The Dendritic Cell-based Vaccine PROTEXI leverages Antiviral CD4 T cell Memory to boost anti-tumor immune responses in mice

Abstract

The efficacy of dendritic cell (DC) cancer vaccines is linked to poor immunogenicity of tumor-associated antigens and failure to elicit robust MHC class II-restricted CD4⁺ T-cell responses. Here, we introduce PROTEXI, a DC vaccine platform that optimizes tumor immunity by co-presenting tumor-specific CD8⁺ T-cell epitopes alongside CD4⁺ T helper epitopes from the SARS-CoV-2 Spike protein, leveraging widespread anti-viral immunity. In preclinical mouse models of melanoma and breast cancer, PROTEXI significantly reduces tumor growth and improves survival by promoting robust T cell infiltration into immune-cold tumors, increasing cytotoxic T cell responses via epitope spreading, and activating genes linked to optimal DC, NK cell, and T cell function. Furthermore, PROTEXI elicits superior responses when combined with other immunotherapy agents in models of therapy-resistant tumors. Finally, in a humanized mouse model of melanoma, PROTEXI vaccine co-presenting CD4⁺ T-specific Spike epitopes with CD8⁺ T cell-restricted PRAME and MAGE-A3 antigens, significantly reduces tumor burden. Thus, these data underscore the potential of harnessing pre-existing viral-specific immunity to enhance the efficacy of DC vaccines in immune-cold tumors.

Clinical efficacy of dendritic cell (DC)-based cancer vaccines has been limited due to the inability to effectively engage both CD4+ and CD8 + T cell-mediated immune responses. Here, the authors present PROTEXI, a DC vaccine platform that leverages pre-existing antiviral CD4+ helper T cells to boost tumor-specific CD8+ cytotoxic T cell response, achieving a significant reduction in tumor burden and prolonged animal survival in preclinical tumor mouse models.

Introduction

Dendritic cells (DC), initially recognized by Steinman and Cohn, are pivotal antigen-presenting cells (APC) that initiate and activate antigen-specific T cell responses¹. The remarkable functional plasticity of DC makes them a desirable platform for developing immunotherapeutic strategies². ~200 clinical trials have been conducted using DC vaccines in cancer patients, demonstrating their safety and ability to elicit robust tumor-specific T cell responses^3,4. However, the clinical efficacy of DC-based cancer vaccines has been limited, with only ~15% of patients showing objective responses^5,6. During cancer vaccination and immunotherapy, dendritic cells play a pivotal role in broadening the T cell response spectrum to endogenous tumor antigens through a phenomenon known as epitope spreading. Epitope spreading, mediated by mechanisms such as the bystander effect, cross-presentation, and cross-dressing^7–9, is reported as a crucial mechanism for successful therapeutic responses to immune checkpoint blockade and is associated with improved outcomes in melanoma patients with low tumor mutational burden^10,11.

The indispensable role of tumor-specific T cells in achieving effective tumor rejection has been well established, making them a central target of modern immunotherapy. CD4⁺ T helper cells play an active role in recruiting CD8⁺ T cells to the tumor site, where they engage dendritic cells¹² and “license” them to fully activate and promote the formation of antigen-specific CD8⁺ memory T cells¹³. The functional importance of CD4⁺ T helper cells was demonstrated decades ago, through a study that revealed co-administration of a CD4⁺ T helper epitope together with a CD8⁺ T cell epitope peptide significantly prolonged the survival of mice bearing MHC II-deficient tumors¹⁴. More recently, cancer vaccination studies targeting neoantigens–newly formed antigens that arise from tumor-specific genetic mutations that are not present in normal tissues–have demonstrated that effective tumor regression can be achieved only when dendritic cells “simultaneously” present both MHC-I and MHC-II neoantigens, even when treating MHC-II-deficient tumors^15,16. Thus, including tumor-specific CD4⁺ T helper epitopes in cancer vaccine designs has been viewed as essential, regardless of the level of tumor MHC-II expression. However, identifying MHC-II-restricted tumor-specific neoantigens using computer-based prediction algorithms remains a challenge due to the unique characteristics of the MHC-II open binding pocket and the diversity of MHC-II molecules^17,18.

The concept that SARS-CoV-2 vaccination might benefit the response to cancer immunotherapy has been substantiated by recent reports examining the impact of COVID-19 vaccination on response to immune checkpoint inhibitor (ICI) therapy. Advanced lung or skin cancer patients who received a COVID-19 mRNA vaccine within 100 days of initiating ICI therapy (e.g., anti-PD1 or anti-PDL1) exhibited significantly improved overall survival compared to those who were not vaccinated¹⁹. Another study demonstrated that the COVID-19 vaccine significantly promotes antitumor immunity via tumor-irrelevant CD4⁺T and CD8⁺ T bystander cells, which are abundant and retain a functional memory phenotype against viral epitopes in the tumor microenvironment (TME)²⁰. These observations strongly support the concept that immunostimulatory epitopes derived from viral molecules may be used in immune-based cancer therapies.

To address the limitations of current dendritic cell vaccines, we develop PROTEXI, which leverages exogenous, non-self CD4⁺ T epitopes derived from the Spike protein of SARS-CoV-2 co-loaded with tumor antigen peptides onto dendritic cells, thereby bypassing the limited availability of tumor-specific MHC-II help. In this hybrid antigen strategy, Spike epitopes ensure robust CD8⁺ T cell recruitment via Spike-specific CD4⁺T cell help, while triggering CD4⁺T-mediated licensing of antigen-presenting cells. This synergistic interplay drives the development of polyfunctional cytotoxic T memory cells and broadens the T cell repertoire through epitope spreading, both of which are essential for durable antitumor immunity.

In the proof-of-concept studies, we show that PROTEXI engages with non-tumor-related CD4⁺T helpers and leads to robust antitumor efficacy against low-immunogenic or therapy-resistant tumors. Importantly, response to the PROTEXI platform is further enhanced by combination with immune modulators such as anti-PD-1 or Vactosertib, a TGF-β receptor antagonist. Moreover, feasibility testing using human PBMC-derived PROTEXI in humanized mice reconstituted with a human immune system from COVID-19-vaccinated donor blood – leveraging pre-existing CD4⁺ T-cell immunity to SARS-CoV-2 Spike epitopes – provides compelling evidence that the PROTEXI regimen enhances antigen-specific T-cell activation against human tumors. In contrast to traditional dendritic cell vaccines that present only MHC class-I restricted peptides, the viral-tumor hybrid antigen presentation strategy of PROTEXI results in a more robust, coordinated, and durable antitumor immunity, which is a significant mechanism in the treatment of immune ‘cold’ and therapy-resistant tumors.

Results

DC pulsed with non-tumor-related MHC class II-restricted peptide significantly enhanced tumor-specific T cell responses and long-term antitumor immunity

To interrogate the potential of the PROTEXI platform to induce effective tumor suppression, we prepared bone marrow-derived dendritic cells (BMDC) expressing maturation markers significantly increased on the BMDC surface which include CD80, CD86, and MHC-II and secreting inflammatory cytokines, IL-12 and TNF, essential for T cell proliferation (Supplementary Fig. 1). The MHC class II (I-A^b)-restricted OVA_323-339 peptide, used as a surrogate for CD4⁺T-restricted COVID-19 epitopes, was loaded onto mature BMDC together with MHC class I (H-2^K)-restricted tumor-associated antigens (TAA): M30-11 and/or Trp2 (Trp2_188-196)^21,22, (peptide sequences can be found in Supplementary Table 1). The M30-11 neoantigen was identified by B16F10 mutanome mapping and stimulates a CD8⁺ T response^16,21. The functional competence of PROTEXI, fully loaded with both CD4⁺ and CD8⁺T cell peptides, was demonstrated by assessing the capacity of PROTEXI to stimulate T cells, comparable to that of BMDC pulsed with either individual CD4⁺ or CD8⁺T cell-specific peptides (Supplementary Fig. 2). In addition, we confirmed that PROTEXI, consisting of peptide-pulsed DC (CD45.1⁺), retained migratory capacity and preferentially homed to the draining lymph nodes (dLN) of recipient mice (CD45.2⁺), where they were detected as CD45.1⁺CD4⁺CD11c⁺ cells, comprising ~1–3% of total lymph node leukocytes. (Supplementary Fig. 3).

To examine the functional improvement of PROTEXI compared to traditional DC vaccines, we first set up the B16F10 melanoma mouse model with treatment groups consisting of (1) control (PBS), (2) Primed OVA₃₂₃-CD4T + DC_TAA (M30-11 and Trp2), and (3) Primed OVA₃₂₃-CD4T + PROTEXI-OVA₃₂₃/TAA groups (Fig. 1a). To expand OVA₃₂₃-specific CD4⁺T helper cells in advance, recapitulating COVID-19 CD4⁺T immunity, mice were primed with OVA₃₂₃- BMDCs (DC_OVA323) at -d12 before B16F10 tumor cell injection (a.k.a. Primed OVA₃₂₃-CD4⁺T or primed-CD4⁺T). Subsequently, either the PROTEXI or DC_TAA vaccine was administered twice at 2-week intervals on d0 and d14. The PROTEXI group showed a significant reduction in tumor growth compared to either the control group or the DC_TAA group, the traditional DC vaccine strategy targeting only tumor-specific CD8⁺ T cells (Fig. 1a). The PROTEXI group also exhibited a substantial improvement in survival rate. In contrast, the DC_TAA group showed a moderate effect (Supplementary Fig. 4).

Fig. 1. PROTEXI vaccination effectively restrained the tumor growth and promoted long-term tumor suppression.

a The B16F10 experimental groups included a PBS-treated control, DC_TAA, and PROTEXI pulsed with TAA peptides (Trp2 and M30-11) and OVA₃₂₃ peptide. CD4⁺ T cells were primed with DC_OVA₃₂₃ at day −14 (n = 5/group). DC vaccines were administered at d0 and d14 (1 × 10⁶/inj. S.Q.) following the injection of B16F10 (1 × 10⁵/inj.) at d0. One-way ANOVA, Turkey’s multiple comparison test. *p = 0.0164. b The PROTEXI effect was assessed by comparing it to treatment with admixed DC_TAA + DC-OVA₃₂₃ in the presence of OVA₃₂₃-primed CD4T cells (n = 5/group). Multiple unpaired t-test. *p = 0.0027, ***p = 0.0003 (c) After adoptive transfer of OTII-CD4T cells (1 × 10⁶/inj. i.v.) at -d2, the efficacy of PROTEXI (Trp2 + OVA₃₂₃) was compared with DCs pulsed with B16F10 tumor cell lysate (n = 3-5/ group). Welch’s t-test. *p = 0.0334. d The effect of PROTEXI with or without OVA₃₂₃-primed CD4T was evaluated by comparing tumor growth in the control and PROTEXI (OVA₃₂₃ and gp70-AH1)-only group in the 4T1 breast cancer model (n = 3–5/group). Paired t-test, two-tailed, *p = 0.0155,**p = 0.0071. e Following OVA₃₂₃-CD4T priming at -d7, the PROTEXI (OVA₃₂₃+Trp2) was evaluated over DC pulsed with both the CD4-specific M379 neoantigen and Trp2 peptides. Tumor weight was measured on d16 post injection (n = 5/group). *p = 0.02, ***p < 0.001, paired t-test, two-tailed (f) TIL of the treated groups in (e) were re-stimulated ex vivo with peptides (M379, Trp2, and OVA₃₂₃) overnight, followed by CD4⁺ IFN⁺, CD8⁺IFN^+, and CD8⁺Gzmb⁺ T cell counts by Flow. One-way ANOVA, Turkey’s multiple comparison test. Left: *p < 0.0139, ****p < 0.0001, ***p < 0.0008, **p < 0.0098. Middle: **p < 0.0027, ns not significant, Right: **p < 0.0012, *p < 0.0217 (Trp2 vs Protexi), *p < 0.024 (M739+Trp2 vs Protexi) (g) In the prophylactic model, B16F10 tumor growth was measured. 5 out of 7 PROTEXI group mice maintained <200 mm³ size at d26 (n = 6–7/group). (h) Surviving mice treated with PROTEXI were re-challenged with B16F10 on d66, and monitored for one month (n = 3-4/group). Paired t-test, one-tailed. *p < 0.0412 (i) Splenocytes(SP) from PROTEXI-treated mice at d100 were adoptively transferred into naïve C57BL/6 recipients, followed by B16F10 challenge (n = 4/group). Wilcoxson test, one-tailed. *p = 0.0312. Data are presented as mean ± SEM for all graphs. The illustration was created in BioRender. Han, E. (https://BioRender.com/7c5or4s).

Next, we assessed whether PROTEXI-OVA₃₂₃/Trp2, DC co-pulsed with both CD4⁺ and CD8⁺T cell epitopes, exerts a superior antitumor effect compared to an admix of individually loaded DC vaccines (DC_Trp2 + DC_OVA323). The group receiving an admixture of DC_Trp2 and DC_OVA323 did exhibit limited tumor suppression and survival benefit compared to the PROTEXI-treated group, indicating the importance of co-presentation of CD4⁺ and CD8⁺T epitopes on antigen-presenting cells, PROTEXI (Fig. 1b, Supplementary Fig. 4). To demonstrate the key role of CD4⁺T cell engagement by PROTEXI, we designed an adoptive transfer of OTII-CD4⁺T expressing the OVA₃₂₃-specific TCR at -d2 before tumor injection. PROTEXI-OVA₃₂₃/TAA also showed a robust antitumor response in the B16F10 model, which was distinct from tumor-lysate-loaded DC vaccine models frequently examined in clinical settings (Fig. 1c). This indicates that OTII CD4⁺T cells aided PROTEXI-induced antitumor immunity similarly to priming OVA₃₂₃-CD4⁺T with DC_OVA323. A detailed analysis of the impact of OTII CD4⁺T helper cells is shown in Fig. 4.

Fig. 4. A combination of OTII-CD4⁺ T cells and PROTEXI enhanced immune rejection of B16F10 melanoma.

a Treatment groups consisted of Control (PBS), DC_TAA, DC_OVA323, and PROTEXI with OTII-CD4⁺ T cell co-administration. Vaccination (1 × 10⁶/inj. S.Q.) was conducted only once at d0, followed by monitoring tumor growth. b The average tumor growth in each group is compared over time up to day 36 (n = 5/group). One-way ANOVA, Dunnett’s multiple comparison test *p < 0.0216(Control vs DC_TAA),*p < 0.0299(Control vs DC_OVA323), **p < 0.0048(Control vs Protexi). c The tumor size of individual mouse in panel b is shown (n = 5/group). Numbers in figure shows mice with >200 mm³ size tumor. d The survival curve demonstrates the survival benefit in the OTII-CD4⁺T + PROTEXI group. Mantel-Cox test. ****p < 0^.0001 (Control vs PROTEXI), e The emergence of antigen-specific T cell clones in the spleen was demonstrated by conducting an ELISPOT (IFN-γ) assay. The indicated peptides (2 µg/ml) were added to splenocytes (5 × 10⁶/well) isolated from the vaccinated mice for IVS and maintained in culture for 2 weeks. Then, the splenocytes (3 × 10⁴/well) were subjected to the ELISPOT assay. f IFN-γ spots were counted using an ELISPOT reader. n = 2/group) (g) The levels of epitope-specific TCR⁺ cells in the IVS splenocytes were measured by flow cytometry. h The percentage of CD8⁺ TCR⁺ T cells specifically binding to tetramers loaded with Trp2 or Luc2 epitopes was compared among the treated groups. Data were generated with 3–4 replicates. One-way ANOVA, Dunnett’s ‘s multiple comparisons test. ns: not significant (i) Tumor staining was performed to compare the tumor-infiltrating lymphocytes (CD4⁺, CD8⁺, CD11c⁺ (DC), Magenta) among the groups, and the overall tumor structures were visualized with H&E staining. The scale bars in the CD4, CD8 and CD11c staining are shown at the corner (50 µm). In the H&E staining, scale bars indicate control (2.5 mm), DC_TAA (2 mm), DCOVA323 (2 mm), and PROTEXI (1 mm), respectively. The images are magnified 40x. j The tumor-infiltrating immune cells (CD4⁺ T, CD8⁺ T, CD11c, NK (NKp46⁺) cells) were counted in 10 random fields of three tumors (30 areas total) in each group (n = 3/group) using Image J software. One-way ANOVA, Tukey’s multiple comparisons test,. Data are presented as mean ± SEM for all graphs.

We further confirmed the critical role of PROTEXI in engaging CD4⁺ T helpers in the 4T1 breast cancer model, comprising the following treatment groups: control, PROTEXI-OVA₃₂₃/gp70-AH1, and Primed-CD4⁺T + PROTEXI-OVA₃₂₃/gp70-AH1. The gp70-AH1 is an H2-L^d-restricted immunopeptide derived from the murine leukemia virus (MuLV) envelope, expressed widely in murine tumors, including 4T1, but not in normal tissues²³. Compared to the PROTEXI-only groups, the Primed-CD4⁺T + PROTEXI group significantly delayed 4T1 tumor growth (Fig. 1d) and achieved over 75% survival at day 45 (Supplementary Fig. 4).

The therapeutic benefit of PROTEXI, leveraging non-tumor-related CD4⁺T helper cells, was further compared with DC co-loaded with a CD4⁺ T-specific neoantigen (M739) and a CD8⁺ T-specific Trp2 epitope. The M739 (a.k.a. M30) neoantigen, a 27-mer immunopeptide identified through mapping of non-synonymous mutations in the B16F10 melanoma genome, has been reported to elicit a robust CD4⁺ T-cell response in the B16F10 melanoma model¹⁶. All four treatment groups were equally pre-primed with OVA₃₂₃-CD4⁺ T cells at -d7 and vaccinated twice at d3 and d10 after tumor injection (Fig. 1e). Interestingly, significant tumor reduction was observed only in the PROTEXI group at d16 (Fig. 1e, bottom). The follow-up analysis with tumor-infiltrating lymphocytes (TIL) revealed that the mice in the PROTEXI group showed a significant induction of antigen-specific CD4⁺T (IFN-γ⁺) and cytotoxic CD8⁺T cells (IFN-γ⁺, Gzmb⁺), which were greater than the T cell responses in the DC_M739/Trp2 group, further signifying the superior therapeutic potential of PROTEXI. (Fig. 1f). To test whether PROTEXI can enhance immune memory formation, a prophylactic vaccination model was established by priming OVA₃₂₃-CD4T on -d21, followed by PROTEXI vaccination on -d14 and -d7 prior to B16F10 inoculation. The PROTEXI group showed robust tumor suppression, thereby presenting <200 mm³ size tumors in 5 out of 7 mice at d26, with significantly increased survival at d66 (Fig. 1g, and Supplementary Fig. 4). We validated PROTEXI-mediated immune memory formation by re-challenging the surviving mice with B16F10 melanoma at the opposite site of the original location. B16F10 growth was nearly completely controlled for a month in the mice that survived the PROTEXI group. The follow-up study with adoptive transfer of splenocytes strongly supports PROTEXI-enhanced immune recognition of the same B16F10 tumor, showing the potential to confer protection against tumor relapse (Fig. 1i). These results indicate that PROTEXI, by engaging non-tumor-related CD4⁺T helper cells, significantly enhances immune-mediated tumor suppression compared to traditional DC vaccine approaches and has potential in preventing tumor relapse by long-term immune memory formation.

PROTEXI promotes a gene expression program in the TME that is associated with robust antitumor immunity

Given the immunosuppressive nature of the tumor microenvironment (TME), we hypothesized that PROTEXI vaccination could shift the TME to a more immunopermissive state. To explore the mechanisms through which PROTEXI achieves such a conversion of the TME, we compared the transcriptional signature of B16F10 tumors in the PROTEXI group with that of the control and DC_TAA-vaccinated groups (Fig. 2a). At day 13 post-vaccination, total RNA was extracted from the tumors and subjected to differential gene expression profiling using the NanoString PanCancer Immune Profiling panel, consisting of 755 genes. The multidimensional scattered plot showed that PROTEXI exhibited distinct gene expression patterns across treatment groups (Fig. 2b) and significantly elevated a gene signature associated with antitumor immune responses, encompassing T, B, NK, and DC functions and antigen presentation (Fig. 2c, d and Supplementary Fig. 5 for the gene list). Furthermore, immune transcriptomic analysis revealed a marked increase in the diversity and abundance of CD45⁺ immune cell infiltration in the PROTEXI-treated tumors, including robust enrichment of Th1 helper cells, cytotoxic CD8⁺T lymphocytes, natural killer (NK) cells, and professional antigen-presenting cells (e.g. DC and macrophage) (Fig. 2e). Specifically, PROTEXI upregulated gene sets associated with T cell activation (e.g., MHC molecules, Cd74, Cd86, Cd40lg)^24,25, cytotoxicity (e.g., Gzma, Gzmb)²⁶, and migration (e.g., Cxcl9, Cxcl10, Ccl5, Ccl19, Ccl21a)²⁷ which showed at least a 2-8-fold increase in expression when compared to the traditional DC_TAA vaccine group (Fig. 2f). The PROTEXI group also showed strong induction of gene signatures related to the formation of effector/memory cells (e.g., Sell, Itk, Spn, tnfsf18)^28–31. Additionally, PROTEXI vaccination led to elevated gene signatures associated with regulatory T cells and exhausted T cells, suggesting the potential benefits of combining PROTEXI with either immune checkpoint blockade or with regulators of Treg development and function. In total, these transcriptional signature analyses strongly indicate that PROTEXI vaccination promotes antitumor immunity by converting the TME to an immune-permissive environment, thereby enhancing immune cell recruitment, cytotoxic T cell activation, and the formation of effector/memory cells within the TME. These immune profiling results support that the PROTEXI platform increases antitumor immunity and could serve as an advanced treatment for immune-cold tumors.

Fig. 2. PROTEXI induced robust transcriptional changes associated with antitumor immune cell function.

a Control, priming OVA₃₂₃-CD4⁺T + DC_TAA, and priming OVA₃₂₃-CD4⁺T + PROTEXI cohorts were vaccinated at d3 and d10 and subjected to total RNA preparation from the tumor at d13 post-injection. Differential gene expression profiling was conducted using the PanCancer Immune Profiling panel (NanoString) and analyzed using the nSolve and Rosalind software platforms (n = 2/group). b The multidimensional scattered (MDS) plot visualized the similarity and dissimilarity among treatment groups. c The heatmap demonstrates the differential pathway score of the indicated immune cell function. d Differential gene expression related to T-cell function (65 genes), NK cell-related gene expression (21 genes), and antigen presentation (16 genes). e Cell type profiling of the control, DC_TAA, and PROTEXI-treated groups. f The bar graphs demonstrate the transcriptional changes related to T-cell function, including activation, cytotoxicity, migration, effector/memory formation, and T-cell exhaustion. The data presented differential gene expression (PROTEXI vs DC_Trp2) that showed at least 2-fold induction and statistical significance using NanoString’s nSolve software (Fold change (Log2), p < 0.05).

Depletion of CD4⁺ T cells abolished the tumor response to PROTEXI

To demonstrate the critical role of CD4⁺ T helper cells in the induction of tumor immunity by PROTEXI, we depleted CD4⁺ T cells by administering an anti-CD4 (αCD4) antibody before priming OVA₃₂₃-CD4⁺ T cells and subsequent PROTEXI (OVA₃₂₃+ Trp2) vaccination (Fig. 3a). The αCD4 antibody injections resulted in long-lasting CD4⁺ T cell depletion, where the αCD4 + PROTEXI group exhibited only 1% CD4⁺ T cells, compared to 16% in the IgG+PROTEXI group on day 20 (Supplementary Fig. 7). While the control group showed aggressive tumor growth, the IgG+PROTEXI group displayed significantly delayed tumor growth (Fig. 3a, bottom). In contrast, the αCD4 + PROTEXI group lost its tumor-suppressive potential due to the depletion of CD4⁺ T helper cells. Additionally, αCD4 antibody-mediated CD4⁺ T cell depletion led to a compensatory increase in the CD8⁺ T cell population to 78%, whereas the IgG and control groups showed 56% and 30%, respectively (Supplementary Fig. 7).

Fig. 3. CD4⁺ T cell depletion abrogated PROTEXI-mediated tumor rejection in the B16F10 model.

a B16F10 cells (1x10⁵/inj.) were injected into three treatment groups (n = 5/group): (1) Control+PBS, (2) IgG+ priming OVA₃₂₃-CD4⁺T + PROTEXI, and (3) αCD4+priming OVA₃₂₃-CD4⁺T + PROTEXI. CD4⁺ T cell depletion was induced by injecting αCD4 antibody (250 µg/inj.) or IgG at -d10 and -d7, followed by OVA₃₂₃-CD4⁺T priming with DC_OVA323 (1 × 10⁶/inj.) at -d7. PROTEXI (1 × 10⁶/inj.) was administered at d0. Tumor sizes were measured at 3-day intervals up to d19 post-tumor injection. Paired t-test, one-tailed. *P < 0.05 b Splenic T cells from each group, after 1-week IVS, were re-challenged with mDC loaded with OVA323/Trp2 peptide (2 µg/ml) for 1 hour, followed by Golgi-stop for 4 hours. The Trp2-specific CD8⁺ T cell activity was determined by measuring IFNγ, TNF, and IL-2 production in each group. Representative dot plots of CD8⁺ T cells are shown. c OVA323-specific CD4⁺ T and Trp2-specific CD8⁺ T cells were highly activated only in the IgG-treated group. Treatment groups were compared using paired t-tests (CD4⁺T) or One-way ANOVA with Dunnett’s multiple comparison test(CD8⁺T). Data were generated with 5 replicates. *P < 0.05, **<0.01, ****p < 0.0001. d CD4⁺ and CD8⁺ T memory cells were compared by demonstrating CD44⁺ and CD62L⁺ populations using flow cytometry. Naïve memory T cells (T_M naïve, CD44^-CD62L + ), central memory T cells (T_CM, CD44⁺ CD62L⁺), effector memory T cells (T_EM, CD44⁺ CD62L^-). e Treatment groups were compared using a paired t-test, two-tailed (CD4⁺T) or a One-Way ANOVA with Dunnett’s multiple-comparison test (CD8⁺T). Data were generated with 5 replicates. *P < 0.05, **<0.01, ****p < 0.0001. Data are presented as mean ± SEM for all graphs.

Subsequently, the splenocytes were subjected to in vitro stimulation (IVS) with OVA₃₂₃ and Trp2 peptides, followed by analysis of the expansion of epitope-specific T cell subsets (Fig. 3b, c). Activation of CD4⁺ T cells by the OVA₃₂₃ epitope in the IgG+PROTEXI group significantly increased secretion of multiple cytokines (IFN-γ, TNF, and IL-2). On the other hand, Trp2-specific CD8⁺ T cells (IFN-γ⁺, TNF⁺, and IL-2⁺) cells were exclusively observed in the IgG+PROTEXI group. Notably, the depletion of CD4⁺ T cells substantially abrogated PROTEXI-mediated activation of Trp2-specific CD8⁺ T cells, reducing the response to levels comparable to the control, despite the augmented frequency of CD8⁺ T cells in the αCD4-depleted group. Thus, the activation and expansion of tumor-epitope-restricted CD8⁺ T cells in the PROTEXI group strongly depended on CD4⁺ T cells.

It is well documented that CD4⁺ T cells play a pivotal role in the formation of memory T cells, which are crucial for long-lasting tumor rejection and the prevention of tumor relapse following immunotherapy¹¹. Based on our observations in the B16F10 re-challenge model (Fig. 1g–i), we speculated that PROTEXI would enhance CD4⁺T cell-dependent CD8⁺T cell memory formation. As such, we investigated the impact of CD4⁺ T cells on CD8⁺ memory T cell formation in the PROTEXI group with and without CD4⁺ T cell depletion. Flow cytometric analysis revealed that the IgG+PROTEXI group exhibited a significantly higher proportion of CD8⁺ T effector memory cells (75% of CD8⁺CD44⁺CD62L^-), approximately 10-fold higher than in the control or CD4⁺ T-depleted group (Fig. 3d, e). In contrast, CD4⁺ T cell depletion markedly altered the PROTEXI response, resulting in a significant reduction in the number of effector memory CD8⁺ T cells (7.8% of CD8⁺CD44⁺CD62L^-), with the majority being naïve CD8⁺ T memory cells (54% of CD8⁺CD44^-CD62L⁺), similar to the control group. CD4⁺ effector memory T cells were also significantly increased in the IgG+PROTEXI group, comprising ~40% of total CD4⁺T cells (Fig. 3e, bottom). These findings clearly indicate that OVA₃₂₃-CD4⁺ T cells play a crucial role in the formation of CD4⁺/CD8⁺ effector memory T cells, thereby contributing to long-term tumor regression and survival. Furthermore, enhancement of memory T cell formation by PROTEXI is an important mechanism for long-lasting control of tumor relapse in the clinic.

Co-administration of OTII-CD4⁺ T cells potentiates PROTEXI-mediated tumor control in the B16F10 mouse model

To further validate the essential role of non-tumor-specific CD4⁺ T helper cells in the induction of tumor immunity through PROTEXI, we co-injected PROTEXI (OVA₃₂₃/M30-11/Trp2) with OTII-CD4⁺ T cells expressing the OVA₃₂₃-specific TCR in the B16F10 model³². We hypothesized that if OTII-CD4⁺ T cells could directly serve as helpers, the combination of OTII-CD4⁺ T cells and PROTEXI would enhance tumor rejection, similar to the findings in primed OVA₃₂₃-CD4T and PROTEXI studies, whereby augmenting tumor-specific cytotoxic T cell responses and immune cell infiltration (Figs. 1 and  2). As such, four treatment groups of mice bearing B16F10-Luc2 reporter cells were randomly divided into (1) Control, (2) OTII-CD4⁺T + DC_TAA (M30-11, Trp2), (3) OTII-CD4⁺T + DC_OVA323, and (4) OTII-CD4⁺T + PROTEXI-OVA₃₂₃/TAA (Fig. 4a). Consistent with previous results, the group receiving OTII-CD4⁺T + PROTEXI exhibited striking tumor suppression, with none of the treated mice developing tumors exceeding >200 mm³ (Fig. 4b, c). In addition, the OTII-CD4⁺T + PROTEXI group exhibited significantly improved survival in 100 % of mice until day 40, whereas OTII-CD4⁺T + DC_TAA and OTII-CD4⁺T + DC_OVA323 groups displayed a 40% and 15% survival rate, respectively. (Fig. 4d). This observation suggests that engaging non-tumor-specific OTII-CD4⁺ T helpers via PROTEXI is a crucial mechanism driving strong epitope-specific CD8⁺ T cell responses and survival.

Subsequently, splenocytes were subjected to a 2-week IVS with M30-11, Trp2, and Luc2 peptides. Remarkably, in the OTII-CD4⁺T + PROTEXI group, M30-11/Trp2-specific cytotoxic T cells were significantly increased (~10-fold) upon epitope re-stimulation in the IFN-γ ELISPOT assay (Fig. 4e, f). Similarly, Luc2 peptide-specific IFN-γ-secreting T cells were also substantially expanded in the OTII-CD4⁺T + PROTEXI group, indicating that PROTEXI can induce enhanced “epitope spreading” to unvaccinated Luc2 epitopes, thereby contributing to immune-mediated repression of B16F10 cells expressing the luciferase protein (Fig. 4e, f).

Consecutively, we aimed to more precisely define the emergence of Trp2/Luc2-specific TCR⁺ CD8⁺ T cell subsets in each group by using H2-K^b tetramers loaded with the respective Trp2 or Luc2 epitope (Fig. 4g, h). The tetramer assay revealed that Trp2-specific TCR⁺ CD8⁺ T-cell clones were ~2-fold higher in the PROTEXI group than in the DC_TAA or DC_OVA323 groups, consistent with the robust tumor suppression phenotype observed. The DC_OVA323 group moderately induced Trp2/Luc2-TCR⁺ T cells and partial tumor reduction, a result that may be attributed to DC properties increasing endogenous epitope spreading or non-tumor-related bystander T cell activation^9,20. Importantly, we discovered that PROTEXI efficiently activated tumor-specific CD8⁺ T cells by engaging OTII-CD4⁺ T helper cells and inducing epitope spreading to endogenous tumor antigens, resulting in significant suppression of B16F10 melanoma growth.

Histopathologic analysis revealed marked tumor regression in the OTII-CD4⁺T +  PROTEXI group, characterized by prominent necrotic areas and pale eosinophilic regions with diminished hematoxylin nuclear staining, indicative of extensive tumor cell death³³. (Fig. 4i). Furthermore, the TME of the OTII-CD4⁺ T + PROTEXI group included a significant number of tumor-infiltrating lymphocytes (TILs) compared to tumors harvested from the other vaccinated groups (Fig. 4i, j). Consistent with the PanCancer immune profiling analysis (Fig. 2), staining with the NKp46 antibody demonstrated that PROTEXI vaccination significantly enhanced the recruitment of active NK cells (Fig. 4j, and Supplementary Fig. 6). This finding is consistent with previous reports highlighting the pivotal role of NK cells in the dendritic cell – innate immune axis within the TME^34,35. Therefore, these results demonstrate that PROTEXI, when combined with OTII-CD4⁺ helper T cells, reprograms the tumor microenvironment from an immune-suppressive to an immune-permissive state by promoting TIL recruitment, ultimately driving effective immune-mediated rejection of B16F10 tumors. These results provide a great insight that PROTEXI platform will be a significantly improved DC vaccine in clinic by directly engaging well-defined, highly immunogenic viral CD4⁺ T helper cells such as SARS-CoV-2.

The combination of PROTEXI with an immune checkpoint blockade demonstrated a synergistic therapeutic effect in ICB-resistant melanoma models

The use of immune checkpoint blockade (ICB) targeting the PD-1/PD-L1 pathway has revolutionized cancer treatment by overcoming molecular barriers that prevent the expansion of tumor-reactive T cells^36,37. However, only a subset of cancer patients benefit from anti-PD1/PD-L1 therapy, and overcoming the ICB resistance remains a clinical challenge³⁸. To evaluate if PROTEXI can further enforce the immunotherapeutic potential in combination with anti-PD1, we utilized the model of anti-PD1 therapy-resistant B16F10 melanoma³⁹. The treatment groups consisted of 1) control, 2) primed-CD4⁺T + PROTEXI, 3) anti-PD1, and 4) primed-CD4⁺T+anti-PD1 + PROTEXI group, where the primed-CD4⁺T was induced by DC_OVA323 injection at -day7 (Fig. 5a). The PROTEXI and anti-PD1 combination was conducted with a therapeutic schedule after allowing tumor formation for a week. Notably, while B16F10 melanoma did not respond to anti-PD1 monotherapy as reported previously³⁹, tumor progression was consistently delayed in the primed-CD4⁺T + PROTEXI group. The combination of anti-PD1 and PROTEXI resulted in even more pronounced tumor suppression and increased survival rates (Fig. 5b, c).

Fig. 5. PROTEXI, in combination with anti-PD1 or Vactosertib, enhanced Ag-specific T cell responses in a therapy-resistant model, with a concomitant increase in epitope spreading.

a The treatment cohorts included control, PreVax+PROTEXI, anti-PD1, and PreVax+PROTEXI + anti-PD1. The anti-PD1 antibody (250 µg/kg) was injected intraperitoneally at d7, d9, and d11(n = 5–7/group). b The average tumor burden in each treatment group was monitored following PROTEXI vaccination and/or anti-PD1 treatment (n = 5–7/group). Data are mean ± SEM with P values. 2-way ANOVA, Dunnett’s multiple comparison test, *P = 0.0153, ****P < 0.0001. c The Kaplan-Meier survival curve shows the probability of survival for each cohort. Data are mean ± SEM with P values. Mantel-Cox test: *p < 0.0198(Control vs PROTEXI), ***p = 0.0005(αPD1 vs PROTEXI + αPD1), *p = 0.0435(PROTEXI vs PROTEXI + αPD1). d TIL isolated from tumors of treated groups showed the relative difference of population, number per gram tumor, of CD4T, CD4Tex (CD44⁺PD1⁺), CD4Tex (IFN-γ⁺), and CD4Tex (Gzmb⁺) after re-activation with anti-CD3/CD28 stimulation. Data were generated from 3-5 replicates. Unpaired t-test, two-tailed. Data are mean ± SEM with P values. e The number of CD8T, CD8Tex (CD44⁺PD1⁺), CD8Tex (IFN-γ⁺), and CD8Tex (Gzmb⁺) in TIL after re-activation with anti-CD3/CD28 stimulation. Data were generated from 3-5 replicates. Unpaired t-test, two-tailed. f After PreVax at -d7, B16F10 was implanted at d0, followed by PROTEXI vaccination at d7 and d14. The treatment cohorts consisted of control, PROTEXI (OVA323+Trp2), Vactosertib, and PROTEXI (OVA323+Trp2) + Vactosertib (n = 7–9/group). g The average tumor sizes (mm³) of the treated groups are shown(n = 7–9/group). 2-way ANOVA, Dunnett’s multiple comparison test, *P = 0.021, ****P < 0.0001. h The Kaplan-Meier survival curve of each group is plotted. Mantel-Cox test: *p = 0.00298 (PROTEXI vs PROTEXI+Vacto), **p = 0.0028 (Vactosertib vs PROTEXI-Vacto), ***p = 0.0004 (Control vs PROTEXI-Vacto). i, j Splenocytes from each group were subjected to 1-week IVS with the indicated peptides, followed by an ELISPOT assay with DCs pulsed with each peptide (DC:T = 1:10). The Trp2 epitope, but not Luc2 and gp70 epitopes, was used for PROTEXI vaccination. Data were generated from 3–6 replicates. Two-way ANOVA, Tukey’s multiple comparisons test: **p = 0.0026(gp70-490), ***p = 0.001(Luc2), ***p = 0.0005(gp70-621) ****p < 0.0001, ns not significant. Data are presented as mean ± SEM for all graphs.

As a means of evading immune surveillance, malignant tumors induce the differentiation of dysfunctional or exhausted T cells, leading to decreased production of effector cytokines (IFN-γ, TNF, Gzmb) and increased expression of inhibitory receptors (PD-1, CTLA-4, TIM-3, LAG-3, TIGIT)^40,41. To assess whether the PROTEXI regimen, alone or in combination with anti-PD1 therapy, can reinvigorate these exhausted T cells within the B16F10 tumor microenvironment, we analyzed intratumoral T-cell responses. Remarkably, treatment with PROTEXI combined with anti-PD1 induced a robust infiltration of both CD4⁺ and CD8⁺ T cells, including an increased frequency of CD4⁺ and CD8⁺ T cells exhibiting an exhausted memory (Tex: CD45⁺CD3⁺CD44⁺PD1⁺) phenotype^40,41 (Fig. 5d, and Supplementary Fig. 8). To evaluate their functional capacity, tumor-infiltrating T cells containing exhausted memory T cells were stimulated ex vivo with anti-CD3/CD28 antibodies. While the control group had a significantly decreased number of CD4⁺CD8⁺ Tex cells secreting IFN-γ and granzyme B (Gzmb), the PROTEXI and anti-PD1 combination significantly expanded functional IFN-γ⁺ and Gzmb⁺ CD4⁺ Tex cells, with a more modest increase observed in CD8⁺ T cells (Fig. 5d, e). Notably, the majority of exhausted T cells (Tex) fell within either the intermediate (CD44⁺PD-1⁺CXCR5⁻CD69⁻) or terminal (CD44⁺PD-1⁺CXCR5⁻CD69⁺) phenotypic subsets. These findings suggest that PROTEXI may improve the efficacy of anti-PD1 therapy by preserving the re-activation capacity of exhausted T cells, possibly through engagement of pre-primed, non-tumor-specific CD4⁺ T-cell help, thereby overcoming resistance to immune checkpoint blockade and amplifying antitumor immunity. Consequently, combining PROTEXI with anti-PD-1 substantially enhances the rejuvenation of exhausted T cells, a key mechanism for overcoming ICB resistance in cancer patients.

Inhibition of the TGF-β type I receptor with Vactosertib reconditions the tumor microenvironment to potentiate PROTEXI efficacy

The abundance of TGF-β in the TME profoundly hinders antitumor T cell responses by directly suppressing the activation of cytotoxic T cells⁴² thereby creating an immune-desert environment characterized by an increase in fibroblasts with a TGF-β response signature (F-TBRS)⁴³. As such, targeting TGF-β signaling in the TME has been a priority in developing therapeutic strategies to improve immunotherapy outcomes. Vactosertib, a type I TGF-β receptor inhibitor, is currently being evaluated in clinical trials as a single agent or combined with leading immuno-oncology drugs^44,45. Since PROTEXI vaccination significantly amplified antitumor T cell responses, we investigated the therapeutic benefit of PROTEXI combined with Vactosertib in the B16F10 therapeutic model. The treatment groups include 1) control, 2) primed-CD4⁺T + PROTEXI, 3) Vactosertib, and 4) primed-CD4⁺T+Vactosertib +PROTEXI groups (Fig. 5f). Following CD4⁺ T-cell priming with DC_OVA323, mice were inoculated with B16F10 melanoma cells on day 0, followed by PROTEXI vaccination and oral administration of Vactosertib using a 5-days-on, 2-days-off dosing schedule for two consecutive weeks. Although B16F10 melanoma exhibited resistance to Vactosertib monotherapy, its combination with PROTEXI significantly delayed tumor progression, extended overall survival, and achieved durable tumor-free survival in 25% of treated mice for over 50 days (Fig. 5g, h). The PROTEXI and Vactosertib combination group showed stronger Trp2-specific T-cell responses in the IFN-γ ELISPOT assay. In particular, consistent with the previous epitope-spreading observation, T cell responses to non-vaccinated epitopes derived from the MuLV env gp70²³ were amplified in the combination group (Fig. 5i, j). These findings strongly support conclusion that PROTEXI vaccination, in combination with Vactosertib, reshapes the TME to suppress tumor progression by enhancing T-cell activation and promoting epitope spreading to endogenous tumor antigens, thereby significantly extending survival in tumor-bearing mice. Enhanced anti-tumor immunity following PROTEXI and vactosertib treatment underscores the importance of targeting TGF-β signaling in the TME and supports the translational potential of this combination strategy.

Spike-specific CD4⁺ T helper cells potentiate tumor-specific immunity via PROTEXI in humanized NSG MHC-I/II dko mice

It is estimated that over 80% of the United States and 65% (>5.1 billion) of the worldwide population are fully vaccinated for SARS-CoV-2 Spike immunogens, and it is believed that a majority of unvaccinated individuals have developed immunity through natural infections. The SARS-CoV-2 Spike protein-derived epitope(s) are highly immunogenic and induce robust CD4⁺ T cell activation in patients, leading to CD8⁺ T reactivity and anti-Spike antibody production⁴⁶. We hypothesized that pre-existing CD4⁺ T-cell immunity to SARS-CoV-2 Spike epitopes would confer a substantial advantage for amplifying tumor antigen-specific CD8⁺ T-cell responses when co-delivered via the PROTEXI vaccination platform. Thus, to assess the feasibility of PROTEXI engaging SARS-CoV-2 Spike epitope-specific human CD4⁺ T helper cells to effectively boost CD8⁺ T cell responses to tumor antigens, we selected clinically proven CD4⁺ T-restricted Spike epitopes^47–50 and tested them in the blood of four healthy donors whose HLA-type was defined (Supplementary Table 4). All healthy donors exhibited HLA-DR/DP/DQ-dependent CD4⁺ T cell immune responses to the selected Spike epitopes, indicating strong immune recognition (Supplementary Fig. 9). However, the influenza HA epitope (H306) did not induce a robust T cell response in the tested blood samples. This may be attributed to individual donors’ lack of up-to-date influenza immunity, whether through vaccination or prior infection, during the COVID-19 pandemic. These findings suggest the potential to leverage global immunity to SARS-CoV-2 in cancer vaccines, setting the stage for the PROTEXI platform. The immune responses in the donor blood were Spike peptide dose-dependent (Fig. 6a, b), which were determined as Spike-specific CD4⁺T cell (IFN-γ⁺) responses by flow cytometry (Fig. 6c). Subsequently, we utilized humanized NSG-MHC-I/II dko mice⁵¹ infused with PBMCs from a COVID-19 vaccinated healthy donor (HLA-DR/DQ⁺ and HLA-A2⁺), followed by implantation of HLA-compatible human MeWo (HLA-A2⁺) melanoma cells⁵². Mice were then administered either 1) PBS (control), 2) DC-TAA or 3) PROTEXI-Spike/TAA, loaded with HLA-restricted Spike epitopes and HLA-A2-restricted TAAs derived from PRAME and MAGE-A3 (Fig. 6d). Human mature dendritic cells (mDC) were differentiated from PBMC-derived CD14⁺ monocytes⁵³ and characterized by high surface expression of CD40, CD80, CD86, and HLA class II, along with secretion of key proinflammatory cytokines, including IL-12 and IL-6 (Supplementary Fig. 10). These PBMC-derived mDC effectively stimulated CD4⁺ T-cell proliferation in an allogeneic mixed lymphocyte reaction (allo-MLR) assay (Supplementary Fig. 11). In the humanized mice, the PROTEXI group exhibited greater delay of MeWo (HLA-A2⁺/HLA-II⁻, and PRAME⁺⁺⁺/ MAGEA3⁺) cell growth whereas the DC-TAA group exhibited a limited effect (Fig. 6d, e). ELISPOT assays performed at necropsy on human peripheral blood from humanized mice revealed a 3 to 4-fold increase in T-cell responses to Spike epitopes in the PROTEXI-treated group compared to controls (Fig. 6f). Importantly, IFN-γ⁺ T cell responses to PRAME and MAGE-A3 epitopes were generally elevated in the PROTEXI group over control or DC-TAA groups, demonstrating that PROTEXI effectively enhanced cytotoxic T cell responses to PRAME and MAGE-A3 antigens in the presence of Spike-specific CD4⁺ T helper cells in the human immune system (Fig. 6g). Thus, PROTEXI loaded with Spike epitopes has the potential to drive robust activation of tumor-reactive cytotoxic T cells, enabling effective immune recognition of HLA-restricted melanoma within a reconstituted human immune system. In essence, the PROTEXI DC vaccine technology offers compelling clinical potential and promises significant benefits for a broad spectrum of cancer patients.

Fig. 6. PROTEXI loaded with Spike and CTA epitopes induced Ag-specific T responses and tumor reduction in NSG-MHC I/II dko mice infused with human PBMC.

PROTEXI with Spike CD4⁺ T epitopes significantly boosted TAA-specific CD8⁺ T cell response and tumor reduction in a humanized mouse model. a CD4-specific Spike epitope responses (S31, S946, S955) were examined in human peripheral blood (healthy donor #282) using IFN-γ ELISPOT assay. b The dose response to Spike peptides was plotted to assess the specificity of the CD4⁺ T cell response. Data are generated from duplicated five-concentration points. c CD4⁺ T cells were identified as the subset that produced IFN-γ in response to Spike epitopes. Data were generated from triplicates per Spike epitope. One-way ANOVA, Dunnett’s multiple comparison test. ****p < 0.0001 (d) Human immunity was reconstituted by transferring PBMC from a COVID-19 vaccinated donor to NSG MHC-I/II dko mice at -d32 (n = 4–5/group). Following MEWO melanoma injection at d0, the control, DC_TAA, and PROTEXI (Spike +TAA) groups were vaccinated on days 7 and 14, and tumor growth was monitored until day 26. Ordinary one-way ANOVA, Dunnett’s multiple comparison test. *P = 0.043 (e) The PROTEXI-vaccinated mice displayed significantly reduced tumor weights. Data were generated from 3-5 replicates. Ordinary one-way ANOVA, Sidak’s multiple comparison test. *P = 0.0126 (f) The PROTEXI vaccinated group showed significantly increased T cell responses to Spike-epitopes (S31 and S955). Data were generated from 2–3 replicates. Welch’s t-test, one-tailed. g The T cell (IFN-γ+ ELISPOT) responses specific to PRAME or MAGEA3 epitopes were not significant but tend to be increased by PROTEXI vaccination. Data were generated from triplicates per peptide. Data were generated from triplicates per peptide. Two-way ANOVA, Dunnett’s multiple comparison test. Data are presented as mean ± SEM for all graphs.

Discussion

In malignant cancers, the functions of CD4⁺ and CD8⁺ T cells are often impaired or suppressed due to the lack of tumor immunogenicity or immune-suppressive conditions in the TME. Consequently, many immune-mediated therapeutic approaches have focused on enhancing robust T cell responses, expanding the repertoire of tumor antigen-specific T cells, and converting immune-cold tumors into inflamed, ‘hot’ tumors. The success of immunotherapy heavily relies on the generation of effective and durable CD8⁺ cytotoxic T cell responses, which are greatly dependent on CD4⁺ T helper signals. However, identifying tumor-specific CD4⁺ T cell epitopes that can provide sufficient helper function is challenging due to limitations of existing screening algorithms.

A recent study by Alspach et al. demonstrated that immunotherapy-induced antitumor responses necessitate the involvement of both tumor-antigen-specific CD8⁺ and CD4⁺ T cells, even in tumors lacking MHC-II molecule expression¹⁵. The authors proposed a plausible mechanism in which antigen-presenting cells, mainly DCs in the TME, capture CD4⁺ and CD8⁺ T cell epitopes, thereby eliciting effective immune responses. This finding is consistent with other studies demonstrating the effectiveness of vaccination with co-presented CD4 and CD8 T cell epitopes by the same antigen-presenting cells^14,54. These studies provide valuable insights for designing an autologous dendritic cell-mediated vaccination strategy that presents highly immunogenic viral epitopes to CD4⁺ T helper cells and tumor-specific CD8⁺ T cell epitopes. In addition, because tumor-specific antigens, such as neoantigens and TAAs, are rare and limited in availability, it is crucial to elicit optimal CD8⁺ T cell responses for each epitope. Consequently, researchers have devoted efforts to developing universal epitopes for CD4⁺ T helper cells derived from highly immunogenic epitopes found in viral or bacterial proteins^55–58; however, this immunotherapeutic approach has not been approved yet for clinical application.

To overcome the insufficiency of CD4⁺ and CD8⁺ T-cell-mediated antitumor immunity, we designed PROTEXI, a DC vaccine platform that harnesses pre-existing CD4⁺ helper T cells specific for SARS-CoV-2 Spike epitopes – widely prevalent across diverse HLA backgrounds – to potentiate the priming of tumor-specific CD8⁺ cytotoxic T cells. This strategy would yield stronger and more durable antitumor immunity than traditional dendritic cell vaccine approaches that focus solely on cytotoxic T lymphocyte (CTL) activation, thereby addressing a critical roadblock in the effort to achieve durable immune responses in immune-cold or therapy-refractory tumors. In addition, this platform promotes epitope spreading^7–9, whereby dendritic cells capture and cross-present endogenous tumor antigens, recruit polyclonal CD8⁺ T cells via CD4⁺T helper cells¹², and broaden immune responses beyond the vaccine-delivered epitopes. Through this hybrid-antigen mechanism, PROTEXI has the potential to establish robust, diversified, and long-lasting antitumor immunity, particularly in sarcoma and other poorly immunogenic cancers.

The data presented here show that autologous DCs loaded with non-tumor-specific CD4⁺ T cell epitopes promote the expansion of tumor-specific CD8⁺ T cells. As a proof-of-principle, we showed that the PROTEXI vaccine, loaded with OVA₃₂₃ and CD8⁺ T cell epitopes, along with either upfront priming of OVA₃₂₃-CD4⁺ T cells or adoptive transfer of OTII CD4⁺T cells (to establish pre-existing OVA₃₂₃-CD4⁺T cells), substantially delayed tumor progression in both B16F10 and 4T1 tumor models. The PROTEXI group exhibited significantly greater tumor reduction and survival than the traditional DC vaccine group, which presented only tumor-specific CD8⁺ T cell epitopes. This finding aligns with a previous report showing that co-injection of influenza hemagglutinin (HA)-restricted CD4⁺ T cells with suboptimal HA-CD8⁺ T cells resulted in enhanced CD8⁺ T cell responses and tumor rejection of a malignant tumor (MHC-I⁺/MHC-II^-) with exogenous HA expression⁵⁹.

PanCancer immune profiling analysis of vaccinated tumors revealed that PROTEXI, with priming of OVA₃₂₃-CD4T, significantly influenced the expression of genes related to antitumor immune cell function compared with a traditional DC-TAA vaccine. Mechanistically, by reprogramming immune-related gene expression within the TME, the PROTEXI regimen significantly enhanced tumor repression, accompanied by diverse immune cell infiltration, including CD11c⁺ dendritic cells, CD4⁺ and CD8⁺ T cells, and NKp46⁺ natural killer cells. Depletion of CD4⁺ T cells before priming further confirmed that OVA₃₂₃-specific CD4⁺ T cells were necessary for optimal CD8⁺ T cell expansion and CD8⁺ effector memory T cell formation, which strongly supports long-term survival and tumor rejection demonstrated in the tumor re-challenge model. Moreover, PROTEXI engaged with OTII CD4⁺T helper cells not only to expand Trp2-specific TCR⁺ CD8⁺ T-cell subsets but also to broaden the antitumor T-cell repertoire–likely through epitope spreading–as evidenced by immune reactivity against unvaccinated luciferase-derived peptides and endogenous retroviral MuLV gp70 epitopes. This observed epitope spreading is mechanistically supported by the ability of PROTEXI to recruit polyclonal CD8⁺ T cells to the TME through chemokine-mediated signaling–most notably via the CCL2/CCL3–CCR5 axis–toward the site of active CD4⁺ T cell-DC interaction¹². This axis facilitates a permissive immune niche that enhances cross-priming and broadens the cytotoxic T-cell repertoire, as demonstrated by Castello et al. in their analysis of helper T cell-driven spatial orchestration of optimal CD8⁺T cell immune responses and memory formation¹². Expanding T-cell clonality and diversifying tumor-specific T-cell responses are critical determinants for durable antitumor immunity, as demonstrated by melanoma patients who remain tumor-free for over four years following therapeutic cancer vaccination¹¹. PROTEXI exhibited increased immune inhibitory genes (e.g., PD-L2 and Lag3), which is notable given prior studies showing CD4⁺ T helper signals activate CTLs and downregulate coinhibitory molecules⁵⁵. Thus, PROTEXI, combined with immune checkpoint blockade (ICB) or with reconditioning the TME to be immune-permissive, may enhance the effectiveness of tumor remission. Indeed, PROTEXI synergized with anti-PD1 in an anti-PD1-resistant melanoma model, and combining PROTEXI with Vactosertib, an antagonist of TGF-β signaling, led to a further reduction in tumor progression. These results suggest that PROTEXI monotherapy may offer therapeutic benefit in immune-cold tumors; however, its combination with anti–PD-1 therapy or pre-conditioning of the tumor microenvironment with Vactosertib may be more effective in overcoming immunotherapy resistance.

Building on the proof-of-principle studies mentioned above, we explored the potential of the PROTEXI vaccine platform in a humanized mouse model, in which the human immune system is infused into NSG-MHC-I/II dko mice. The CD4⁺ T helper epitopes in this model were comprised HLA-compatible Spike epitopes, leveraging the globally developed immunity against SARS-CoV-2. Notably, the Spike-specific CD4⁺ T cell response was preserved in healthy COVID-19 vaccinated donors and further augmented by PROTEXI in humanized NSG-MHC-I/II dko mice. These Spike-mediated CD4⁺ T helper signals largely enhanced the CD8⁺ T cell responses specific to PRAME/MAGE-3A following PROTEXI vaccination. These compelling results indicate that PROTEXI holds immense potential for immunotherapeutic success in clinical settings by harnessing the CD4 Spike antigens to trigger enforced therapeutic responses against rare TAAs/TSAs in personalized cancer vaccines.

This approach is further supported by a recent study demonstrating the potential of tumor-irrelevant SARS-CoV-2 Spike-RBD epitopes to inhibit tumor progression by inducing T cell bystander effects and therapy-induced epitope spreading, especially when combined with immune checkpoint blockade²⁰. PROTEXI, however, extends beyond classical bystander activation. By incorporating SARS-CoV-2 Spike–derived MHC-II epitopes, PROTEXI intentionally redirects robust pre-existing antiviral helper immunity to potentiate tumor-specific cytotoxic T-cell responses, broaden the T-cell repertoire via epitope-spreading, and remodel the TME toward immune activation (Fig. 4). This strategy circumvents the need for patient-specific CD4 neoantigen discovery and supports broad population coverage across HLA haplotypes—an important translational advance for future IND-enabling development and first-in-human trials.

In line with the therapeutic advancement represented by PROTEXI, a recent seminal study reported that SARS-CoV-2 mRNA vaccination unexpectedly conferred clinical benefit in cancer patients¹⁹. Mechanistically, SARS-CoV-2 mRNA vaccine stimulates anti-viral innate and adoptive immune responses and induces robust cytokine secretion, which may contribute to initial suppression of tumor progression. However, tumor cells can evade these prime responses through upregulation of PD-L1 expression. Retrospective analyses conducted by Grippin et al.¹⁹ demonstrated that patients with advanced lung or skin cancers who received COVID-19 Spike mRNA vaccines within 100 days of initiating anti-PD-1/PD-L1 therapy exhibited significantly improved overall survival compared to those who did not receive the vaccine. Interestingly, patients with resectable stage III tumors who received a pneumonia or influenza vaccine within 100 days of ICB initiation did not demonstrate a comparable improvement. These findings highlight the unique potential of SARS-CoV-2 Spike-specific immune memory, including CD4⁺ memory T cells, to synergize with immune checkpoint blockade and enhance anti-tumor immunity. Accordingly, we demonstrated that the combination of PROTEXI and anti-PD-1 therapy further amplifies anti-tumor immunity and overcomes resistance to immune checkpoint blockade by reshaping the TME toward a more immune-active state, highlighting its potential applicability across diverse cancer types and patient populations.

In conclusion, the PROTEXI cancer vaccine platform shows significant potential for broad clinical translation, particularly in patients with immune-cold tumors. PROTEXI is, to our knowledge, the unique cancer vaccine platform leveraging high-affinity CD4⁺ T-cell immunity against the SARS-CoV-2 Spike protein as a universal helper signal. This PROTEXI DC vaccine strategy reframes the use of widespread viral immunity as a powerful tool in cancer immunotherapy.

Methods

Cell line and culture

B16F10 melanoma cells expressing Luciferase were cultured in DMEM media (Thermo Fisher Scientific, #11995073) supplemented with 10% Fetal Bovine Serum (Thermo Fisher Scientific, #26-140-079), 100 U/ml penicillin, 100 µg/ml streptomycin (Thermo Fisher Scientific, #15140122), and 10 µg/ml Blasticidin (Sigma, #15205). 4T1 murine breast cancer cells were maintained in RPMI1640 (Thermo Fisher Scientific, #11875093) containing 10% FBS, 100 U/ml penicillin, and 100 µg/ml streptomycin. Cells were incubated at 37˚C in a 5% CO₂ chamber with media replenishment every 3-4 days.

Synthetic peptides

The peptides used in the study were synthesized by Peptide 2.0 and were endotoxin-free, with >90% purity. The peptides were dissolved in DMSO (Sigma, #D8418) at the indicated concentration and stored at –80 °C until use. The lists of peptides are posted in Supplementary Tables 1-3.

Mouse PROTEXI production with Bone marrow-derived dendritic cell

Bone marrow-derived monocytes were isolated from the tibia and tubercular Bone of mice. 6x10⁵ cells per well were seeded in a 6-well plate with 3 ml DC media (RPMI1640 supplemented with 10% FBS, 50 nM β-mercaptoethanol (Thermo Fisher Scientific, #21985023), 1% nonessential amino acids (NEAA, Thermo Fisher Scientific, #11140050), 10 mM HEPES(Thermo Fisher Scientific, #15630130), 1% penicillin and streptomycin) supplemented with 20 ng/ml GM-CSF (Peprotech, #315-03) and 10 ng/ml IL-4 (Peprotech, #214-14). After 3 days, the media was completely replaced with fresh media. Then, 2 days later, 3 ml fresh media was added. After 2 days, half of the media was replaced with fresh media. The next day, suspension cells were harvested and characterized for further experiments (Supplementary Fig. 12b). DC was pulsed with MHC-I and/or MHC-II-restricted synthetic peptides (2-5 µg/ml) for an hour (Supplementary Table 1) and stimulated overnight with an inflammatory molecule/cytokine cocktail for DC maturation: (20 µg/ml Poly(I:C) (InvivoGen, Cat.# tlrl-pic), 5 ng/ml TNF (Peprotech, #315-01 A), 1000 U/ml IFN-γ (Peprotech, #315-05), 100 ng/ml LPS (Sigma, #L4391)). Gating strategies used for mature DC markers (CD11c, CD80, CD86, MHC-II) are shown in Supplementary Fig. 12b.

Animal experiments

Mice were housed in a pathogen-free animal facility maintained by the Case Western Reserve University (CWRU) Animal Resource Center, at the Wolstein Research Building. Mice were kept in 12 h light/12 h dark cycle, average 20 °C room temperature, and 40–60% humidity. Experimental and control mice (6-week-old female C57BL/6 J, Jackson Laboratory, Strain # 000664) were co-housed for in vivo experiments. For melanoma tumor establishment, 2 × 10⁵ B16F10-Luc cells were injected subcutaneously into the right flank of mice. Tumor size was measured using a Vernier caliper three times a week, and tumor volume was calculated using the traditional formula (sagittal dimension (mm) X (cross dimension (mm))²/2). Animals developing tumors >2.0 cm in diameter or >20% of their pre-injection bodyweight were euthanized by CO₂ inhalation, followed by cervical dislocation.

For vaccination, matured DC (1 × 10⁶) loaded with OVA_323-339 peptide (5 µg/ml) were injected to induce OVA₃₂₃-specific CD4⁺ T cells 7 days before B16F10 inoculation. DC loaded with tumor-specific peptides (2 µg/ml) and/or OVA_323-339 (5 µg/ml) were injected subcutaneously according to the vaccination schedule. We initially compared vaccination schedules with one- and two-week intervals and found they yielded similar outcomes. Consequently, we adopted the one-week interval protocol for all subsequent experiments. For CD4⁺ T cell depletion, an anti-CD4 antibody (250 µg, BioXcell, #BE0003-1) was injected intraperitoneally into mice at -d10 with 3-day intervals before the pre-vaccination. The IgG control group was treated with an equal amount of corresponding IgG antibody (250 µg, BioXcell, #BE0090) in parallel.

For combination treatment with anti-PD1 antibody, 200 µg/mouse of αPD1(BioXcell, #RMP1-14,) was injected intraperitoneally (i.p.) at days 7, 10, and 13. The DC vaccine was administered on days 7 and 14. For Vactosertib co-treatment, Vactosertib (40 mg/kg, Medpacto)^44,45 was administered orally (p.o.) with a 5 days on/2 days off schedule for 2 weeks, starting on day 0. The DC vaccine was injected on days 7 and 14.

Our research complies with all relevant ethical regulations, as determined by the Case Western Reserve University (CWRU) under approved protocols of Institutional Animal Care and Use Committee (IACUC, #2016-0289) and the CWRU Institutional Biosafety Committee (#IBC-2019-335).

Immunohistochemistry

Fixed tissues were embedded in paraffin, sectioned (5-µm serial sections; 200 µm apart), and stained with hematoxylin and eosin (Histoserv, Inc.). For immunohistochemistry, tissue sections were stained using the following antibodies: mouse anti-NKp46 (R&D Systems, #AF2225), mouse anti-CD11c (Cell Signaling, #97585), anti-CD4 (Abcam, #ab183685), and anti-CD8a (Invitrogen, #14-0808-80), visualized with a Magenta chromogen. Images were captured with a digital microscope and analyzed using ImageJ software.

Tetramer staining

For the production of epitope-specific tetramers, MHC class I (H-2Kb) tetramers conjugated with P.E. were incubated with a peptide according to the manufacturer’s manual (QuickSwitch Quant tetramer kit, MBL International, #TB-7400-K1). After Fc blocking with CD16/32 antibody (Cytek, #70-0161), cells were incubated with a peptide-loaded tetramer, a live/dead staining dye (Thermo Fisher Scientific, #L34966), and an anti-CD8 antibody (clone KT15, Thermo Fisher Scientific, #MA5-16759) for 30 minutes at room temperature. After washing with FACS buffer (PBS with 2% FBS), the cells were fixed with a fixation buffer. Epitope-specific TCR-positive CD8⁺ T cells were evaluated by flow cytometry. Gating strategies used for Tetramer (Trp2, Luc2) are shown in Supplementary Fig. 8b.

Antigen-specific T cell in vitro stimulation

To expand antigen-specific T cells with IVS, 1 × 10⁷ splenocytes per well were cultured with 2 µg/ml peptides in 12-well plates. Initially, culture media (RPMI 1640 supplemented with 10% FBS, 50 nM β-mercaptoethanol, 10 mM HEPES, 1% penicillin/streptomycin) contained IL-7 (20 ng/mL, Pepprotech, #217-17). After 3 days, IL-2 (20 U/ml, Peprotech, #212-12) was added. Every 3 days, half of the media was replaced with fresh media containing IL-7 and IL-2. The 2-week IVS T cells were harvested and counted for further experiments. To measure intracellular cytokines produced by antigen-specific T cells, DCs were loaded with 2 µg/ml peptide and matured with 100 ng/ml LPS overnight. These DCs were co-cultured with 2-week IVS T cells (1 × 10⁵) in 96-well plate. After 1 hr, GolgiStop (0.7 ng/ml, B.D. Biosciences, #BDB555029) was added to block extracellular cytokine release for 4 hours. As a positive control, 50 ng/ml PMA(Peprotech, #1652981) and 1 µg/ml Ionomycin(Peprotech, # 5608212) were used. The cells were then stained using the intracellular cytokine staining method and analyzed by flow cytometry.

IFN-γ enzyme-linked immunospot assay

IFN-γ ELISPOT assays were performed according to the manufacturer’s instructions (Mabtech, mouse #3321-2H, human #3420-2H, 96-well plate #3654-WP-10). For the mouse ELISPOT assay, matured mouse DC were pulsed with peptide (2 µg/ml) for an hour and washed with serum-free medium. The peptide-loaded mDCs were co-cultured with either splenocytes (2.5x10⁵), lymphocytes (5x10⁴), or T cells (3x10⁴) in an anti-IFNγ antibody-coated Multiscreen 96-well plate (1:10 ratio) at 37˚C in 5% CO₂ chamber overnight.

For the human ELISPOT assay, the IVS T cells (2 × 10⁴) were co-cultured with PBMC (2 − 4 × 10⁵) pulsed with individual peptides (2 µg/ml) for 1 hour in serum-free RPMI 1604 medium at 37 °C in a CO₂ chamber overnight. PBMCs (2 × 10⁵) were incubated for 1 hour with 10, 1, 0.1, 0.01, or 0.001 µg/mL of the individual peptide for the dose-dependent response test, and then co-cultured with the IVS T cells (2 × 10⁴). Humanized mouse blood samples were collected in EDTA-coated tubes via retro-orbital bleeding. Red blood cells were lysed with ACK lysing buffer (Thermo Fisher Scientific, # A10492-01) 2 times before the ELISPOT assay. The number of IFNγ spots was analyzed using CTL-Immunospot S6 Universal Analyzer from Cellular Technology Limited.

Differential gene expression profiling analysis

RNA samples were prepared from freshly isolated tumors and analyzed for the differential gene expression with the PanCancer Immune Profiling Panel (NanoString, #XT_PGX_MmV1_CancerImm_CSO). Sample quality control was conducted according to the manufacturer’s instructions to ensure reliable results. Background levels were determined for each sample using the mean plus two standard deviations of the included negative control counts. Background-subtracted counts were normalized using geometric means of 16 housekeeping genes. Gene lists were extracted from the NanoString panel annotation for cell types and immune response categories. Gene lists with fewer than three associated genes were excluded to ensure statistical significance. Finally, a heatmap and Log2 fold-induction of genes were generated using the nCounter and Rosalind platform, which allows visualization and analysis of the gene expression data.

Humanized mouse model

To establish a humanized mouse model, freshly prepared HLA-A2-positive human PBMC (1 × 10⁷ cells/mouse) was injected into NSG-MHC-I/II double knockout mice (The Jackson Laboratory, Strain #025216) via retro-orbital injection 32 days prior to the tumor cell injection. On day 0, MeWo cell (1 × 10⁶ cells/mouse) was implanted via subcutaneous injection. Human PBMC-derived PROTEXI from the donor pulsed with spike (S31 and S955), PRAME (P100, P141, P300, P424, P434), and MAGE A3 (MA3-112) peptides and stimulated for O/N with DC maturation cocktails. The PROTEXI (2-8x10⁵ cells/mouse) were injected on day 7 and day 14, on the same side as the tumor cells were injected. Mouse blood samples were obtained by retro-orbital bleeding to confirm human blood reconstitution and peptide-specific T cell responses.

Human PROTEXI production with human PBMC-derived Dendritic cells

Deidentified healthy donor blood was provided by the CWRU Hematopoietic Biorepository and Cellular Therapy Core under an approved IRB protocol (IRB#09-90-195). All recruited volunteers provided written informed consent and signed the consent form. Human PBMCs were isolated by Ficoll gradient, and CD14-positive cells were isolated from the PBMCs using CD14 MicroBeads (Miltenyi Biotec, Cat# 130-050-201). Isolated cells are resuspended with Cell Genix DC medium (Cell Genix, #20801-0500). 6x10⁵ CD14⁺ cells plated in the 6-well tissue culture plate were cultured in the media supplemented with 1000 IU/ml GM-CSF (Peprotech, #AF-300-03) and 1000 IU/ml IL-4 (Peprotech, #AF-200-04). On day 3, 3 mL of fresh DC medium supplemented with 1000 IU/ml GM-CSF and 1000 IU/ml IL-4 is added. On Day 5 and Day 7, 50% of the medium is changed with 1000 IU/ml GM-CSF and 1000 IU/ml IL-4, including fresh DC media. Suspension cells and loosely attached cells are carefully harvested and characterized (Suppl. Fig. 12a). DC_TAA was pulsed with TAA (PRAME and MAGE-A3 peptides for DC_TAA), while PROTEXI was pulsed with Spike, PRAME, and MAGE-A3 peptides (2-5 µg/ml) for 1-2 hours (Supplementary Tables 2 and 3). And then, the DCs were treated overnight with maturation cocktail: 1000 U/ml IFN-γ (Peprotech, #AF-300-02), 5 ng/ml TNF (Peprotech, #AF-300-01A), 25 ng/ml IL-1ß (Peprotech, #AF-200-01B), and 2 µg/ml Poly (I:C) (Invivogen, # tlrl-pic-5). Harvested DC cells were characterized with surface marker expressions using flow cytometry and FlowJo software. Gating strategies used for human matured DC markers (CD14, CD11c, CD80, CD86, HLA-II) are shown in supplementary Fig. 12a.

Flow cytometry analysis

Cell type-specific markers for human PBMC-derived DC cells, human blood reconstituted in humanized mouse model using NSG-MHC-I/II double knockout mice, intracellular cytokines, and other immune cells were stained with antibodies listed in the supplementary Table 5 below. For mouse flow cytometry analysis, mouse antibodies are listed in supplementary Table 6. Fc receptors were blocked using an anti-mouse CD16/CD32 antibody (Cytek, #70-0161) for 10 mins at 4˚C. Further, all experiments included live/dead staining using Invitrogen (Cat# L34966) or eBioscience (Cat# 65-0865-18). For intracellular cytokine staining, harvested cells were stimulated with respective peptide (2 µg/ml) or PMA (50 ng/ml, Peprotech, #1652981) & ionomycin (1 μg/ml, Peprotech, #5608212) as a positive control, followed by 10 mg/ml of GolgiStop (B.D. Bioscience, # 554724) for 4 hours. Next, the cells were fixed and permeabilized (B.D. Biosciences, #554714) for cytokine staining with anti-IFNγ, TNF, and IL2 antibodies. For Treg identification, cells were fixed and permeabilized with Foxp3/Transcription factor staining kit (eBioscience, # 00-5523-00). Finally, the cells were counted and analyzed using Attune NxT (Thermo Fisher) and FlowJo software. Gating strategies used in this study are shown in Supplementary Figs. 2, 3, 8, 9, and 12.

Identification of Spike epitope-specific T cells derived from PBMC of healthy donors

The human PBMCs were maintained in RPMI-1640 medium supplemented with L-glutamine (Gibco, Cat# 11875-093), 1% nonessential amino acids (Gibco, #11140-050), 10 mM HEPES (Gibco, # 15630-080), 50 nM β-mercaptoethanol, 1 mM sodium pyruvate (Gibco, #11360-070), 1% penicillin/streptomycin (Gibco, #15140-122), and 10% heat-inactivated human serum A.B. (Gemini Bio, #100-512-100). For in vitro stimulation (IVS) of antigen-specific T cells, 1.5 ~ 2 × 10⁷ cells were cultured in 10 cm tissue culture plate with pooled peptides (2 µg/ml) in the presence of IL-7 (20 ng/ml, Peprotech, #AF-200-07). On day 3, low-dose IL-2 (2 ng/ml, Peprotech, #AF-200-02) was added. A 50% medium change and cytokine supplementation were performed every 3 days. After 1 or 2 weeks of IVS, established T cell lines were tested for antigen specificity by interferon (IFN)-γ ELISPOT assay.

Statistical analyses

Statistical analysis was performed using GraphPad Prism software version 9. Significance was determined with a paired, two-tailed Student’s t-test, and in some of the studies analyzed by the one-way ANOVA, Tukey’s, or Dunnett’s multiple comparisons test. The Gehan-Breslow-Wilcoxon or Mantel-Cox test is used for the Kaplan-Meier survival curve. In all figures, data are presented as mean ± SEM, unless stated otherwise. A p value less than 0.05 was considered to be statistically significant.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

T.P., L.L., S.K., J.L., and S. Lim designed and framed the study. J.M.K., E.H.H, and S. Lim designed the methodology. J.M.K., E.H.H, J.K.C., S.Y. and S. Lim performed the investigation. T.P., L.L., S.K., J.L., and S. Lim analyzed and interpreted the data. J.M.K., E.H.H, and S. Lim performed the statistical analysis. J.M.K., E.H.H, and S. Lim wrote the manuscript. All authors revised the manuscript and approved the submission. All authors had full access to all the data and the final responsibility to submit for publication.

Peer review

Peer review information

Nature Communications thanks Eyad Elkord, who co-reviewed with Haozhe Cui,Olivier Adotévi and Michal Lotem for their contribution to the peer review of this work. A peer review file is available.

Funding

J.L. discloses support for this research from the Jane and Lee Seidman Chair in Pediatric Cancer Innovation and the Angie Fowler Adolescent & Young Adult Cancer Institute. S.L. and all other authors disclose that this work and its publication are fully supported by Cell therapy division of Celloram Inc.

Data availability

The NanoString data generated in this study have been deposited in the NCBI Gene Expression Omnibus (GEO) database under accession code GSE329637. All data are included in the Supplementary Information or available from the authors. The raw numbers for charts and graphs are available in the Source Data file. Source data are provided with this paper.

Competing interests

T.P., L.L., S.K., and S. L. are named inventors on a pending patent application relating to PROTEXI and its use for the treatment of cancer. T.P., L.L., J.K.C., and S. Lim are employees of Celloram Inc. T.P., J.L., and S.K. serve as board members of Celloram Inc. All other authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-74891-3.