Heterologous prime-boost with KISIMA-TAA and VSV-GP-TAA induces long-term immunity
To characterize the heterologous combination of KISIMA-TAA with VSV-GP-TAA oncolytic vaccine, we first investigated the schedule of administration using OVA. Priming with subcutaneous injection of KISIMA-OVA followed by intramuscular boost with VSV-GP-OVA elicited higher proportion (frequency and number) of circulating OVA-specific CD8+ T cells compared to VSV-GP-OVA prime and KISIMA-OVA boost (Fig. 1a). Importantly, the OVA-specific T cell population further expanded after a second boost with KISIMA-OVA but not with VSV-GP-OVA (Fig. 1a). Thus, for subsequent experiments an alternating regimen was selected starting with KISIMA-TAA priming, VSV-GP-TAA boost followed with a second KISIMA-TAA boost (KVK regimen). Next, different routes of virus administration were compared. A single intravenous (i.v.) administration of VSV-GP-OVA induced the highest frequency of antigen-specific CD8+ T cell response compared to intraperitoneal (i.p.), subcutaneous (s.c.) or intramuscular (i.m.) injection for both OVA (Supplementary Fig. 1a) and VSV-N (Supplementary Fig. 1b) antigens. Subsequently, immune response elicited upon boost with either i.v. or i.m. VSV-GP-OVA was assessed using the KVK regimen. Intravenous boost resulted in a significantly higher proportion of OVA-specific peripheral CD8+ T cells, which further expanded following the KISIMA-OVA boost. 140 days post prime, OVA-specific T cells were still present in the i.v. group suggesting the formation of immunological memory (Fig. 1b). Consistently, a higher number of OVA-specific CD8+ T cells were found in spleen (Fig. 1c) and bone marrow (Fig. 1d) following KVK vaccination using i.v. in contrast to i.m. route. Since the phenotypical composition in general and the antigen-specific memory precursor effector cells expressing CD127 in particular are important for generation of long-lasting memory27, the presence of CD127+ memory precursors and KLRG1+ effector cells among the antigen-specific T cells was assessed (Supplementary Fig. 1c). Systemic administration of VSV-GP-OVA in KVK regimen resulted in a higher number of both memory and effector OVA-specific CD8+ T cells in circulation (Fig. 1e), spleen (Fig. 1f) and bone marrow (Fig. 1g) compared to i.m. immunization, confirming its ability to generate immunological memory.
a C57BL/6J mice (n = 5) were immunized against ovalbumin with either KISIMA-OVA given s.c. (K) or VSV-GP-OVA (V) administered i.m. on days 0, 7, and 14 (dotted lines) or left untreated (mock). The fraction of OVA-specific cells among CD8+ T cells in peripheral blood is shown. Two-way ANOVA with Tukey’s multiple comparisons (*, p < 0.05; **, p < 0.01). b–g C57BL/6J mice were immunized with KISIMA-OVA (s.c.) on day 0 and 14 and VSV-GP-OVA was administered either i.m. or i.v. on day 7 as indicated by dotted lines (n = 5 per treatment group, n = 2 for mock). b The frequency of OVA-specific CD8+ T cells in circulation is depicted. Two-way ANOVA with Sidak’s multiple comparison (**, p < 0.01). c, d The number of OVA-specific CD8+ T cells in the (c) spleen and (d) bone marrow of immunized mice was measured 19 weeks post prime. One-way ANOVA with Tukey’s multiple comparison (**, p < 0.01). e–g The number of OVA-specific CD8+ T cells with effector (KLRG1+CD127−) and memory precursor (CD127+) phenotype in (e) peripheral blood (cells/ml), (f) spleen and (g) bone marrow 19 weeks after prime is shown. Two-way ANOVA with Sidak’s multiple comparison (****, p < 0.0001). (h, i) C57BL/6J mice were vaccinated against (h) Adpgk and Reps1 neoantigens or (i) HPV-E7 with either (h) KISIMA-Mad24 or (i) KISIMA-HPV given s.c. (K) or (h) VSV-GP-Mad24 or (i) VSV-GP-HPV administered i.v. (V) on day 0 and as indicated with dotted lines. The frequency of circulating CD8+ T cells specific for (h) Adpgk (n = 5 per group) and (i) HPV-E7 (n = 7 per group) is depicted. Two-way ANOVA (h) and one-way ANOVA (i) with Tukey’s multiple comparison was performed (*** p < 0.001; ****p < 0.0001) and significance compared to mock is shown. All data shown as mean ± SEM. Studies (a, c–g) were performed once, studies (b, i) were independently repeated once, study (h) was repeated once with the listed groups except VKK, which was only included once. Source data and p-values are provided in the Source Data File.
The immunogenicity of KVK regimen was further assessed against neoantigens and viral oncoprotein as target antigens. For targeting neo-epitopes, KISIMA-Mad24, a KISIMA-derived vaccine bearing the previously described neoantigens Adpgk (ADP-dependent glucokinase) and Reps1 (RalBP1-associated Eps domain-containing protein 1) which are expressed in the murine colorectal carcinoma model MC-3828 and the corresponding VSV-GP-Mad24 were used. Additionally, KISIMA-HPV bearing HPV-derived E7 oncoprotein as antigen was used in combination with VSV-GP-HPV. The latter contained HPV-derived E7 and in addition E6 and E2. Priming with KISIMA followed with an i.v. VSV-GP-TAA boost elicited the highest frequency of antigen-specific T cells in the periphery for both antigen models (Fig. 1h, i). Targeting E7, heterologous KVK vaccination resulted in significantly higher antigen-specific T cell responses compared to both homologous vaccinations, inducing an over 30-fold increase in circulating HPV-E7-specific CD8+ T cell frequency (Fig. 1i). Though subsequent boosting with KISIMA-HPV did not further increase HPV-E7-specific CD8+ cells frequency, it did prevent them from undergoing dramatic contraction. Consistent with the OVA model, HPV-specific CD8+ T cells induced by heterologous prime-boost vaccination as shown in Supplementary Fig. 2a persisted in the periphery (Supplementary Fig. 2b), bone marrow (Supplementary Fig. 2c) and spleen (Supplementary Fig. 2d) for up to 5 weeks after last immunization and displayed an effector memory phenotype (Supplementary Fig. 2e–g).
Comparing antiviral with anti-target immune responses, KVV heterologous prime-boost treatment (Supplementary Fig. 2h) not only enhanced tumor antigen-specific immunity but also dampened the antiviral response (Supplementary Fig. 2i) compared to homologous VSV-GP-OVA vaccination (VVV) or priming with VSV-GP-OVA (VKV). In addition, an inverse correlation between proportion of virus- and OVA-specific CTLs in blood was observed (Supplementary fig. 2j). This reversal in the ratio of antitumor and antiviral T cells was also reproduced for the tumor antigen E7 (Supplementary Fig. 2k). Thus, heterologous prime-boost vaccination using KVK regimen induces a potent CD8+ T cell response against model, tumor-associated and tumor-specific antigens and favors the development of immunological memory while dampening antiviral immunity.
Priming with KISIMA-TAA improves functionality of tumor-specific T cells
Some key properties of cancers are immune exclusion and suppression, which allow tumor cells to counter cytotoxic CD8+ T cell infiltration and function. Thus, we assessed the ability of therapeutic KVK heterologous prime-boost vaccination to overcome this constraint in the immunologically ‘cold’ TC-1 tumor model. TC-1 cells are transformed murine lung epithelial cells expressing the HPV-derived oncoproteins E6 and E729. Once the tumors were palpable, mice were primed with KISIMA-HPV or VSV-GP-HPV followed 7 days later with a VSV-GP-HPV boost. Tumor-infiltrating cells were analyzed one week after boost (Fig. 2a). Consistent with the results in non-tumor-bearing animals, KISIMA-HPV prime followed by VSV-GP-HPV boost resulted in significantly higher frequency (Fig. 2b) and absolute numbers (Fig. 2c) of HPV-E7-specific CD8+ T cells in the periphery, compared to homologous VSV-GP-HPV treatment. Both vaccination regimens were able to induce high infiltration of CD8+ T cells within the tumor, about 60% of which were found to be HPV-E7-specific by multimer staining (Fig. 2d, e). In contrast to the periphery, there were no differences within the tumor between the two vaccine regimens in HPV-E7-specific CD8+ T cells frequency (Fig. 2d) and numbers (Fig. 2e). As the immunosuppressive tumor microenvironment is well known to induce a rapid exhaustion of T cells, we next assessed the phenotype of circulating and tumor-infiltrating antigen-specific CD8+ T cells. While only a small portion of HPV-E7-specific CTLs displayed an exhausted phenotype in the periphery (Fig. 2f), characterized by the expression of PD-1 and Tim-3, the majority of tumor-infiltrating CD8+ T cells expressed both markers – suggesting their exhaustion (Fig. 2g). Interestingly, a higher proportion of intratumoral PD-1+Tim-3+ CD8+ T cells in KV vaccinated mice still expressed the early activation marker KLRG1, suggesting a less advanced exhaustion status compared to the homologous VV treated mice. Since T cell exhaustion is a progressive process, which initiates with the expression of markers and continues with loss of function and eventually cell death, we assessed CD8+ T cells functionality by measuring cytokine secretion after ex vivo restimulation. A significantly higher proportion of splenic HPV-E7-specific CD8+ T cells in KV vaccinated mice expressed IFN-γ, TNF-α and the degranulation factor CD107a compared to VV treated mice (Fig. 2h). Additionally, higher frequency of granzyme B producing CTLs were detected in KV vaccinated mice compared to VV vaccinated mice (Fig. 2j). In accordance to the results from the spleen, KV vaccination induced a significantly higher proportion of multifunctional HPV-E7-specific CD8+ T cells in the tumor compared to VV treatment, in particular IFN-γ+TNF-α+CD107a+ triple-positive cells (Fig. 2i), highlighting a highly cytotoxic, less exhausted phenotype of KV elicited antigen-specific CD8+ T cells.
a–j C57BL/6J mice were injected s.c. with TC-1 cells on day 0 and vaccinated with KISIMA-HPV (s.c.) or VSV-GP-HPV (i.v.) on day 7 and 14. Blood, spleen and tumors were harvested on day 21 for flow cytometric analysis. a Schematic of experimental plan. b, d Frequency and (c, e) number of HPV-E7-specific CD8+ T cells were measured by flow cytometry in (b, c) blood and in (d, e) tumors. Two-tailed Mann–Whitney test (*, p < 0.05). f, g Proportions of (f) peripheral and (g) intratumoral HPV-E7-specific CD8+ T cells expressing activation and exhaustion markers are depicted (n = 5 for mock, n = 4 for each treatment group). Two-way ANOVA with Sidak’s multiple comparison (***, p < 0.001; ****, p < 0.0001). h, i Frequencies of HPV-E7-specific CD8+ T cells secreting different cytokines upon ex vivo restimulation among (h) splenic (n = 5 for mock and KV, n = 6 for VV) and (i) intratumoral (n = 5 for mock, n = 4 for each treatment group) CD8+ T cells are shown. Two-way ANOVA with Sidak’s multiple comparison (**, p < 0.01; ***, p < 0.001; ****, p < 0.0001). j Frequencies of granzyme B expressing cells among splenic CD8+ T cells is shown (n = 5 for KV and n = 6 for VV). One-way ANOVA with Tukey’s multiple comparisons (****p < 0.0001). All data shown as mean ± SEM. Studies (b–e, j) were repeated once. Studies (f–i) were performed once. Source data and p-values are provided in the Source Data File.
In addition, the virus-specific immune response was also monitored in the periphery and within the tumor (Supplementary Fig. 3a–e). Similarly to the HPV-specific CD8+ T cells response, a small proportion of peripheral antiviral CD8+ T cells expressed exhaustion markers or secreted cytokines (Supplementary Fig. 3f, h). The differences in peripheral responses (Supplementary Fig. 3b, c) did not correlate with the intratumoral response (Supplementary Fig. 3d, e), where higher numbers of VSV-N-specific CD8+ T cells were observed in VV compared to KV vaccinated mice (Supplementary Fig. 3d, e). However, intratumoral VSV-N-specific CD8+ T cells mostly expressed PD-1 on their cell surface (Supplementary Fig. 3g) and showed low functionality (Supplementary Fig. 3i), suggesting that they were bystander cells.
Overall, priming with KISIMA-HPV and boosting with VSV-GP-HPV not only supports induction of higher magnitude of tumor-specific CD8+ T cells, but also promotes their recruitment into the tumor and enhances their functionality compared to homologous viral vaccination.
Heterologous vaccination reverses immunosuppression in TME
After heterologous KV vaccination, dramatic changes in the TC-1 TME were observed upon transcriptome analysis (Supplementary Data 1 and Fig. 3a, b). 64.9% of all panel genes were upregulated in KV treated tumors, compared to 36.8% after homologous VV vaccination; indicating stronger activation of multiple immune pathways (Supplementary Data 2 and Fig. 3a). While 244 of these genes could be attributed to the immune activating effects of VSV-GP-HPV, a set of 243 genes was upregulated only in the heterologous vaccination group (Supplementary data 2 and Fig. 3c). The genes uniquely upregulated in KV treatment are involved in both innate and adaptive immune responses (Supplementary Table 1). Interestingly, heterologous vaccination also negatively regulated the expression of 35 genes (Supplementary data 2 and Fig. 3b, d) including Cdkn1a and Msln which are involved in cancer progression (Supplementary Table 2). In addition, heterologous KV vaccination activated multiple immune genes associated with cytotoxic T cells (Fig. 3e), dendritic cells (DCs) (Fig. 3g), cytokines (Fig. 3f), chemokines (Fig. 3h) and antigen processing and presentation (Fig. 3i). Hierarchical clustering revealed that tumors from mice receiving a specific vaccine combination had a similar transcriptome and thus were more likely to cluster together. Biologically, increased CTL infiltration along with elevated levels of cytotoxic genes such as granzymes (Grzma, Grzmb and Grzmk) and perforin (Prf1) (Fig. 3e) and antigen presentation (Fig. 3i) suggested enhanced tumor cell killing as a result of heterologous vaccination. Besides, more genes indicative of DC function and maturation including cross-presentation were upregulated in heterologous treated tumors (Fig. 3g). All vaccine combinations upregulated components of the antigen processing machinery but homologous VV vaccination had a stronger impact on genes encoding MHC I and MHC II molecules; whereas KV vaccination positively regulated non-classical MHCs (Fig. 3i).
a–i C57BL/6 J mice bearing TC-1 tumors were immunized as in Fig. 2 or left untreated (mock). Tumors were harvested on day 23 post tumor implantation for transcriptome analysis using NanoString® technology (n = 7 for mock, KK, and VV, n = 10 for KV). a, b Gene expression in TC-1 tumors from each vaccination group was normalized to mock tumors and the proportion of (a) significantly upregulated (fold change [FC] >2 and p < 0.05) and (b) significantly downregulated (negative reciprocal of FC < −2 and p-value < 0.05) genes is displayed. c, d Venn diagrams depict the total number of significantly (c) upregulated and (d) downregulated genes after different vaccine regimens and the overlap between each gene set. e–i Heatmaps display relative gene expression as z-scores (scaled to each gene) and hierarchichal clustering (Euclidean distance, average linkage) was applied to sample data with each column representing one individual tumor. Expression of typical genes associated with (e) cytotoxic T cells, (f) cytokines, (g) dendritic cells, (h) chemokines and (i) antigen presentation is shown. 7–10 mice analyzed for each treatment group, p-values were calculated using two-tailed t test and false discovery rate (FDR) adjusted p-values calculated using Benjamini–Yekutieli procedure are reported. The study as shown was performed once. Groups mock and KV were repeated in a separate study. Source data are provided in the Source Data File.
Both proinflammatory and anti-inflammatory cytokines were upregulated as a result of the immune activating effect of KV vaccination, including elevated levels of type I and type II interferons (Fig. 3f). Also, cytokines such as Ifng and Tnf important for T cell effector functions were elevated in TC-1 tumors after heterologous vaccination (Fig. 3f). Importantly, some of the cytokines and chemokines upregulated in the tumor were also elevated in the plasma of mice receiving heterologous KV vaccine, including increased levels of IFN-γ, CCL5, CXCL10, CCL2, IL-6, CXCL1, and IL-1β one day after VSV-GP-HPV boost (Fig. 4a).
Mice bearing palpable TC-1 tumors were immunized as in Fig. 2. a Cytokine and chemokine levels in plasma were quantified on day 15 and are shown as mean ± SEM (n = 6 for mock, VV, and KV, n = 4 for KK). One-way ANOVA with Tukey’s multiple comparison (*p < 0.05, ** p < 0.01; ***p < 0.001, ****p < 0.0001). Dotted lines indicate the limit of quantification. b, c Tumors were harvested on day 26 and tumor-infiltrating leukocytes were characterized by flow cytometry (n = 3 for mock and VV, n = 2 for KK and KV). b Total CD45+ leukocytes (mean) and (c) relative proportions of various immune cell subsets among all leukocytes is shown. d Representative immunohistochemistry images show T cell infiltration (CD8) in TC-1 tumors day 23 post tumor implantation after different vaccinations. Lower panel shows higher magnification view of boxed area from upper panel. Scale bars: 500μm upper row, 50 μm lower row. Study (a) was performed twice, studies (c, d) were performed once. Source data and p-values are provided in the Source Data File.
Observations from transcriptome analysis were further supported by the analysis of the number (Fig. 4b) and types (Fig. 4c, Supplementary Fig. 4) of tumor-infiltrating leukocytes (TILs). TILs from untreated TC-1 tumors are predominantly (>80%) composed of immunosuppressive cells such as M2-like tumor-associated macrophages (TAM-2) and myeloid-derived suppressor cells (MDSCs), while T cells constitute only 1% of the infiltrate (Fig. 4c). Therapeutic vaccination induced deep changes in the TILs, with a striking influx of both CD8+ and CD4+ T cells populations and a drastic decrease of TAM-2, resulting in an enhanced TAM-1/TAM-2 ratio suggesting repolarization. Furthermore, heterologous KV vaccination promoted the strongest influx of CD8+ T cells (>25%) (Fig. 4c). Thus, while both vaccination regimens promoted trafficking of immune cells into the tumor, KV vaccination attracted the highest proportion of CTLs, CD4+ T helper cells and increased TAM-1/TAM-2 ratio thereby favorably remodeling the TME.
Next, immunohistochemistry was performed to confirm the location of immune infiltrates. CD8 staining of tumors harvested 9 days post boost confirmed the general immune-excluded phenotype of untreated TC-1 tumors with few CD8+ T cells confined to the tumor margin (Fig. 4d). While CD8+ T cell infiltration increased with homologous KK and VV vaccine regimen, the heterologous combination KV displayed a massive cytotoxic T cell presence in the deepest parts of the tumor.
Efficacy and synergy with checkpoint blockade
Susceptibility to oncolytic viruses varies between tumors, and continued viral propagation is often limited. Heterologous prime-boost with oncolytic vaccines may address such limitations in tumors with known antigenic targets. Hence, heterologous KISIMA-TAA/VSV-GP-TAA prime-boost was assessed in a selection of tumor models that show resistance or very limited responses to oncolytic VSV-GP monotherapy. Murine lymphoma cells E.G7-OVA are resistant to VSV-GP induced oncolysis in vitro (Supplementary Fig. 5a). TC-1 and B16-OVA tumors are susceptible to infection and lysis in vitro, but are protected by IFN-mediated innate antiviral responses (Supplementary Fig. 5b, c). MC-38 tumors on the other hand show full in vitro susceptibility even in the presence of IFN, indicating impaired antiviral defense (Supplementary Fig. 5d). Of note, this in vitro susceptibility does not translate to efficacy of single VSV-GP application in vivo (Supplementary Fig. 6a, b), and intratumoral virus activity is strongly diminished within the first three days after treatment (Supplementary Fig. 6c). TC-1 tumors also do not respond to single VSV-GP treatment (Supplementary Fig. 6d, e) and VSV-GP replication in vivo is limited to an initial infection period (Supplementary Fig. 6f).
Therapeutic vaccination with both KISIMA-OVA and VSV-GP-OVA homologous as well as heterologous prime-boost as shown in Fig. 5a, significantly delayed growth of E.G7-OVA tumor cells, resulting in an increased median survival (Fig. 5b,c). Interestingly, a strong regression of large established tumors was only observed following boost in KVK vaccinated mice. The tumor regression correlated with the potent OVA-specific CD8+ T cells in the periphery (Fig. 5e, Supplementary Fig. 7e). In contrast, in MC-38 colorectal cancer model, in which the KISIMA-Mad24 homologous and heterologous prime-boost vaccinations (Fig. 6a) showed a low therapeutic efficacy, VSV-GP-Mad24 homologous vaccination was highly efficient, resulting in over 50% of complete tumor regression (Fig. 6b, c). This effect exemplifies the response to the oncolytic component of VSV-GP, as similar results were obtained using multiple VSV-GP (Vϕ) while the amplitude of the antigen-specific CD8+ T cells response did not correlate with antitumoral effect (Fig. 6e, Supplementary Fig. 7j). Results in TC-1 tumor model (Fig. 6f) were similar to the ones obtained in the oncolytically resistant E.G7-OVA model, with all vaccination schedules resulting in delayed tumor growth and increased median survival (Fig. 6g, h). Although frequency of HPV-specific CD8+ T cells in the periphery is higher in KVK-treated mice compared to homologous VSV-GP-HPV treated mice (Fig. 6j), they were similar in the tumor-infiltrating leukocytes (Fig. 2b–e). This might explain why homologous VSV-GP-HPV was equally effective as heterologous prime-boost schedule despite lack of correlation between circulating HPV-E7-specific and tumor-size (Supplementary Fig. 7o). In order to address the role of a KISIMA-HPV prime, VSV-GP-HPV treatment at 14 days post tumor (time of boost) was assessed with or without KISIMA-HPV prime (Supplementary Fig. 8a). While virus treatment alone led to the slowing of tumor growth, no remission was observed (Supplementary Fig. 8a, c). In contrast, virus treatment following KISIMA-HPV prime led to a complete remission in all tumors; even in large tumors (Supplementary Fig 8b, c). This strongly indicates that priming with KISIMA-HPV is essential to induce tumor regression of sizeable tumors treated with virus two weeks after grafting. Whether this tumor antigen-specific KISIMA prime affects the intratumoral virus activity was assessed by daily measurements of virally encoded luciferase reporter gene activity (VSV-GP-Luc). Importantly, as the VSV-GP-Luc virus does not express the E7 tumor antigen cassette, any effects are predominantly based on tumor infection and lysis. Priming was performed at day 7 post tumor with vehicle, a non TC-1 related antigen prime (KISIMA-OVA prime), or the TC-1 tumor-specific prime (KISIMA-HPV prime). As shown above with VSV-GP, no effect on tumor growth kinetic was observed in any of the VSV-GP-Luc treated tumors (108 TCID50 i.v.), compared to untreated mock control (Supplementary Fig. 9a). Importantly, daily bioluminescence measurements revealed no differences in antigen-specific, non-specific or buffer-prime intratumoral luciferase signals (Supplementary Fig. 9b–d). Together, these data suggest that KISIMA-HPV prime lays the immunological foundation for the strong tumor remission that follows VSV-GP-HPV boost in the TC-1 tumor model.
a–e C57BL/6J mice were injected s.c. with E.G7-OVA cells and vaccinated with KISIMA-OVA (K) s.c. or VSV-GP-OVA (V) i.v. on days indicated in schematic (a). 200 µg αPD-1 antibody was given i.v. twice weekly and blood was drawn for tetramer analysis as shown (n = 7). b Tumor growth and (c, d) survival is depicted with red numbers indicating long-term remissions within a group. Pairwise Log-rank test was performed (*p < 0.05; **p < 0.01; ***p < 0.001). e Frequency (mean ± SEM) of OVA-specific CD8+ T cells in circulation is shown. One-way ANOVA with Tukey’s multiple comparison was performed and significance displayed only for each vaccinated group vs. mock (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001). The experiments (b–d) were repeated once with i.m. application of VSV-GP-OVA. Study (e) was performed once. Source data and p-values are provided in the Source Data File.
C57BL/6J mice were subcutaneously injected with (a–e) MC-38 cells or (f–j) TC-1 cells. Mice were immunized with (a–e) KISIMA-Mad24 or VSV-GP-Mad24 or (f–j) KISIMA-HPV or VSV-GP-HPV on days indicated in the schematic (a, f), respectively. Additionally, mice received 200 µg of (a–e) αPD-L1 antibody i.p. or (f–j) αPD-1 antibody i.v. as indicated in the schematic (a, f), respectively. b, g Tumor growth curves and (c, d, h, i) survival of the animals is depicted with red numbers indicating long-term remissions within a group. Pairwise Log-rank test was performed (*p < 0.05; **p < 0.01; ***p < 0.001). e, j The frequency of circulating (e) Adpgk-specific or (j) HPV-E7-specific CD8+ T cells is depicted as mean ± SEM. 5–7 mice analyzed per treatment group as indicated in (b, g). One-way ANOVA with Tukey’s multiple comparison was performed and significance displayed only for each treatment group vs. mock (****p < 0.0001). Studies (b–d) have been repeated twice, once with PD-1 checkpoint combination treatment, once without checkpoint combination. T cell analysis in e has been repeated once. Studies (g–i) were repeated independently in three additional experiments, including once with PD-1 checkpoint combination. T cell analysis in (j) has been repeated once. Source data and p-values are provided in the Source Data File.
Despite KVK heterologous prime-boost inducing significant tumor remission in both E.G7-OVA and TC-1 tumor models, tumors were able to relapse at later time points, despite the presence of high levels of circulating antigen-specific CD8+ T cells. In order to understand the underlying mechanism of tumor escape, relapsing TC-1 tumors were harvested on day 42 post-implantation (Supplementary fig. 10a) for in-depth characterization of antigen-specific TILs. Similar to the analysis performed 7 days after the first boost (Fig. 2, Supplementary Fig. 3), both HPV-E7- (Supplementary Fig. 10b–e) and VSV-N- (Supplementary Fig. 10f–i) specific CD8+ T cells were more abundant within the tumor (Supplementary Fig. 10d, e, h, i) compared to the periphery (Supplementary Fig. 10b, c, f, g), irrespective of the vaccine regimen. While the frequency of intratumoral HPV-E7-specific CD8+ T cells was similar to the one at the earlier time point, a higher proportion of these cells co-expressed PD-1 and Tim-3 exhaustion markers (Supplementary Fig. 10l). Further, the functionality of HPV-E7-specific CD8+ T cells was strongly reduced, with only a small proportion of cells still producing IFN-γ and TNF-α or expressing the CD107a degranulation marker (Fig. 2d, Supplementary Fig. 10n). In contrast, peripheral HPV-specific CD8+ T cells (Supplementary Fig. 10j) and antiviral CD8+ T cells (Supplementary Fig. 10k, m) did not progressively upregulate Tim-3 at the time of tumor relapse.
The transcriptome of TC-1 tumors undergoing relapse was compared to that of TC-1 tumors responding to therapeutic vaccination. Principal component analysis showed that responding tumors from KV vaccinated mice clustered together and had a distinct gene expression pattern compared to the other samples (Supplementary data 1 and Supplementary Fig. 11). Notably, the relapsing tumors both from VV and KV immunized groups clustered closer to untreated and KK treated tumors, suggesting loss of immune activation. This was further confirmed as few panel genes were upregulated (Supplementary data 3 and Supplementary Fig. 12a) and most genes were downregulated (Supplementary data 3 and Supplementary Fig. 12b) in relapsing tumors when compared to responding tumors from the same treatment group. This was also reflected in the overall reduction in gene signatures associated with cytotoxic T cell infiltration (Supplementary Fig. 12c), DC function (Supplementary Fig. 12e) and loss in antigen presentation (Supplementary Fig. 12g). Additionally, cytokines (Supplementary Fig. 12d) and chemokines (Supplementary Fig. 12f) necessary for attracting T cells and other immune cells into the tumors were drastically decreased in relapsing tumors. Together, these data highlight that tumor relapses are linked to an unsustained and fading immune activation, which in turn could be addressed by concomitant application of immune checkpoint inhibitor (CPI) compounds.
Therefore, the combination of KVK heterologous prime-boost with checkpoint blocking antibodies was assessed in order to limit intratumoral T cell exhaustion and avoid tumor relapses. Strong synergy was observed between checkpoint blockade and heterologous vaccination in all 3 tumor models (Figs. 5, 6). PD-1 blockade alone had no effect on tumor growth or median survival of mice bearing either E.G7-OVA (Fig. 5b, d) or TC-1 tumors (Fig. 6g, i). However, when combined with heterologous vaccination, αPD-1 treatment prevented tumor relapse after complete regression, resulting in a high number of long-term survivors while median survival was not reached (Figs. 5d, 6i). In contrast, αPD-L1 antibody monotherapy delayed tumor growth in MC-38 bearing mice, which may be due to high mutational burden and presence of endogenous tumor-reactive T cells. However, in combination with KVK heterologous vaccination, αPD-L1 treatment strongly increased vaccine efficacy, resulting in long-lasting complete regression in over 70% of animals (Fig. 6b,d). Additionally, combination with CPI also promoted expansion of vaccine induced OVA- (Fig. 5e, Supplementary Fig. 7a), Adpgk- (Fig. 6e, Supplementary Fig. 7f) and HPV-E7-specific (Fig. 6j, Supplementary Fig. 7k) CD8+ T cells in the periphery. Surprisingly, circulating antiviral CD8+ T cells were unaffected by checkpoint inhibition in all three tumor models (Supplementary Fig. 7b,c,g,h,l,m) but the ratio of antitumor to antiviral CD8+ T cells in circulation was not greatly enhanced by combining checkpoint blockade antibodies with KVK vaccination (Supplementary fig 7d, i, n). Interestingly, αPD-1 therapy did not lead to further expansion of circulating OVA-specific CD8+ T cells in mice bearing B16-OVA tumors when combined with heterologous vaccination (Supplementary Fig. 13d). This might explain why αPD-1 therapy failed to further enhance the therapeutic effect of KVK vaccination (Supplementary Fig. 13a) in this tumor model, reflected in delayed tumor growth (Supplementary Fig. 13b) and prolonged survival (Supplementary Fig. 13c).
Tumor re-challenge of long-term survivors was performed in E.G7-OVA, MC-38 and TC-1 tumor models. KVK heterologous prime-boost (+/- CPI) developed an effective memory response, as almost all the re-challenged mice rapidly rejected the newly implanted tumor (Supplementary Table 3–5). Interestingly, in TC-1 bearing mice, only 60% of homologous VSV-GP-HPV treated long-term survivors were protected against re-challenge, possibly reflecting the reduced formation of memory precursor cells compared to the heterologous vaccination. Similarly, only 75% of long-term survivors which had successfully rejected MC-38 tumors upon homologous VSV-GP-Mad24 vaccination remained tumor-free after rechallenge.
Taken together, the data strongly support the combined application of KISIMA-TAA cancer vaccine and VSV-GP-TAA oncolytic vaccine in a heterologous prime-boost regimen. This approach leads not only to significantly enhanced peripheral and intratumoral T cell levels, but also to a profound reshaping of the TME towards a more immune-supportive composition.
