Figure 8a). Five mice injected with tumor cells and receiving sham vaccinations and isotype control for the anti-PD-1 antibody were used as comparators. Despite a significant difference in the weight between the groups with small and large tumors, no difference in the percentage of macrophages (CD68+/F4/80+) infiltrating small and large tumors was detected (Figure 8b). However, a significant percentage of these macrophages also expressed CD206 and interleukin-10 (IL-10), features that are associated M2 macrophages, whereas the percentage of macrophages negative for CD206 and IL-10 expression but expressing tumor necrosis factor-α was significantly higher in mice bearing small tumors. In addition, larger tumors had more regulatory T cells (CD3+CD4+CD25+FoxP3+) than smaller tumors, whereas the proportion of CD3+CD4+ T cells was not significantly affected (Figure 9a). No differences in the proportion of myeloid-derived suppressive cells (CD11b+LY-6C+LY-6G+/−) were found (Figure 9b). Moreover, TILs extracted from mice which did not receive the vaccine contained a significantly higher proportion of “exhausted” (CD3+CD8+PD-1+Tim3+) CD8+ T cells (Supplementary figure 1).

Effect of additional vaccinations and anti-PD1 antibody therapy after tumor implantation on pre-vaccinated animals. Mice (n = 10/group) were all vaccinated three times (once per week) with ImmunoBody-HAGE, thereafter all animals received HHDII+DR1+HAGE+Luc+ B16 cells and a further three injections of the vaccine with/without the addition of anti-PD-1. (a-d) The tumour size as measured by total flux for the following groups: control (a), prohpylactic (b), prophylactic and therapeutic (c) and prophylactic and therapeutic in combination with anti-PD1 (d). (e) The survival curves for all groups. Results demonstrate the benefit of further vaccination after the tumor injection with a significant delay before mice had to be culled because of tumor size. The addition of anti-PD-1 did not, however, further improve this outcome. ns, not significant; PD-1, programmed cell death protein 1. **P-value < 0.01.

M2-type macrophages are significantly increased in large tumors. Tumor-infiltrating lymphocytes (TILs) from all tumors (with and without anti-PD-1) were extracted and phenotypic analysis was performed using flow cytometry. (a) The tumor grouping according to their weights. (b) The percentage of TILs positive for different macrophage-related markers. Results demonstrate that while no differences could be found in the proportion of macrophages, the proportion of M2 macrophages (CD206+IL-10+) was significantly increased in larger tumors. MHC, major histocompatibility complex; PD-1, programmed cell death protein 1. TNF-α, tumor necrosis factor-α. All data presented are mean ± SEM. *P-value < 0.05, **P-value < 0.01, ***P-value < 0.001 or ****P-value < 0.0001.

Large tumors contain proportionally more regulatory T cells than smaller ones, but no more myeloid-derived suppressor cells (MDSCs). Tumor-infiltrating lymphocytes (TILs) from small, large and tumor only groups were assessed by flow cytometry. Graphs show proportions of immune cell populations such as (a) effector and Tregs and (b) MDSCs found within HHDII+DR1+HAGE+ B16 tumors. Although no significant difference could be found in the proportion of CD3+CD4+, the percentage of CD25+FoxP3+ regulatory T cells (Treg) within this population was significantly higher in tumors with the largest weight. No differences were found in the MDSC compartments. All data presented are mean ± SEM. **P-value < 0.01 and ****P-value < 0.0001.

DISCUSSION

Recent years have seen remarkable advances in the field of cancer immunotherapy. Active immunotherapy mainly focuses on eradication of cancerous cells by induction of T-cell-mediated antitumor responses. Antigens that are differentially expressed or are selectively induced/arise in tumor tissues, and which can trigger antitumor immune responses, represent potential immunotherapeutic targets. The induction of potent antitumor immune responses is critically dependent on the choice of target antigen and although tumor-associated antigens are easier to identify and are shared by many patients, vaccines based on these have not yet been proven to be effective in clinical trials. In addition, as they are not necessarily tumor specific, there is the potential for severe adverse events owing to off-target effects. By contrast, neoepitopes derived from mutated antigens are tumor specific, more immunogenic and also patient specific. CTAs represent a compromise between these two categories of antigens because of their restricted expression pattern, almost exclusively in tumor, with little to no normal tissue expression (except for testis and placenta) and are, when combined with an appropriate adjuvant/delivery system, immunogenic.

HAGE is a CTA (CT13), which was first identified by Martelange et al.,25 and found to be expressed 100-fold higher in tumor tissues than in normal tissues, except testis. HAGE messenger RNA expression in many malignant tumors along with the expression of BAGE, MAGE and GAGE indicates the potential role of CTAs as targets and/or diagnostic markers.26 HAGE has also been shown to promote the proliferation of malignant cells and it has also been shown to be involved in the process of tumorigenesis in multiple cancer types.27 HAGE has also been shown to be expressed in 43% of locally advanced primary TNBC tumors before anthracycline combination Neo-ACT. Following anthracycline combination Neo-ACT, a pathological complete response was achieved in 48% of HAGE+ tumors, but in only 14% (8/56) of HAGE– TNBC tumors.4

Although HAGE protein has previously been shown to be immunogenic, no specific peptide region responsible for triggering and driving immunogenicity was identified.28 Melief and van der Burg have previously demonstrated that the use of long peptide fragments (15–30) rather than the entire protein is better at inducing CD8+ T cells.18 Therefore, in this study, an immunogenic region within the HAGE protein has been identified and its potential to generate specific immune responses against tumors expressing HAGE was investigated. The study began with the identification of a 30-amino-acid sequence predicted to encompass several HLA-restricted immunogenic epitopes by in silico SYFPEITHI analysis. Splenocytes derived from HAGE-derived 30mer (either peptide with CpG/IFA or as an ImmunoBody DNA construct) vaccinated HHDII+DR1+ mice were able to induce strong IFNγ responses. Our results showed that although the combination of peptide and CpG/IFA was better than peptide with either adjuvant alone, ImmunoBody-HAGE was significantly better at generating an IFNγ response. The efficiency of DNA vaccines to deliver antigenic epitopes within a human antibody IgG1 scheme has previously been demonstrated for other antigens.10, 11 The HLA-A2 peptides, peptide 5 (YLMPGHFIL) and peptide 6 (YLMPGFIHLV), which only differ from each other by an additional valine residue, and the HLA-DR1-restricted peptide, peptide 7 (GKTCLYLMPGHFILV) which contains both of these peptide sequences, generated strong immune responses, as assessed by the high number of IFNγ producing cells. Interestingly, this region is also predicted to include an HLA-DR4 T-cell epitope, which suggests a certain HLA promiscuity of this sequence.

More importantly, splenocytes from animals immunized with the ImmunoBody-HAGE vaccine recognized and responded to HAGE-expressing target cells. Indeed, a significantly higher number of vaccine-induced splenocytes cocultured with T2 cells pulsed with HAGE-derived peptides or HHDII+DR1+HAGE+Luc+ B16 cells produced IFNγ than when cocultured with non-HAGE expressing/pulsed target cells. Furthermore, vaccine-induced immune cells were able to kill HAGE+/pulsed target cells, as assessed by in vitro 51chromium release cytotoxicity assays. These results confirmed the superior immunogenicity of the ImmunoBody-HAGE vaccine.

The in vivo antitumor efficacy of this vaccine was then confirmed by in vivo tumor challenge studies using HHDII+DR1+ mice models in both prophylactic and therapeutic settings. Immunization with the ImmunoBody-HAGE vaccine significantly delayed the growth of both subcutaneously and intravenously injected HHDII+DR1+HAGE+Luc+ B16 tumors. However, the vaccine on its own was not able to influence the growth of tumor cells when these were injected first (therapeutic settings).

Murine B16 melanoma cells are aggressive cells that grow quickly once injected in vivo and by the time the vaccine generates a strong immune response the now well-established tumor cells create an immunosuppressive microenvironment. In addition, 100% of B16 cells express PD-L1 molecules which will bind to the PD-1 expressed by activated T cells and induce their death. We have previously shown that over 80% of TILs express PD-1, while T cells found in the spleen of the same animals do not.29

The use of immune checkpoint inhibitors can also have “adjuvant” effects by enhancing the efficacy of the vaccine, as has been demonstrated in preclinical models combining SCIB1 and SCIB2 ImmunoBody vaccines with anti-PD-1.30, 31 Interestingly, the schedule chosen to administer both checkpoint inhibitor and vaccine will influence the efficacy of the vaccine. Indeed, although the efficacy of both the GVAX prostate-specific antigen prostate cancer vaccine and the TG4010 (Muc-1) vaccine for advanced non-small-cell lung cancer vaccine were improved by administering anti-CTLA-4 and anti-PD-1, respectively, the prostate-specific antigen-targeted DNA vaccine was found to be most effective if given at the same time as the anti-PD-1.32-34

We propose that the future clinical potential for this vaccine will involve its administration to TNBC patients who have received conventional treatment(s) and have been declared tumor free, given that relapse is known to occur in some of these patients within 2 years following the end of their treatment. Immunized mice were, therefore, challenged with a lethal dose of tumor cells and additional vaccinations with or without the addition of anti-PD-1 administered. Our results demonstrate that the additional vaccinations significantly prolonged the time to death compared with the prophylactic setting, but that anti-PD-1 had no influence on this prolongation of survival. We hypothesized that additional suppressive mechanism(s) must be operating, and therefore profiled TILs using flow cytometry. Our results clearly show that M2 (CD68+F4/80+CD206+IL-10+) macrophages predominate in large tumors and also that large tumors were more highly infiltrated by regulatory T cells (CD3+CD4+CD25+FoxP3+). No differences were found in the level of myeloid-derived suppressor cells (CD11b+LY6C+LY6G+) between the two tumor types.

In conclusion, our studies indicate that the HAGE antigen is immunogenic and the 30mer peptide region within HAGE is efficient at generating high frequencies of high-avidity T cells that can induce potent antitumor responses against HAGE-expressing tumors. The ImmunoBody-HAGE vaccine could, therefore, be translated into the clinic for the treatment of patients with TNBC expressing HAGE after they have received chemotherapeutic treatment with the intention of preventing relapse. Targeting additional antigens as well as including approaches to re-direct M2 macrophages toward M1 macrophages might further enhance the therapeutic effect. Moreover, it is possible that patients with TNBC with no/few TILs (immune desert) would benefit from first receiving an antigen-specific vaccine (such as ImmunoBody-HAGE) to induce a strong and, at least partially, protective antitumor immunity followed by additional therapies aimed to alleviate immunosuppressive effect induced by the tumor cells.

METHODS Animals

HHDII+DR1+ double transgenic mice expressing human α1 and α2 chains of HLA-A*0201 chimeric with the α3 chain of the H-2Dd allele (HHDII), also expressing HLA-DRB*0101 and knocked out for the expression of murine MHC class I (H-2b) and II (I-Ab), were provided by Dr Lone (CNRS, Orléans, France). The use of protocols and procedures in this study and animal care were employed in accordance with EU Directive 2010/63/EU and UK Home Office Code of Practice for the housing and care of animals bred, supplied or used for scientific purposes. Experiments to assess the vaccine immunogenicity were performed at least once with three female mice aged 6–8 weeks. A minimum of 10 mice per test group were used for the in vivo tumor growth models.

Cell lines

The TAP-deficient T2 cell line (i.e. HLA-A2 lymphoblastoid suspension cells) was a gift from Dr J. Bartholomew (Paterson Institute, Manchester, UK). The human TNBC cell lines, MDA-MB-231 and MDA-MB-468, were purchased from American Type Culture Collection and cultured in Leibovitz (L15, SLS/Lonza, Nottingham, UK) medium supplemented with 10% v/v fetal bovine serum (FBS, Warington, UK) and 1% v/v l-glutamine (SLS/Lonza, UK).

The murine melanoma cell line B16F1, which was knocked out for both murine endogenous MHC class I and II by zinc finger nuclease technology and stably transfected with plasmids to express both HHDII and HLA-DR1 molecules (hereafter referred to as HHDII+DR1+HAGE+Luc+ B16), was a generous gift from Professor Lindy Durrant (Faculty of Medicine and Health Sciences, Nottingham University and Scancell Ltd). These cells were grown in Roswell Park Memorial Institute (RPMI, SLS/Lonza, UK) 1640 medium supplemented with 10% v/v fetal bovine serum, 1% v/v l-glutamine, 300 μg mL−1 hygromycin (Sigma-Aldrich, Merck LifeSciences, Dorset, UK) and 500 μg mL−1 geneticin (Sigma-Aldrich, Merck LifeSciences, Dorset, UK).

Identification of immunogenic and MHC binding predictions

The SYFPEITHI database algorithm was used to identify an immunogenic HAGE-derived sequence (30mer) based on the presence and position of anchor residues that influence their binding affinities to different HLA alleles [HLA-A*0201 (HLA-A2) and HLA-DRB*0101 (HLA-DR1)]. Table 1 summarizes peptides derived from the 30mer and their binding scores to different HLA haplotypes.

ImmunoBody-HAGE DNA vaccine

ImmunoBody vaccines are designed to generate potent T-cell responses capable of a broad antitumor effect. They are DNA vaccines that encode a protein in the form of an antibody, but the parts of the antibody that would normally bind to the target protein are replaced with epitopes from a cancer antigen.10, 11 The ImmunoBody-HAGE vaccine incorporates the complementary DNA sequence for the HAGE287–316 (QTGTGKTLCYLMPGFIHLVLQPSLKGQRNR) 30mer peptide region inserted into the CDRH2 heavy variable of the human IgG1 chain alongside the light human kappa chain encoded within the double expression vector pDCOrig. The ImmunoBody heavy and light chains are within separate expression cassettes, both of which are under the control of a cytomegalovirus promoter, as described previously.10,