S-propargyl-cysteine remodels the HCC microenvironment and potentiates anti-PD-1 therapy through CSE/H₂S-linked vascular and immune reprogramming

Abstract

Background

Hepatocellular carcinoma (HCC) is characterized by aberrant angiogenesis and an immunosuppressive tumor microenvironment, both of which limit durable responses to PD-1 blockade. Strategies that concurrently target vascular dysfunction and immune inhibition may improve therapeutic efficacy.

Methods

The interaction between S-propargyl-cysteine (SPRC) and cystathionine-gamma-lyase (CSE) was assessed by surface plasmon resonance. In human umbilical vein endothelial cells, NF-κB p65 and STAT3 phosphorylation, VEGFA expression, extracellular H₂S production, CSE enzymatic activity, and tube formation were evaluated. In HepG2 and activated Jurkat-cell co-cultures, PD-L1 expression, T-cell phenotypes, and granzyme B secretion were analyzed. H3K27me3 enrichment at RELA and STAT3 was examined by ChIP-qPCR. Antitumor efficacy was assessed in an HCC mouse model treated with anti-PD-1 alone or in combination with SPRC. Tumor tissues were further analyzed for intratumoral H₂S levels, CSE enzymatic activity, NF-κB/STAT3 activation, VEGFA and PD-L1 expression, CD31-associated vascular density, immune-cell infiltration, metabolomic remodeling, and oxidative-stress status.

Results

SPRC bound recombinant CSE with nanomolar affinity and enhanced CSE-dependent H₂S generation and enzymatic activity. In endothelial cells, SPRC reduced NF-κB p65 and STAT3 phosphorylation, decreased VEGFA expression, and impaired tube formation, whereas these effects were attenuated by CSE knockdown. In co-culture, SPRC reduced PD-L1 expression on HepG2 cells and promoted a more cytotoxic T-cell phenotype, as indicated by increased CD8α, reduced FoxP3, and elevated granzyme B secretion. Rescue experiments showed that constitutively active STAT3 restored PD-L1 expression despite SPRC treatment, whereas p65 overexpression reversed SPRC-mediated suppression of VEGFA expression and endothelial tube formation. SPRC also increased H3K27me3 enrichment at the RELA and STAT3 promoters. In vivo, SPRC enhanced the antitumor efficacy of PD-1 blockade, increased intratumoral CD8⁺ T-cell infiltration, decreased Treg abundance, and reduced CD31-associated vascular density. Mechanistically, SPRC plus anti-PD-1 increased intratumoral H₂S levels and CSE enzymatic activity, suppressed NF-κB p65 and STAT3 phosphorylation, and decreased VEGFA and PD-L1 expression in tumor tissues. Metabolomic profiling further linked an amino-acid-centered metabolic module, particularly the arginine-ornithine-citrulline axis, to immune and vascular remodeling, accompanied by reduced MDA and increased SOD activity.

Conclusions

SPRC activates a functional CSE/H₂S-linked program that suppresses NF-κB/STAT3 signaling, inhibits angiogenesis, alleviates PD-L1-associated immune suppression, and enhances the efficacy of PD-1 blockade in HCC. These findings support SPRC as a promising adjunct strategy for microenvironment-targeted immunotherapy in HCC.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12967-026-08438-x.

Introduction

Hepatocellular carcinoma (HCC) is a highly lethal malignancy often diagnosed at advanced stages, with limited response to monotherapies [1]. Immune checkpoint inhibitors targeting the programmed cell death protein 1 (PD-1) and programmed cell death ligand 1 (PD-L1) axis have reshaped first-line therapy for unresectable HCC, yet only a subset of patients achieve durable responses [2]. A central barrier to immunotherapy efficacy is the immunosuppressive tumor microenvironment (TME) of HCC, characterized by aberrant angiogenesis, hypoxia, and infiltration of suppressive immune cells [3]. As a typically hypervascular tumor, HCC is driven by pro-angiogenic factors like vascular endothelial growth factor (VEGF), leading to disorganized, leaky vasculature that elevates interstitial pressure and physically hinders cytotoxic T lymphocyte infiltration [4]. Moreover, VEGF signaling and hypoxia jointly reinforce an immunosuppressive niche by impairing dendritic cell function and recruiting regulatory T cells (Tregs) and M2-polarized tumor-associated macrophages (TAMs) [5, 6]. Hypoxia-inducible factors (HIF) further exacerbate immune evasion through upregulation of PD-L1 on tumor and stromal compartments [7]. Notably, inflammatory transcriptional circuits that include nuclear factor kappa B (NF-κB) and signal transducer and activator of transcription 3 (STAT3) are frequently activated in HCC and converge on proangiogenic and immune evasive outputs [8]. Functional crosstalk between these pathways has been implicated in PD-L1 regulation, providing a mechanistic link between inflammatory signaling and T cell dysfunction. Together, these observations highlight the need for therapeutic strategies that simultaneously normalize vascular abnormalities and restore effective antitumor immunity.

Hydrogen sulfide (H₂S) is an endogenous gasotransmitter produced by enzymes including cystathionine-gamma-lyase (CSE), and it has increasingly been recognized as a regulator of inflammation, redox balance, and cell death pathways that are relevant to cancer biology [9–12]. H₂S signaling can influence immune and inflammatory networks through redox-dependent mechanisms and protein persulfidation, and these effects have been connected to modulation of NF-κB activity in experimental systems [13–17]. In parallel, H₂S-based therapeutic strategies, including controlled donor platforms, are being actively explored as tools to overcome therapy resistance and reshape tumor-associated stress responses [18]. These properties motivate investigation of H₂S donors as candidates to remodel the HCC TME at the intersection of vascular dysfunction, inflammatory signaling, and immune suppression.

S-propargyl-cysteine (SPRC) is a cysteine-based H₂S donor that has been studied in cardiovascular and inflammatory settings and has been linked to endothelial and vascular biology through CSE-related pathways [15, 19]. Although SPRC has been associated with angiogenesis regulation in noncancer contexts, its capacity to reprogram the vascular immune axis that limits immune checkpoint inhibitor (ICI) responses in HCC has not been clearly defined [20]. A clearer mechanistic understanding is needed, particularly regarding whether SPRC can enhance functional output of the CSE/H₂S pathway, suppress NF-κB/STAT3 signaling, downregulate PD-L1-mediated immune inhibition, and promote a TME that supports cytotoxic T-cell activity.

In this study, we examined whether SPRC remodels the HCC TME through a CSE/H₂S-dependent mechanism. Surface plasmon resonance identified a direct interaction between SPRC and CSE, and we further assessed extracellular H₂S production and CSE enzymatic activity to determine whether this interaction was accompanied by enhanced functional output of the CSE/H₂S axis. We then investigated the impact of SPRC on NF-κB and STAT3 signaling and on downstream effectors implicated in angiogenesis and immune evasion, including VEGF and PD-L1. Endothelial function was evaluated using human umbilical vein endothelial cell (HUVEC) tube-formation assays, while immune modulation was assessed in HepG2 and activated Jurkat-cell co-cultures with granzyme B as a functional cytotoxicity readout. To determine whether chromatin regulation contributes to pathway modulation, we quantified histone H3 lysine 27 trimethylation (H3K27me3) enrichment at the RELA and STAT3 promoter regions and combined CSE knockdown with rescue experiments targeting STAT3 and NF-κB pathway activity to strengthen mechanistic inference. Finally, we integrated untargeted tumor metabolomics with in vivo treatment using SPRC in combination with anti-PD-1 therapy to link metabolic rewiring and redox-associated changes to improved antitumor efficacy. This integrated strategy was designed to connect target engagement with vascular normalization, immune activation, and metabolic adaptation in HCC.

Materials and methods

Reagents and cell culture

SPRC was dissolved in dimethyl sulfoxide (DMSO) as a 100 mM stock solution. For all in vitro experiments, the final DMSO concentration was maintained at ≤ 0.1% (v/v) in culture medium. HUVECs and HepG2 cells were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA) and cultured under standard conditions at 37 °C in a humidified incubator with 5% CO₂. HUVECs were maintained in endothelial growth medium, and HepG2 cells were cultured in DMEM supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin. Jurkat cells, a human acute T-cell leukemia/lymphoblastic T-cell line, were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin under standard culture conditions. Cells were routinely tested for mycoplasma contamination and used at low passage.

Small interfering RNA (siRNA) knockdown and plasmid overexpression

siRNA targeting human CSE and a non-targeting control siRNA were transfected into HUVECs using a lipid-based transfection reagent according to the manufacturer’s protocol. Knockdown efficiency was verified by quantitative reverse transcription polymerase chain reaction (qRT-PCR) and western blotting. For CSE rescue experiments, a CSE expression plasmid or the corresponding empty vector control was introduced into CSE-silenced HUVECs prior to SPRC exposure. For pathway-rescue experiments, HepG2 cells were transfected with a constitutively active STAT3 expression plasmid (STAT3C) or the corresponding empty vector control to evaluate PD-L1 regulation, whereas HUVECs were transfected with an NF-κB p65 overexpression plasmid or empty vector control to assess VEGFA expression and tube-formation phenotypes. All plasmid transfections were performed using the same lipid-based transfection reagent according to the manufacturer’s instructions.

Endothelial tube formation assay

Tube formation was performed on growth factor-reduced Matrigel (Corning, USA). A 96-well plate and tips were pre-chilled on ice for 30 min. Matrigel (50 µL/well) was added and polymerized at 37 °C for 30 min. HUVECs were detached, filtered to obtain a single-cell suspension, and seeded at 5.0 × 10⁴ cells/well in assay medium. SPRC or vehicle was added immediately after seeding. After incubation at 37 °C for 12 h, capillary-like structures were imaged by an inverted microscope. Representative tube-formation endpoints, including total branch length, number of junctions, and total meshes, were quantified quantified using ImageJ (Angiogenesis Analyzer plugin) by an investigator blinded to group allocation. For p65 rescue experiments, HUVECs were transfected with p65 overexpression plasmid or vector control prior to SPRC treatment, and tube formation was analyzed using the same procedure.

qRT-PCR

Total RNA was isolated using TRIzol reagent and reverse-transcribed using a commercial reverse-transcription kit. Quantitative PCR was performed using SYBR Green chemistry on a real-time PCR system. Relative mRNA expression was calculated using the 2^−ΔΔCt method with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as the reference gene [21].

Western blotting

Cells or tumor tissues were lysed in RIPA buffer supplemented with protease and phosphatase inhibitors. For tumor-tissue analyses, freshly excised tumor samples from the vehicle control, anti–PD-1, and SPRC plus anti–PD-1 groups were homogenized on ice in RIPA lysis buffer, followed by centrifugation at 12,000 × g for 15 min at 4 °C. Protein concentrations were determined by BCA assay, and equal amounts of total protein were separated by SDS–PAGE and transferred to PVDF membranes. Membranes were blocked in 5% non-fat milk and incubated overnight at 4 °C with primary antibodies against phosphorylated NF-κB p65 (Ser536), total NF-κB p65, phosphorylated STAT3 (Tyr705), total STAT3, VEGFA, PD-L1, CSE, and GAPDH, followed by HRP-conjugated secondary antibodies. For rescue experiments, the same procedure was used to assess VEGFA expression after CSE or p65 modulation in HUVECs and PD-L1 expression after STAT3C overexpression in HepG2 cells. Signals were developed using enhanced chemiluminescence and quantified by densitometry (ImageJ) where appropriate.

Jurkat-cell activation and HepG2 co-culture

For co-culture experiments, Jurkat cells were stimulated with phorbol 12-myristate 13-acetate (PMA, 50 ng/mL) plus ionomycin (1 µM) for 6 h to induce T-cell activation. HepG2 cells were pretreated as indicated and then co-cultured with activated Jurkat cells at an effector-to-target (E: T) ratio of 10:1 for 48 h. After co-culture, tumor-cell PD-L1 expression and T-cell phenotypes were analyzed by flow cytometry, and co-culture supernatants were collected for granzyme B measurement.

Enzyme-linked immunosorbent assay (ELISA) for granzyme B quantification

Supernatants from activated Jurkat–HepG2 co-cultures were centrifuged to remove debris and analyzed using a granzyme B ELISA kit according to the manufacturer’s instructions. Absorbance was measured at 450 nm, and concentrations were calculated from a standard curve.

Surface plasmon resonance (SPR) binding assay

The interaction between SPRC and recombinant human CSE was measured using an OpenSPR™ instrument (Nicoya) with a carboxylated gold sensor chip. CSE was immobilized by amine coupling, and SPRC was injected as a serial dilution in running buffer with DMSO matching. Sensorgrams were double-referenced, and binding kinetics were fit using a 1:1 Langmuir model to derive kinetic parameters and the equilibrium dissociation constant (K_D).

H₂S measurement

HUVECs were seeded in culture plates and subjected to the indicated treatments, including vehicle, SPRC, SPRC + small interfering RNA negative control (si-NC), SPRC + small interfering RNA targeting CSE (si-CSE), and si-CSE alone. After incubation, culture supernatants were collected and cleared by centrifugation to remove cellular debris. Extracellular H₂S levels were measured using a commercially available hydrogen sulfide detection kit or methylene blue-based assay according to the manufacturer’s instructions. H₂S concentrations were calculated from a standard curve and expressed as µM. Intratumoral H₂S levels were further measured in tumor tissues collected from the vehicle control, anti–PD-1, and SPRC plus anti–PD-1 groups. Tumor samples were weighed, homogenized in ice-cold assay buffer, and centrifuged to obtain clarified supernatants. H₂S levels were then quantified according to the manufacturer’s protocol and normalized to tissue weight or total protein concentration.

CSE enzymatic activity assay

In parallel with extracellular H₂S measurement, cell lysates were prepared from the same treatment groups for assessment of CSE enzymatic activity. Briefly, cells were lysed in ice-cold lysis buffer, and the supernatants were collected after centrifugation. CSE activity was determined using a cell lysate-based enzymatic activity assay according to the manufacturer’s protocol. Enzymatic activity was normalized to total protein concentration and expressed as nmol H₂S/min/mg protein. For in vivo validation, CSE enzymatic activity was also measured in tumor tissues from the vehicle control, anti–PD-1, and SPRC plus anti–PD-1 groups. Fresh tumor tissues were homogenized on ice in the assay buffer provided by the kit, and the homogenates were centrifuged to collect supernatants. CSE activity was determined according to the manufacturer’s instructions and normalized to total protein concentration.

Chromatin immunoprecipitation PCR (ChIP-PCR) for H3K27me3 enrichment

ChIP was performed in HUVECs after the indicated treatments using a standard ChIP protocol. Briefly, cells were cross-linked with formaldehyde, lysed, and sonicated to generate fragmented chromatin. The chromatin lysates were immunoprecipitated with an antibody against H3K27me3 or with normal IgG as a negative control. After reversal of cross-links and DNA purification, the precipitated DNA was analyzed by qPCR using primers targeting the promoter regions of RELA and STAT3, as described in recent promoter-centered chromatin studies with appropriate modifications [22, 23]. ChIP-qPCR results were normalized to input DNA and presented as relative enrichment.

Flow cytometry

For in vitro assays, HepG2 cells and activated Jurkat cells from the co-culture system were harvested after the indicated treatments and stained with fluorochrome-conjugated antibodies against PD-L1, CD8α, and FoxP3, together with a viability dye where appropriate. For STAT3 rescue experiments, HepG2 cells transfected with vector or STAT3C in the presence or absence of SPRC were analyzed for PD-L1 surface expression by flow cytometry. For in vivo assays, single-cell suspensions from tumors were prepared by mechanical dissociation and enzymatic digestion, followed by filtration and red blood cell lysis when required. Cells were stained with fluorochrome-conjugated antibodies against CD8α, PD-L1, and FoxP3, together with a viability dye. Data were acquired on a BD FACSCanto II and analyzed using FlowJo. Compensation controls and fluorescence minus one (FMO) controls were included to define gating boundaries.

Mouse tumor model and in vivo treatments

A total of eighteen 6–8-week-old C57BL/6 mice (20 ± 2 g; equal numbers of males and females) were acclimatized for one week under standard specific-pathogen-free conditions (controlled temperature and humidity, 12 h light/dark cycle) with ad libitum access to food and water. Mice were randomly assigned to three groups (n = 6 per group): vehicle control, anti–PD-1 inhibitors, and SPRC plus anti–PD-1 inhibitors. Tumors were established by subcutaneous inoculation of Hepa1-6 cells suspended in a 1:1 mixture of PBS and Matrigel at a final concentration of 2.5 × 10⁷ cells/mL. Each mouse received 0.2 mL of the suspension into the right flank under isoflurane anesthesia. Once tumors became palpable, mice in the anti–PD-1 group received intraperitoneal injections of an anti–PD-1 antibody (10 mg/kg per dose), while mice in the combination group received the same anti–PD-1 regimen plus SPRC (50 mg/kg per dose). Tumor size was measured at regular intervals using calipers, and tumor volume was calculated as V (mm³) = (L × W²)/2, where L is the longest diameter and W is the shortest diameter. This in vivo study was designed as a terminal mechanistic experiment with a predefined endpoint to permit collection of tumor tissues for downstream flow cytometry, immunohistochemistry, metabolomic profiling, and oxidative-stress analyses. At the experimental endpoint, tumors were excised for downstream analyses including flow cytometry, immunohistochemistry, and metabolomics.

Immunohistochemistry (IHC) for CD31

Tumor tissues were fixed in 10% neutral buffered formalin, paraffin-embedded, and sectioned at 4 μm. After deparaffinization, antigen retrieval, and blocking, sections were incubated overnight at 4 °C with an anti-CD31 antibody, followed by HRP-conjugated secondary antibody and DAB development. Images were acquired under identical settings, and CD31 immunoreactivity was quantified using ImageJ in a blinded manner.

Liquid chromatography–mass spectrometry (LC–MS)-based untargeted metabolomics

Tumor tissues were homogenized in ice-cold methanol-based extraction solvent, centrifuged, and supernatants were processed for LC–MS analysis. Samples were analyzed on an UHPLC system coupled to a high-resolution mass spectrometer in both positive and negative ion modes. Pooled QC samples were injected periodically to monitor instrument stability. Raw data were processed for peak picking, alignment, and annotation using standard metabolomics workflows, followed by statistical testing to define differential metabolites between SPRC + anti–PD-1 and anti–PD-1 alone groups. To link treatment-associated metabolic remodeling with biological phenotypes, an integrative correlation analysis incorporating intratumoral CD8⁺ T cells, Treg cells, and CD31 was also performed. Correlation analysis, Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment, metabolomics pathway analysis (MetPA), and metabolite set enrichment analysis (MSEA) were conducted using MetaboAnalyst (version 5.0) with appropriate multiple-testing correction.

Oxidative stress assays

To evaluate the effect of SPRC alone on intratumoral oxidative-stress status, oxidative-stress assays were performed using tumor tissues from a separate pharmacodynamic mouse experiment. In this separate experiment, tumor-bearing mice were divided into two groups: a vehicle control group and an SPRC-treated group receiving SPRC at 50 mg/kg. At the experimental endpoint, tumor tissues were collected for oxidative-stress analysis. Intratumoral lipid peroxidation was assessed by measuring malondialdehyde (MDA) levels using a TBARS assay and normalized to total protein concentration. Total superoxide dismutase (SOD) activity was measured using a WST-based assay and normalized to protein concentration according to the manufacturer’s definition of one unit of activity.

Statistical analysis

Data are presented as mean ± SD. Two-group comparisons were performed using an unpaired two-tailed Student’s t-test. Comparisons among three or more groups were analyzed by one-way ANOVA followed by Tukey’s post hoc test. For metabolomics, adjusted P values were calculated using false discovery rate correction where applicable. Statistical analyses were performed using GraphPad Prism and/or R. A two-sided P < 0.05 was considered statistically significant.

Results

SPRC enhances CSE/H₂S functional output and suppresses NF-κB/STAT3–VEGFA signaling in endothelial cells

To investigate the effect of SPRC on pro-angiogenic signaling in endothelial cells, we first examined the NF-κB and STAT3 pathway activation in HUVECs after SPRC exposure. Western blot analysis showed that SPRC reduced the levels of phosphorylated NF-κB p65 (p-p65) and phosphorylated STAT3 (p-STAT3), whereas total NF-κB p65 and total STAT3 protein levels remained largely unchanged (Fig. 1A). These findings indicate that SPRC suppresses NF-κB and STAT3 pathway activation rather than altering total protein expression. Consistent with the inhibition of these signaling pathways, VEGFA expression was also decreased following SPRC treatment (Fig. 1A). In Matrigel tube-formation assays, SPRC-treated HUVECs formed visibly impaired capillary-like networks relative to vehicle controls (Fig. 1B). The mesh number and total branch length were both significantly decreased (P < 0.01; P < 0.05; Fig. 1C), highlighting the ability of SPRC to disrupt endothelial tubulogenesis. To test whether these effects were CSE-dependent, HUVECs were transfected with CSE siRNA, resulting in efficient knockdown at the mRNA and protein levels (P < 0.001; Fig. 1D-E). Under CSE depletion, the SPRC-associated reduction in VEGFA and pathway phosphorylation was attenuated, supporting a CSE-linked mechanism (Fig. 1A). SPR binding assays confirmed a direct interaction between SPRC and recombinant CSE, with an equilibrium dissociation constant (K_D) of 78.4 nM (Fig. 1F). Functionally, SPRC significantly increased extracellular H₂S levels in HUVEC culture supernatants compared with the vehicle group, and this increase was preserved in the SPRC + si-NC group but markedly attenuated in the SPRC + si-CSE group (Fig. 1G). In parallel, SPRC also significantly increased CSE enzymatic activity, whereas CSE knockdown substantially blunted this SPRC-induced increase (Fig. 1H).

Fig. 1.

SPRC binds CSE and enhances functional output of the CSE/H₂S axis while suppressing NF-κB/STAT3 signaling and VEGFA expression in HUVECs. (A) Representative western blot showing phosphorylated NF-κB p65 (p-p65, Ser536), total NF-κB p65, phosphorylated STAT3 (p-STAT3, Tyr705), total STAT3, and VEGFA in HUVECs treated with vehicle or SPRC (10, 20, or 50 µM). A CSE-silenced condition treated with SPRC (si-CSE + 20 µM SPRC) is included to assess target dependence. GAPDH served as the loading control for each blot. (B) Representative images of Matrigel tube formation in HUVECs under control conditions and after SPRC treatment (20 µM). (C) Quantification of tube formation metrics, including mesh number and total branch length, comparing control and SPRC (20 µM) groups. (D) qRT-PCR quantification of CSE mRNA levels in HUVECs transfected with si-CSE or si-NC. (E) Representative western blot validating CSE knockdown at the protein level, with GAPDH as the loading control. (F) SPR sensorgrams showing concentration-dependent binding of SPRC to recombinant CSE. (G) Quantification of extracellular H2S levels in HUVEC culture supernatants under the indicated treatment conditions, including vehicle, SPRC, SPRC + si-NC, SPRC + si-CSE, and si-CSE alone. (H) Quantification of cell lysate–based CSE enzymatic activity in HUVEC lysates under the indicated treatment conditions, normalized to total protein. Abbreviations: SPRC, S-propargyl-cysteine; CSE, cystathionine-gamma-lyase; H2S, hydrogen sulfide; NF-κB, nuclear factor kappa B; STAT3, signal transducer and activator of transcription 3; VEGFA, vascular endothelial growth factor A; HUVECs, human umbilical vein endothelial cells; qRT-PCR, quantitative reverse transcription polymerase chain reaction; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; siRNA, small interfering RNA; si-CSE, CSE siRNA; si-NC, negative control siRNA; SPR, surface plasmon resonance. Notes: *P < 0.05; **P < 0.01; ***P < 0.001

SPRC downregulates tumor-cell PD-L1 and promotes a cytotoxic T-cell phenotype in co-culture

To assess whether SPRC modulates the PD-L1 axis and T-cell effector status, pretreated HepG2 cells were co-cultured with activated Jurkat cells and analyzed by flow cytometry. Representative histograms and quantitative analysis showed that SPRC treatment significantly reduced PD-L1 surface expression on HepG2 cells compared with control conditions (P < 0.05; Fig. 2A and B). In parallel, SPRC induced a pronounced shift in the T-cell compartment toward a cytotoxic effector phenotype, as evidenced by a significant upregulation of CD8α expression and reduced FoxP3 expression (P < 0.05; Fig. 2A–B). Importantly, pharmacological inhibition of STAT3 using Stattic elicited similar immunophenotypic changes, including significantly reduced PD-L1 expression together with enhanced CD8α and diminished FoxP3 expression, thereby supporting the involvement of STAT3-dependent signaling in SPRC-mediated immune modulation (P < 0.01; Fig. 2A–B). Consistent with improved T-cell cytotoxic activity, granzyme B (GZMB) levels in the co-culture supernatants were significantly increased following SPRC treatment (P < 0.01; Fig. 2C), indicating enhanced tumor-directed immune effector function.

Fig. 2.

SPRC downregulates tumor-cell PD-L1 and promotes a cytotoxic T-cell phenotype in co-culture. (A) Representative flow cytometry histograms showing PD-L1 on HepG2 cells and FoxP3 and CD8α signals in the activated Jurkat-cell compartment under control, SPRC (20 µM), or Stattic treatment. (B) Quantification of flow cytometry signals for PD-L1 on HepG2 cells and for CD8α and FoxP3 in activated Jurkat-cell compartment across the three conditions. (C) ELISA quantification of GZMB concentration in activated Jurkat–HepG2 co-culture supernatants under control and SPRC (20 µM) conditions. Abbreviations: SPRC, S-propargyl-cysteine; PD-L1, programmed death ligand 1; FoxP3, forkhead box P3; CD8α, cluster of differentiation 8 alpha; ELISA, enzyme-linked immunosorbent assay; GZMB, granzyme B; FITC, fluorescein isothiocyanate. Notes: *P < 0.05; **P < 0.01

SPRC remodels promoter-associated chromatin features and functionally links STAT3-dependent PD-L1 regulation with NF-κB-dependent antiangiogenic signaling

To explore the upstream epigenetic regulation of key NF-κB/STAT3 signaling genes, we performed ChIP assays in HUVECs using an anti-H3K27me3 antibody and quantified enrichment at the promoter regions of RELA (NF-κB p65) and STAT3. ChIP-PCR showed clear amplification of the expected promoter fragments in H3K27me3 immunoprecipitates, whereas IgG controls remained at background levels (Fig. 3A). Consistently, ChIP-qPCR demonstrated that SPRC treatment significantly increased H3K27me3 enrichment at both the RELA and STAT3 promoters compared with vehicle-treated controls (P < 0.01; Fig. 3B), indicating SPRC-induced remodeling of promoter-associated chromatin at these loci. To determine whether SPRC-mediated suppression of downstream angiogenic signaling requires CSE, we manipulated CSE expression and assessed VEGFA protein abundance. Consistent with an anti-angiogenic phenotype, SPRC (20 µM) reduced VEGFA protein in HUVECs. However, CSE knockdown attenuated this reduction, resulting in higher VEGFA levels than those observed with SPRC treatment alone. Re-expression of CSE in CSE-silenced cells partially restored the SPRC-associated decrease in VEGFA (Fig. 3C), supporting a CSE-dependent mechanism for VEGFA regulation.

Fig. 3.

SPRC increases promoter-associated H3K27me3 at RELA and STAT3 loci and functionally links STAT3-dependent PD-L1 regulation with NF-κB-dependent antiangiogenic signaling. (A) ChIP-PCR analysis showing H3K27me3 enrichment at the promoter regions of RELA (NF-κB p65) and STAT3 in HUVECs after SPRC treatment. IgG was used as the negative control, and specific IP yielded detectable bands at the expected sizes. (B) ChIP-qPCR quantification of H3K27me3 enrichment at RELA and STAT3 promoter regions in control and SPRC-treated HUVECs. (C) Representative western blot showing VEGFA protein levels in HUVECs under control conditions, after SPRC treatment (20 µM), under si-CSE plus SPRC, and after oe-CSE rescue conditions, with GAPDH as the loading control. (D) Representative western blot showing PD-L1 protein levels in HepG2 cells transfected with vector or constitutively active STAT3 (STAT3C) in the presence or absence of SPRC, with GAPDH as the loading control. (E) Representative western blot showing VEGFA protein levels in HUVECs transfected with vector or p65 overexpression plasmid in the presence or absence of SPRC, with GAPDH as the loading control. (F) Representative images of Matrigel tube formation in HUVECs transfected with vector or p65 overexpression plasmid and treated with SPRC as indicated. Abbreviations: SPRC, S-propargyl-cysteine; H3K27me3, histone H3 lysine 27 trimethylation; NF-κB, nuclear factor kappa B; p65, NF-κB subunit p65; STAT3, signal transducer and activator of transcription 3; STAT3C, constitutively active STAT3; VEGFA, vascular endothelial growth factor A; PD-L1, programmed death ligand 1; CSE, cystathionine-gamma-lyase; ChIP, chromatin immunoprecipitation; ChIP-PCR, chromatin immunoprecipitation polymerase chain reaction; ChIP-qPCR, chromatin immunoprecipitation quantitative polymerase chain reaction; IgG, immunoglobulin G; IP, immunoprecipitation; HUVECs, human umbilical vein endothelial cells; siRNA, small interfering RNA; si-CSE, CSE siRNA; oe-CSE, CSE overexpression; GAPDH, glyceraldehyde-3-phosphate dehydrogenase. Notes: *P < 0.05; **P < 0.01

To further assess whether SPRC-mediated immune modulation is mechanistically dependent on STAT3 activity, we performed rescue experiments using constitutively active STAT3 (STAT3C) in HepG2 cells. Under vector control conditions, SPRC reduced PD-L1 protein expression (Fig. 3D). In contrast, STAT3C markedly restored PD-L1 expression despite SPRC treatment (Fig. 3D). This rescue was further supported by flow-cytometric analysis, which showed that STAT3C substantially reversed the SPRC-induced reduction in PD-L1 surface expression (Supplementary Fig. 1). These findings indicate that SPRC-mediated PD-L1 downregulation is largely STAT3 dependent. We next evaluated whether the antiangiogenic effect of SPRC was functionally dependent on NF-κB activity. In HUVECs, SPRC suppressed VEGFA expression under vector control conditions, whereas p65 overexpression substantially reversed this inhibitory effect (Fig. 3E). Consistent with the molecular findings, Matrigel tube-formation assays showed that SPRC impaired endothelial network formation in vector-transfected cells, while p65 overexpression partially restored capillary-like tube formation despite SPRC treatment (Fig. 3F).

Therapeutic efficacy of SPRC in combination with PD-1 blockade in vivo

Building on the vascular- and immune-modulatory effects of SPRC observed in vitro, we next evaluated its antitumor activity in an HCC mouse model treated with anti–PD-1 alone or in combination with SPRC. Tumor-bearing mice were randomized to receive vehicle, anti–PD-1 inhibitors, or SPRC plus anti–PD-1 inhibitors. At endpoint, gross tumor specimens (Fig. 4A) and volumetric quantification (Fig. 4B) showed that anti–PD-1 monotherapy significantly reduced tumor volume relative to the control group (P < 0.05), whereas the combination regimen achieved a further decrease compared with both control (P < 0.01) and anti–PD-1 alone (P < 0.05). To characterize intratumoral immune remodeling, single-cell suspensions from excised tumors were analyzed by flow cytometry. Representative gating and summary statistics demonstrated an increased frequency of CD3⁺CD8⁺ T cells in the anti–PD-1 group (P < 0.05; Fig. 4C), with a further enrichment in the SPRC plus anti–PD-1 group (P < 0.01; Fig. 4C). In parallel, the proportion of Treg cells (FoxP3⁺ cells within CD4⁺ T cells) decreased under anti–PD-1 treatment (P < 0.05; Fig. 4D) and was further reduced by the combination therapy (P < 0.01; Fig. 4D), indicating attenuation of immunosuppressive T-cell compartments.

Fig. 4.

SPRC enhances the antitumor efficacy of anti-PD-1 therapy and remodels intratumoral immune composition in vivo. (A) Images of excised tumors from mice treated with control, anti-PD-1, or SPRC plus anti-PD-1 at the experimental endpoint. (B) Tumor volume quantification across treatment groups. (C) Representative flow cytometry gating plots and summary quantification of intratumoral CD3⁺CD8⁺ T cells, expressed as a percentage of total cells in dissociated tumor samples. (D) Representative flow cytometry gating plots and summary quantification of Tregs, expressed as a percentage of FoxP3+ cells within the CD4+ T-cell population. Abbreviations: SPRC, S-propargyl-cysteine; PD-1, programmed cell death protein 1; CD3, cluster of differentiation 3; CD8, cluster of differentiation 8; Tregs, regulatory T cells; FoxP3, forkhead box P3; CD4, cluster of differentiation 4. Notes: *P < 0.05; **P < 0.01

SPRC activates the intratumoral CSE/H₂S axis and suppresses NF-κB/STAT3–VEGFA/PD-L1 signaling in vivo

Given the close interplay between vascular abnormalities and immune exclusion, we assessed tumor vascularization by CD31 immunohistochemistry. Compared with controls, CD31 staining intensity and the corresponding quantitative index were reduced by anti–PD-1 treatment and most prominently diminished in tumors from mice receiving SPRC plus anti–PD-1 treatment (P < 0.01; P < 0.001; Fig. 5A), supporting the ability of SPRC to enhance vascular remodeling in the setting of PD-1 blockade. To determine whether the underlying mechanism observed in vitro was recapitulated in tumor tissues, we performed additional mechanistic analyses using tumor lysates from each treatment group. Western blot analysis showed that SPRC plus anti–PD-1 markedly reduced the phosphorylation levels of NF-κB p65 and STAT3 compared with the control and anti–PD-1 monotherapy groups, whereas total NF-κB p65 and total STAT3 levels were not obviously altered (Fig. 5B). Importantly, the reduction in p-p65 and p-STAT3 was accompanied by decreased expression of VEGFA and PD-L1, indicating that suppression of inflammatory transcriptional signaling was linked to attenuation of both angiogenic and immune-evasive effectors in vivo (Fig. 5B). We next measured intratumoral H₂S levels and CSE enzymatic activity to evaluate whether SPRC enhanced functional output of the CSE/H₂S axis in tumor tissues. Anti–PD-1 monotherapy did not significantly alter intratumoral H₂S levels or CSE activity compared with the control group (P > 0.05, Fig. 5C-D). In contrast, SPRC plus anti–PD-1 significantly increased both intratumoral H₂S production and CSE enzymatic activity compared with the control and anti–PD-1 groups (P < 0.001, Fig. 5C–D).

Fig. 5.

SPRC activates the intratumoral CSE/H₂S axis and suppresses NF-κB/STAT3–VEGFA/PD-L1 signaling during anti-PD-1 therapy. (A) Representative CD31 IHC images of tumor sections at 200× and 400× magnification, with semi-quantitative analysis of CD31 immunoreactivity across groups. (B) Representative western blot analysis of phosphorylated NF-κB p65, total NF-κB p65, phosphorylated STAT3, total STAT3, PD-L1, and VEGFA in tumor lysates from the indicated treatment groups. GAPDH served as the loading control. (C) Quantification of intratumoral H₂S levels in the indicated treatment groups. (D) Quantification of CSE enzymatic activity in tumor tissues from the indicated treatment groups. Abbreviations: SPRC, S-propargyl-cysteine; CSE, cystathionine-γ-lyase; H₂S, hydrogen sulfide; NF-κB, nuclear factor kappa B; STAT3, signal transducer and activator of transcription 3; VEGFA, vascular endothelial growth factor A; PD-L1, programmed death-ligand 1; CD31, platelet endothelial cell adhesion molecule-1; IHC, immunohistochemistry. Notes: *P < 0.05; **P < 0.01; ***P < 0.001

Metabolomic profiling indicates that SPRC reshapes intratumoral metabolic coordination, links to immune and vascular phenotypes, and alleviates oxidative stress

To characterize metabolic alterations associated with SPRC in the context of PD-1 blockade, we performed untargeted LC–MS metabolomics on tumor tissues and compared the SPRC + anti–PD-1 group with anti–PD-1 alone. Differential correlation analysis revealed a markedly reorganized metabolite–metabolite association structure, with a modular correlation pattern apparent on the heatmap and extensive positive and negative connections in the correlation network (Fig. 6A–B), indicating coordinated metabolic rewiring rather than isolated changes in individual metabolites. Pathway enrichment of differential metabolites further highlighted broad remodeling across amino-acid and transport/lipid-related programs. KEGG analysis prioritized pathways including lysine degradation, ABC transporters, central carbon metabolism in cancer, biosynthesis of amino acids, and alanine/aspartate/glutamate metabolism, together with lipid-associated pathways such as glycerophospholipid metabolism, ether lipid metabolism, and choline metabolism in cancer (Fig. 6C). Consistently, the pathway–metabolite network underscored multi-pathway connectivity and additionally implicated redox-relevant metabolism within the enriched landscape (Fig. 6D). MetPA further prioritized pathways with combined statistical significance and higher pathway impact (Fig. 6E). In parallel, MSEA ranked neuroactive ligand–receptor interaction and arginine and proline metabolism among the most enriched sets, along with multiple carbohydrate and lipid metabolic processes, supporting widespread metabolic reprogramming under combination therapy (Fig. 6F).

Fig. 6.

Tumor metabolomics reveals coordinated metabolic rewiring, links to immune and vascular phenotypes, and improved oxidative-stress status following SPRC treatment. (A) Differential metabolite correlation network comparing tumors from the SPRC plus anti-PD-1 group versus the anti-PD-1 group, illustrating altered metabolite–metabolite association structure. (B) Correlation heatmap of differential metabolites for SPRC plus anti-PD-1 versus anti-PD-1 alone. (C) KEGG pathway enrichment analysis of differential metabolites between SPRC plus anti-PD-1 and anti-PD-1 groups. (D) KEGG pathway–metabolite interaction network constructed from enriched pathways and mapped differential metabolites. (E) MetPA pathway topology plot showing pathway impact and statistical significance for enriched metabolic pathways. (F) MSEA enrichment overview of the top metabolite sets. (G) Integrative metabolite–phenotype correlation matrix incorporating intratumoral CD8⁺ T cells, Treg cells, and CD31, highlighting associations between treatment-associated metabolic remodeling and immune/vascular readouts. (H) Tumor MDA levels in the control and 50 mg/kg SPRC groups. (I) Tumor SOD activity in the control and 50 mg/kg SPRC groups. Abbreviations: SPRC, S-propargyl-cysteine; PD-1, programmed cell death protein 1; KEGG, Kyoto Encyclopedia of Genes and Genomes; MetPA, metabolomics pathway analysis; MSEA, metabolite set enrichment analysis; Treg, regulatory T cell; CD31, platelet endothelial cell adhesion molecule 1; MDA, malondialdehyde; SOD, superoxide dismutase. Notes: *P < 0.05; **P < 0.01

To better connect these metabolomic changes with therapeutic phenotypes, we performed an integrative correlation analysis incorporating intratumoral CD8⁺ T cells, Treg cells, and CD31 into the metabolite–phenotype matrix (Fig. 6G). This analysis identified an amino-acid-centered metabolic module, including methionine, glutamic acid, glutamine, ornithine, citrulline, valine, histidine, arginine, isoleucine, and phenylalanine, which showed strong positive intercorrelation and was generally positively associated with intratumoral CD8⁺ T-cell abundance, while showing an inverse relationship with Treg and CD31. By contrast, Treg and CD31 tended to cluster together and displayed the opposite correlation pattern relative to this amino-acid-enriched metabolite module. Among the correlated metabolites, the arginine–ornithine–citrulline axis was particularly notable because it showed a coherent relationship with both immune and vascular readouts. Functionally, these metabolic shifts were accompanied by an improved oxidative-stress profile in vivo. Compared with the control group, 50 mg/kg SPRC significantly reduced intratumoral MDA levels (P < 0.05; Fig. 6H) while significantly increasing SOD activity (P < 0.01; Fig. 6I). These findings are consistent with attenuated lipid peroxidation and enhanced antioxidant capacity.

Discussion

In this study, we demonstrated that the H₂S-donating small molecule SPRC remodels key components of the HCC TME and enhances antitumor efficacy when combined with PD-1 inhibitors. By integrating target engagement assays with endothelial and immune functional studies, epigenetic profiling, and tumor metabolomics, our study supports a convergent model in which SPRC simultaneously attenuates angiogenesis, relieves tumor-intrinsic immune inhibition, and reshapes metabolic and redox homeostasis. This multi-axis activity aligns with current therapeutic concepts that emphasize tumor microenvironment normalization as a prerequisite for effective immune-mediated tumor control.

A central mechanistic observation is that SPRC directly engages CSE with nanomolar binding affinity. Importantly, our results demonstrated that SPRC increased extracellular H₂S levels and CSE enzymatic activity in HUVECs, and both effects were substantially attenuated by CSE knockdown, indicating that the downstream anti-angiogenic and immunomodulatory phenotypes should be interpreted in the context of enhanced functional output of the CSE/H₂S axis rather than a binding event alone. Although CSE-dependent H₂S augmentation has been proposed as a core feature of SPRC pharmacology, quantitative target validation in the HCC context has been limited. Our binding data provide a molecular basis for CSE pathway activation and are supported by functional experiments showing that SPRC-mediated VEGFA suppression is attenuated by CSE silencing and partially restored by CSE re-expression. These findings are consistent with prior evidence that SPRC elevates endogenous H₂S and that its biological effects can be blunted by disrupting CSE activity [24]. They also reinforce the broader concept that H₂S signaling can yield antitumor outcomes when its level, kinetics, and cellular context are appropriately controlled [25, 26].

The role of H₂S in cancer remains context dependent, in part because sulfur metabolism is frequently dysregulated in tumors. Some cancers upregulate CSE or cystathionine beta synthase to support growth, angiogenesis, and bioenergetics, whereas exogenous H₂S donors can disrupt redox balance and induce stress responses that are unfavorable for tumor survival [12]. In HCC, slow releasing H₂S donors have been reported to suppress STAT3 activity and promote apoptosis, supporting a mechanism in which H₂S interferes with oncogenic survival pathways under defined conditions [27, 28]. In our study, the inhibition of STAT3 phosphorylation by SPRC is concordant with these observations and provides a mechanistic bridge between CSE engagement, downstream signaling repression, and reduced immune evasion.

A major downstream consequence of SPRC exposure is the concomitant inhibition of NF-κB and STAT3 signaling, which has direct implications for angiogenesis and immune suppression. These pathways are frequently co-activated in inflammation associated malignancies and cooperate to sustain expression of pro-tumorigenic mediators [8]. In HCC, NF-κB contributes to angiogenic programs that include VEGFA, while STAT3 regulates immunosuppressive outputs, including PD-L1 and cytokines such as IL-10 [29, 30]. Consistent with this framework, SPRC reduced phosphorylation of NF-κB p65 and STAT3 in endothelial cells and was accompanied by decreased VEGFA expression and impaired tube formation. Moreover, the rescue experiments strengthen causal inference in our system: constitutively active STAT3 restored PD-L1 expression despite SPRC treatment, whereas p65 overexpression substantially reversed the SPRC-mediated suppression of VEGFA expression and endothelial tube formation. Accordingly, our results support a model of largely STAT3-dependent PD-L1 regulation and NF-κB-dependent anti-angiogenic signaling in the presence of SPRC. Consistent with previous studies indicating that STAT3 and NF-κB functionally cooperate and engage in signaling crosstalk in cancer, our findings support a similar signaling relationship in this system [31, 32]. These findings also extend prior reports that H₂S-related signaling can interfere with NF-κB cascades [17, 33] and support a model in which SPRC disrupts inflammatory signaling that otherwise reinforces a proangiogenic microenvironment.

Our epigenetic analyses further suggest that SPRC influences transcriptional regulation at the level of promoter-associated chromatin. We observed increased H3K27me3 enrichment at the RELA and STAT3 promoter regions after SPRC treatment, indicating promoter-associated chromatin remodeling at these loci. Because H3K27me3 is a canonical Polycomb-associated repressive chromatin mark, increased H3K27me3 enrichment at the RELA and STAT3 promoter regions is mechanistically compatible with transcriptionally repressive promoter states [34]. Accordingly, these data should be interpreted as evidence for an additional epigenetic layer that may contribute to restraint of NF-κB/STAT3 signaling, rather than as merely the inverse of the phosphorylation data. Additional studies will be needed to define the upstream chromatin modifiers involved and to clarify whether these chromatin changes contribute to sustained pathway suppression.

Beyond vascular signaling, SPRC exerted immunologically relevant effects at the tumor cell interface. Tumor intrinsic PD-L1 is a major determinant of T cell dysfunction, and in HCC it can be induced by inflammatory cytokines and oncogenic signaling circuits that include STAT3-related pathways [29, 30]. Using an activated Jurkat–HepG2 co-culture system, we found that SPRC reduced PD-L1 expression on tumor cells and shifted the T-cell compartment toward a more cytotoxic phenotype, as reflected by increased CD8α-associated features, reduced FoxP3-associated regulatory characteristics, and elevated granzyme B release. Together with the similarity between SPRC treatment and pharmacologic STAT3 inhibition, as well as the STAT3C rescue data, these findings support the interpretation that SPRC-mediated PD-L1 downregulation is largely STAT3 dependent. Functionally, the increase in granzyme B indicates that the immune effect is not limited to marker changes, but is accompanied by enhanced cytotoxic activity. Collectively, these results support a model in which SPRC mitigates a tumor intrinsic inhibitory checkpoint program and thereby facilitates productive T cell effector function.

The in vivo findings provide translational relevance by demonstrating that SPRC can act as an immune adjuvant to PD-1 blockade. Combination therapy achieved stronger tumor growth inhibition than PD-1 monotherapy and was accompanied by increased intratumoral CD8 positive T cells and reduced Tregs. In parallel, CD31-indexed vascular density was decreased most prominently in tumors receiving SPRC plus anti–PD-1. Importantly, the tumor-tissue analyses further extend these phenotypic observations by directly validating the proposed CSE/H₂S–NF-κB/STAT3–VEGFA/PD-L1 axis in vivo. SPRC plus anti–PD-1 significantly increased intratumoral H₂S levels and CSE enzymatic activity, whereas anti–PD-1 alone did not substantially affect either parameter, indicating that activation of the CSE/H₂S pathway was primarily associated with the SPRC-containing regimen. This was accompanied by reduced phosphorylation of NF-κB p65 and STAT3, with no obvious changes in their total protein levels, supporting pathway inhibition at the activation level rather than nonspecific loss of pathway components. The concomitant downregulation of VEGFA and PD-L1 in tumor tissues provides a mechanistic bridge between CSE/H₂S activation and the observed vascular and immune remodeling. These findings are consistent with the clinical principle that improving the vascular and immune contexture can enhance the effectiveness of checkpoint blockade in HCC, as established by antiangiogenic and ICI combinations in patients [35]. Importantly, SPRC differs conceptually from direct VEGF neutralization. Rather than acting as a single axis anti-VEGF agent, SPRC appears to couple modulation of angiogenic signaling with suppression of tumor cell PD-L1 and broader stress response remodeling, which could widen the therapeutic leverage across multiple resistance nodes within the TME.

Our metabolomic analysis provides an additional mechanistic layer by linking SPRC to coordinated metabolic rewiring and improved redox status in vivo. In the integrative correlation analysis, we incorporated CD8-positive T cells, Treg cells, and CD31 into the metabolite–phenotype matrix, which provided a more direct link between treatment-associated metabolic remodeling and the biological phenotypes observed in earlier sections. An amino-acid-centered metabolic module, including methionine, glutamic acid, glutamine, ornithine, citrulline, valine, histidine, arginine, isoleucine, and phenylalanine, was positively associated with intratumoral CD8-positive T-cell abundance and inversely associated with Treg abundance and CD31. Among these metabolites, the arginine–ornithine–citrulline axis was particularly notable because it showed a coherent relationship with both immune and vascular readouts. These associations are biologically plausible because arginine metabolism is a central determinant of T-cell metabolic fitness and myeloid immunosuppression, whereas endothelial metabolism is increasingly recognized as an active driver of angiogenesis rather than a passive bystander [36, 37]. More broadly, rather than isolated metabolite shifts, we observed coordinated reorganization of metabolite-correlation structure and pathway-level enrichment across amino-acid, lipid, and transporter-related programs. The reduction in MDA together with the increase in SOD activity further supports a shift toward a less oxidatively stressed tumor milieu, which may favor preservation of antitumor immune function [38, 39]. Nevertheless, these metabolomic findings should be interpreted as mechanistically plausible associations rather than definitive causal drivers of SPRC efficacy, and further targeted validation will be required to determine which metabolites are functionally upstream of the therapeutic effect.

From a translational standpoint, SPRC represents an attractive candidate for combination strategies because it is a small molecule with a multifaceted mechanism that targets vascular signaling, immune inhibition, and metabolic stress responses. Such pleiotropic activity may help overcome compensatory adaptations that frequently limit the durability of monotherapy in HCC. More broadly, our results support the feasibility of using a pharmacologic H₂S-donor framework to reprogram the TME and improve the therapeutic index of immunotherapy.

Several limitations should be considered. First, the present in vivo design did not include an SPRC-alone treatment group. Future studies using a complete four-arm design, including vehicle control, SPRC alone, anti-PD-1 alone, and SPRC plus anti-PD-1, together with formal interaction modeling, will be necessary to define the additive or synergistic nature of the combination. Second, because H₂S biology is highly context dependent, it will be important to relate treatment response to baseline sulfur metabolism, including CSE and CBS expression and tumor oxygenation status. Third, deeper immune profiling is warranted to determine how SPRC affects myeloid subsets, antigen presentation programs, and exhaustion associated transcriptional states within CD8 positive T cells. Fourth, validation across additional HCC models with distinct molecular and phenotypic features will be essential to establish robustness and generalizability. Finally, the in vivo study was designed as a terminal mechanistic experiment and therefore did not include survival analysis. Despite these limitations, the concordance between the in vitro co-culture findings and the in vivo efficacy of the combination treatment supports a coherent mechanism whereby SPRC alleviates both vascular- and checkpoint-mediated constraints on antitumor immunity.

Conclusion

SPRC engages a functional CSE/H₂S-linked program that suppresses NF-κB p65 and STAT3 phosphorylation, couples antiangiogenic activity with reduced PD-L1-associated immune inhibition, and enhances cytotoxic immunity. SPRC augments PD-1 blockade in HCC with coordinated immune, vascular, metabolic, and redox remodeling, supporting its development as a promising adjunct strategy to overcome the vascular-immune barrier.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 1 ^{(2MB, docx)}

Author contribution

Y.L., D.Z., and H.Y. were responsible for study conceptualization, data analysis, drafting and revision of the manuscript. Q.Y. and Y.Z. were responsible for study conceptualization, validation, and revision of the manuscript. Y.S., Y.C., and M.W. were responsible for validation and revision of the manuscript. All authors contributed to the article and approved the submitted version.

Funding

This study was supported by Research Project of Science and Technology Department of Sichuan Province (2024YFHZ0358).

Data availability

The data supporting the findings of our study are available from the corresponding author under reasonable requirements.

Declarations

Ethics approval and consent to participate

The animal study protocol was approved by the Ethics Committee of Sichuan Cancer Hospital (KY-2025-020). All animal experiments were conducted in accordance with institutional guidelines for the care and use of laboratory animals.

Consent to participate

Not applicable.

Conflict of interest

The authors declare no potential conflicts of interest.