In situ DC-primed vaccine combined with CD137 agonist elicits long-lasting antitumor immunity through cDC1-mediated tumor antigen-specific CD8+ T cell responses

Abstract

Background

Intratumoral vaccines offer a promising avenue in cancer immunotherapy by harnessing the tumor microenvironment to stimulate immune responses. However, challenges persist in maximizing their effectiveness and addressing immune suppression within tumors.

Methods

Conventional in situ vaccine (CisVac) and α-CD137 antibody were administered in implanted mouse models of lung and colon cancer to evaluate therapeutic efficacy. The initiation mechanism of CD8⁺ T cells was determined using antibody blockade and transgenic mouse models. Single-cell RNA sequencing was used to further characterize the activation mechanism of T cells.

Results

Our findings indicate that this synergistic approach markedly inhibits tumor growth while eliciting a robust antitumor immune response, characterized by heightened activation and cytotoxic differentiation of CD8⁺ T cells, as well as the polarization of conventional dendritic cells (cDC1). Mechanistically, it promotes the cDC1-dependent proportion and cytokine production of tumor antigen-specific CD8⁺ T cells, mobilizes and establishes enduring systemic immune memory. Our single-cell transcriptome analyses revealed that the therapy facilitates a functional remodeling of regulatory T cells (Tregs) with upregulated inflammatory genes, potentially attenuating immune suppression. Cell–cell communication analyses highlighted interactions between CD4⁺ Th1-like/Th17 cells and monocytes/DCs through the CD40L-CD40 pathway, indicating potential intercellular regulatory mechanisms.

Conclusions

Our results highlight the transformative potential of the CisVac/α-CD137 combination in cancer immunotherapy, paving the way for further exploration of its clinical utility and long-term efficacy.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• In situ cancer vaccines are a promising class of immunotherapeutics that aim to activate immune responses directly within the tumor microenvironment (TME). Vaccines using damage-associated molecular patterns like HMGN1 and toll-like receptor (TLR) agonists such as 3M-052 have been explored for their ability to stimulate dendritic cells (DCs) and trigger antitumor immunity. Despite encouraging preclinical results, the effectiveness of these vaccines is often limited by the immunosuppressive properties of the TME, insufficient DC activation, and challenges in sustaining durable immune responses. Combining TLR agonists with immune checkpoint inhibitors or co-stimulatory signals, such as CD137 agonists, has shown promise but remains underexplored and underdeveloped.

WHAT THIS STUDY ADDS

• This study demonstrates that the combination of a conventional in situ vaccine (CisVac), consisting of HMGN1 and 3M-052, with a CD137 agonist (α-CD137) significantly improves antitumor immune responses in murine models. The treatment promotes robust activation of CD8⁺ T cells, enhances the maturation and migration of conventional DC1, and reprograms Tregs into an effector-like phenotype. Additionally, the therapy fosters the generation of systemic immune memory, resulting in durable tumor regression with minimal toxicity. These findings highlight a novel strategy for amplifying immune activation and overcoming TME-induced immune suppression.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• This study has important implications for cancer immunotherapy, particularly in the development of combination strategies that modulate both innate and adaptive immune responses within the TME. The demonstrated synergy between CisVac and α-CD137 provides a compelling basis for future clinical investigations, potentially leading to more effective treatments for various cancers. The findings suggest a new therapeutic avenue for overcoming immune suppression, which may improve patient outcomes, guide clinical trial design, and influence treatment paradigms by incorporating CD137 agonists into immunotherapeutic regimens.

Introduction

In situ cancer vaccines offer a unique immunotherapeutic approach by stimulating local immune responses directly within the tumor site. Unlike traditional cancer vaccines that primarily aim to elicit systemic immunity, in situ vaccines modulate the tumor microenvironment (TME) to activate immune cells, particularly dendritic cells (DCs), to initiate robust antitumor immunity.1,4 However, significant challenges remain, especially in achieving sufficient DC activation and sustaining durable immune response for effective tumor control.⁵ Clinical translation has also been hindered by limitations in overcoming the various barriers that reduce vaccine efficacy.^{6 7}

DCs are essential antigen-presenting cells, activated on recognizing pathogen-associated molecular patterns or damage-associated molecular patterns (DAMPs) through pattern recognition receptors, such as toll-like receptors (TLRs). Conventional in situ vaccine (CisVac) strategies often use the adjuvant effect of targeting TLRs to mobilize endogenous DCs.^{1 8} For instance, HMGN1, a DAMP molecule, has been shown to bind to TLR4, effectively inducing DC maturation and migration.9,12 This effect is further synergistically enhanced when combined with TLR7/8 agonists like MEDI9197 (3M-052),^{13 14} which is a lipophilic TLR7/8 agonist designed for prolonged retention at the injection site. This retention amplifies localized immune activation while minimizing systemic toxicity, representing a potential for clinical application.^{15 16} Nonetheless, despite these adjuvant effects, the overall anti-tumor efficacy remains limited, necessitating further enhancements to overcome existing barriers.¹

The TME poses significant hurdles due to its immunosuppressive properties, characterized by a high density of inhibitory receptors on T cells, including PD-1, CTLA-4, TIGIT, and LAG-3, which collectively drive T cell exhaustion and attenuate antitumor activity.^{17 18} Beyond this, co-stimulatory pathways like the 4-1BB (CD137) axis are critical for reinvigorating T cells and reversely activating DCs,¹⁹ thereby promoting robust antitumor activity. Activation of 4-1BB can initiate strong immune responses,^{20 21} and combining 4-1BB agonists with immune checkpoint inhibitors (ICIs) has shown synergistic effects by both activating effector T cells and alleviating immunosuppression.^{22 23} However, activated DCs within the TME often exhibit reduced expression of 4-1BBL (CD137L), limiting their ability to stimulate T cells effectively and sustaining immune tolerance.24,26 While earlier CD137 agonists faced safety issues,^{27 28} newer generations with improved profiles are currently undergoing clinical trials.29,32 The feasibility of combining CD137 agonists with TLR agonists remains underexplored,³³ despite its potential to synergistically enhance immune activation and reprogram the TME.

Here, we investigated the therapeutic potential of combining a CisVac, composed of HMGN1 and 3M-052, with a CD137 agonist (α-CD137). We demonstrated that this combination effectively suppressed tumor growth in murine models of lung and colon cancer. The therapy promoted the maturation and migration of conventional DCs (cDC1) to tumor-draining lymph nodes (TDLNs), activated both CD4⁺ and CD8⁺ T cells, and reprogrammed the intratumoral Treg population toward an inflammatory phenotype. Additionally, the combination boosted the functionality and frequency of antigen-specific T cells, facilitating memory cell dynamics and establishing durable systemic immune memory with minimal toxicity. Our findings highlighted a promising strategy for achieving potent and long-lasting antitumor immunity.

Result

CisVac synergizes with α-CD137 to inhibit tumor progression with minimal toxicity

To evaluate the efficacy of combining the HMGN1 and 3M-052 in situ vaccine (CisVac) with an α-CD137 agonist, we established subcutaneous tumor models in immunocompetent C57BL/6 mice (6–8 weeks of age), using Lewis lung cancer cells (LLC), KRAS mutation/TP53 homozygous knockout (mKRAS/TP53^−/−; KP13) lung cancer cells, or colon cancer cell (MC38), respectively. Treatment was started when the tumor reached 8–10 mm, designating this day as day 0. CisVac, containing HMGN1 (0.5 µg) and 3M-052 (20 µg), was administered intratumorally on days 0, 4, and 8, with or without α-CD137, which was injected intraperitoneally on days 1 and 6 post-treatment initiation (figure 1A).

Figure 1. Combination of CisVac and α-CD137 inhibits tumor growth with minimal toxicity. C57BL/6J mice were subcutaneously inoculated with 1×10⁶ tumor cells of LLC, KP13, and MC38, respectively. CisVac (HMGN1: 0.5 µg; 3 M-052: 20 µg) and/or an α-CD137 agonist (200 µg) were administered. (A) A schematic diagram outlining the treatment schemes for the different tumor-bearing mice. (B–D) The average volumes of LLC (B), KP13 (C), or MC38 (D) tumors in the PBS (phosphate-buffered saline was used as negative control in all experiment), α-CD137, CisVac, and CisVac/α-CD137 treatment groups, respectively. (E–G) The survival curves of LLC (E), KP13 (F), or MC38 (G) tumor-bearing mice in different treatment groups, respectively. (H–K) Serum levels of ALT (alanine aminotransferase), AST (aspartate aminotransferase), CREA (creatinine), and UA (uric acid) at the end of MC38 tumor model experiment (n=5). Data were shown as mean±SEM. (B–D) were compared between groups employing two-way ANOVA test. (E, F) were compared between the PBS group and the therapy group using the log-rank test. (H–K) were compared between groups using one-way ANOVA. ***p<0.001; ****p<0.0001; ns, not significant. ANOVA, analysis of variance; CisVac, conventional in situ vaccine; LLC, Lewis lung cancer cell.

In the LLC and KP13 tumor models, treatment with CisVac or α-CD137 alone effectively curtailed tumor growth rate and volume. Importantly, the combination of CisVac and α-CD137 exerted a synergistic effect, markedly inhibiting tumor progression (figure 1B,C, online supplemental figure S1A,B). Among LLC-bearing mice, two of the fifteen treated with the combination achieved complete tumor regression (combination index, CI=0.82) and remained tumor-free for up to 3 months (figure 1E, online supplemental figure S1A). Nevertheless, this therapeutic effect was not observed in the KP13 model (CI=0.77) (figure 1F, online supplemental figure S1B). In the MC38 model, characterized by higher immunogenicity than LLC and KP13, both CisVac and α-CD137 displayed antitumor activity individually. The combination treatment resulted in complete tumor regression and dramatic extended survival, achieving a 60% reduction in tumor volume relative to other groups (CI=0.78) (figure 1D and G). In the combination-treated group, eight out of fifteen mice remained tumor-free and were monitored for another 3 months, while the rest succumbed to tumor burden within 40 days of treatment initiation (online supplemental figure S1C). Therefore, in these aggressive primary tumors, the combined CisVac and α-CD137 treatment effectively suppressed tumor growth. However, due to inter-strain variations in tumor immunogenicity, tumors with lower immunogenicity may exhibit resistance to eradication.

During the treatment of LLC, KP13, and MC38 tumor-bearing mice, no significant changes in body weight were observed (online supplemental figure S2A–C). To further assess safety, serum levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST), creatinine (CREA), and uric acid (UA) were measured. The results showed no significant differences in these liver and kidney functional markers between treated and untreated groups (figure 1H–K). To thoroughly evaluate the potential impact of treatment, we performed HE staining on major organs (lungs, kidneys, liver) across all groups. No significant abnormalities were observed in these vital organs within any treatment group (online supplemental figure S2D), further supporting the favorable safety profile of the CisVac/α-CD137 treatment strategy in this model. In summary, CisVac/α-CD137 can effectively inhibit tumor growth without eliciting significant toxic side effects, demonstrating both efficacy and safety in the experimental model.

The efficacy of synergistic therapy depends on the efficient cross-presentation of antigens by cDC1

CD103⁺ DCs (cDC1) have been shown to be the primary antigen-presenting cells in TDLNs that drive CD8⁺ T cell responses.34,36 To elucidate the synergistic antitumor mechanism of CisVac/α-CD137 agonist, DC subpopulations in TDLN were evaluated and characterized via CD103 and CD11b expression on day 13 following tumor inoculation (figure 2A).^{34 37} The data indicated a stable total DC count in the TDLN (figure 2B). Nevertheless, in contrast to the phosphate-buffered saline (PBS) group, the CisVac/α-CD137 group showed a marked increase in cDC1 proportion alongside a decrease in cDC2 (figure 2C–F), suggesting dynamic shifts in DC populations following CisVac/α-CD137 treatment. Further analysis demonstrated that both cDC1 and cDC2 exhibited significantly higher expression of migratory marker CCR7 in the combination treatment group (figure 2G and I), suggesting heightened migratory potentials to TDLNs. However, cDC1, instead of cDC2, exhibited higher expression of the activation marker CD80 (figure 2H and J) in CisVac-treated groups, indicating activation of cDC1s in response to CisVac-based treatment. We further characterized the distribution patterns of DCs 1 day post-treatment (day 19). The results revealed a significant reduction in the proportion of total DCs within the TME (online supplemental figure S3A), concurrent with marked increases in the TDLN and spleen following CisVac/α-CD137 combination therapy (online supplemental figure S3E,F). This redistribution of DCs likely reflects the mobilization of activated DCs to TDLN, where they potentiate T cell priming, followed by robust T cell expansion in the TME—thereby diluting DC proportions among the CD45⁺ lymphocyte infiltrates. Notably, the DC1 subsets, but not DC2, exhibited pronounced increased proportion within the TME (online supplemental figure 3B–D), suggesting DC1 potential involvement in anti-tumor immunity.

Figure 2. The combination therapy promotes Batf3-dependent cDC1 polarization. (A) A schematic diagram illustrated the analysis of DC subsets in TDLN on day 13 after inoculation of MC38 tumor cells in mice. The proportion of MHC-II⁺CD11c⁺ cells (DCs) among living cells (n=4) was shown in (B), while (C) presented a representative diagram of the proportion of cDC1 and cDC2 in the DC subgroup, as determined by flow analysis. (D, E) The proportion of cDC1 and cDC2 cells in the DC subgroup (n=4), respectively. (F) The ratio of cDC1 to cDC2 in TDLN (n=4). The mean fluorescence intensities (MFI) of CCR7 (G) and CD80 (H) in cDC1 (n=4) were presented, with the mean fluorescence intensities of CCR7 and CD80 in cDC2 shown in (I, J), respectively. The data were obtained from four independent samples. (K) Schematic representation of the inoculation of 1×10⁶ MC38 tumor cells into the right subcutaneous tissue of wild-type WT and Batf3^−/− mice, respectively. (L) Growth curves and (M) survival curves of WT and Batf3^−/− mice in the PBS group and the Therapy group (CisVac/α-CD137) (n=5). Data were shown as mean±SEM. Two-way ANOVA was used for statistical analysis in (L), log-rank tests were used for statistical analysis in (M), and one-way ANOVA was used for statistical analysis in all the bar graphs. *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001; ns, not significant. ANOVA, analysis of variance; cDC, conventional dendritic cell; CisVac, conventional in situ vaccine; TDLN, tumor-draining lymph node; WT, wild-type.

To further elucidate cDC1’s role in the combined therapy, we assessed tumor growth in both wild-type (WT) and Batf3^−/− mice, where the ability of cDC1 cells to effectively present antigens and activate T cells is impaired (figure 2K).^{35 38} Results showed that tumor growth was no longer controlled, and the antitumor effect of CisVac/α-CD137 synergy was abolished in Batf3^−/− mice (figure 2L,M). In summary, CisVac/α-CD137 enhances the anti-tumor immune response by promoting cDC1 polarization and activation, facilitating their migration to TDLN, thus underscoring cDC1’s indispensable role in antigen presentation and potentially CD8⁺ T cell-mediated antitumor immunity.

CisVac/α-CD137 therapy relies on the activation of T cells

In order to investigate the role of T cells in the treatment of CisVac/α-CD137, we established an MC38 tumor-bearing model with Foxp3-YFP-Cre mice, allowing for precise tracking of Treg dynamics via flow cytometry. On the second day post-treatment, we analyzed the T cell subsets and their functional status in both TDLN and tumor tissue (figure 3A, online supplemental S4A). In TDLN, CisVac/α-CD137 significantly increased the proportion of CD8⁺ T cells (CI=0.46) while reducing CD4⁺ T cells and Tregs among TCR-positive cells (figure 3B–D), resulting in a significant rise of CD8⁺ T cells to Tregs ratio (CI=0.003) (figure 3E). Similarly, in tumors, CD8⁺ T cell levels were significantly elevated in the α-CD137 and CisVac/α-CD137 groups compared with the PBS and CisVac groups, whereas CD4⁺ T cells and Foxp3⁺ Tregs were markedly reduced (online supplemental figure S4B–E). Further analysis highlighted increased CD25 expression, a T cell activation marker, on CD8⁺ T cells (CI=0.22) and Foxp3⁻CD4⁺ T cells (CI=0.33) within TDLN in the CisVac/α-CD137 treatment group (figure 3F–G). These findings suggested effective activation and expansion of CD8⁺ T cells in TDLN by CisVac/α-CD137. However, CD25 expression on CD8⁺ T cells and Foxp3⁻CD4⁺ T cells within tumor tissue showed no significant changes (online supplemental figure S4F–G), possibly due to these T cells reaching a stable or terminal differentiation state at this stage, limiting further changes in CD25 expression. At the same time point, we quantified cytokine secretion within TME, and the results demonstrated a significant upregulation of GM-CSF, IL-1β, and IFN-γ within the TME (online supplemental figure S5A–H).

Figure 3. The combination therapy promotes the activation of T cells. (A) A schematic diagram illustrating the analysis of T cell subsets from MC38 tumor-bearing mice engineered with Foxp3 YFP-Cre on day 19. (B, C) The proportions of CD8⁺ T cells and CD4⁺ T cells in the TDLN, respectively. (D) The proportion of Tregs in TCR⁺ cells and CD4 cells. (E) The mean ratio of CD8⁺ T cells to Treg cells. (F, G) The proportions of CD25 expression in Foxp3^-CD4⁺ T cells and CD8⁺ T cells, respectively. The diagram on the left showed a flow cytometry diagram, while the right diagram presented a statistical chart. (H) A schematic representation of the monoclonal antibody cloning and treatment of α-CD4, α-CD8, and α-NK1.1. (I, J) The average tumor growth curves and survival of mice in the different monoclonal antibody and therapy (CisVac/α-CD137) groups, respectively. Data were shown as mean±SEM. Two-way ANOVA was used for statistical analysis in (I), log-rank tests were used for statistical analysis in (J), and one-way ANOVA was used for statistical analysis in all the bar graphs. *p<0.05; **p<0.01; ****p<0.0001; ns, not significant. ANOVA, analysis of variance; CisVac, conventional in situ vaccine; TDLN, tumor-draining lymph node.

To further clarify the role of T cells in CisVac/α-CD137 combined therapy, we depleted specific T cell subsets in mice using α-CD4 and α-CD8 antibodies (figure 3H). Given that α-CD137 agonists can also enhance NK cell activation and cytotoxicity, we administered α-NK1.1 antibodies to assess NK cell involvement in this therapy (figure 3H). CD4⁺ T cells and CD8⁺ T cells were confirmed to be fully depleted on days 14 and 20, whereas NK cells exhibited rapid recovery, leading to no significant differences among the groups regarding NK cell depletion (online supplemental figure S6A–C). To further confirm the complete depletion of NK cells, we quantified the proportion of NK cells in tumor, TDLNs, and spleen on day 19. The results revealed that following administration of the α-NK1.1 antibody, NK cells were near-completely depleted in vivo (online supplemental figure S7A–C). In the α-CD8/Therapy (CisVac/α-CD137) group, tumor growth was partially inhibited compared with the IgG control, yet this group failed to mount an effective antitumor response due to CD8⁺ T cell depletion, resulting in later tumor recurrence and 0% survival (0/5) (figure 3I,J). In contrast, the α-CD4/Therapy group showed slightly compromised tumor control compared with IgG/Therapy group, but significantly prolonged survival compared with the IgG controls, with two mice achieving cure (figure 3I,J), likely due to impaired Tregs’ immunosuppression by the α-CD4 antibody. In the α-NK1.1/Therapy group, no significant difference in tumor growth or survival was observed compared with the IgG/Therapy group, with three mice achieving cure (figure 3I,J). These results highlighted that the CisVac/α-CD137 combination therapy predominantly relies on T cell response in vivo. CD8⁺ T cells exhibited a more pronounced response than CD4⁺ T cells, further emphasizing the critical role of CD8⁺ T cells in mediating antitumor immunity within the context of combination therapy.

CisVac/α-CD137 efficacy requires Batf3-dependent DC1-mediated tumor antigen-specific T cells cross-priming

To determine whether the phenotypic differences observed in T cells could translate into functional variations, we conducted a comprehensive evaluation of T cell responses to tumor antigens. Irradiated tumor cells were used as antigen sources, presenting a wide array of tumor-derived antigens to activate diverse immune cells and elicit responses from various tumor antigen-specific T cells in vivo. Flow cytometry analysis of spleen cells stimulated by irradiated tumor cells for 20 hours revealed that tumor antigen-specific CD8⁺ T cells exhibited markedly elevated secretion of IFN-γ (CI=0.11) and Granzyme B (CI=0.09) following the combined treatment (figure 4A,B, online supplemental figure S8A). Similarly, a notable increase in IFN-γ (CI=0.06) and Granzyme B (CI=0.09) secretion was observed in tumor antigen-specific CD4⁺ T cells (figure 4C,D). Furthermore, ELISpot assays demonstrated a substantial rise in IFN-γ levels in the CisVac/α-CD137 group compared with the PBS group, highlighting an enhancement in the functional activity of activated tumor antigen-specific T cells (figure 4E). To determine whether Batf3-dependent cDC1 could activate T cells through antigen cross-presentation in the treatment group, MC38 tumor cells were subcutaneously inoculated into WT and Batf3^−/− mice, with subsequent evaluation of IFN-γ secretion by T cells in PBS and therapy (CisVac/α-CD137) groups on the second day post-treatment. The results indicated a marked reduction in IFN-γ secretion by endogenous T cells in the Batf3^−/− tumor-bearing mice treated group compared with their WT counterparts (figure 4F). The presence of IFN-γ in Batf3^−/− mice, yet not significant compared with untreated Batf3^−/− mice, may result from direct activation of T cells by the α-CD137 agonist rather than classical DC antigen cross-presentation.

Figure 4. CisVac/α-CD137 enhances cDC1-mediated tumor antigen-specific T cells cross-priming. (A–D) On the second day following treatment of MC38-OVA tumor-bearing mice, spleen samples from the various groups were collected and prepared into single-cell suspensions. The spleen cells were stimulated with irradiated tumor cells, and the cytokines secreted by T cells were detected by flow cytometry. The statistical analysis of IFN-γ and Granzyme B secretion by activated CD8⁺ T cells was shown in (A, B). The statistical analysis of IFN-γ and Granzyme B secretion by activated CD4⁺ T cells was presented in (C, D). The corresponding flow chart was displayed on the left-hand side, with the corresponding statistical chart displayed on the right. Left: representative flow cytometric plots; Right: a statistical chart. (E) Enzyme-linked immunosorbent spot (ELISpot) analysis of IFN-γ spot-forming cells in spleen cells of WT mice stimulated by irradiated MC38-OVA tumor cells. 1 day after treatment, IFN-γ secretion in the PBS and therapy (CisVac/α-CD137) group of WT and Batf3^−/− tumor-bearing mice was detected using ELISpot. (F) Irradiated MC38 tumor cells stimulate spleen cells in tumor-bearing mice, leading to the secretion of IFN-γ by tumor-specific response cells. Left: average counts; Right: representative images for each group. (G) Carboxyfluorescein diacetate succinimidyI ester (CFSE) staining was used to detect cytotoxicity. Left: the representative flow histogram. Right: the in vivo cytotoxicity killing rate. Data were shown as mean±SEM. One-way ANOVA was used for statistical analysis in all the bar graphs. *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001; ns, not significant. ANOVA, analysis of variance; cDC, conventional dendritic cell; CisVac, conventional in situ vaccine; WT, wild-type.

These data suggest enhanced antigen-specific cytotoxic T cell (CTL) activity against tumor cells following combined therapy. To further evaluate this, we assessed the in vivo lytic capacity of different treatment groups on target cells. At the end of the treatment, MC38-OVA tumor-bearing mice received spleen cells labeled with two different concentrations of carboxyfluorescein diacetate succinimidyI ester (CFSE) via tail vein injection to monitor specific target cell lysis (online supplemental figure S8B,C). The CisVac/α-CD137 treated mice exhibited significantly enhanced killing of OVA_SIINFEKL (N4)-specific target cells labeled with high concentration of CFSE within TDLN, accompanied by a substantial increase in cytotoxicity (figure 4G). No significant difference was observed in the spleen (online supplemental figure S8D). Collectively, these findings suggest that CisVac/α-CD137 combined therapy activates tumor antigen-specific T cells through cDC1 antigen cross-presentation, thereby enhancing cytotoxicity and effectively targeting tumor cells.

CisVac/α-CD137 therapy could potently augment tumor antigen-specific T cell responses

To further track the response of tumor antigen-specific T cells, we conducted an assay using spleen cells from transgenic OT-1 mice, in which the OT-1 cells are tumor antigen-specific CD8⁺ T cells that respond to OVA antigens. CFSE-labeled spleen cells from CD45.1 OT-1 mice were adoptively transferred to WT mice (CD45.2) bearing MC38-OVA tumors via the tail vein on day 13 post-inoculation. Three days later (early stage), the proliferative capacity of OT-1 cells in TDLN and spleen of treated mice was analyzed using flow cytometry (figure 5A, online supplemental figure S9A). The results showed that, compared with the monotherapy or control groups, the combination therapy group not only increased the proportion of antigen-reactive OT-1 cells (figure 5B,C), but also significantly enhanced proliferation in TDLN (figure 5D). In contrast, no significant difference was observed in the spleen (online supplemental figure S9C). These results suggested that the combination therapy effectively induced an early-stage antigen-specific CD8⁺ T cell response through antigen cross-presentation specifically in the TDLN. To further explore the differentiation dynamics of antigen-reactive OT-1 cells, the CD44 and CD62L levels of OT-1 cells from spleen and TDLN were analyzed 7 days (late-stage) after adoptive transfer of spleen cells from OT-1 mice on day 9 (figure 5E, online supplemental figure S9B). Following adoptive transfer, we observed a non-significant trend wherein the proportion of OT-1 cells among CD8⁺ T cells decreased in the peripheral blood and increased within the tumor tissue of the combination therapy group compared with controls at late stage (online supplemental figure S9D). Contrary to the early stage, a significant increase in OT-1 cell proportion following combined treatment was observed in the spleen (CI=0.11, figure 5F) instead of TDLN (online supplemental figure S9E), potentially due to trafficking of antigen-reactive OT-1 cells from TDLN to spleen in late stage. Notably, within OT-1 cells, there was a prominent accumulation of antigen-specific CD44⁺CD62L⁺ central memory T cells in both TDLN and spleen (figure 5G, online supplemental figure S9F–G), known for their ability to self-renew, persist long-term in circulation, and rapidly respond on antigen rechallenge.³⁹ This phenomenon is consistent with the observed increase in the proportion of DCs in the TDLN and spleen (online supplemental figure S3E,F), further suggesting that the interaction between antigen-presenting cells and effector T cells plays a critical role in modulating the antitumor immune response. These findings highlighted that CisVac/α-CD137 combination therapy not only elicited an antigen-specific CD8⁺ T cell response but also facilitated the dynamic repository of memory cells, thereby potentially enhancing long-term immune memory against tumor antigens in mice.

Figure 5. CisVac/α-CD137 potently augments tumor antigen-specific T cell proliferation and memory cell mobilization. On the 13th day following the inoculation of MC38-OVA (with treatment commencing on the 7th day), spleen cells from CD45.1 OT-1 mice stained with CFSE (2 µM) for 10 min, and then infused into CD45.2 MC38-OVA tumor-bearing mice via the tail vein. After 3 days, the proliferation of OT-1 cells was analyzed. (A) Schematic diagram. (B) A flow chart showing the proportion of OT-1 cells among TCR⁺CD8⁺ T cells in different treatment groups. (C) The statistical analysis of OT-1 cell proportions in TDLN was presented. (D) The proliferation of OT-1 cells. Left: Flow chart representation. Right: The division index of OT-1 cells in different treatment groups. On the 9th day after inoculating with MC38-OVA (with treatment commencing on the seventh day), 2×10⁶ spleen cells from OT-1 mice were infused via the tail vein, and the proportion of OT-1 cells was analyzed 7 days later. (E) Schematic diagram. (F, G) The proportion of OT-1 cells in the spleen and the proportion of central memory T cells within the OT-1 cell population in different treatment groups, respectively. The flow chart was presented on the left, while the statistical chart is shown on the right. Data were shown as mean±SEM. One-way ANOVA was used for statistical analysis in all the bar graphs. *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001; ns, not significant. ANOVA, analysis of variance; CisVac, conventional in situ vaccine; TDLN, tumor-draining lymph node.

The combined therapy induces systemic and long-term tumor-specific immune memory

We further investigated whether the combined therapy initiates systemic immune response and induces long-term, tumor-specific immune memory. To verify this, we established a bilateral MC38 tumor model and monitored tumor growth at both sites (figure 6A). The combined therapy not only significantly suppressed primary tumor growth but also markedly inhibited the growth of distal tumors (figure 6B), indicative of a robust systemic antitumor immune response, although survival could not be assessed in this bilateral model due to the mandated humane endpoints related to total tumor volume. Next, the mice cured by combined therapy, who have undergone a 100-day observation period without recurrence, were rechallenged by homologous MC38 tumor cells at the primary tumor site and heterologous PANC02 tumor cells simultaneously on the contralateral side as a positive control (figure 6C). The results revealed no regrowth of MC38 tumors at the original site, whereas the PANC02 tumors on the contralateral side grew unaffected (figure 6D). To exclude the factor of inoculation site, we subsequently injected PANC02 tumor cells at the primary tumor site and MC38 tumor cells on the contralateral side (figure 6E). The results showed that six out of eight mice did not develop MC38 tumors on the contralateral side, while PANC02 tumor growth remained unaffected (figure 6F). Collectively, these data suggested that CisVac/α-CD137 combination therapy not only established localized tumor-specific immune protection at the primary site, but also extended this effect systemically and formed long-term immune memory, supporting a more effective and comprehensive immune response.

Figure 6. CisVac/α-CD137 establishes remote protection and long-term tumor-specific immune memory. (A) A total of 1×10⁶ MC38 tumor cells (primary tumor) were inoculated subcutaneously on the right flank, while 0.5×10⁶ tumor cells (distal tumor) were inoculated on the left flank. Once the primary tumor had reached a diameter of 8–10 mm, CisVac was administered intratumorally, with the distal tumor not receiving treatment. α-CD137 was then injected intraperitoneally for treatment. (B) The mean growth curve of the right primary tumor (Right) and distant tumor (Left) (n=5). (C) Schematic diagram of ipsilateral inoculation of MC38 and contralateral inoculation of PANC02 tumor cells. (D) The average tumor growth curves of tumor-regressing (red line) and naïve (black line) mice for contralateral PANC02 (Right) and primary site MC38 tumors (Left) (n=5). (E) A schematic depicts contralateral inoculation with MC38 and ipsilateral inoculation with PANC02 cells. (F) The average growth curves for contralateral MC38 (Right) and ipsilateral PANC02 tumors (Left) (n=7). Data were shown as mean±SEM. Two-way ANOVA was used for statistical analysis in all the tumor growth curves. *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001; ns, not significant. ANOVA, analysis of variance; CisVac, conventional in situ vaccine.

CisVac/α-CD137 treatment promotes CD8⁺ T and NK cell cytotoxicity and remodeled TME

To verify the antitumor efficacy of the CisVac/α-CD137 treatment strategy, we conducted single-cell RNA sequencing (scRNA-seq) of sorted CD45⁺ tumor-infiltrating immune cells from MC38 tumor-bearing mice on day 19 following CisVac/α-CD137 treatment or PBS control. Eight primary clusters were identified, including B, CD4⁺ T, CD8⁺ T, γδT, DC, Ma/Mo, NK, and NKT cells (online supplemental figure S10A). Notable increases in myeloid cells and NK cells were accompanied by a decrease in CD8⁺ T cell population following CisVac/α-CD137 treatment compared with PBS control (online supplemental figure S10B). Further analysis of CD8⁺ T cell subsets revealed a striking reversal in the ratio of effector memory T cells (C3) to cytotoxic T cells (C6) (figure 7A,B). Clusters 3, 4, and 6 shared heightened expression of activation and effector (Cd44, Cxcr6, Gzmb) markers (figure 7C, online supplemental figure S10C). Cluster 3, as compared with cluster 4, exhibited higher expression of genes associated with T cell survival and trafficking (Pim1 and S1pr1), and lower levels of interferon-inducible genes (Ifit1 and Ifit3), indicating early-stage effector function (figure 7D). In contrast, cluster six was characterized by genes linked to cytotoxic function (Ifng, Ccl3, Prf1, Tnfrsf9) and a propensity towards exhaustion (Havcr2, Pdcd1, Lag3, and Tox) (online supplemental figure S10C). Consistently, cluster six showed enhanced gene modules associated with cytotoxicity and cytokines, as well as higher signature scores related to anti-tumor immunity (figure 7E, online supplemental figure S10D). Additionally, cluster 3 of CisVac/α-CD137 treatment was more immune-active, as shown by upregulated inflammatory response, cytokine production, and phagocytosis pathways (figure 7F, online supplemental figure S10E). NK cells were also robustly activated and expanded, producing abundant cytokines (online supplemental figure S10F). These findings suggested that CisVac/α-CD137 treatment promoted the activation and differentiation of CD8⁺ T cells and NK cells towards a cytotoxic phenotype.

Figure 7. CisVac/α-CD137 enhances CD8⁺ T activation and remodels TME. (A) Uniform manifold approximation and projection (UMAP) visualization of CD8⁺ T populations in MC38 tumor-bearing mice colored according to cluster classification. (B) Cell components of CD8⁺ T populations in PBS control and CisVac/α-CD137 treatment group. (C) Dot plot showing expression levels of selected genes in CD8⁺ TIL clusters. (D) UMAP visualization of transcriptional expression of indicated immune genes as determined by scRNA-seq. (E) Violin plot illustrating comparisons of indicated signature scores among CD8⁺ TIL clusters. (F) Gene set enrichment analysis (GSEA) illustrating upregulated pathways in CD8⁺ effector memory T cells. (C3) following CisVac/α-CD137 combined treatment, compared with PBS control. (G) UMAP visualization of the CD4⁺ T populations in MC38 tumor-bearing mice colored according to cluster classification. (H) Dot plot showing expression levels of selected genes in CD8⁺ TIL clusters. (I) Violin plot illustrating indicated gene expression in CD4⁺ Tregs following CisVac/α-CD137 combined treatment compared with PBS control with log₂ fold change in parentheses. (J) UMAP visualization of the myeloid cells in MC38 tumor-bearing mice colored according to cluster classification, split by treatment groups. (K) GSEA illustrating up-regulated pathways in cDC1s (C3) following CisVac/α-CD137 combined treatment, compared with PBS control. (L) Heatmap illustrating the relative importance of each cell group identity based on the four network centrality metrics of CD40L-CD40 signaling. (M) Chord diagram visualizing cross-talks from other immune cells to Tregs, the edge width refers to the weight strength of indicated cells to Tregs (Cross group and between group comparisons in (E, I) were performed using Kruskal-Wallis test and Dunn’s test with Bonferroni correction, respectively. *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001; ns, not significant). CisVac, conventional in situ vaccine; scRNA-seq, single-cell RNA sequencing; TME, tumor microenvironment.

CD4⁺ T cells comprised four clusters, representing central memory, Th1-like, Th17, and Treg phenotypes (figure 7G,H). Despite a decline in overall proportion in the CisVac/α-CD137 treatment group (online supplemental figure S10G), the Th1-like subtype exhibited an enhanced type I immune phenotype (online supplemental figure S10H). Interestingly, Tregs, which typically play an immunosuppressive role in the TME, displayed an increased proportion among CD4⁺ T cells, yet expressed lower levels of Foxp3 and Il2ra (encodes CD25), and higher levels of pro-inflammatory genes and transcription factor Tbx21, as compared with that in the control group (figure 7I). Previous studies have reported that certain Tregs can transition to a cytotoxic phenotype (termed exTregs) functionally resembling Th1-like or Th17 cells under inflammatory circumstances.⁴⁰ The downregulation of Foxp3 and CD25 manifested an unstable Treg phenotype following CisVac/α-CD137 treatment. The paradox in proportion and status of Tregs implied that CisVac/α-CD137 may functionally remodel tumor-infiltrating Tregs, potentially weakening suppression of CD8⁺ T cell responses.

The myeloid populations expanded dramatically (figure 7J), especially the cDC1, but not cDC2, which exhibited significantly upregulated TLR, interferon, and chemokine signaling (figure 7K), highlighting the role of cDC1 over cDC2 in the context of CisVac/α-CD137 combined treatment. Additionally, Ly6c2^hiCcr2^hi and Cxcl2^hi inflammatory monocytes were abundant, which are pivotal for immune cells' recruitment and activation in the context of infection and tumor contexts.^{41 42} Indeed, these monocytes were the most important participants that interacted with CD8⁺ T cells (online supplemental figure S10I) in CisVac/α-CD137 treatment, as evidenced by cell-cell crosstalk analysis. The Cxcl2^hi inflammatory monocytes and C1qa^hi macrophages engaged closely with effector memory (C3) and cytotoxic (C6) CD8 T cells (online supplemental S10J) via chemokine signaling, underscoring their roles in T cell activation and differentiation. Furthermore, Th1-like and Th17 CD4⁺ T cells acted as the primary activators for DCs and Cxcl2^hi inflammatory monocytes via CD40L-CD40 ligand-receptor pathway (figure 7L), while the reprogrammed Tregs received signals mainly from cDC1 and Cxcl2^hi inflammatory monocytes (figure 7M). Collectively, these findings suggested that CisVac/α-CD137 treatment reshaped the TME and enhanced the interactions between myeloid cells and T cells, thereby promoting effective tumor control.

Discussion

This study elucidates the efficacy and underlying immunological mechanisms of an innovative in situ vaccine strategy targeting DCs in combination with α-CD137, highlighting its potential application in MC38 tumor-bearing murine model. A key feature of this approach involves DCs, especially CD103⁺ DCs with high expression of CCR7, which are essential for transporting tumor antigens to TDLNs and activating T cells.35,3743 Our findings align with this role, indicating that cDC1 are indispensable for antitumor immunity following CisVac/α-CD137 combined therapy, as shown in figure 2K–M. Additionally, the enhanced secretion of IFN-γ and granzyme B from tumor-specific CD8⁺ T cells, coupled with increased OT-1 cell proliferation, underscores the effective T cell activation mediated by α-CD137 and highlights the critical role of cDC1s in tumor antigen presentation of in situ vaccine.^{20 21} Notably, the treatment establishes long-lasting immune memory and maintains a favorable safety profile, underscoring the potential of CisVac/α-CD137 as a transformative therapeutic strategy in cancer immunotherapy with the capacity to harness and amplify the immune system’s natural ability to combat tumors.

Our study also revealed that significant findings regarding the roles of CD8⁺ and CD4⁺ T cells in the CisVac/α-CD137 combined therapy, especially their intricate interactions with myeloid cells, including DCs and monocytes. Consistent with previous studies, the enhanced differentiation from early-stage effector memory to cytotoxicity in CD8⁺ T cell population and significantly compromised tumor control on depletion underscore their pivotal role in mediating antitumor immunity. However, our findings extend this understanding by highlighting the potential role of CD4⁺ T cells in antitumor immunity, noting their partial contribution to tumor control, an area less emphasized in prior studies.44,46 A recent study also highlights the spatial location and interactions between CD4⁺ T cells and CD8⁺ T cells on the same DC, which form trinity clusters, for success of tumor elimination.⁴⁷ Notably, our single-cell transcriptome analyses revealed a prominent interaction between CD4⁺ Th1-like/Th17 cells and monocytes/cDC1 via the CD40L-CD40 pathway, facilitating cross-presentation of tumor antigens, as well as recruitment and activation of T cells. This finding underscores the complexity of the immune response and highlights the collaborative roles of distinct T cell populations in the therapeutic context, which require future validation through approaches including flow cytometry, functional depletion assays, and spatial imaging (eg, immunofluorescence).

Beyond the traditional views that depict Tregs as immunosuppressive entities in infectious disease and tumor formation, our results reveal a notable significant reprogramming of Treg function following combined therapy. Specifically, we observed Treg instability and a shift toward an effector-like phenotype, characterized by reduced Foxp3 expression and enhanced proinflammatory signaling with Th1-like/Th17 transcriptional profiles.⁴⁸ This finding suggested that these unstable Tregs may adopt functions that attenuate immunosuppression or even enhance immune response, highlighting potential for therapeutic strategies that not only engage effector T cells but also modulate Treg behavior to support antitumor immunity. The combination therapy increased the ratio of CD8⁺ T cells to Tregs in the TME, reshaping the tumor into a “hot tumor” with higher immunogenicity. ICIs, such as PD-1 antibodies, could further enhance the anti-tumor immune responses against “cold” tumors exhibiting poor immune responsiveness to immunotherapy. Notably, the recent clinical trial demonstrated that the GVAX vaccine in combination with CD137 agonists and anti-PD-1 antibody exhibited potent antitumor immunity alongside manageable toxicity in pancreatic cancer, often recognized as “cold” tumor,⁴⁹ providing a promising strategy for less immunogenic tumors that warrants further investigation.

In conclusion, our study demonstrates that the CisVac/α-CD137 combination therapy represents a promising immunotherapeutic strategy for treating aggressive tumors. However, CD137 agonists may have adverse effects on antibody and B cell memory responses during vaccination, while limited B cells may weaken antitumor immune responses.⁵⁰ Due to the high heterogeneity of B cells, targeting regulatory B cells (Bregs) may become a new strategy to enhance the efficacy of immunotherapy.^{51 52} Future work should characterize the B cell response to fully understand the interplay between cellular and humoral immunity in this therapeutic setting.

Materials and methods

Mice

In this study, female C57BL/6J mice were purchased from Zhaoqing Huaxia Kaiqi Biological. Batf3^−/− mice (both male and female) were provided by Dr. Yingping Xu of Guangdong Dermatology Hospital. Female CD45.1 OT-1 mice (8–10 weeks old) and Foxp3-YFP-Cre mice (8–10 weeks old, both sexes) were purchased from The Jackson Laboratory (USA) and bred in Zhaoqing Huaxia Kaiqi Biological. All mice were transported to the Animal Center of Guangzhou Medical University under controlled conditions before the experiment and maintained under specific pathogen-free conditions.

Cell lines

The cell lines used in this study included the LLC lung cancer cell line (purchased from Wuhan Punosai Life Science and Technology), the KP13 lung cancer cell line (provided by Dr. Hongbin Ji from Shanghai Institute of Biochemistry and Cell Biology), the MC38 colon cancer cell line (provided by Dr Chenqi Xu from Shanghai Institute of Biochemistry and Cell Biology), the PANC02 pancreatic cancer cell line (provided by Dr Hui Liu from the Fifth Affiliated Hospital of Guangzhou Medical University), and the MC38-OVA transfected cell line (provided by Dr Liufu Deng from Shanghai Jiao Tong University). All cell lines were cultured in DMEM medium (Gibco) supplemented with 10% fetal bovine serum (FBS, BI) and 1% penicillin-streptomycin (Life Technologies). To ensure that all the cell lines were free from mycoplasma contamination, PCR-based mycoplasma testing was performed periodically.

Tumor models and treatments in vivo

For the primary tumor treatment experiment, 1×10⁶ LLC, KP13, MC38, or 0.5×10⁶ MC38-OVA tumor cells were injected subcutaneously into the right flank of C57BL/6J mice. When the tumors reached 8–10 mm in diameter, the mice were randomly assigned to treatment groups. Each group of mice was placed in a cage (n=5–8). Treatment commenced on day 7 for the MC38-OVA bearing mice, on day 9 for the MC38 or KP13 bearing mice, and on day 10 for the LLC bearing mice. The therapeutic regimen (defined as starting on day 0) included intratumoral injections of a conventional in situ DC-primed vaccine (abbreviated as “CisVac”) containing HMGN1 (0.5 µg, R&D Systems) and 3M-052 (20 µg, MedChemExpress) on days 0, 4, and 8. Additionally, a CD137 agonist (200 µg, clone 3H3, BioxCell) was administered intraperitoneally on days 1 and 6. For the tumor rechallenge experiment, mice that were initially inoculated with MC38 cells and achieved complete tumor regression on combined therapy were monitored for another 100 days after the treatment ended to ensure no tumor recurrence. Subsequently, tumor cells were reinoculated on the original site, with PANC02 cells simultaneously inoculated on the opposite flank as a control. For the lymphocyte depletion experiment in MC38 tumor-bearing mice, rat IgG2a isotype control (200 µg, clone 2A3, BioxCell), antibody against CD4 (200 µg, clone GK1.5, BioxCell) and CD8 (200 µg, clone Lyt.2.1, BioxCell) were administered intraperitoneally on days 12, 15, 18, and 21. Additionally, the α-NK1.1 antibody (200 µg, clone PK136, BioxCell) was administered on days 12, 16, 18, 20, and 22 post-tumor inoculation by intraperitoneal. For the bilateral tumor model experiment, C57BL/6J mice were injected subcutaneously with 1×10⁶ MC38 cells on the right flank and 0.5×10⁶ MC38 cells on the left flank. On days 10, 14, and 18, CisVac was administered intratumorally to the primary tumor, leaving the distal tumor untreated. On days 11 and 16, 200 µg of α-CD137 was administered intraperitoneally. All the mice were age-matched and sex-matched. Throughout the treatment period, tumor length (a) and width (b) were measured every 3 days, and tumor volume was calculated using the formula $V=\frac{a\times b^2}{2}$. When the tumor volume reached 2000 mm³, we sacrificed the mice. Complete tumor regression (CR) is defined as maintaining the tumor volume below 50 mm³ throughout the experiment period (5–6 weeks or until animal welfare consideration). Tumor cure is defined as the absence of tumor recurrence during a subsequent observation period of around 100 days.

Preparation of single cell suspension and flow cytometry

Following the experiment, 0.1–0.2 mg of tumor tissue was carefully resected, washed, and thoroughly minced on ice, followed by placement in RPMI 1640 medium (Gibco) containing collagenase IV (0.5 mg/mL, Merck) and DNase I enzyme (0.05 mg/mL, Merck) and digested in a metal bath at 37°C for 40 min. Subsequently, the digestion was terminated using 0.2% bovine serum albumin (BSA)/PBS (FACS buffer), and the cell suspension was filtered through a 70 µm nylon cell filter. Spleen tissue was subjected to a similar processing protocol but was initially lysed with a red blood cell lysate for 5 min, with the reaction stopped with five times the volume of FACS buffer. TDLNs were prepared by grinding and filtering only. Peripheral blood from mice was processed solely with erythrocyte lysate. The tissue suspensions obtained were centrifuged at 1200 rpm to remove the supernatant and then resuspended in FACS buffer to obtain the corresponding single-cell suspensions for subsequent experiments.

Once the single-cell suspensions were prepared, an appropriate quantity was transferred to a flow cytometry tube. Surface staining was performed by incubating the cells with labeled antibodies (table 1) at 4°C in the dark for 30 min. The cells were then washed with 1 mL of FACS buffer, centrifuged, and resuspended in 0.3 mL of FACS buffer. For intracellular staining, cells were stimulated with irradiated tumor cells for 16 hours, followed by blocking with BFA (2.5 µg/mL, YEASEN) for 3 hours. Controls included a negative control (unstimulated) and a positive control stimulated with PMA/Ionomycin (0.05 µg/1.25 µM, YEASEN). The stimulation was terminated with FACS buffer. After extracellular markers were stained, cells were fixed and permeabilized using BD Cytofix/Cytoperm solution (554714, USA) according to the manufacturer’s instructions. Subsequently, intracellular staining for IFN-γ and granzyme B was performed. The stained cells were collected using a BD Biosciences flow cytometer and analyzed with FlowJo software (V.10.8.1-CL).

Table 1. The antibody models mentioned in the article.

Antibody	Clone	Conjugate	Company	Catalog#
CCR7	4B12	APC	Biolegend	120 107
CD3	145–2 C11	BV421	Biolegend	100 336
CD4	GK1.5	APC	Biolegend	100 412
CD4	RM4-5	FITC	TONOBO biosciences	11-0042-82
CD4	RM4-5	PerCP-Cy5.5	TONOBO biosciences	45-0042-82
CD8a	53–6.7	BV510	Biolegend	100 752
CD8a	53–6.7	PE	Biolegend	100 708
CD11b	M1/70	PE/Cy7	Biolegend	101 215
CD11c	N418	PerCP-Cy5.5	Biolegend	117 327
CD25	PC61.5	PE/Cy7	TONOBO biosciences	25-0251-82
CD44	NIM-R8	FITC	Biolegend	156 008
CD44	IM7	PE/Cy7	eBioscience	25-0441-82
CD45	BRA-55	BV510	eBioscience	25-0441-82
CD45.1	A20	APC	Biolegend	110 713
CD45.1	A20	PE	Biolegend	110 708
CD45.2	104	APC	Biolegend	109 814
CD45.2	QA18A15	PE	Biolegend	111 104
CD62L	MEL-14	PE/Cy7	eBioscience	25-0621-82
CD80	16–10 A1	FITC	eBioscience	11-0801-82
CD90.2	QA20B09	PerCP-Cy5.5	Biolegend	117 110
CD103	2E7	PE	Biolegend	121 406
NK1.1	PK136	PerCP-Cy5.5	BD Bioscience	553 305
NK1.1	PK136	APC	BD Bioscience	557 751
IFN-γ	XMG1.2	PE	eBioscience	12-7311-82
Granzyme B	GB11	APC	eBioscience	17-8898-82
MHC Class II	M5/114.15.2	BV450	eBioscience	M5/114.15.2
TCR		PerCP-Cy5.5	Biolegend	109 228

Cytokine and chemokine quantification

Proteins were extracted from tumor tissue homogenate by lysing in RIPA buffer. Antibody-conjugated microspheres were prepared (magEasyQPlex Mouse 7-plex Flow Assay Kit, Laizee Biotech), and standards/samples were incubated with the microspheres overnight at 4°C. Following incubation, the mixture was washed, and a biotinylated detection antibody was added for incubation at room temperature. Subsequently, the SA-PE solution was added for an additional incubation step. Finally, the microspheres were resuspended in Reading Buffer, and fluorescence signals were detected by flow cytometry (Attune nxt, ThermoFisher). Protein concentrations were quantitatively analyzed by normalizing to BCA results for standardization.

In vivo killing assay

After treatment, CFSE-labeled target cells were injected into the tail vein for an in vivo killing assay as described previously.⁵³ Briefly, spleen cells from C57BL/6J wild-type (WT) mice were pulsed with the specific peptide SIINFEKL (the immunodominant H-2K^b-restricted epitope of OVA, abbreviated as N4) and the non-specific peptide GGFNFRTL (the immunodominant H-2K^b-restricted epitope of LAMA4 protein) for 1 hour, then labeled with 5 µM or 0.5 µM CFSE, respectively, according to the manufacturer’s instructions. The labeled cells were then mixed at a 1:1 ratio, and 1×10⁷ mixed cells were transferred into MC38-OVA tumor-bearing mice via the tail vein. The following day, the percentage of labeled target cells in the spleens and TDLNs was determined by flow cytometry to assess the killing activity of CFSE-labeled target cells. The percentage of target cell killing was calculated using the formula: $\%targetcellkilling=\left[1-\left(\frac{\%{CFSE}_{high}}{\%{CFSE}_{low}}\right)\right]\times 100$.

ELISpot

The day before the experiment, MC38-OVA tumor cells were pretreated with recombinant mouse IFN-γ (30 ng/mL, Novoprotein) for 12 hours, followed by irradiation at 60 Gy. On the second day post-treatment, spleens from mice bearing the MC38-OVA tumor were collected and processed into single-cell suspensions. A total of 4×10⁵ spleen cells were added in each well of a 96-well plate (Millipore Multiscreen-HA), which had been coated with antibodies for 12 hours at 4℃, and co-cultured with 4×10⁴ tumor cells for 30 hours in a 37 ℃ cell culture incubator. The unstimulated group served as a negative control, while the ConA (MedChemExpress) stimulated group was used as a positive control. ELISpot assays were performed according to the instructions provided by BD Biosciences IFN-γ ELISPOT kit. The resulting spots were analyzed and quantified using the ImmunoSpot analyzer (Cellular Technology Limited).

Proliferation of OVA-specific CD8⁺ T cells in vivo

On day 13 of the experiment, 2×10⁶ CD45.1 OT-1 mouse spleen cells labeled with 3 µM CFSE were infused into MC38-OVA tumor-bearing mice via the tail vein. 3 days later, the CFSE fluorescence intensity of OT-1 cells was detected by flow cytometry using CD45.1 and CD8 double-positive gating to evaluate cell proliferation, and the division index was calculated using FlowJo software. On day 9 of the experiment, spleen cells from CD45.1 OT-1 mice were re-infused into MC38-OVA tumor-bearing mice. Seven days later, the proportion of OT-1 cells in the tumor, blood, spleen, and TDLNs was determined by flow cytometry. The division index is calculated following the formula: $DivisionIndex=D/N$, where $D$ refers to the total number of divisions, $N$ refers to the total number of cells at start of culture.

Drug toxicity test in vivo

At the end of the experiment, blood samples and major organs such as the lung, liver, and kidney were collected. A serum biochemical analyzer (Chemray 240/420/800, Shenzhen Leidu Life Science and Technology, China) was used to measure liver and kidney function markers, including ALT, AST, CREA, and UA. The lung, liver, and kidney tissues were embedded in paraffin and stained with hematoxylin and eosin (Servicebio). The results were analyzed using K-Viewer software.

scRNA-seq library preparation

Live tumor-infiltrating CD45⁺ cells were FACS-sorted from CisVac/α-CD137 or PBS-treated mice bearing MC38 tumors (three biological replicates per group). Single-cell separation and library preparation were performed using the BD Rhapsody microwell-based scRNA-seq platform with the BD Rhapsody Whole Transcriptome Analysis reagent kit (BD Biosciences), according to the manufacturer’s instructions. Briefly, sorted live cells were captured by random distribution across >200,000 microwells. The cells, mixed with oligonucleotide barcode beads, were loaded on the BD Rhapsody cartridge for single-cell separation. Cell-lysis buffer was added so that poly-adenylated RNA molecules were hybridized to the beads. After reverse transcription within microwells containing uniquely barcoded beads, cDNA was enriched, purified, and amplified following the manufacturer’s instructions. The libraries were quantified using a High Sensitivity DNA chip (Agilent) on a Bioanalyzer 2200 and the Qubit High Sensitivity DNA assay (Thermo Fisher Scientific). Sequencing for scRNA-seq libraries was conducted using the DNBSEQ-T7 Sequencer (MGI, Shenzhen, China) on a 150 bp paired-end run.

scRNA-seq and scTCR-seq data analysis

Raw scRNA-seq data were processed using the NovelBrain Cloud Analysis Platform (NovelBio). This involved filtering the adaptor sequence, removing the low-quality reads via fastp (V.0.21.0), aligning the reads to the prebuilt transcriptome reference (mm10/GRCm38 version 100, Ensembl) via STARsolo (V.2.7.10a). The output consisted of a digital matrix detailing gene expression per cell using unique molecular identifiers (UMIs). The downstream analysis of gene expression was conducted in R (V.4.3.3) using the Seurat (V.5.1.0) package. Initially, several filtering steps were implemented to exclude low-quality data. Data were filtered based on criteria: cells with more than 200 detected genes and genes detected in at least three cells. Additionally, cells were retained if they had less than 20% of mitochondrial genes and hemoglobin genes, more than 5% of ribosomal genes, and a ratio of log₁₀-transformed genes to that of UMIs greater than 0.8. Potential doublets were identified and removed using the DoubletFinder (V.2.0.4) package. Subsequent analysis employed Seurat functions: filtered read counts from each sample were normalized and scaled independently using the ‘NormalizeData’ and ‘ScaleData’ respectively, with regression to mitigate mitochondrial contamination and cycling effects. Principal component analysis was applied on the top 2000 highly variable genes identified by ‘FindVariableFeatures’ using the ‘vst’ method. Batch correction and integration across samples were accomplished via the Harmony (V.1.2.1) package.⁵⁴ Uniform manifold approximation and projection reduction was then computed based on the harmony dimensions. A shared nearest neighbor graph was further constructed using 30 nearest neighbors and a resolution of 1.0 for the Louvain algorithm to cluster cells. We employed two complementary strategies to assign cell type identities to each query cluster: (1) marker gene-based annotation leveraging literature-derived and scRNA-seq analysis-derived markers, identified as top-ranked of differentially expressed genes (DEGs) within each cluster; (2) CellTypist (V.1.6.3),⁵⁵ using built-in immune cell models for supervised classification based on machine learning approaches labeled scRNA-seq datasets as references. DEGs between two designated groups were identified using the Wilcoxon rank sum test with Bonferroni correction. Gene set expression scores at the single-cell level were calculated using the UCell (V.2.8) package. T cell-associated gene signatures during antitumor immunity were sourced from a prior study. Intercellular communication networks were quantitatively analyzed using CellChat v2 toolkit.⁵⁶

Statistical analysis

GraphPad Prism V.0.5 (USA) was used for statistical analysis, and the data were presented as the mean±SEM, report checklist reference ARRIVE 2.0.⁵⁷ Unless otherwise specified, comparisons between single factors were conducted using one-way analysis of variance (ANOVA), while multifactor comparisons were performed using two-way ANOVA. The survival time of tumor-bearing mice was analyzed using the Kaplan-Meier method. For the assessment of synergistic effect, a CI using the Bliss Independent Effect Model was applied. CI <1 indicates synergy, CI=1 additivity, and CI >1 antagonism.⁵⁸ A significance level of α=0.05 (p<0.05) was established, and the results were interpreted as follows: ns: not significant, *p<0.05, **p<0.01, ***p<0.001, and ****p<0.0001.

Data availability statement

Data are available on reasonable request.