GPC3-based circular RNA vaccine suppresses hepatocellular carcinoma progression by activating adaptive immune responses

Abstract

Background and Aims:

Messenger RNA (mRNA) vaccine is a promising approach for cancer therapy. However, the development of mRNA cancer vaccines encounters several challenges, including mRNA instability, inefficient delivery systems, and potential biosafety concerns. Addressing these issues requires optimizing mRNA design, developing novel delivery vectors, and enhancing immune responses. We aimed to develop a circular RNA (circRNA)-based cancer vaccine targeting tumor-associated antigen Glypican-3 (GPC3), a promising target in hepatocellular carcinoma (HCC), to improve antigen stability and anti-tumor immune responses.

Approach and Results:

We designed a circRNA-based vaccine encoding GPC3 for HCC and evaluated the therapeutic efficacy and safety of the circGPC3 vaccine. The circGPC3 vaccine demonstrated sustained antigen production, overcoming the inherent limitations of traditional mRNA vaccines and resulting in more potent and durable anti-tumor immune responses. In addition, we incorporated a Toll-like receptor 4 (TLR4) agonist as an adjuvant to further enhance immune responses. CircGPC3 vaccine plus TLR4 agonist effectively suppressed tumor progression in HCC. We employed multiplex immunofluorescence, single-cell sequencing, spatial transcriptomics, and mass cytometry to characterize the tumor microenvironment (TME) in mice following combination therapy and to elucidate the mechanism. Mechanistically, the circGPC3 vaccine significantly enhanced immunoproteasome-mediated antigen presentation and strengthened the interaction between cDC1 and CD8⁺ T cells through the MHC-I pathway, therefore facilitating a more effective initiation of adaptive immune responses and reprogramming TME.

Conclusions:

The circGPC3 cancer vaccine demonstrated superior efficacy compared with the mRNA cancer vaccine. When further combined with a TLR4 agonist, it disrupted immune tolerance in HCC, offering a promising translational treatment strategy for HCC.

INTRODUCTION

Nowadays, diverse immunotherapy approaches have been applied to treat malignant tumors.^1,2 Therapeutic cancer vaccine, an innovative type of immunotherapy, has gained increasing interest in recent years.³ It exhibits considerable specificity, safety, and tolerability in malignancy treatment. However, stabilizing antigen production remains a key challenge, which limits the efficacy of cancer vaccines. With the success of mRNA platforms during the COVID-19 pandemic, several mRNA-based cancer vaccines have entered clinical trials.^4,5 mRNA vaccine offers a novel alternative compared with the DNA vaccine, which carries the risk of integrating into the host genome, and the peptide vaccine, which faces challenges such as HLA restriction and complex preparation.^6,7 It allows encoding of tumor antigens and/or immunomodulatory molecules, a critical advantage for inducing both adaptive and innate immune responses.⁸ Despite these benefits, the mRNA vaccine still faces challenges, including instability, the possibility of immune tolerance, and inefficient delivery systems. In addition, mRNA production via in vitro transcription (IVT) has a relatively short half-life in cells, necessitating modifications such as nucleoside substitutions, methylguanosine caps, and codon optimization to improve stability. However, these modifications significantly increase production costs without greatly improving stability. Moreover, systematic modifications (eg, n1-methylpseudouridine, m1Ψ) can induce ribosomal frameshifting during translation, as seen in patients who received COVID-19 mRNA vaccines.⁹ Thus, alternative methods that enhance RNA stability and extend protein expression duration may unlock the full potential of RNA vaccines in cancer therapy.

Circular RNA (circRNA) is a class of covalently closed non-coding RNA. Compared with mRNA, its circular structure enhances stability by protecting against exonuclease degradation, thus prolonging its persistence in vivo and improving antigen expression.^10,11 Sustained antigen expression is key to eliciting long-lasting and effective anti-tumor immune responses in cancer immunotherapy. Recent studies have shown that circRNA could mediate ribosomal initiation and protein translation through internal ribosome entry sites (IRESs) despite lacking a 5′ cap.¹² This makes circRNA a promising vehicle for delivering immunogens. The Wei team developed a circRNA-RBD vaccine platform, which offered robust protection against SARS-CoV-2 variants in mice and rhesus macaques without causing symptoms, demonstrating the safety and tolerability of circRNA-based vaccines.¹³

Identifying antigens with high immunogenicity and precise target specificity is crucial to induce potent and durable immune responses. Glypican-3 (GPC3), a member of the heparan sulfate proteoglycan family, is anchored to the cell membrane via glycosylphosphatidylinositol (GPI). GPC3 is highly expressed in hepatocellular carcinoma (HCC), making it a specific biomarker.¹⁴ The sensitivity and specificity for diagnosing HCC are 68% and 92%, respectively.¹⁵ Currently, multiple GPC3-based therapy candidates are undergoing clinical trials, including monoclonal antibodies, bispecific antibodies, and CAR-T cell therapies.^16–18 However, no GPC3-targeted drugs have been approved for clinical application worldwide.

Clinical trials have shown that GPC3-based peptide vaccine had limited anti-tumor efficacy as a monotherapy.^19,20 This was likely due to the limited antigen fragments, which thus restricted the overall immune responses. Consequently, there is an urgent necessity for improved immunotherapy approaches. GPC3-based RNA cancer vaccine offers an appealing alternative. RNA vaccine has the ability to encode the complete antigen, which includes all immune epitopes, thereby inducing more comprehensive immune responses. Furthermore, the production of RNA vaccines is more simple and more cost-effective. However, there are no RNA cancer vaccines targeting GPC3-encoded antigens.

In this study, we aimed to develop a circRNA-based cancer vaccine targeting the tumor-associated antigen (TAA) GPC3 to improve antigen stability and anti-tumor efficacy. In addition, we incorporated immune adjuvants to further enhance immune responses. We also provided insight into the underlying mechanism of the circGPC3-based cancer vaccine. Our study may provide a promising strategy to improve the efficacy of cancer immunotherapy for HCC.

METHODS

Study approval

This study was performed in accordance with the guidelines of ethical regulation for human samples under approved protocols and was approved by the Ethics Committee of the First Affiliated Hospital, Zhejiang University School of Medicine (2024-1347).

All animal studies and experiments were approved by the Animal Research Committee of the First Affiliated Hospital, Zhejiang University School of Medicine (2024-1672/2024-1700). All the mice were bred and maintained in a specific pathogen-free laboratory and used in accordance with In Vivo Experiments (ARRIVE) guidelines [developed by the National Centre for the Replacement, Refinement, and Reduction of Animals in Research (NC3Rs)].

Regents

Cholesterol, DSPC (1,2-Distearoyl-sn-glycero-3-phosphorylcholine), DMG-PEG2000 (1,2-Dimyristoyl-sn-glycero-3-methoxypolyethylene glycol 2000), and SM102 (HUO) were purchased from SINOPEG (Xiamen, China). TLR3 agonist [poly(I:C), HY-107202], TLR4 agonist (RS 09, HY-P1439), TLR5 agonist (TH1020, HY-116961), TLR7 agonist (Vesatolimod, HY-15601), TLR8 agonist (Motolimod, HY-13773), and TLR9 agonist (CPG 1826, HY-146245) were purchased from MCE (USA).

Cell lines and cell culture

The Hepa1-6 cell line was obtained from American Type Culture Collection (ATCC CRL-1830, Manassas, VA). The hepatoma-22 cell line was from China Infrastructure of Cell Line Resources (CVCL_H613, Beijing, China). They were routinely cultured in DMEM medium (Biological Industries, Israel) or RPMI-1640 medium (Biological Industries, Israel) supplemented with 10% fetal bovine serum (Biological Industries, Israel), 100 μg/mL streptomycin (Sigma-Aldrich, Germany) and 100 U/mL penicillin (Sigma-Aldrich, Germany) at 37 °C with a humidified incubator (Thermo Fisher Scientific, USA) of 5% CO₂.

ELISpot assay

The ELISpot assay was performed according to the manufacturer’s instructions. Briefly, precoated plates from Mouse IFN-γ Precoated ELISPOT kit (strips) (DAKEWE, 2210004) were activated with RPMI-1640 medium at room temperature for 5–10 minutes. Then, the medium was discarded. The adjusted cell suspension was added to the microwells, followed by the addition of stimuli. Plates were then covered with lids and incubated at 37 °C in a 5% CO₂ incubator for 16–24 hours. Following incubation, the cell suspension was carefully discarded, and ice-cold deionized water was added to induce hypotonic cell lysis. Plates were placed at 4 °C for 10 minutes. Then the wells were washed 6 times with washing buffer. Biotinylated detection antibody was added, and plates were incubated at 37 °C for 1 hour, followed by 6 washes with Wash Buffer. Streptavidin-HRP solution was then added, and plates were incubated at 37 °C for 1 hour, after which wells were washed 5 times with Wash Buffer. Freshly prepared AEC substrate solution was then added to each well, and plates were incubated at room temperature, protected from light, for 5–30 minutes to allow spot development. The chromogenic reaction was stopped by rinsing with deionized water. Spots were finally scanned and quantified using a Mabtech IRIS Analyzer.

Animal study

C57BL/6J mice, BALB/c mice, and B6-hHLA-A2.1/hB2M mice (Strain No. T064344) were purchased from Gempharmatech Co., Ltd (Nanjing, China). The male C57BL/6J mice and BALB/c mice, aged 6 weeks, were utilized in this study unless otherwise specified.

For subcutaneous tumor models, 5×10⁶ Hepa1-6 cells or 4×10⁶ hepatoma-22 cells suspended in 0.1 mL PBS were subcutaneously injected into the right flank of C57BL/6J or BALB/c mice. When tumors reached a volume of ~50–100 mm³, typically around 7 days after inoculation, this time point was designated as day 0. Mice were then randomly assigned to different groups and received RNA vaccine administration on days 0, 3, and 6. The tumor size was measured and recorded every 3 days using calipers, and the tumor volume was calculated according to the formula = (length × width²)/2. The body weight of the mice was also measured. Approximately 2–3 weeks after the injection, mice were sacrificed. Peripheral blood was collected to detect biochemical indicators as well as cell cytokines. Mouse tissues were collected for histological and flow cytometric analyses.

Additional HCC models are presented in the Supplemental Material, http://links.lww.com/HEP/K317.

Statistical analysis

Unpaired Student t tests were used to compare pairs of groups, and 2-way ANOVA was used when there were more than 2 groups. All the data are presented as the mean ± SD, unless otherwise stated. Statistical analysis was carried out using SPSS (V23, IBM Corp, USA) and GraphPad Prism software (V9.0.2, GraphPad Inc., USA). Flow cytometry data were analyzed using FlowJo.10 (TreeStar, Ashland, OR, USA) and values of p<0.05 were considered statistically significant (*p<0.05, **p<0.01, ***p<0.001).

RESULTS

GPC3 serves as a potential target for cancer vaccines in HCC

Investigating proper tumor antigen candidates is essential for designing novel RNA-based vaccines. Characterized by its genomic landscape, GPC3 could potentially be a TAA, which is highlighted by its capacity to raise the immunotherapy efficiency within HCC.^17,21 On the basis of datasets from the public library, the pan-cancer analysis (GTEx, TCGA, and GEO) of GPC3 was conducted. We identified that GPC3 was specifically upregulated in HCC (Figures 1A, B and Supplemental Figures S1A–D, http://links.lww.com/HEP/K317), making it an ideal target for cancer vaccine development.

FIGURE 1.

GPC3 could be a potential target for a cancer vaccine in HCC. (A) Based on integrated pan-cancer data from GTEx and TCGA, the radar chart revealed a distinct overexpression of GPC3 in liver cancer compared with other malignancies. (B) Single-cell RNA-seq analysis from the TISCH dataset (GSE146115 and GSE166635) revealed that GPC3 is specifically overexpressed in HCC cells. (C) Distribution of immune cell subsets in all HCC patient groups merged, colored by cell types. (D) Stacked bar plot displaying the frequency of immune cell subsets in HCC patients with high (GPC3^high, n=11) and low (GPC3^low, n=6) GPC3 expression. The protein expression of GPC3 was identified in tumor tissues through IHC staining. Patients were categorized into the GPC3^high or GPC3^low group based on the IHC staining score. (E) Boxplots showing the frequencies of the cDC (left) and CD8⁺ T cell clusters (right) in tumor samples categorized as GPC3^high and GPC3^low. (F) Expression levels of representative markers on CD8⁺ T cell clusters in GPC3^high and GPC3^low tumors. (G, H) KLRG1 expression in tumors with high and low GPC3 expression. Abbreviations: ACC, adrenocortical carcinoma; BLCA, bladder urothelial carcinoma; BRCA, breast invasive carcinoma; CESC, cervical squamous cell carcinoma and endocervical adenocarcinoma; CHOL, cholangiocarcinoma; COAD, colon adenocarcinoma; cDC, conventional dendritic cells; DLBC, lymphoid neoplasm diffuse large B-cell lymphoma; ESCA, esophageal carcinoma; GBM, glioblastoma multiforme; GPC3, Glypican-3; HNSC, head and neck squamous cell carcinoma; IHC, immunohistochemistry; ILC, innate lymphoid cells; KICH, kidney chromophobe; KIRC, kidney renal clear cell carcinoma; KIRP, kidney renal papillary cell carcinoma; LAML, acute myeloid leukemia; LGG, brain lower grade glioma; LIHC, liver hepatocellular carcinoma; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; MESO, mesothelioma; NK, natural killer cells; OV, ovarian serous cystadenocarcinoma; PAAD, pancreatic adenocarcinoma; PCPG, pheochromocytoma and paraganglioma; pDC, plasmacytoid dendritic cells; PRAD, prostate adenocarcinoma; READ, rectum adenocarcinoma; SARC, sarcoma; SKCM, skin cutaneous melanoma; STAD, stomach adenocarcinoma; TGCT, testicular germ cell tumors; THCA, thyroid carcinoma; THYM, thymoma; UCEC, uterine corpus endometrial carcinoma; UCS, uterine carcinosarcoma; UVM, uveal melanoma.

To investigate whether GPC3 expression correlates with changes in the immune microenvironment in HCC, we performed mass cytometry profiling of 17 collected tumor samples (Figure 1C and Supplemental Figure S1E, http://links.lww.com/HEP/K317). Interestingly, we found that HCC tumors with high GPC3 expression showed increased infiltration of conventional dendritic cells (cDCs), which are known to play a key role in antigen presentation and the priming of CD8⁺ T cell responses (Figures 1D, E).²² However, this enhanced cDC infiltration was not accompanied by a concomitant increase in overall CD8⁺ T cell infiltration within GPC3^high tumors (Figure 1E). Further analysis revealed that elevated GPC3 expression was associated with an inhibition in the terminally differentiated effector KLRG1⁺CD8⁺ T cells (Figures 1F–H).

Taken together, these observations associated GPC3 with distinct immunological features in the HCC microenvironment. Therefore, we further investigated GPC3 as a potential vaccine target to elucidate its role in the tumor microenvironment and to overcome the immune tolerance in HCC.

GPC3-based mRNA cancer vaccine induces anti-tumor immunity

GPC3 is a prospective candidate for mRNA vaccine development in HCC. However, whether the GPC3-based mRNA vaccine would effectively activate immune responses and enhance anti-tumor activity requires further investigation. Here, we developed an mRNA–GPC3 vaccine using SM102-based lipid nanoparticles (LNPs). The transfection efficiency of the mRNA vaccine in tumor cells was confirmed (Supplemental Figures S2A, B, http://links.lww.com/HEP/K317), followed by evaluation of the vaccine-induced immune responses (Figures 2A, C). Compared with the control group (48.0%) and the LNP group (52.3%), the mRNA–GPC3 vaccine significantly increased the proportion of CD80⁺CD86⁺ DCs (75.7%) (Figure 2B). This maturation effect was further supported by elevated interleukin-12 (IL-12) secretion (Figure 2H). These findings indicated that the mRNA–GPC3 vaccine promoted and sustained DC activation. Moreover, co-culture of mRNA–GPC3–transfected DCs with autologous CD8⁺ T cells enhanced the proliferation and activation of these T cells, along with increased tumor necrosis factor α (TNF-α) secretion (Figures 2D–G). Taken together, these observations demonstrated that mRNA–GPC3–transfected DCs could elicit both innate and adaptive immune responses, with adaptive anti-tumor immunity possibly reliant on the classical translation of the particular mRNA. To further investigate the anti-tumor effects of the mRNA–GPC3 vaccine in vivo, we employed a subcutaneous HCC mouse model. Tumor burden was notably reduced in mice following mRNA–GPC3 vaccination (Figures 2I–K and Supplemental Figure S2C, http://links.lww.com/HEP/K317). Consistently, we observed increased infiltration of cytotoxic T cells in the tumor microenvironment (TME), indicating that the mRNA vaccine effectively elicited T cell responses (Figures 2L, M and Supplemental Figure S2D, http://links.lww.com/HEP/K317). In addition, the mice tolerated mRNA–GPC3 administration well throughout the study, with no notable effects on body weight (Supplemental Figure S2E, http://links.lww.com/HEP/K317). Consequently, we proposed that GPC3 could be an effective antigen for a cancer vaccine in HCC.

FIGURE 2.

Immunogenicity assessment of mRNA vaccine effects both in vitro and in vivo. (A) A schematic diagram illustrated the in vitro evaluation of the effect of mRNA vaccine on DC maturation. (B) The percentage of CD80⁺CD86⁺ DCs in CD11c⁺MHC-II⁺ DCs in response to mRNA vaccination was assessed by flow cytometry analysis. The right panel displayed the statistically analyzed data. (C) Following a 48-hour pre-treatment with the mRNA vaccine, DCs were cocultured with autologous T cells for an additional 24 hours to evaluate T cell activation. (D) The percentage of CD44⁺CD62⁻CD8⁺ T cells in CD8⁺ T cells was assessed by flow cytometry analysis. The right panel displayed the statistically analyzed data. (E, F) Pre-treated DCs were cocultured with CFSE-labeled spleen CD8⁺ T cells from C57BL/6J mice for 3 days. The level of CFSE in CD8⁺ T cells was assessed by flow cytometry. (G, H) Detection of TNF-α and IL-12 secretion in culture medium by ELISA. (I, J) Representative image of tumors in C57BL/6J mice and statistical analyses of tumor weights (n=5). (K) Tumor growth of mice receiving the mRNA vaccine was monitored. (L) Representative immunofluorescence staining of tissue sections showed the tumor infiltrated CD8⁺ T cells after different treatments. Scale bar, 50 μm. (M) The proportion of CD8⁺ T cells in CD45⁺ cells within the tumor was assessed by flow cytometry. Abbreviations: CFSE, carboxyfluorescein succinimidyl ester; DCs, dendritic cells; ELISA, enzyme-linked immunosorbent assay; GPC3, Glypican-3; IL-12, interleukin-12; LNP, lipid nanoparticle; mRNA, messenger RNA; TNF-α, tumor necrosis factor α.

CircRNA proves superior to mRNA as a cancer vaccine candidate

The efficacy of RNA vaccines is largely influenced by antigen expression.²³ CircRNA exhibits greater stability than mRNA due to its covalently closed-loop structure.²⁴ To evaluate its therapeutic potential, we introduced both mRNA–GPC3 and circGPC3 into Hepa1-6 cells. After 48 hours, circGPC3 showed significantly higher mRNA and protein expression levels than its linear counterpart (Supplemental Figure S3A, B, http://links.lww.com/HEP/K317). Moreover, circGPC3 demonstrated greater resistance to RNase R digestion and remained more stable under actinomycin D treatment (Figures 3A, B). In vivo imaging further confirmed that circRNA outperformed mRNA in terms of translation efficiency and expression stability (Figure 3C). In addition, EGFP-labeled circGPC3 maintained higher and more prolonged expression, whereas linear GPC3-EGFP expression declined rapidly after transfection (Figure 3D). These findings suggested that circRNA might enhance the efficacy of RNA cancer vaccines by providing improved stability and prolonged expression.

FIGURE 3.

Circular RNA shows greater potential than mRNA as a cancer vaccine platform. (A, B) The abundance of circGPC3 and mRNA–GPC3 was detected by qPCR after RNase R (3 U/μg) or actinomycin D (10 μg/mL) treatment, respectively. (C) Representative images of bioluminescence intensity in mice treated with mRNA–GPC3–Luc@LNP and circGPC3–Luc@LNP for 6 hours. (D) Representative images showing EGFP expression in the Hepa1-6 cells transfected with mRNA–GPC3–EGFP@LNP and circGPC3-EGFP@LNP at different time points. Scale bar, 20 μm. (E) Flow cytometry analysis of the percentage of CD80⁺CD86⁺ DCs in CD11c⁺MHC-II⁺ DCs after treatment with mRNA–GPC3 vaccine and circGPC3 vaccine. (F) Following a 24-hour co-culture with the DCs in (E), the percentage of CD44⁺CD62⁻ T cells in CD8⁺ T cells was analyzed by flow cytometry analysis. (G) Detection of TNF-α secretion in co-culture medium in (F) by ELISA. (H) Representative image of tumors in C57BL/6J mice receiving different dosages of circGPC3 vaccine and statistical analyses of tumor weights (n=5). (I) Tumor growth of mice receiving different dosages of circGPC3 vaccine was monitored. (J) Schematic representation of the administration strategy for assessing anti-tumor efficacy of circGPC3 vaccine (10 µg per mouse) and mRNA–GPC3 vaccine (30 µg per mouse) in subcutaneous C57BL/6J mice tumor model. (K) Representative image of tumors in C57BL/6J mice receiving circGPC3 vaccine or mRNA–GPC3 vaccine and statistical analyses of tumor weights (n=5). (L) Tumor growth of mice receiving the circGPC3 vaccine or the mRNA–GPC3 vaccine was monitored. (M) The percentage of CD8⁺ T cells in the tumor in (K) was assessed by flow cytometry. (N) Representative immunofluorescence staining of tissue sections showed the tumor infiltrated CD8⁺ T cells after circGPC3 vaccination or mRNA–GPC3 vaccination. Scale bar, 50 μm. Abbreviations: circGPC3, circular GPC3; EGFP, enhanced green fluorescent protein; GPC3, Glypican-3; LNP, lipid nanoparticle; Luc, luciferase; mRNA, messenger RNA; qPCR, quantitative PCR.

Sufficient highly mature DCs are critical for effective immune activation. These DCs activate T cells by presenting antigenic peptides and secreting tumor-suppressive cytokines including IL-12 to induce robust cytotoxic responses.²⁵ To evaluate the immune responses of the circGPC3 vaccine, we next examined its impact on DC maturation and T cell activation. The proportion of CD80⁺CD86⁺ DCs in the circGPC3 group reached 83.5%, significantly higher than that in the mRNA–GPC3 group (75.1%) (Figure 3E and Supplemental Figure S3H, http://links.lww.com/HEP/K317). When these DCs were cocultured with T cells, circGPC3-induced DCs activated a greater percentage of CD44⁺CD62⁻ T cells (32.8%) compared with the mRNA group (23.3%) (Figure 3F and Supplemental Figure S3I, http://links.lww.com/HEP/K317). This enhanced T cell activation was further validated by elevated TNF-α secretion (Figure 3G). These results indicated that the circGPC3 vaccine promoted DC maturation and activated T lymphocytes.

To explore the viability of circGPC3 as an innovative cancer vaccine, we subsequently established a subcutaneous Hepa1-6 tumor model. Each mouse was administered 10 µg, 20 µg, or 30 µg of circGPC3 every 3 days for a total of 3 doses. We observed that the tumor suppression efficiency reached up to 88.5% at the 30 µg dose, compared with the 10 µg (75.6%) and 20 µg (82.0%) (Figures 3H, I and Supplemental Figure S3D, E, http://links.lww.com/HEP/K317). It indicated a clear dose–response relationship between circGPC3 vaccine dosage and tumor suppression.

To further validate the superiority of the circGPC3 vaccine, we compared its lowest dose (10 µg) with the highest dose of mRNA–GPC3 (30 µg) (Figure 3J). Both vaccines significantly suppressed tumor growth. However, circGPC3 demonstrated stronger therapeutic efficacy even at a lower dose (Figures 3K, L and Supplemental Figures S3F, G, http://links.lww.com/HEP/K317). On day 15, the tumor inhibition rate in the circGPC3 group reached 77.3%, compared with 63.6% in the mRNA–GPC3 group (Figure 3L). Moreover, circGPC3 vaccination considerably increased the proportion of CD8⁺ T cells within TME (Figure 3M), as further supported by immunofluorescence analysis (Figure 3N and Supplemental Figure S3J, http://links.lww.com/HEP/K317).

Collectively, our findings revealed that the circGPC3 vaccine elicited more potent anti-tumor responses than the mRNA–GPC3 vaccine at a lower dosage. Therefore, we selected a circRNA vaccine in our subsequent experiments to further optimize therapeutic efficacy.

Combination of the circGPC3 vaccine and Toll-like receptor 4 agonist elicits robust immune responses in mice

The immunogenicity of circRNA vaccines can be enhanced when paired with optimized adjuvants. Among them, Toll-like receptor (TLR) agonists are particularly effective adjuvants. They can rapidly upregulate pro-inflammatory cytokines and costimulatory molecules, which are critical in priming adaptive immune responses.^26–28 We evaluated several TLR agonists to identify the most suitable combination (Supplemental Figure S4A, http://links.lww.com/HEP/K317). CircGPC3 vaccine combined with different TLR agonists suppressed tumor growth in all groups, with TLR4 agonist (RS 09) showing the most pronounced effect (Supplemental Figures S4B–G, http://links.lww.com/HEP/K317). Flow cytometry further confirmed increased CD8⁺ T cell infiltration with TLR4 agonist combination therapy (Supplemental Figure S4E, http://links.lww.com/HEP/K317). Furthermore, the circGPC3 vaccine plus TLR4 agonist led to a notable change in cytokine concentrations in the serum. There was a substantial rise in the levels of inflammatory cytokines (TNF-α, IFN-γ, IL-2, and GM-CSF) and chemokines (CCL3, CCL4, CCL5, and CXCL10) (Supplemental Figure S4H, http://links.lww.com/HEP/K317). This enhanced immune activation may help improve anti-tumor efficacy. Given this, we chose a TLR4 agonist as an adjuvant for further studies.

In the subcutaneous HCC model, circGPC3 vaccine plus TLR4 agonist slowed down tumor growth with a 94% inhibition rate (Figure 4A and Supplemental Figures S5A–D, http://links.lww.com/HEP/K317). In contrast, the application of either TLR4 agonist or circGPC3 vaccination as a single treatment only led to tumor inhibition rates of 33% or 49%, respectively. Particularly, the circular EGFP encoding an irrelevant antigen failed to show such an enhanced effect. This highlighted the antigen specificity of GPC3 rather than a structural effect of circRNA. Moreover, consistent anti-tumor effects were observed in both the orthotopic HCC model and the NRAS/c-Myc–driven model across different mouse strains (Figures 4B–H and Supplemental Figure S6, http://links.lww.com/HEP/K317). We additionally evaluated the safety of the vaccine in vivo (Supplemental Figures S5E, F, http://links.lww.com/HEP/K317). There were no statistically significant differences in body weight or biochemical parameters among the treatment groups, indicating a favorable safety profile. And no severe symptoms or apparent evidence of tissue damage were observed following the injection.

FIGURE 4.

Combination with a TLR4 agonist enhances the efficacy of the circular RNA vaccine. (A) Tumor growth of mice receiving circGPC3 vaccine plus RS 09 in subcutaneous C57BL/6J mice tumor model was monitored (n=5). (B, C) Representative image of hepatic orthotopic tumors in C57BL/6J mice and statistical analyses of tumor weights (n=5). (D) The percentage of CD8⁺ T cells in the tumor in (B) was assessed by flow cytometry. (E, F) Representative image of tumors in the NRAS/c-Myc–induced HCC model in C57BL/6J mice and statistical analyses of the ratio of liver weight to body weight (n=5). (G) The percentage of CD8⁺ T cells in the tumor in (E) was assessed by flow cytometry. (H) Heatmap of serum cytokines in (E). (I, J) Representative image of hepatic orthotopic tumors in humanized HLA-A2.1 transgenic mice receiving different treatments and statistical analyses of tumor weights (n=5). (K, L) Representative IFN-γ ELISpot images and quantification of the spot count per 4 × 10⁵ splenic lymphocytes, isolated from the mice shown in (I). (M, N) Flow cytometry analysis of ELFDSLFPV-tetramer staining of tumor-infiltrating CD8⁺ T cells. Numbers indicated the percentage of tetramer⁺ cells within the gated CD8⁺ T cell population. Tumor-infiltrating lymphocytes were isolated from mice shown in (I). Abbreviations: circGPC3, circular GPC3; LNP, lipid nanoparticle; TLR, Toll-like receptor; TLR4, Toll-like receptor 4.

To further assess the immunogenicity of the circGPC3 vaccine and enhance the translational significance of our findings, we evaluated epitope-specific immune responses in humanized HLA-A2.1 transgenic mice. Consistent with prior findings, circGPC3 vaccination markedly suppressed tumor growth, whereas mice receiving circRNA encoding an irrelevant antigen exhibited rapid tumor progression (Figures 4I, J). Notably, the circGPC3 vaccination demonstrated superior anti-tumor efficacy compared with its linear RNA counterpart. To identify immunodominant epitopes, we subsequently stimulated splenocytes from vaccinated mice with a panel of truncated peptides of GPC3 in vitro. Among these, GPC3_169–177 peptide (ELFDSLFPV) elicited robust antigen-specific CD8⁺ T cell responses, as evidenced by increased IFN-γ production (Figures 4K, L). In vivo, antigen-specific circGPC3_169–177-tetramer⁺ cytotoxic T lymphocytes (CTLs) were significantly expanded in circGPC3-vaccinated mice, further confirming its potent immunogenicity (Figures 4M, N).

Taken together, the circGPC3 vaccine plus RS 09 demonstrated potent anti-tumor efficacy across multiple models. Notably, it effectively induced tumor-specific T cell responses against the ELFDSLFPV epitope. These results emphasized the potential of the circGPC3 vaccine in combination with a TLR4 agonist as a prospective candidate for cancer vaccine-based immunotherapy.

CircGPC3 vaccine combined with TLR4 agonist reprograms the immunogenic tumor microenvironment

To investigate the immune effects of the circGPC3 vaccine combined with the TLR4 agonist, we employed 36-parameter mass cytometry by time-of-flight (CyTOF) to profile the immune landscape of HCC under different treatments. We identified 10 distinct immune cell subsets and calculated the proportion of each immune cell type using the dimensionality reduction algorithm t-SNE and the clustering algorithm PhenoGraph (Figures 5A, B and Supplemental Figure S7, http://links.lww.com/HEP/K317). Combination therapy markedly increased CD8⁺ T cells, conventional dendritic cells (cDCs), and natural killer (NK) cells within the TME, while reducing the frequency of myeloid-derived suppressor cells (MDSCs) (Figure 5C).

FIGURE 5.

Immune cell populations in tumor samples are detected by CyTOF analysis. (A) t-SNE map of CD45⁺ immune cells collected from tumors in different treatment groups (n=4). (B, C) Frequency of each immune cell subset as in (A). (D) Heatmap showing the normalized expression profiles of the T cell clusters as in (A). (E) The distributions of Th1 cells, effector CD8⁺ T cells, DNT cells, effector memory CD8⁺ T cells, naïve CD8⁺ T cells, Th2 cells, Treg cells, and naïve CD4⁺ T cells were analyzed in all 4 groups. (F) Mean expression of indicated functional markers in selected T cell clusters. (G) t-SNE map of DCs as in (A) colored by PhenoGraph for 9 clusters. (H) Stacked bar plot of (G). Abbreviations: cDC, conventional dendritic cell; CyTOF, cytometry by time-of-flight; DCs, dendritic cells; G-MDSCs, granulocyte-like myeloid-derived suppressor cells; ILC cells, innate lymphoid cells; Macro cells, macrophages; Momf cells, monocyte-derived macrophages; NK cells, natural killer cells; pDC, plasmacytoid dendritic cell.

Given the crucial role of cytotoxic T cells in anti-tumor responses, we next analyzed the expression profiles of various T cell clusters (Figures 5D, E and Supplemental Figure S8, http://links.lww.com/HEP/K317). In the combination therapy group, effector CD8⁺ T cells showed elevated levels of cytotoxic molecules such as granzyme B and activation markers like CD27 and CD69 (Figure 5F). This suggested enhanced infiltration of activated effector CD8⁺ T cells. Notably, these CD8⁺ T cells also exhibited high expression of Ly6C, which was associated with a more activated and cytotoxic state (Figure 5F).²⁹

We also observed the changes in the cDC populations within the TME (Figures 5G, H and Supplemental Figure S9, http://links.lww.com/HEP/K317). Specifically, the cDC1 subset (cluster 4), which was critical for antigen presentation, was significantly increased, whereas the immunosuppressive cDC2 subset (cluster 5) showed reduced infiltration (Supplemental Figure S9B, http://links.lww.com/HEP/K317).

In summary, our CyTOF analysis demonstrated that combination therapy promoted immune infiltration, particularly of cDC1, enhanced CD8⁺ T cell activation, and suppressed immunosuppressive phenotypes in the TME. These changes collectively contributed to a more immunogenic environment and strengthened the anti-tumor immune responses.

CircGPC3 vaccine combined with TLR4 agonist enhances the interaction between cDCs and CD8⁺ T cells

To better understand the mechanism of the circGPC3 vaccine, we applied single-cell RNA sequencing (scRNA-seq) to characterize the transcriptional profiles of various cell types within the TME from tumors under different treatments in mice. We obtained 110,240 transcriptomes of single cells from 12 mice and identified 6 major cell clusters (Figures 6A–C). Further subclustering identified 7 T cell subtypes (Figures 6D, E) and 5 myeloid cell subtypes (Figures 6F, G). Ro/e scoring revealed that cDC1 was enriched in both the circGPC3 vaccine monotherapy and combination therapy group, indicating enhanced tumor antigen cross-presentation (Figure 6H). Moreover, we noticed that effector CD8⁺ T cells, proliferating T cells, and naïve T cells were most abundant in the combined therapy group. The simultaneous elevation of these 3 T cell subsets suggested a multi-phase immune response against the tumor. Naïve T cells demonstrated an adequate supply ready for activation, proliferating T cells exhibited active immune expansion, and effector CD8⁺ T cells reflected functional T cells attacking cancer cells. cDC1 bridged innate and adaptive immunity by directly activating T cells through antigen presentation. CircGPC3 vaccine presumably enhanced this process by recruiting and activating cDC1, while the adjuvant further boosted T cell responses across multiple stages. In addition, we found that the pro-cancer pathway represented by Wnt was significantly downregulated in the tumors after circGPC3 vaccination, while pathways related to apoptosis and necrosis were upregulated (Figure 6I). This was consistent with the overall tumor data we previously observed. These observations implied that the circGPC3 vaccine plus TLR4 agonist could facilitate cDC1 infiltration in HCC, thereby enhancing adaptive immune responses.

FIGURE 6.

Single-cell transcriptome map of immune cells in the tumor after circGPC3 vaccination. (A) UMAP plots of 110,240 single cells from 12 individuals, including 6 major clusters. Each cluster was shown in different colors (n=3). (B) Dot plots showing the expression of marker genes in the indicated cell clusters. The dot size represented the percentage of cells expressing the genes in each cluster. The expression intensity of markers was shown in color. (C) The proportion of 6 major cell types in different groups was shown in bar plots. (D) UMAP plots of T_NK cell subsets in (A). (E) The proportion of 7 subsets in (D) was shown in bar plots. (F) UMAP plots of myeloid cell subsets in (A). (G) The proportion of 5 subsets in (F) was shown in bar plots. (H) After subdividing the major cell types in (A), the identified subsets were projected onto all major cell types for Ro/e scoring. (I) ssGSVA analysis within the differentially expressed genes in tumors among different groups. (J) WikiPathway analysis of differentially expressed genes in cDC1 between the control group and the combined group. (K) Volcano plot of DEGs in (J). Significantly upregulated DEGs were shown in red, while downregulated DEGs were shown in blue. Abbreviations: cDC1, conventional dendritic cell 1; cDC2, conventional dendritic cell 2; CD8_Teff, effector CD8⁺ T cell; circGPC3, circular GPC3; DEGs, differentially expressed genes; GD_T, γδ T cell; GPC3, Glypican-3; MAIT, mucosal-associated invariant T cell; pDC, plasmacytoid dendritic cell; proliferation_T cell, proliferating T cell; ssGSVA, single-sample gene set variation analysis; Tn, naïve T cell; Treg, regulatory T cell; UMAP, uniform manifold approximation and projection.

WikiPathway analysis revealed that proteasome degradation, as well as the p38 MAPK signaling pathway, were significantly enriched in cDC1 after the vaccination of circGPC3 and the immune adjuvant (Figures 6J, K). Recent studies have demonstrated that the proteasome degradation pathway is essential for processing endogenous antigens into peptides presented to CD8⁺ T cells through MHC class I molecules.³⁰ The p38 MAPK pathway functions as a fundamental regulator of cellular responses to diverse stimulation.³¹ Its activation augmented the capacity of cDCs to respond to tumor stress, ensuring their crucial role in immune defense.^32,33 Consequently, the application of the vaccine may motivate cDCs to more efficiently process antigens and deliver them to T cells.

We next evaluated cDC1–CD8⁺ T cell interactions using CellChat analysis. Significant interactions were found between cDC1 and all the 3 types T cell mentioned above in combination therapy group (Figure 7A). Ligand–receptor (L–R) analysis further showed enhanced cDC1–CD8⁺ T cell interactions, especially through the upregulated MHC-I antigen presentation pathway (Supplemental Figure S10A, B, http://links.lww.com/HEP/K317). CircGPC3 vaccination also increased output signals from cDC1 and input signals to CD8⁺ T cells (Figure 7B). Consistently, we further confirmed the enrichment of CD86⁺ DCs with adjacent locations to Ki67⁺CD8⁺ T cells in the combined therapy group by multi-color immunofluorescence staining (Figure 7C). Among MHC-I pathway–related L–R pairs, H2–K1–CD8a showed higher interaction potentials in the combined group (Figure 7D). The H2–K1–CD8a axis enhances cytotoxic responses by facilitating antigen recognition and CD8⁺ T cell activation.³⁴ It plays a critical role in initiating and sustaining anti-tumor immunity within TME. Taken together, these results revealed that the circGPC3 vaccine plus TLR4 agonist significantly enhanced immunoproteasome-mediated antigen presentation and strengthened the interaction between cDC1 and CD8⁺ T cells through the MHC-I signaling pathway.

FIGURE 7.

CircGPC3 vaccine combined with TLR4 agonist enhances the interaction between cDCs and CD8⁺ T cells. (A) CellChat analysis showing the interaction between cDC1 subsets and CD8⁺ T cell subsets. (B) The input and output signal intensities of cDCs and T cells in (A). (C) Immunofluorescence images showing the interaction between DCs and CD8⁺ T cells. Scale bars, 20 μm. (D) Bubble heatmap showing the MHC-I pathway–related L–R pairs between cDC1s and CD8⁺ T cells. (E) Unbiased clustering of ST spots and identification of cell types in each cluster. (F) MHC-I–associated molecule (H2–K1) in tissue sections. (G) Representative image of tumors in NRAS/c-Myc–induced HCC model in C57BL/6J mice receiving the treatment of anti-PD-1 and/or circGPC3@LNP vaccine plus RS 09 (n=5). (H) Statistical analyses of the ratio of liver weight to body weight in (G). (I) The percentage of CD8⁺ T cells in the tumor in (G) was assessed by flow cytometry. Abbreviations: cDCs, conventional dendritic cells; cDC1, conventional dendritic cell 1; circRNA, circular RNA; LNP, lipid nanoparticle; L–R, ligand–receptor; MAIT, mucosal-associated invariant T cell; pDC, plasmacytoid dendritic cell; ST, spatial transcriptomics; TLR4, Toll-like receptor 4.

We then explored the spatial heterogeneity characteristics of the TME after combination therapy using spatial transcriptomics (ST) technology. We performed deconvolution on the marker genes of cell subpopulations obtained from scRNA-seq to assess the composition and changes of major cell types in different ST regions. The results showed that immune cell infiltration significantly increased following vaccination, and the infiltrating immune cells gradually migrated toward the center of the tumor (Figure 7E). In addition, MHC-I molecule expression was upregulated in the tumor region, further supporting the potential of tumor vaccines in promoting anti-tumor immune responses (Figure 7F and Supplemental Figure S10C, http://links.lww.com/HEP/K317).

Recent studies have shown that RNA vaccines combined with immune checkpoint inhibitors (ICIs) exert significant anti-tumor effects.^35,36 We finally studied the potential synergistic anti-tumor effect of the vaccine plus TLR4 agonist and anti-PD-1 antibody using an orthotopic HCC model (Figures 7G–I). It was observed that the monotherapy significantly inhibited tumor growth, while combination therapy exhibited superior anti-tumor therapeutic effects and further enhanced the infiltration of CD8⁺ T cells.

DISCUSSION

The mRNA vaccine is considered a promising cancer therapy. However, the inherent instability and insufficient protein expression duration of mRNA constrain the efficacy and widespread application of the vaccine. CircRNA vaccine presents a promising alternative with significant therapeutic potential in cancer therapy.³⁷ In this study, we developed a circRNA-based cancer vaccine targeting the tumor-associated antigen GPC3 to enhance antigen stability and anti-tumor immune responses. Our findings provided valuable insights into the functional validation of circRNA vaccines and their therapeutic potential in HCC.

Compared with a conventional mRNA vaccine, we discovered that circRNA showed far greater persistence in protein expression than mRNA. This enhanced stability is likely attributed to the covalently closed-loop structure of circRNA, which confers resistance to exonuclease degradation and supports sustained protein expression. Prolonged antigen exposure is critical for eliciting effective cytotoxic T cell responses.³⁸ Previous strategies to extend antigen exposure mainly involved using slow-release nanoparticles or engineered dendritic cells.^39,40 However, our research demonstrated that cyclizing linear mRNA offered a simpler and more effective alternative for sustained antigen presentation. In HCC mouse models, circRNA enabled prolonged antigen expression, which in turn led to more robust cytotoxic T cell activity than that induced by mRNA.

A pro-inflammatory immune environment is essential for potent cytotoxic T cell responses.^41–43 TLR agonists help establish this condition, thereby enhancing anti-tumor immune responses.⁴⁴ As adjuvants in various vaccine platforms, TLR agonists have demonstrated promising safety and efficacy.⁴⁵ In our study, we compared the effects of activating different TLRs on the anti-tumor efficacy of the circGPC3 vaccine and found that adding a TLR4 agonist maximally stimulated the vaccine’s therapeutic potential.

The circGPC3 vaccine had great potential to eliminate HCC in mice. Our subcutaneous HCC model demonstrated the therapeutic efficacy of the circGPC3 cancer vaccine, but the immune microenvironment in subcutaneous tissues differs from that in the liver.⁴⁶ To address this, we employed various orthotopic HCC models in different mouse strains and humanized HLA-A2.1 transgenic mice that more accurately reflect the tumor microenvironment, enhancing the precision of therapeutic vaccine evaluation. Mechanistically, the circGPC3 vaccine significantly improved immunoproteasome-mediated antigen presentation and reinforced the crosstalk between cDC1 and CD8⁺ T cells through the MHC-I signaling pathway.

Our study also indicated that the circRNA vaccine did not induce side effects or pathological impairment in vaccinated mice. These findings provided encouraging evidence for the potential advancement of circRNA-based therapeutics.

The circRNA vaccine used in our study was encapsulated in LNPs. Notably, the naked circGPC3 vaccine without LNP encapsulation also exhibited a modest tumor-suppressive effect. This finding suggested that circRNA might reduce the need for complex delivery systems and warranted further investigation. It encourages us to explore simpler, less toxic delivery materials to minimize the immunogenicity or side effects that the materials themselves might cause.

Combining the AFP-targeted vaccine with ICIs has recently shown promising results in HCC.³⁶ We subsequently evaluated the therapeutic benefit of combining our circGPC3 RNA vaccine plus TLR4 agonist and anti-PD-1 antibody. This combination resulted in markedly enhanced tumor suppression and increased T cell infiltration into the tumor microenvironment, suggesting a synergistic effect between innate immune activation and checkpoint blockade. However, the efficacy of combining RNA vaccines with other ICIs remains to be determined. Moreover, the underlying mechanisms to drive the synergistic effect require further investigation, particularly regarding the interaction between innate and adaptive immune pathways.

Personalized cancer vaccine based on tumor-specific antigen (TSA) has obtained increasing attention due to its strong specificity and potential therapeutic advantages.⁴⁷ Currently, several clinical trials focusing on personalized cancer vaccines are ongoing.^5,48 However, the high cost and complexity of personalized treatments pose significant challenges for large-scale clinical application. In this study, we have demonstrated the efficacy of a circRNA vaccine based on tumor-associated antigen GPC3 in HCC. Our encouraging results suggested a potential strategy that could be expanded to target various tumor-associated antigens, such as MAGE and NY-ESO-1, in other solid tumors and offer a universal immunotherapy approach for a wider range of malignancies. Moreover, frequently mutant genes such as TP53 and TERT could also be incorporated into the design of a polyvalent vaccine to enhance immune responses. However, more detailed comparative studies are required in the future to assess the efficacy and safety of different antigen combinations. Our study will provide theoretical and practical guidance for developing more broadly applicable cancer immunotherapies.

Supplementary Material

hep-84-672-s001.pdf ^{(8.6MB, pdf)}