XIAP inhibition augments personalized neoantigen vaccine-driven antitumor immunity in orthotopic hepatocellular carcinoma

Abstract

Background

Personalized neoantigen vaccines offer a rational immunotherapeutic strategy for hepatocellular carcinoma (HCC), but their efficacy is limited by insufficient tumor immunogenicity and an immunosuppressive tumor microenvironment. Tumor-intrinsic factors that constrain vaccine-induced immune responses in HCC are not well defined. X-linked inhibitor of apoptosis protein (XIAP) is frequently overexpressed in HCC and has been linked to immune evasion, raising the possibility that targeting XIAP may improve the effectiveness of neoantigen-based vaccination.

Methods

XIAP expression is examined in human HCC specimens using transcriptomic analysis and multiplex immunofluorescence. A personalized neoantigen vaccine (NeoVAC) is generated and evaluated in combination with Birinapant, a clinically investigated second mitochondria-derived activator of caspases (SMAC) mimetic, and IAP antagonist, in orthotopic HCC and KPC-OVA-ICC mouse models. Antitumor efficacy is assessed by bioluminescence imaging and survival analysis. Immune landscapes are characterized by flow cytometry, immunohistochemistry, multiplex immunofluorescence, and single-cell RNA sequencing. Neoantigen-specific immune responses are further assessed using functional assays.

Results

XIAP is markedly upregulated in human HCC and inversely associates with intratumoral CD8+T cell infiltration and recurrence-free survival. Birinapant treatment increases major histocompatibility complex class I (MHC-I) expression, reduces PD-L1 levels, and facilitates immune-cell infiltration, but only has modest antitumor effects as monotherapy. In contrast, combining Birinapant with NeoVAC substantially improves tumor control. A notable proportion of treated mice achieves complete tumor regression, accompanied by durable protection against tumor rechallenge and metastatic dissemination. These effects are associated with the expansion of neoantigen-specific Ly6C⁺CD8⁺ T cells exhibiting effector-memory features, strong cytotoxic capacity, and limited exhaustion, together with enhanced dendritic-cell antigen presentation and a shift of tumor-associated macrophages toward a pro-inflammatory M1-like phenotype.

Conclusions

These data identify XIAP expression as a clinical correlate of an immune-refractory HCC phenotype and show that pharmacological IAP antagonism with Birinapant enhances tumor immunogenicity and supports antigen-specific T-cell immunity. The findings provide a preclinical rationale for combining the SMAC mimetic Birinapant with neoantigen vaccine-based immunotherapy in HCC.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Hepatocellular carcinoma (HCC) has an immunosuppressive tumor microenvironment that limits the efficacy of current immunotherapies. Personalized neoantigen vaccines can induce tumor-specific T-cell responses, but their therapeutic activity may be restricted by tumor-intrinsic mechanisms that reduce immune recognition and persistence of antitumor responses.

WHAT THIS STUDY ADDS

• This study shows that the second mitochondria-derived activator of caspases mimetic Birinapant enhances the antitumor activity of personalized neoantigen vaccination in orthotopic HCC models. The combination promotes tumor immunogenicity, increases antigen presentation, enhances tumor-infiltrating T-cell responses, and expands neoantigen-specific Ly6C⁺CD8⁺ T cells with effector-memory features. BiriNeo treatment also provides durable protection against tumor recurrence and metastasis in preclinical models.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• These findings support further evaluation of inhibitor of apoptosis protein inhibition combined with personalized neoantigen vaccination as a potential strategy to improve antitumor immunity in HCC. This approach may provide a rationale for developing combination immunotherapies for tumors with limited responses to current immunotherapeutic approaches.

Introduction

Neoantigen vaccines, designed based on immunogenic peptides arising from tumor-specific genetic variation, represent a highly promising avenue for personalized cancer immunotherapy. These vaccines, including peptide, mRNA, and DNA-based neoantigen vaccines, are uniquely capable of eliciting robust, tumor-specific immune responses without violating central tolerance, offering the prospect of highly selective and effective anticancer activity.^{1 2} Neoantigen-specific T cells activated by these vaccines can eliminate tumor cells not only through direct recognition and cytotoxic killing but also via bystander mechanisms, including the release of cytotoxic mediators (TNF-α) and activation of innate immune effectors within the tumor microenvironment (TME). Notably, an early-phase study in advanced hepatocellular carcinoma (HCC) showed that a personalized DNA neoantigen vaccine co-administered with plasmid IL-12 by electroporation was feasible and elicited neoantigen-specific T-cell responses.³ However, despite their theoretical advantages, the clinical efficacy of neoantigen vaccines in HCC remains unsatisfactory.^{4 5} This is largely attributable to the intrinsically high heterogeneity and immune evasiveness of HCC, which manifests as a profoundly immunosuppressive TME and multiple sophisticated resistance mechanisms.⁶ Tumorous immune evasion is driven by a complex interplay of factors: tumor cells frequently evade neoantigen-specific T cell recognition and killing through downregulation or loss of MHC class I expression, rendering them invisible to CD8⁺ T cells; simultaneously, they upregulate anti-apoptotic pathways to resist cytotoxic assault, and overexpress inhibitory immune checkpoints such as PD-L1 to dampen T cell activation and function.⁷ These multifaceted escape mechanisms collectively hinder the infiltration, persistence, and cytotoxic function of neoantigen-specific T cells, ultimately limiting the therapeutic impact of vaccine monotherapy. Even combinations with immune checkpoint blockade (such as PD-1 or PD-L1 inhibitors) have yielded limited improvements, with objective response rates in HCC still unsatisfactory, as they address only one layer of the complex immune evasion network.³ Therefore, innovative combination strategies are urgently needed to boost neoantigen vaccine efficacy by simultaneously targeting multiple escape pathways and transforming immunologically “cold” HCC tumors into “hot” tumors with robust antigen presentation, T-cell infiltration, and sustained cytotoxic activity.

X-linked inhibitor of apoptosis protein (XIAP), the most potent member of the inhibitor of apoptosis protein (IAP) family, orchestrates tumor immune evasion through multiple mechanisms: it directly suppresses apoptosis by inhibiting caspase-3/7/9 activity, enabling tumor cell survival under cytotoxic assault; simultaneously, it promotes immunosuppression by facilitating cIAP1/2-mediated NF-κB activation, which drives the secretion of immunosuppressive cytokines including IL-6 and IL-10.^{8 9} Recent breakthrough studies have revealed that XIAP inhibition can significantly upregulate MHC class I expression on tumor cell surfaces, thereby enhancing antigen presentation and rendering tumor cells more vulnerable to T cell-mediated cytotoxicity.¹⁰ Moreover, XIAP targeting has been shown to reshape the tumor immunopeptidome, improve neoantigen presentation, and increase tumor immunogenicity.¹¹ Second mitochondria-derived activator of caspases (SMAC) mimetics, such as Birinapant, which antagonize XIAP and other IAP family members, have demonstrated remarkable potential in remodeling the TME.^12–14 Beyond inducing tumor cell apoptosis, these agents can enhance bystander killing effects by promoting the release of damage-associated molecular patterns and activating innate immune responses.¹⁵ Clinical evidence suggests that high XIAP expression correlates with immunosuppressive cell infiltration and poor immunotherapy responses, underscoring its role as a critical immune checkpoint.^{16 17} Importantly, recent preclinical studies have shown that SMAC mimetics can synergize with immune checkpoint inhibitors, targeted agents, chemotherapeutic drugs, or adoptive T cell therapies (such as CAR-T cells), significantly improving antitumor efficacy compared with monotherapy approaches.^18–24 To address these multilayered immune escape mechanisms, targeting XIAP emerges as a particularly promising strategy.^25–28 Given the complementary effects of neoantigen vaccines in activating tumor-specific immunity and SMAC mimetic-mediated IAP antagonism in lowering tumor-cell resistance to immune killing, we hypothesized that combining NeoVAC with Birinapant could overcome key limitations of vaccine monotherapy in HCC.

To test this hypothesis, we conducted a comprehensive preclinical investigation combining NeoVAC with the SMAC mimetic Birinapant in HCC. We first analyzed the correlation between XIAP expression and CD8⁺ T-cell infiltration in human HCC specimens, then evaluated the therapeutic efficacy of this combination strategy in an orthotopic murine HCC model. Using single-cell RNA sequencing and multiplex immunofluorescence histochemistry, we systematically characterized the immune remodeling and antitumor responses following combination therapy. Our findings provide the first preclinical evidence that this novel combination can effectively enhance neoantigen-specific T-cell responses and significantly improve antitumor efficacy, establishing a strong foundation for clinical translation in HCC immunotherapy.

Materials and methods

Human specimens

HCC tissues and paired para-tumor tissues were obtained from the bio-bank of Mengchao Hepatobiliary Hospital, Fujian Medical University. All enrolled patients were pathologically diagnosed with HCC and had not received any prior relevant treatment before surgery.

Mice and cell lines

The C57BL/6 male mice (6–8 weeks old) were purchased from SLAC Laboratory Animal (China) and raised under specific pathogen-free conditions. The Hepa1-6 cell line was kindly provided by Professor Hongyang Wang’s laboratory at the Eastern Hepatobiliary Surgery Hospital, Naval Medical University, Shanghai, China, in February 2022 (RRID: CVCL_0327, ATCC). To establish a stable cell line expressing luciferase (Hepa1-6-Luc), the Hepa1-6 cells were infected with luciferase-expressing lentivirus following standard viral infection protocols. Stable transfectants were selected using puromycin, and the expression of luciferase was confirmed by fluorescence microscopy. The resulting Hepa1-6-Luc cell line was used for in vivo imaging and tumorigenesis studies. KPC-ICC cells were purchased from Quicell Biology (China). KPC-ICC is a genetically engineered mouse intrahepatic cholangiocarcinoma cell line with liver-specific Cre recombinase expression, Kras_G12D overexpression, and PTEN deletion. This cell line exhibits stable in vitro expansion and robust in vivo tumor-forming capacity. Then, KPC-ICC cells were infected with chicken ovalbumin antigen OVA_{250-264 (SGLEQLESIINFEKL)} lentivirus following standard viral infection protocols. Stable transfectants were selected using puromycin. All cells were cultured in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (Excell) and 1% penicillin/streptomycin (Gibco), and maintained at 37°C in a 5% CO₂ atmosphere. All cell lines used in this study were routinely tested for mycoplasma contamination using a commercial Mycoplasma PCR Detection Kit (Beyotime, C0301S) and confirmed to be mycoplasma-free before experiments.Cells were cryopreserved using CELLSAVING Cell Freezing Medium (NCM Biotech, C40050).

Detailed procedures for additional experimental methods, including in vivo tumor models, cell viability, flow cytometry, western blotting, and immunohistochemical analyses, are provided in the Supplementary Methods(online supplemental file 1).

Tissue microarrays construction

Formalin-fixed, paraffin-embedded tissue blocks from 93 patients with HCC, pathologically confirmed at our institution, were collected for the construction of tissue microarrays (TMAs). The inclusion criteria were as follows: (1) age between 18 and 75 years, (2) China Liver Cancer Clinical Staging Ia to IIa, (3) treatment-naïve patients with no prior chemotherapy or locoregional therapy, (4) R0 resection, and (5) availability of complete follow-up data. Patients with concurrent intrahepatic cholangiocarcinoma were excluded.

Multicolor immunofluorescence staining

Paraffin-embedded tissues were preheated at 60°C for 10 min to soften the paraffin for core sampling. Target tissue regions were precisely extracted using a TMA punch needle. The tissue cores were gently transferred to the corresponding holes in the recipient block, with sample identifiers recorded to ensure alignment with patient information. After completing the array, the recipient block was heated at 60°C for 10 min and immediately cooled at 4°C for 10 min. This thermal cycling process was repeated three times to enhance the stability of the TMA by ensuring tight integration between tissues and paraffin. Sample identifiers, patient information, and array coordinates were cross-verified to guarantee accuracy. The TMA block was sectioned into 4 µm slices using a microtome, preserving tissue integrity for subsequent analyses.

Multicolor immunofluorescence was performed according to the Opal 6-Plex Detection Kit (Akoya Biosciences, NEL821001KT). After baking slides at 65°C for 1 hour, tissue sections were sequentially immersed in xylene, absolute ethanol, 95% ethanol, 85% ethanol, 75% ethanol, and 55% ethanol, followed by rinsing with distilled water. Antigen retrieval was performed by incubating sections in EDTA-based antigen retrieval buffer under pressurized steam (boiling for 3.5 min after achieving stable steam venting in a pressure cooker). After cooling to room temperature, slides were washed three times (5 min each) with phosphate-buffered saline (PBS) on a decolorizing shaker. Sections were treated with 3% hydrogen peroxide (H₂O₂) for 15 min at room temperature in the dark, followed by three PBS washes. Normal goat serum was applied for 30 min at room temperature to block nonspecific binding. Primary antibodies were added, and slides were incubated overnight at 4°C in a light-protected humidified chamber. After washing, HRP-conjugated goat anti-mouse/rabbit universal secondary antibodies (species-matched to primary antibodies) were applied and incubated for 10–20 min at room temperature in the dark, followed by three PBS washes. Ready-to-use TYR fluorescent dye (200 µL) was directly applied for 10 min at room temperature, and sections were washed three times with PBS. For subsequent cycles of staining, the above steps were repeated. Finally, slides were washed three times with PBS, counterstained with DAPI in the dark for 10 min at room temperature, rinsed, and mounted with an anti-fade medium. Immune-stained slide images were acquired and processed using the PhenoCycler-Fusion 2.0 platform (Akoya Biosciences, USA). Subsequently, the phenoptrReports R package (downloaded from Akoya Biosciences’ official website) was used to perform statistical analysis on the multiplex imaging data. The primary antibodies for multiplex immunofluorescence were used as follows: CD4 (RRID:AB_2941893, Abcam, ab288724), CD8 (RRID:AB_3107047, Abcam, ab316778), XIAP (RRID:AB_2215009, proteintech, 10037–1 Ig), PD-L1 (RRID:AB_2687655, CST, 13684), GZMB (RRID:AB_2860567, Abcam, ab255598). TUNEL (Roche, 12156792910).

Neoantigen selection and vaccine formulation

The identification and immunogenicity validation of tumor neoantigens were performed as previously described in our published work. Briefly, candidate neoantigens were identified through integrated genomic and transcriptomic analyses of Hepa1-6 tumor cells, followed by prediction of MHC-binding potential and prioritization of candidate mutant peptides. The immunogenicity of selected peptides was evaluated by IFN-γ ELISPOT following peptide stimulation of splenic T cells, as described previously.²⁹ In this study, seven identified neoantigen peptides (4 mg each) were individually dissolved in 50 μL dimethyl sulfoxide (DMSO), vortexed until fully dissolved to prepare stock solutions at 80 µg/µL(online supplemental file 1, table S1). These seven stock solutions were pooled into a single 350 μL mixture. For each mouse, a therapeutic dose of 140 μg personalized neoantigen vaccine (NeoVAC, total neoantigen peptide mixture) was combined with 50 μL Toll-like receptor 3 agonist polyinosinic:polycytidylic acid(Poly (I: C)) adjuvant (1 mg/mL), diluted to 200 μL with sterile saline, and administered via subcutaneous injection in the subaxillary region.

Single-cell RNA sequencing library construction and bulk RNA sequencing data analysis

Fresh tumor specimens were harvested from orthotopic HCC-bearing mice treated with distinct regimens (PBS, NeoVAC, Birinapant, or BiriNeo). The tissues were enzymatically dissociated into single-cell suspensions using 2 mL of sCelLive Tissue Digestion Solution (Singleron, China). Cell viability, assessed by trypan blue exclusion, exceeded 80%. After adjusting the cell density to 3×10⁵ cells/mL in PBS, single cells were isolated, and mRNA was captured using the Singleron Matrix single-cell processing system. ScRNA-seq libraries and single-cell targeted RNA sequencing were simultaneously performed using the FocuSCOPE Single Cell Multiomics mRNA×Mouse Liver Mutation Detection Kit (Singleron, China). Libraries that passed quality control were sequenced on an Illumina HiSeq 6000 system (150 bp paired-end mode). The custom sequencing panel included seven neoantigen-derived somatic mutations.

Raw single-cell RNA sequencing reads were initially analyzed using Celescope (https://github.com/singleron-RD/CeleScope) with the GRCm38 reference genome to construct gene expression matrices. Subsequent data processing was conducted using the Seurat R package. Following cell quality control, including contamination removal via decontX, 9000 cells per sample were randomly subsampled for downstream analyses. Batch effects were mitigated through data integration using the Harmony R package. Cell clusters were identified via Seurat’s graph-based clustering algorithm, and automated cell type annotation was performed with the SingleR package. Final cell type assignments were manually refined by assessing cluster-specific marker gene expression patterns.

To delineate tumor cells from normal counterparts, copy number variation analysis was performed via SCEVAN (https://github.com/AntonioDeFalco/SCEVAN). Cells co-identified as tumor cells by SCEVAN and annotated as hepatocytes in prior cell typing were classified as bona fide tumor cells for subsequent analyses. The expression levels of predefined pathways in tumor cells across samples were quantified using Seurat’s AddModuleScore function. T cells were subsetted based on prior annotations and re-analyzed through an analogous pipeline to stratify distinct subsets, with each subset annotated via its unique gene expression profile. Seurat’s AverageExpression function was employed to compute mean expression values of selected genes across T cell subsets.

Single-cell targeted RNA sequencing data focusing on seven neoantigen-derived mutations were analyzed using Celescope to resolve mutation distribution at single-cell resolution. Shared cell barcodes between targeted sequencing and scRNA-seq datasets were retained to exclude cells lacking transcriptional profiles prior to downstream analyses. Genotyping of the seven mutations identified cells harboring heterozygous (0/1) or homozygous (1/1) genotypes as mutation-expressing cells. This genotypic information was integrated into the single-cell metadata, and the cellular prevalence of each neoantigen was quantified as the proportion of cells expressing the corresponding mutation.

Cell-cell interaction analysis

To investigate the impact of immunotherapy on intercellular communication among distinct cell clusters, CellChat was used to infer biologically relevant cell-cell interactions. Ligand-receptor pairs with statistically significant differences (p<0.05) were retained as key communication pathways. Analyses were conducted independently for each experimental group, and the results were aggregated for cross-group comparison to identify the signaling networks.

Statistical analysis

Statistical analysis of data was analyzed by t-test, one-way, or two-way analysis of variance(ANOVA). Survival curves were estimated using the Kaplan–Meier method and compared using the log-rank (Mantel–Cox) test. Results are presented as mean±SEM from at least three experiments by the software of GraphPad Prism (San Diego, USA).

Results

XIAP overexpression negatively regulates CD8⁺ T-cell infiltration and associates with poor prognosis in HCC

To establish the clinical relevance of XIAP in HCC immunity, we first analyzed bulk RNA sequencing data from an institutional cohort of 97 paired HCC and para-tumor tissues. XIAP was significantly overexpressed in tumor tissues compared with para-tumor tissues (figure 1A). Correlation analysis revealed an inverse association between XIAP expression and CD8⁺ T-cell infiltration (r=−0.35, p=0.00016), whereas no significant correlation was observed with CD4⁺ T-cell infiltration (figure 1B,C). Kaplan-Meier analysis further showed that patients with high XIAP expression had significantly shorter recurrence-free survival (RFS) (p=0.023, figure 1D), whereas overall survival (OS) was not significantly different between the two groups (online supplemental figure S1A). To validate these findings at the protein level, we performed multiplex immunofluorescence analysis using an independent TMA containing 93 paired HCC and para-tumor tissue specimens. XIAP expression was increased in tumor tissues (online supplemental figure S1B), accompanied by reduced CD8⁺ and CD4⁺ T-cell infiltration (online supplemental figure S1C, D). The 93 HCC tumors from the TMA cohort were analyzed for XIAP expression and immune-cell infiltration. Representative images and quantification of XIAP expression, CD8⁺ T-cell infiltration, and CD4⁺ T-cell infiltration are shown in online supplemental figure S1E–G, respectively, with each parameter stratified into high and low groups according to its respective median value. XIAP-high tumors showed lower CD8⁺ and CD4⁺ T-cell infiltration (figure 1E), and patients with high XIAP expression had poorer OS and RFS (figure 1F,G). Conversely, higher CD3⁺CD8⁺ T-cell infiltration was associated with improved OS and RFS (figure 1H,I), whereas CD3⁺CD4⁺ T-cell infiltration showed a weaker association with prognosis (figure 1J,K). Spatial analysis further showed an inverse correlation between XIAP expression and CD3⁺CD8⁺ T-cell density (r=−0.36, p=0.0007, figure 1L), while no correlation was observed with CD3⁺CD4⁺ T cells (online supplemental figure S1H). XIAP expression was also positively associated with PD-L1 expression in the bulk RNA-seq cohort (r=0.25, p=0.015, figure 1M). Collectively, these clinical data show that elevated XIAP expression is associated with reduced cytotoxic T-cell infiltration and increased PD-L1 expression, a pattern consistent with an immunosuppressive TME in HCC.

Figure 1. Discovery and validation of the association between XIAP expression, immune infiltration, and clinical outcome in HCC. (A–D, M) Analyses based on the institutional bulk RNA-seq cohort (97 paired HCC and para-tumor tissue specimens). (A) The expression levels of XIAP in 97 paired HCC and para-tumor tissues. (B, C) The correlation analysis between XIAP expression and CD8⁺ or CD4⁺ T-cell infiltration. (D) Kaplan-Meier analysis of recurrence-free survival (RFS) according to XIAP expression. (E–L) Validation in an independent tissue microarray (TMA) cohort containing 93 paired HCC and para-tumor specimens. (E) Representative multiplex immunofluorescence images and quantification of CD4⁺ and CD8⁺ T cell infiltration in XIAP-high and XIAP-low tumors. (blue: DAPI, green: XIAP, purple: CD4, yellow: CD8, n=93). Scale bar= 50 µm. (F, G) Kaplan-Meier analysis of overall survival (OS) and RFS according to XIAP expression. (H, I) Kaplan-Meier analysis of OS and RFS according to CD3⁺CD8⁺ T-cell infiltration. (J, K) Kaplan-Meier analysis of OS and RFS according to CD3⁺CD4⁺ T-cell infiltration. (L) Correlation between XIAP expression and CD3⁺CD8⁺ T-cell density. (M) The correlation analysis between XIAP and PD-L1 expression. Data are presented as the mean±SEM. Statistical significance was determined by one-way ANOVA. ANOVA, analysis of variance; HCC, hepatocellular carcinoma; RFS, recurrence-free survival; XIAP, X-linked inhibitor of apoptosis protein.

Birinapant enhances tumor immunogenicity and promotes T-cell infiltration in vivo

To evaluate whether pharmacological disruption of IAP-mediated survival signaling could enhance HCC immunogenicity, we first assessed the effects of Birinapant, a SMAC mimetic/IAP antagonist, in vitro. Birinapant alone had little effect on Hepa1-6 cells and HCC patient-derived organoid(PDO) viability(figure 2A and online supplemental figure S2B). Since TNF-α can potentiate Birinapant-mediated cytotoxicity, Hepa1-6 cells and HCC PDOs were treated with TNF-α and increasing concentrations of Birinapant, resulting in dose-dependent growth inhibition (figure 2A and online supplemental figure S2A–C). Birinapant treatment also increased MHC class I expression on Hepa1-6 cells (p<0.0001, figure 2B), consistent with enhanced antigen-presentation capacity. We next evaluated its immunomodulatory effects in an orthotopic HCC model in immunocompetent C57BL/6 mice (figure 2C). Birinapant inhibited tumor growth and prolonged survival relative to PBS (figure 2D,E). Immunohistochemical analysis showed reduced XIAP abundance after Birinapant treatment (figure 2F and online supplemental figure S2E). At day 10, Birinapant-treated tumors had more CD3⁺CD4⁺ (p=0.0315, figure 2G and online supplemental figure S2F) and CD3⁺CD8⁺ T cells (p=0.0005, figure 2H and online supplemental figure S2G) and showed lower PD-L1 expression (figure 2I and online supplemental figure S2H). Together, these findings show that Birinapant treatment increases tumor immunogenicity and immune-cell infiltration in this HCC model.

Figure 2. Birinapant enhances tumor immunogenicity and promotes T-cell infiltration. (A) Dose-response analysis of Hepa1-6 cell viability after Birinapant treatment in the presence or absence of TNF-α using the CellTiter-Glo ATP assay. (B) Quantification of MHC-I expression by flow cytometry (n=3, t-test). (C, D) Bioluminescence monitoring and quantification of tumor burden in PBS-treated and Birinapant-treated mice (n=6, two-way ANOVA). (E) Survival curves after treatment. (F–I) Quantification of XIAP (F), CD4 (G), CD8 (H), and PD-L1 (I) staining in orthotopic HCC tumors (n=5, t-test). (J) Quantification of tumor-infiltrating CD8⁺ T cells over time (n=5, one-way ANOVA). Data are presented as mean±SEM. PBS, phosphate-buffered saline; MHC-I, major histocompatibility complex class I; ANOVA, analysis of variance; HCC, hepatocellular carcinoma.

Longitudinal analysis nevertheless revealed limitations of Birinapant monotherapy. Five mice developed tumor progression after an initial regression following Birinapant treatment. In independent mice analyzed at different time points, intratumoral CD8⁺ T-cell infiltration was lower at day 30 and day 50 compared with day 10 after treatment, coinciding with tumor relapse (figure 2J). This decline suggests that the immune activation induced by Birinapant alone was not sufficient to sustain durable tumor control. These observations provided the rationale for combining pharmacological IAP antagonism with neoantigen vaccination to maintain antigen-specific antitumor immunity.

Combining neoantigen vaccination with Birinapant achieves potent antitumor efficacy

We next investigated whether combining Birinapant with NeoVAC could enhance antitumor immunity. Based on previously characterized immunogenic synthetic neoantigen peptides, we formulated NeoVAC containing seven neoantigen peptides (Mapk3_S284F, Lmf1_F523V, Samd9l_K752M, Traf7_C403W, Dtnb_K40T, Lbr_A341P, and Ptpn2_I383T) and poly(I:C). Given the capacity of Birinapant to increase immune-cell infiltration and tumor immunogenicity, we evaluated the BiriNeo combination (Birinapant plus NeoVAC) in vitro and in vivo. For in vitro validation, neoantigen-pulsed bone marrow-derived dendritic cells were co-cultured with syngeneic splenic T cells and assessed for cytotoxicity against Hepa1-6 cells (figure 3A). BiriNeo increased tumor-cell apoptosis relative to the control and monotherapy groups (p<0.0001, figure 3B,C), accompanied by reduced XIAP abundance and caspase-3 activation (figure 3D, online supplemental figure S3). Increased lactate dehydrogenase release was also observed in the BiriNeo group (p<0.0001, figure 3E). These results indicate that Birinapant potentiates vaccine-elicited T-cell-mediated tumor killing.

Figure 3. Antitumor effects of NeoVAC combined with the SMAC mimetic Birinapant in vitro and in vivo. (A) Schematic of the in vitro experimental workflow. (B, C) Flow cytometric analysis and quantification of Annexin V+/PI+ Hepa1-6 cells across treatment groups (n=5, one-way ANOVA). (D) Western blot analysis of XIAP, caspase-3, and cleaved caspase-3 in Hepa1-6 cells. (E) LDH-release assay across treatment groups (n=5, one-way ANOVA). (F, G) Bioluminescence monitoring and quantification of tumor burden in mice treated with PBS, NeoVAC, Birinapant, or BiriNeo (PBS, NeoVAC, and Birinapant, n=6; BiriNeo, n=8; two-way ANOVA). (H) Survival curves after treatment. (I) Bioluminescence monitoring of tumor recurrence and metastasis in BiriNeo-treated and control mice (n=5). Data are presented as mean±SEM. PBS: phosphate-buffered saline; ANOVA, analysis of variance; CR, complete response; DCs, dendritic cells; LDH, lactate dehydrogenase; SMAC, second mitochondria-derived activator of caspases; XIAP, X-linked inhibitor of apoptosis protein.

To evaluate therapeutic efficacy, mice bearing orthotopic HCC tumors were randomized to PBS, NeoVAC, Birinapant, or BiriNeo. Longitudinal bioluminescence imaging revealed marked differences among treatment groups (figure 3F). PBS-treated mice showed progressive tumor growth with 100% mortality by day 50, whereas the monotherapy groups showed modest improvements (mortality: 2/6 for NeoVAC and 1/6 for Birinapant). BiriNeo produced sustained tumor regression, with complete responses observed in 6 of 8 mice (figure 3G). Kaplan-Meier analysis showed significantly prolonged survival in the combination group compared with the other treatment groups (figure 3H), while biochemical and histological assessments supported an acceptable safety profile (online supplemental figure S4A–I). To examine the broader relevance of the combination in liver cancer, we also evaluated Birinapant plus OVA_250-264 vaccination in an orthotopic KPC-OVA-ICC model. The combination significantly suppressed tumor growth (online supplemental figure S5A, B).

Collectively, these results demonstrate that BiriNeo achieves superior antitumor efficacy compared with either monotherapy. To further evaluate the therapeutic potential of BiriNeo, anti-PD-1 was combined with BiriNeo treatment. Compared with BiriNeo alone, the addition of anti-PD-1 further suppressed tumor growth and increased the frequencies of IFNγ⁺ and GZMB⁺ CD8⁺ T cells in tumor tissues (online supplemental figure S8A, B, C–G), indicating that the antitumor activity of BiriNeo can be further enhanced by PD-1 blockade. To determine whether the antitumor activity of BiriNeo depends on T cells, in vivo CD4⁺ and CD8⁺ T-cell depletion experiments were performed (online supplemental figure S8A, B). Efficient depletion was confirmed by flow cytometry (online supplemental figure S8H, I). CD8⁺ T-cell depletion largely abolished the therapeutic efficacy of BiriNeo, whereas CD4⁺ T-cell depletion had a more limited effect, indicating that BiriNeo-mediated tumor control is primarily dependent on CD8⁺ T cells.

BiriNeo establishes durable immunological memory against HCC recurrence and metastasis

Encouraged by the antitumor efficacy of BiriNeo, we evaluated its capacity to elicit protective immunity against tumor recurrence and metastasis. In orthotopic HCC-bearing mice, BiriNeo treatment resulted in complete tumor regression in all animals (n=5) by day 30, whereas PBS-treated controls showed progressive disease(figure 3I). To mimic distant metastasis, BiriNeo-cured mice and age-matched naïve controls were intravenously challenged with Hepa1-6-Luc cells on day 30. No metastatic lesions were detected in BiriNeo-cured mice on days 40 and 50, whereas all control mice developed lung and brain metastases by day 10, with further progression by day 20. To assess protection against intrahepatic recurrence, BiriNeo-cured mice and new untreated controls (n=5) were rechallenged by subcapsular injection of Hepa1-6-Luc cells into the liver on day 50. Only one mouse in the BiriNeo group developed a small, transient focus that was cleared by day 70; the remaining mice stayed tumor-free. In contrast, all untreated controls developed progressive intrahepatic tumors (figure 3I). These findings show that BiriNeo treatment establishes durable systemic protection against recurrence and metastasis in this model.

BiriNeo combination therapy potentiates antigen presentation and robust tumor-infiltrating T-cell responses

To investigate the mechanism underlying the higher efficacy of BiriNeo, we examined how IAP antagonism by Birinapant interacts with neoantigen vaccination to enhance antitumor immunity. Multiplex immunofluorescence showed reduced XIAP staining in both the Birinapant and BiriNeo groups (figure 4A,C). BiriNeo produced the highest infiltration of CD4⁺ and CD8⁺ T cells (figure 4B, D and E) and the highest granzyme B expression (figure 4F,I), together with the lowest PD-L1 expression (figure 4G,J) and increased tumor-cell apoptosis by TUNEL staining (figure 4H,K). Cytokine profiling showed increased GZMB, IFNγ, TNF-α, IL-2, and IL-12 in BiriNeo-treated tumors (figure 4L), consistent with a pro-inflammatory antitumor microenvironment. BiriNeo also increased MHC-I and MHC-II expression on dendritic cells relative to control (figure 4M,N and online supplemental figure S6A), together with higher frequencies of mature CD11c⁺CD80⁺CD86⁺ dendritic cells in tumor-draining lymph nodes (figure 4O and online supplemental figure S6B), activated CD69⁺CD8⁺ T cells in tumors (figure 4P and online supplemental figure S6C), and CD44⁺CD62L^- effector-memory CD8⁺ T cells in spleen (figure 4Q and online supplemental figure S6D). Ptpn2_376-384:H-2Kᵇ-specific tetramer staining further showed increased neoantigen-specific CD8⁺ T-cell infiltration in BiriNeo-treated tumors relative to NeoVAC or Birinapant alone (figure 4R and online supplemental figure S6C). At day 30, BiriNeo also increased CD44⁺CD62L^- effector-memory CD8⁺ T cells in spleen and peripheral blood compared with NeoVAC alone (online supplemental figure S9). In the KPC-OVA-ICC model, Birinapant plus OVA_250-264 vaccination similarly increased effector-memory and OVA_{250-264-specific} CD8⁺ T-cell responses (online supplemental figure S5C–F), supporting the immunological activity of this combination across two liver tumor models.

Figure 4. BiriNeo combination therapy potentiates antigen presentation and robust tumor-infiltrating T-cell responses. (A, C) Representative images and quantification of XIAP expression in tumor tissues after treatment by IF staining. Scale bar=200 µm (n=5, t-test). (B, D, E) Representative images and quantification of CD4⁺(green), CD8⁺(red) T-cell infiltration in tumor tissues after treatment by IF staining. Scale bar=200 µm (n=5, one-way ANOVA). (F, G, I, J) Representative images and quantification of GZMB (F, I) and PD-L1 (G, J) in tumor tissues after treatment by IF staining. Scale bar=200 µm (n=5, one-way ANOVA). (H, I) Quantification of tumor cell apoptosis across treatment groups by TUNEL assay. Scale bar=200 µm (n=5, one-way ANOVA). (L) ELISA analysis of cytokine production in tumor tissues (n=5, one-way ANOVA). (M–O) The percentage and statistical analysis of CD80⁺CD86⁺ (M) and MHC-I (N), MHC-II (O) in CD11c⁺ (DCs) by flow cytometry (n=5, one-way ANOVA). (P–R) Quantification and statistical analysis of CD69⁺CD8⁺ T cells infiltrating tumor tissues (P), CD44⁺CD62L⁻ effector-memory CD8⁺ T cells in the spleen (Q), and Ptpn2₃₇₆₋₃₈₄:H-2Kᵇ-specific CD8⁺ T cells infiltrating tumor tissues (R) by flow cytometry. (n=5, one-way ANOVA). Data are presented as the mean±SEM. ANOVA, analysis of variance; DCs, dendritic cells; IF, immunofluorescence.

Collectively, these comprehensive analyses demonstrate that BiriNeo combination therapy achieves synergistic antitumor efficacy through coordinated enhancement of antigen presentation, neoantigen-specific T cell priming, immune-cell infiltration, and sustained cytotoxic responses, establishing this approach as a promising therapeutic strategy for HCC.

BiriNeo combination therapy achieves powerful neoantigen-specific cytotoxicity

To dissect the mechanistic basis of BiriNeo efficacy, we performed single-cell RNA sequencing and targeted neoantigen mutation analysis on orthotopic HCC tissues from the PBS, NeoVAC, Birinapant, and BiriNeo groups at day 10 after treatment. After quality control, 9000 randomly selected cells were analyzed by Uniform Manifold Approximation and Projection (UMAP), revealing 11 major populations (figure 5A). SCEVAN-based copy-number profiling identified malignant hepatocytes. BiriNeo-treated tumors contained fewer malignant cells and were enriched for T cells, macrophages, B cells, and monocytes relative to the other groups (figure 5B,C). Targeted analysis of the seven vaccine neoantigens was performed on tumor cells and showed the lowest residual neoantigen expression and mutation frequencies in the BiriNeo group (figure 5D–F), consistent with enhanced elimination of neoantigen-positive tumor cells. Birinapant monotherapy produced variable changes, whereas NeoVAC reduced mutation frequencies across the seven targets. BiriNeo also increased MHC-I and MHC-II module scores in tumor cells relative to NeoVAC (p=0.0023 and p=0.00092, respectively; figure 5G,H), supporting enhanced antigen presentation and tumor-cell recognition. These findings indicate that Birinapant potentiates vaccine-driven elimination of neoantigen-positive tumor cells without assigning this effect to a single IAP family member.

Figure 5. Single-cell RNA sequencing revealed alterations of HCC microenvironment on different treatments. Samples were obtained from orthotopic HCC tumors in all four treatment groups at day 10 post-treatment. (A) UMAP plot showing all the identified cell types within the HCC model. (B) Stacked bar charts showing absolute changes in tumor cells and normal hepatocytes across treatment groups. (C) Absolute quantification of tumor-infiltrating immune cell populations across four treatment groups. (D) UMAP plot showing the expression of tumor-derived neoantigens in tumor cells across treatment groups. Color gradient indicates relative expression levels. Neoantigen expression was analyzed in tumor cells. (E) The proportion of tumor cells co-expressing different numbers of neoantigen sequences. (F) The difference in expression frequency of different neoantigens in each group as indicated. (G, H) Module scores of MHC-I and MHC-II in tumor cells of each group. Data are presented as the mean±SEM. UMAP: Uniform Manifold Approximation and Projection; MHC-I: major histocompatibility complex class I; PBS: phosphate-buffered saline; HCC, hepatocellular carcinoma.

BiriNeo combination therapy induces robust neoantigen-specific Ly6C CD8⁺ T cell expansion and immunological remodeling

Given the central role of T cells in BiriNeo-mediated responses, we performed focused re-clustering analysis of all T-cell populations. Based on canonical marker expression (CD3d, CD3e, CD4, CD8a), we identified eight CD8⁺ T cell subsets, two CD4⁺ T cell subsets, and one γδ T cell subset, which were functionally annotated using integrated marker gene profiles (figure 6A,B). Quantitative analysis revealed that BiriNeo treatment significantly expanded four distinct cell clusters (cluster 1: Ly6C⁺CD8⁺ T cells, cluster 2: CD7⁺CD8⁺ T cells, cluster 5: Itgb1⁺CD8⁺ T cells, cluster 7: Tcf7⁺CD8⁺ T cells) compared with all other treatment groups (figure 6C, online supplemental figure S7A). Among these four expanded populations, comprehensive phenotypic analysis demonstrated that Ly6C⁺CD8⁺ T cells exhibited high expression of effector molecules (Gzmb, Ifng), alongside limited expression of exhaustion markers (Pdcd1, Lag3, Tigit), while the other three clusters displayed relatively weaker effector functions (figure 6D), thereby identifying Ly6C⁺CD8⁺ T cells as the predominant antitumor effector population. Ly6C, a glycosylphosphatidylinositol-anchored protein of the lymphocyte antigen 6 superfamily, critically regulates CD8⁺ T-cell migration, tissue homing, and memory cell survival and reactivation potential, positioning Ly6C⁺CD8⁺ T cells as well-suited effector-memory populations for sustained antitumor immunity.^30–32 The expansion of Ly6C⁺CD8⁺ T cells thus marks a shift toward a more immunologically active TME capable of durable surveillance. Flow cytometry validation confirmed significantly higher proportions of Ly6C⁺CD8⁺ T cells in BiriNeo-treated tumors (figure 6E,F, online supplemental figure S10A), demonstrating effective expansion and infiltration of this critical subset. To validate antigen specificity, we assessed Ptpn2 neoantigen tetramer binding in Ly6C⁺CD8⁺ T cells. BiriNeo-treated mice showed significantly higher percentages of Ptpn2_376-384 :H-2Kᵇ-specific Ly6C⁺CD8⁺ T cells compared with other groups (figure 6G,H, online supplemental figure S10B), confirming that this population exhibits robust neoantigen-specific recognition capacity and that BiriNeo treatment optimally enhances their immunoreactivity. Consistent findings were also obtained in the KPC-OVA-ICC model, where combination treatment similarly increased the frequency of OVA_250-264 tetramer-positive Ly6C⁺CD8⁺ T cells (online supplemental figure S7B). Furthermore, analysis of intercellular interactions revealed extensive crosstalk between Th1 (Ifng⁺CD4⁺) cells and Ly6C⁺CD8⁺ T cells within BiriNeo-treated tumors. Th1 cells enhanced CD8⁺ T-cell activation and recruitment by promoting antigen presentation and chemokine (Ccl3/4/5-Ccr5) signaling. In turn, Ly6C⁺CD8⁺ T cells stabilized the immunological synapse between Ifng⁺CD4⁺ T cells and antigen-presenting cells, promoted Th1 cell differentiation, establishing a mutually reinforcing antitumor circuit (figure 6I). Taken together, BiriNeo combination therapy profoundly remodels the HCC tumor immune microenvironment by enhancing antigen presentation, expanding neoantigen-specific effector-memory CD8⁺ T cells (notably Ly6C⁺CD8⁺), and promoting synergistic Th1-CD8⁺ T-cell interactions. This orchestrated immune reprogramming shifts the TME toward a highly pro-inflammatory, immune-active state, underpinning the potent and sustained antitumor responses seen with BiriNeo.

Figure 6. BiriNeo combination therapy induces infiltration of Ly6C⁺CD8⁺ T cells. Samples were obtained from orthotopic HCC tumors in all four treatment groups at day 10 post-treatment. (A) UMAP visualization of the 11 annotated T-cell subclusters. The cluster numbers and color scheme are shared with B and C. The expression distribution of core T-cell receptor complex genes (Cd3d and Cd3e) and lineage markers (Cd4 and Cd8a) is shown across the annotated T-cell subclusters. (B) Heatmap showing the expression of signature genes across the 11 annotated T-cell subclusters. (C) Absolute numbers of the 11 T-cell subclusters in each treatment group. (E–H) The percentage and statistical analysis of tumor-infiltrated Ly6C⁺CD8⁺ T cells. (E, F) and Ptpn2_376-384:H-2Kᵇ-specific Ly6C⁺CD8⁺ T cells (G, H) in tumor tissues by flow cytometry (n=5, one-way ANOVA). (I) Significant ligand-receptor interactions between Ifng⁺CD4⁺ T cell and Ly6C⁺CD8⁺ T cell subsets as indicated by CellChat. Data are presented as the mean±SEM. ANOVA, analysis of variance; HCC, hepatocellular carcinoma.

BiriNeo combination therapy drives antitumor M1 macrophage polarization

The BiriNeo group demonstrated a notable enrichment of macrophage populations compared with other treatment groups, suggesting increased macrophage infiltration into the TME. Assessment of macrophage cytotoxic potential across treatment groups using GOBP datasets revealed the highest levels of cytotoxic activity in the combination therapy group, indicating enhanced macrophage-mediated tumor killing (p<0.0001, figure 7A). Re-clustering analysis of all macrophages showed a proportional expansion of clusters 1, 2, 3, 6, and 7 in BiriNeo-treated tumors (figure 7B,C). Single-cell RNA sequencing-based M1 macrophage scoring identified clusters 1, 3, and 6 as significantly activated, pro-inflammatory M1-like macrophage subsets (figure 7D). This suggests that the combination therapy promotes M1-like macrophage polarization, a finding supported by flow cytometry, which revealed significantly elevated proportions of CD80⁺F4/80⁺ M1 macrophages and a concurrent reduction in CD206⁺F4/80⁺ M2 macrophages in BiriNeo-treated tumors (figure 7E-H, online supplemental figure S10C). These results suggest that combination therapy facilitates the polarization of antitumor M1 macrophages.

Figure 7. BiriNeo combination therapy drives antitumor M1 macrophage repolarization. (A) Scoring of cell killing in each macrophage cluster. (B) UMAP visualization of the distribution patterns of macrophage subpopulations. (C) Histogram indicating the absolute quantification of macrophage clusters within each group. (D) M1 (upper) and M2 (lower) polarization scores across annotated macrophage clusters (clusters 0–12). (E–H) The percentage and statistical analysis of tumor-infiltrated CD80⁺F4/80⁺ macrophages (E–F) and CD206⁺F4/80⁺ macrophages (G–H) in tumor tissues by flow cytometry (n=5, one-way ANOVA). Data are presented as the mean±SEM. ANOVA, analysis of variance.

Discussion

HCC is characterized by multiple, convergent mechanisms of immune suppression that can limit the efficacy of single-agent immunotherapy. Our study provides preclinical evidence that combining personalized neoantigen vaccination with the SMAC mimetic Birinapant can address complementary barriers to antitumor immunity. NeoVAC elicits tumor-specific T-cell responses, whereas Birinapant antagonizes IAP-mediated survival signaling and enhances tumor-cell susceptibility to immune-mediated killing, accompanied by increased antigen presentation and immune-cell infiltration in our models. The combination produced durable tumor control accompanied by expansion of neoantigen-specific Ly6C⁺CD8⁺ effector-memory T cells, predicted ligand-receptor interactions with Th1 cells, and enrichment of pro-inflammatory M1-like macrophages. These coordinated changes provide a cellular framework for the enhanced efficacy of BiriNeo relative to either monotherapy.

Most previous studies of SMAC mimetics have combined these agents with chemotherapy, radiotherapy, targeted therapy, or immune-checkpoint blockade. Here, combining Birinapant with a neoantigen vaccine increased antigen presentation, reduced PD-L1 expression, promoted immune-cell infiltration, and established immunological memory in preclinical HCC models. Our focus on XIAP was based on its association with poor prognosis and reduced CD8⁺ T-cell infiltration in our HCC cohorts, which provided the clinical and biological rationale for this study. Although the present study used a peptide-based neoantigen vaccine, Birinapant acts on tumor-cell survival and immune sensitivity rather than on a particular vaccine platform; the combination concept may therefore extend to DNA-based or RNA-based neoantigen vaccines. This possibility should be evaluated directly in future studies.

In conclusion, our data provide preclinical support for combining the SMAC mimetic Birinapant with personalized neoantigen vaccination in HCC. This strategy integrates pharmacological disruption of IAP-mediated apoptosis resistance with vaccine-induced, antigen-specific T-cell immunity and provides a rationale for further evaluation in clinically relevant models and future translational studies.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.