Chemotherapy-derived DAMPs drive reprogramming of tumor-associated macrophages toward a pro-inflammatory phenotype in hepatocellular carcinoma

Abstract

Hepatocellular carcinoma (HCC) remains a global health challenge, with treatment efficacy often compromised by the immunosuppressive tumor microenvironment (TME), particularly tumor-associated macrophages (TAMs). While chemotherapy induces immunogenic cell death (ICD) and the release of Damage-Associated Molecular Patterns (DAMPs), the specific impact of these molecules on macrophage plasticity and the underlying signaling mechanisms remain poorly understood. This study investigated the immunomodulatory effects of DAMPs released from HepG2 cells treated with standard chemotherapeutic agents, sorafenib and oxaliplatin. Macrophage polarization was assessed using murine RAW 264.7 and human monocyte-derived macrophages. The current study utilized flow cytometry, multiplex cytokine profiling, immunoblotting, and proteomic analysis to characterize phenotypic shifts and signaling pathways. Both Sorafenib and oxaliplatin induced significant cytotoxicity and the release of HMGB1 and S100A9, while sorafenib specifically triggered HSP90 release. Exposure to these chemotherapy-derived DAMPs effectively reprogrammed IL-4 stimulated M2 macrophages toward an M1-like phenotype, evidenced by downregulation of CD163/CD206, and upregulation of CD86. Mechanistically, this reprogramming was driven by a distinct, non-canonical signaling axis involving the upregulation of p-ERK and the NLRP3 inflammasome, occurring despite the downregulation of the classical NF-κB pathway. Proteomic analysis further revealed a metabolic shift toward cellular stress, ER stress and transcriptional reprogramming. Although the cytokine profile was predominantly pro-inflammatory (TNF-α, IL-1β), the concurrent secretion of IL-10 suggests a complex, mixed activation state. By overriding M2 immunosuppression via an ERK-NLRP3-dependent pathway, sorafenib-derived DAMPs act as an immunological primer to convert cold to hot tumors, providing a molecular rationale for chemo-immunotherapy combinations.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-47480-z.

Introduction

Hepatocellular carcinoma (HCC) remains a major global health challenge, characterized by high incidence and diagnosis often occurring at advanced, metastatic stages. While systemic chemotherapy either alone or in combination with immune checkpoint inhibitors is recommended for advanced HCC, its efficacy is often compromised by the tumor microenvironment (TME). Therapeutic induction of cancer cell injury leads to the systemic release of Damage-Associated Molecular Patterns (DAMPs). These molecules, which normally function in immune induction, debris clearance, and tissue repair, are known consequences of chemotherapy-induced cell stress and death caused by agents such as cisplatin, oxaliplatin, doxorubicin, and paclitaxel^1–4. Specifically, the release of proteins like high mobility group box 1 (HMGB1) and S100A8/A9 is a critical event. These DAMPs act as ligands that engage key pattern recognition receptors, particularly the Receptor for Advanced Glycation End products (RAGE) and Toll-like Receptor 4 (TLR4), which are expressed on both cancer cells and immune infiltrates. This ligand-receptor interaction triggers the activation of critical downstream signal transduction pathways, principally the NF-κB pathway, the Mitogen-Activated Protein Kinase (MAPK) pathways, and the pro-survival PI3K/Akt pathway⁵. The activation of NF-κB subsequently fosters a chronic, sterile inflammatory microenvironment by inducing the expression of cytokines and chemokines, while the activation of PI3K/Akt confers potent survival advantages and promotes chemoresistance by upregulating anti-apoptotic proteins⁶. Collectively, this cascade drives cellular proliferation and facilitates metastatic dissemination, establishing a robust pathogenic feedback loop wherein therapeutic intervention inadvertently promotes tumor progression and resistance.

A major limitation in treating solid tumors is the significant contribution of tumor-associated macrophages (TAMs), which possess potent pro-tumor conditioning abilities, enhancing tumor growth, metastasis, and vascular formation. Current strategies to target TAMs involve either depletion by inhibiting the CSF-1/CSF-1R signaling pathway with agents such as Pexidartinib⁷ or reprogramming to an anti-tumoral, M1-like phenotype using agents like CD40 agonists, or restoring their phagocytic capabilities, for example, by targeting the CD47-SIRPα axis with antibodies like Magrolimab⁸. However, the successful clinical application of these TAM-targeting therapies is greatly challenged by the immense plasticity and heterogeneity of TAMs. The simplistic M1/M2 polarization model often fails to capture the complexity of TAM states, and the TME is known to rapidly induce resistance to reprogramming efforts⁹. This evidence suggests that effective modulation of TAMs is complicated and requires intensive mechanistic study.

Given the established role of chemotherapy-induced cell injury in triggering DAMP release, and the critical regulatory function of these molecules within the TME, understanding their impact is vital. Previous research has primarily focused on intracellular HMGB1, which is highly expressed in tumors and TAMs, enhancing M1 polarization through the RAGE/NF-κB/NLRP3 pathway¹⁰. Furthermore, recombinant human (rh) HMGB1 has been shown to promote M1 polarization through IL-6 secretion via NF-κB and p38 MAPK signaling. Despite this, there is a paucity of data regarding chemotherapy-induced DAMPs in the tumor mass. We hypothesize that these DAMPs represent a key, unexamined mechanistic link between chemotherapy-induced stress and detrimental TAM polarization. Here, we investigated the effects of DAMPs derived from chemotherapeutic agents (sorafenib and oxaliplatin) on macrophage function. We demonstrate that while chemotherapy-associated DAMPs promote an M1-like phenotype, these macrophages paradoxically exhibit distinct, pro-tumoral functions. These findings may help to elucidate the rational for the limited efficacy or inconsistent results observed in clinical trials of TAM-targeted therapies.

Materials and methods

Cell culture and chemotherapeutics reagents

Human hepatocellular carcinoma and mouse macrophage cell lines, HepG2 and RAW 264.7 cells were obtained from the American Type Culture Collection (VA, USA) and maintained in Dulbecco’s Modified Eagle Medium (Gibco, Thermo Fisher Scientific, New York, NY, USA) enriched with 10% fetal bovine serum (FBS) and 1% antibiotic–antimycotic (10,000 U/mL of penicillin, 10,000 µg/mL of streptomycin, and 25 µg/mL of Gibco Amphotericin B) (Gibco, Thermo Fisher Scientific). The cells were cultured at 37 °C in an atmosphere of 5% CO₂ and 95% air.

Sorafenib and oxaliplatin were purchased from MedChemExpress (NJ, USA). The lyophilized was restituted using DMSO and deionized water, respectively for preparation as stock concentration.

Determination of cell cytotoxicity

HepG2 cells were seeded at density 5 × 10³ cell/well into 96 well cell culture plate and incubated overnight for cell adhesion. Sorafenib and oxaliplatin were prepared as working concentration in DMEM containing 1% of FBS and subsequently added into cell culture condition. The maximum concentration containing solvent either DMSO or water were used as control conditions. After incubation for 24 and 48 h, cell culture medium was removed and replaced with MTT reagent for determination for cell viability and calculated as % cell viability and inhibition concentration at 50% cell (IC50).

Preparation of chemotherapy-derived tumor conditioned medium

HepG2 cells were seeded into 90-mm culture dishes at a density of 3 × 10⁷ cells per dish. Upon achieving adherence, the cells were treated with chemotherapeutic agents at their respective half-maximal inhibitory concentrations (IC50) for 24 h. DMEM supplemented with DMSO or water served as the vehicle control. Following treatment, the medium was discarded, and the monolayers were washed three times with phosphate-buffered saline (PBS) to remove residual drugs. Fresh complete medium was added, and the cells were incubated for an additional 24 h. The culture medium was then harvested and centrifuged at 6000×g for 10 min. The resulting supernatants designated as control, oxaliplatin-derived, or sorafenib-derived tumor-conditioned medium were collected for the quantification of candidate DAMPs and macrophage stimulation assays.

Determination of HMGB1, HSP90 and S100A9

The level of HMGB1, HSP90 and S100A9 were determined using ELISA kit from ElabScience (TX, USA). The assay was performed according to the manufacturer’s protocol. Briefly, conditioned medium, standards, and blanks (100 µL/well) were added to the ELISA plate and incubated for 90 min at 37 °C. Following liquid removal, 100 µL of Biotinylated Detection Antibody working solution was added to each well and incubated for 1 h at 37 °C. The wells were subsequently washed three times with 350 µL of wash buffer. Next, 100 µL of HRP Conjugate working solution was applied for a 30-min incubation at 37 °C. After washing the plate five times, color development was initiated by adding 90 µL of Substrate Reagent and incubating for 15 min at 37 °C in the dark. The reaction was terminated with 50 µL of Stop Solution, and the optical density (OD) was immediately measured at 450 nm using a microplate reader.

Isolation and preparation of human monocyte-derived macrophages

Peripheral blood monocytes were isolated from healthy donors using Ficoll-Hypaque and Percoll density gradient centrifugation (GE Healthcare, Freiburg, Germany). The study was conducted with informed consent and approved by the Ethics Committee of Naresuan University (Protocol No. P1-0116/2565). The monocyte fraction was resuspended in RPMI 1640 medium supplemented with 10% autologous serum and 2 mM L-glutamine (Gibco, Thermo Fisher Scientific, MA, USA) and plated for 2 h to allow adherence. Non-adherent cells were removed by washing three times with PBS. Subsequently, the adherent cells were differentiated in complete RPMI 1640 medium containing 40 ng/mL recombinant human macrophage colony-stimulating factor (M-CSF; Miltenyi Biotec, Bergisch Gladbach, Germany) for 7 days. Upon harvesting, the cells were analyzed for CD16, which were observed in more than 80% of the total population.

Flow cytometry

Human monocyte-derived macrophages were seeded into 6-well plates at a density of 3 × 10⁶ cells/well and allowed to adhere overnight. The cells were subsequently treated for 24 h with tumor-conditioned medium and oxaliplatin-, and sorafenib-derived conditioned medium, supplemented with 40 ng/mL recombinant human interleukin-4 (IL-4; ImmunoTools, Friesoythe, Germany). The conditioned medium was diluted prior treatment with fresh RPMI1640 medium at 1:1 ratio. Following incubation, cells were harvested and washed with PBS containing 3% FBS. Staining was performed for 45 min using the following fluorophore-conjugated antibodies: anti-CD16-PE and anti-CD86-PE (ImmunoTools, Friesoythe, Germany); and anti-CD14-FITC, anti-CD163-FITC, and anti-CD206-PE (Abcam, Cambridge, UK). After a final wash, the cells were resuspended in FACS buffer (PBS supplemented with 10% FBS) and analyzed on a Cytoflex FC 500 flow cytometer using CXP software (Beckman Coulter, IN, USA).

Immunoblotting analysis

RAW 264.7 macrophages were seeded into 6-well cell culture plates at a density of 3 × 10⁶ cells/well and incubated overnight to allow for cell attachment. Subsequently, the cells were treated with conditioned medium derived from either sorafenib-treated or untreated tumor cells. Experimental controls included cells incubated with medium alone (control), medium supplemented with IL-4 (50 ng/mL), or medium supplemented with a combination of IFN-γ (20 ng/mL) and LPS (10 ng/mL). All cultures were incubated for 24 h.

Following treatment, cells were washed with ice-cold PBS and lysed in NP40 buffer supplemented with a protease inhibitor cocktail. Lysates were cleared by centrifugation, and protein concentrations were quantified using the Pierce™ BCA Protein Assay Kit (Thermo Scientific, MA, USA). Proteins were resolved by SDS-PAGE and transferred onto PVDF membranes. The membranes were blocked with 5% skim milk and incubated overnight at 4 °C with gentle agitation using primary antibodies against pAkt, Akt, pERK, and ERK (Cell Signaling Technology, MA, USA), as well as NLRP3 and p65 NF-κB (ABclonal, MA, USA). Subsequently, membranes were washed and incubated with HRP-linked anti-rabbit secondary antibody (Cell Signaling Technology, MA, USA) for 1 h at room temperature. Protein bands were visualized using the ChemiDoc™ XRS system and analyzed with Image Lab software (Bio-Rad, CA, USA).

RNA extraction and real-time polymerase chain reaction (PCR)

Total RNA was isolated from macrophages under the indicated experimental conditions using the Total RNA Mini Kit (Bio-Rad, CA, USA) in accordance with the manufacturer’s protocol. First-strand cDNA was synthesized utilizing the Tetro cDNA Synthesis Kit (Bioline, UK). Subsequently, real-time PCR was performed using the specific primer sequences listed in Table S1. Relative gene expression levels were calculated using the 2 − ΔΔCT method, with ACTB serving as the internal normalization control.

Proteomics analysis

RAW 264.7 macrophages were cultured in present of IL-4 with conditioned medium from tumor cells that were either treated with sorafenib or untreated for 24 h. Cell pellets were lysed in 6 M urea and 50 mM ammonium bicarbonate, then homogenized with a bead-based (SWE-FP, ServiceBio). Lysates were purified using centrifugation, and proteins were precipitated using ice-cold methanol, subsequently collected and resuspended in 50 mM ammonium bicarbonate. Protein concentrations were measured using the Bradford assay and standardized to 0.33 mg/mL.

Samples underwent proteolysis through reduction with 10 mM DTT at 65 °C for 30 min, followed by alkylation with 25 mM iodoacetamide at room temperature in the dark, and were subsequently digested overnight at 37 °C using sequencing-grade trypsin. Digestions were neutralized with formic acid, clarified, and transported to LC–MS vials. Forty microliters of each digest were examined on LC-MS/MS (Agilent 6545XT AdvanceBio LC–QTOF) utilizing a Peptide Mapping column (2.1 × 150 mm, 2.7 μm; 60 °C). Peptides were fractionated across an 85-minute gradient at a flow rate of 0.4 mL/min and analyzed in positive ion mode (m/z 100–1700) utilizing data-dependent MS/MS and continuous reference mass calibration (m/z 922.0098).

Raw data were transformed into mzXML format and analyzed in MaxQuant (v2.6.3) utilizing the UniProt Mus musculus proteome, with carbamidomethylation as a fixed modification and methionine oxidation/N-terminal acetylation as variable modifications. Searches employed Trypsin/P criteria, permitted ≤ 2 missed cleavages, and maintained a 1% false discovery rate (FDR). Processed data were examined in MetaboAnalyst 6.0 for transformation, imputation, PCA, grouping, and statistical analysis. Processed datasets were loaded into MetaboAnalyst 6.0 for statistical analysis, encompassing data filtration, logarithmic transformation, imputation of missing values, principal component analysis (PCA), hierarchical clustering, fold-change evaluation, and univariate hypothesis testing.

Statistical analysis

Statistical analyses were conducted using GraphPad Prism software (version 10.4.0; GraphPad Software, Inc., La Jolla, CA, USA). Data are presented as the mean ± Standard Error of the Mean (SEM). Comparisons between two groups were analyzed using unpaired Student’s t-tests, while comparisons among multiple groups were performed using analysis of variance (ANOVA). A р-value of < 0.01 was considered statistically significant.

Results

Sorafenib and oxaliplatin induce cell death and the release of HMGB1 and S100A9

To mimic the release of chemotherapy-derived DAMPs from cancer cells, HepG2 cells were incubated with various concentrations of chemotherapeutic drugs, and cell viability was determined at 24 and 48 h. Both chemotherapeutic agents induced HCC cell death in a dose- and time-dependent manner (Fig. 1A). Sorafenib exhibited stronger cytotoxicity than oxaliplatin, as reflected by the calculated IC50 shown in Fig. 1B. At the concentration causing 50% cytotoxicity (IC50), cell supernatants were collected and measured for the content of candidate DAMPs. The results showed that HMGB1 levels were significantly elevated (2- to 3-fold) in the sorafenib and oxaliplatin treatment groups compared to the control group. Similar to HMGB1, S100A9 levels were significantly higher following treatment with both drugs. In contrast, sorafenib induced HSP90 release, whereas oxaliplatin did not. These results indicate that sorafenib and oxaliplatin potentially induce cell cytotoxicity and subsequently promote DAMP release.

Fig. 1.

Sorafenib and oxaliplatin induce cytotoxicity and DAMP release in HepG2 cells. (A) Cell viability of HepG2 cells treated with varying concentrations of sorafenib and oxaliplatin for 24 and 48 h, assessed by MTT assay. Data indicate a dose- and time-dependent induction of cell death. (B) Comparative IC50 values for sorafenib and oxaliplatin. (C) Quantification of DAMPs (HMGB1, S100A9, and HSP90) in the supernatant of HepG2 cells treated at IC50 concentrations, measured by ELISA. Both agents significantly increased HMGB1 and S100A9 levels, while HSP90 release was specific to sorafenib treatment. Data are presented as mean ± SEM (*p < 0.05, **p < 0.01).

Sorafenib- and oxaliplatin-derived DAMPs promote M1-like macrophage polarization

Next, we investigated whether chemotherapy-induced DAMPs release in the tumor affects M2 macrophages, which are normally present in the tumor mass. Human monocyte-derived macrophages were cultured in either control tumor medium or conditioned medium from sorafenib/oxaliplatin-treated cells in the presence of IL-4 to induce differentiation into M2 macrophages. The phenotypes were assessed by measuring the expression of CD markers after a 24-h incubation. The results demonstrated that the tumor medium control and the tumor medium control with IL-4 exhibited high expression of CD163 and CD206, which are markers of M2 macrophages. Interestingly, conditioned medium derived from sorafenib and oxaliplatin treatments downregulated the expression of CD163 and CD206. Moreover, these conditions significantly enhanced the expression of CD86, a costimulatory molecule of M1 macrophages, when compared to the tumor medium control with IL-4 supplement (Fig. 2). This suggests that chemotherapy drugs induce DAMP release, which is associated with the reprogramming of M2 macrophages into M1-like macrophages.

Fig. 2.

Chemotherapy-derived DAMPs drive repolarization of M2 macrophages toward an M1-like phenotype. Human monocyte-derived macrophages were differentiated into M2 phenotype (IL-4 stimulation) and exposed to conditioned medium (CM) from sorafenib- or oxaliplatin-treated HepG2 cells for 24 h. Flow cytometric analysis shows the expression of (A, B) M2 markers CD163 and CD206, which were significantly downregulated by chemotherapy-derived CM, and the M1 costimulatory marker CD86, which was significantly upregulated compared to controls. Data are presented as mean ± SEM (*p < 0.05, **p < 0.01, ***p < 0.001).

Chemotherapy-derived DAMPs promote macrophage polarization toward an M1 phenotype

Cytokines and chemokines play a critical role in regulating macrophage differentiation, polarization, and antitumor activity. We investigated the secretion profiles of human monocyte-derived macrophages exposed to IL-4 with or without chemotherapy-induced DAMPs conditioned media. Quantitative analysis was performed using a multiplex ELISA.

Incubation with oxaliplatin- and sorafenib-derived media significantly upregulated M1-associated cytokines, including IFN-γ, GM-CSF, TNF-α, IL-1β, IL-6, IL-12, IP-10, IL-8, and MIP-1α; most exhibited a greater than 2-fold increase compared to the control tumor medium. Conversely, chemotherapy-induced DAMPs elicited only a modest response in M2-associated cytokines (IL-1RA and PDGF-BB), though increased production of IL-13, IL-10, FGF-basic, and eotaxin was observed (Fig. 3). These alterations were attributed specifically to chemotherapy-derived DAMPs, as confirmed by the baseline cytokine levels observed in the control and control with IL-4 treatment. Collectively, these findings suggest that chemotherapeutic DAMPs predominantly drive macrophage polarization toward an M1 phenotype, while simultaneously exhibiting minor M2 characteristics.

Fig. 3.

Cytokine secretion profiles reveal a mixed pro-inflammatory phenotype induced by chemotherapy-derived DAMPs. Multiplex ELISA analysis of cytokine and chemokine levels in the supernatant of macrophages treated with oxaliplatin- or sorafenib-derived conditioned medium. (A) Heat map analysis displays a significant upregulation of M1-associated pro-inflammatory cytokines (TNF-α, IL-1β, IL-6, IL-12) and chemokines (IP-10, MIP-1α), alongside a concurrent but modest increase in M2-associated cytokines (IL-10, IL-13). (B) Quantification of selected cytokines analyzed using one-way ANOVA. Data are presented as mean ± SEM (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

Sorafenib-derived tumor conditioned medium induced distinct the molecular pathways in macrophages

To investigate the molecular mechanisms by which chemotherapeutic-induced DAMPs affect macrophages, we performed mechanistic studies. Raw 264.7 cells were treated with either control tumor medium or sorafenib-derived tumor conditioned medium, in the presence or absence of IL-4 or IFN-γ/LPS supplementation. Morphological analysis revealed distinct differences between the treatment groups. Cells treated with control tumor medium exhibited a spindle-shaped (elongated) morphology, which was most prominent in the control and IL-4-treated groups but less pronounced in the IFN-γ/LPS group. In contrast, cells exposed to sorafenib-derived conditioned medium displayed a flattened, round morphology across all groups (Fig. 4A).

Fig. 4.

Sorafenib-derived DAMPs induce an M1-like macrophage phenotype and activate the ERK and NLRP3 signaling pathways. (A) Representative images showing morphological changes in RAW 264.7 cells. Sorafenib-derived conditioned medium induced a transition from a spindle-shaped (M2-like) (Red arrowhead) to a flattened, round (M1-like) (Yellow arrowhead) morphology. Scale bar = 20 μm. (B) Relative mRNA expression of pro-inflammatory genes (TNF-α, IL-6, IL-1β, iNOS) determined by RT-qPCR. Data are presented as mean ± SEM (*p < 0.05, **p < 0.01, ****p < 0.0001). (C) Immunoblot analysis of key signaling proteins. Densitometric quantification (relative to β-actin) confirms the upregulation of p-ERK and NLRP3, and downregulation of p65-NF-κB.

Gene expression patterns aligned with the cytokine profiling results. The expression levels of TNF-α, IL-6, IL-1β, and iNOS genes were higher in the sorafenib-derived conditioned medium group compared to the control medium group. Specifically, sorafenib-derived conditioned medium significantly enhanced the expression of TNF-α and iNOS in the absence of additional stimuli. Moreover, it potentiated the expression of IL-1β and iNOS in cells treated with IFN-γ and LPS (Fig. 4B). In addition, immunoblotting revealed that treatment with sorafenib-derived conditioned medium increased the expression of NLRP3 and p-ERK, while downregulating p65-NF-κB (Fig. 4C). Taken together, these results suggest that sorafenib-derived conditioned medium induces an M1-like phenotype via inflammasome induction, despite the downregulation of the NF-κB signaling pathway.

To further delineate the molecular mechanisms driving the inflammatory phenotype in macrophages, PD98059 (a specific inhibitor of MAPK/ERK kinase) was introduced into the treatment conditions. Subsequently, the expression of key signaling proteins and selected cytokine genes was evaluated using immunoblotting and RT-qPCR. The results demonstrated that PD98059 effectively abolished p-ERK expression, which coincided with the suppression of NLRP3 in cells exposed to sorafenib-derived conditioned medium supplemented with LPS and IFN-γ. Conversely, the expression of p65-NF-κB remained elevated despite the inhibitor. Furthermore, the inhibition of ERK signaling reversed the upregulation of pro-inflammatory cytokine genes, including TNF-α, IL-6, and IL-1β (Fig. 5). These findings confirm that sorafenib-derived DAMPs override M2 immunosuppression primarily via an ERK-NLRP3-dependent pathway.

Fig. 5.

Sorafenib-derived DAMPs drive pro-inflammatory macrophage reprogramming via an ERK-NLRP3-dependent pathway. RAW 264.7 macrophages were treated with either control medium or sorafenib-derived conditioned medium, supplemented with LPS and IFN-γ, in the presence or absence of PD98059. (A) Representative immunoblots and corresponding densitometric analysis of key signaling proteins, including p-ERK, NLRP3, and p65-NF-κB. (B) Relative mRNA expression levels of pro-inflammatory cytokines (TNF-α, IL-6, and IL-1β) were determined by RT-qPCR. Data are presented as mean ± SEM (**p < 0.01, ***p < 0.001, ****p < 0.0001.

The proteomics analysis of macrophages treated with sorafenib-derived condition medium

To investigate the systemic alterations in macrophages exposed to chemotherapy-induced conditions, we employed a proteomic approach to identify upregulated and downregulated proteins. We examined RAW 264.7 cells treated with either control medium or sorafenib-derived conditioned medium in the presence of IL-4, mimicking a tumor-suppressive microenvironment. Proteomic analysis was performed to elucidate the differential protein expression profiles between treatments. Principal Component Analysis (PCA) analysis revealed distinct separation between the control medium and sorafenib-derived conditioned medium groups (Fig. 6A). Among the 149 identified protein groups, 19 exhibited significant differences (p < 0.1) (Fig. 6B). The control group was characterized by higher expression levels of nucleolar and ribosomal proteins including histones, nucleophosmin, ribonucleoproteins, and ribosomal subunits as well as the cell surface marker CD44. Additionally, mitochondrial stress-related proteins, specifically mitochondrial stress-70 protein and mitochondrial heat shock protein, were upregulated in the control group. Conversely, the sorafenib-derived condition medium group exhibited significantly higher levels of endoplasmin and calcium-binding proteins, such as calumenin and EF-hand domain-containing protein D2. Furthermore, DNA-binding proteins, Y-box-binding protein, showed a markedly up-regulation in this group (Fig. 6C-D).

Fig. 6.

Proteomic profiling reveals metabolic and stress-response reprogramming in macrophages treated with sorafenib-derived DAMPs. (A) PCA demonstrates distinct clustering between macrophages treated with control tumor-conditioned medium and those exposed to sorafenib-derived DAMPs, indicating global proteomic divergence. (B) Differential expression analysis identifies 19 significantly altered protein groups (p < 0.1) out of 149 detected proteins. (C) Representative differentially expressed proteins are shown as box plots comparing control (red) and sorafenib-derived (green) conditioned media. These include proteins enriched in the control group, such as nucleolar and ribosomal components (e.g., histones, ribonucleoproteins) and mitochondrial stress-related proteins (Stress-70 and mitochondrial HSPs), as well as proteins elevated in the sorafenib-derived group, including endoplasmin, calcium-binding proteins (Calumenin, EF-hand D2), and DNA-binding proteins (YBX1). (D) Hierarchical heatmap clustering of significant proteins highlights coordinated shifts in protein expression between Class A (control) and Class B (sorafenib-derived) groups.

Discussion

In this study, investigated the immunological consequences of chemotherapy-induced HCC cell death, specifically focusing on the modulation of TAMs. The major findings confirm that the chemotherapeutic agents sorafenib and oxaliplatin induce significant cytotoxicity in HepG2 cells, resulting in the release of DAMPs, including HMGB1, S100A9, and HSP90. These chemotherapy-derived DAMPs effectively reprogram immunosuppressive M2-like macrophages toward a pro-inflammatory M1-like phenotype via a distinct signaling axis involving ERK activation and the NLRP3 inflammasome, independent of the classical NF-κB pathway. Moreover, proteomic data reveals that sorafenib-derived signals drive a major functional shift in macrophages. The cells transition from a metabolic state focused on growth (M2) to an active, pro-inflammatory state. This reprogramming is fueled by ER stress and calcium signaling, which prepares the cellular machinery (NLRP3) to release cytokines.

Sorafenib and oxaliplatin are the targeted and chemotherapeutic agents, which are recommended to use to treatment in HCC patients with advanced stage disease. Especially for sorafenib, it is the first line drug in case of not suitable for immune checkpoint inhibitor treatment¹¹. Sorafenib inhibit RAF/MEK/ERK signaling pathway, inhibits tumor angiogenesis, and induces HCC cell apoptosis, as well as oxaliplatin that cause cell apoptosis mediated by induction of cell cycle arrest^12,13. It has been reported that treatment of sorafenib induced release of HMGB1, ATP and exposure of CRT, which play crucial role in induction of immunogenic cell death by activation of ferroptosis in HCC cell¹⁴. Moreover, oxaliplatin can induce HMGB2, the HMG family translocation and secretion by nuclear exporter XPO1¹⁵. Taking together with our findings, we observed that both sorafenib and oxaliplatin effectively triggered the release of HMGB1 and S100A9. Interestingly, HSP90 release was specific to sorafenib treatment, suggesting drug-specific mechanisms of stress response or membrane permeabilization. This data confirms that standard-of-care agents for HCC do not act solely through direct cytotoxicity but also by altering the extracellular milieu via DAMP accumulation.

The current study further demonstrated that these chemotherapy-derived DAMPs possess the capacity to reprogram IL-4-stimulated M2 macrophages toward an M1-like phenotype. We observed a significant downregulation of the scavenger receptor CD163 and the mannose receptor CD206, alongside a marked upregulation of the costimulatory molecule CD86. However, our cytokine profiling revealed a complex mixed phenotype where immunosuppressive factors such as IL-10 persist, potentially explaining the paradoxical pro-tumoral functions often observed despite therapeutic intervention. Our findings align with established models of immunogenic cell death, where therapeutic stress triggers the release of intracellular molecules into the tumor microenvironment. These molecules act as ligands for pattern recognition receptors like TLR4 and RAGE, which initiates the downstream signaling cascades necessary for immune modulation. The notably DAMPs member such as HMGB1, S100A9 and HSP70/90 can directly bind to receptor on macrophages either TLR2/4 or RAGE^16–20.

The cytokine paradox and the phenotypic surface markers suggested a successful M1 conversion, the cytokine secretion profile revealed a more nuanced reality. A robust increase in classic M1 cytokines, including TNF-α, IL-6, and IL-1β, was observed. However, this was accompanied by the sustained or elevated production of M2-associated cytokines such as IL-10 and IL-13, as well as angiogenic factors like FGF-basic. This mixed profile supports the growing consensus that the binary M1/M2 model is insufficient to capture the plasticity of TAMs in the TME. This complexity is further underscored by recent high-resolution single-cell RNA sequencing atlases of human HCC, such as the one developed by Li et al. which identified ten distinct TAM clusters (Macro1–Macro6 and Kupffer1–Kupffer4). Their findings demonstrate that in vivo TAMs, such as the CXCL10+ (Macro1) cluster, often exhibit overlapping signatures where pro-inflammatory markers coexist with regulatory signals like IL-10. Our observation of a mixed state suggests that chemotherapy-derived DAMPs do not merely switch macrophages to a monolithic M1 state, but rather drive them into a complex, mechanistically relevant spectrum that mirrors the high-resolution heterogeneity seen in human patients²¹. The concurrent presence of IL-10 alongside inflammatory mediators suggests that while DAMPs initiate inflammation, compensatory feedback loops may be simultaneously activated to limit tissue damage, inadvertently preserving a pro-tumoral niche. This aligns with our hypothesis that chemotherapy-associated DAMPs create a “confused” macrophage state that, while phenotypically M1-like, retains functional capabilities that may support tumor survival or metastasis.

Upon activation by ligands, RAGE initiates intracellular signaling via the MEK/ERK, PI3K/AKT, and JAK pathways, alongside ROS production through NADPH oxidase. This signaling activity promotes the activation and translocation of transcription factors including NF-κB, AP-1, and STAT3, resulting in a pro-inflammatory transcriptional program^22,23. TLR4 signaling is characterized by two parallel cascades: a MyD88-dependent pathway and a MyD88-independent pathway. The former facilitates pro-inflammatory cytokine release through the sequential engagement of IRAK, TRAF6, and TAK1, followed by MAPK and NF-κB activation. The latter pathway, mediated by TRIF and TRAM, targets TBK1 and IRF3 to specifically induce the production of type I IFNs²⁴.

We observed that the release of chemotherapeutic-induced DAMPs by sorafenib treatment is sufficient to override M2-like signals and promote M1 polarization. Macrophages exposed to sorafenib-derived conditioned medium displayed a profound morphological transition to a round, flattened state, accompanied by a robust pro-inflammatory gene signature (TNF-α, iNOS). Of mechanistic interest is the signaling route responsible for this activation. While classical macrophage activation often relies heavily on NF-κB signaling, our results indicate a down regulation of p65-NF-κB in these cells. Instead, the upregulation of NLRP3 and p-ERK suggests that sorafenib-induced immunogenic cell death activates macrophages primarily through an ERK-mediated inflammasome pathway. This highlights a specific molecular vulnerability on the NLRP3. Standard cytotoxins, such as taxanes or platinum-based drugs, typically trigger ICD that activates macrophages primarily through the TLR and NF-κB signaling axis^25,26. This classical pathway relies heavily on the nuclear translocation of p65-NF- κB to drive the transcription of pro-inflammatory cytokines. Instead of upregulating κB, sorafenib-derived condition medium induced macrophage activation is driven by an ERK-dependent inflammasome activation rather than classical transcriptional priming. This aligns with previous findings identifying ERK signaling as a critical non-transcriptional primer for the NLRP3 inflammasome, which could activate the inflammation by other promoter such as AP and STAT3^22,23,27. This proposed mechanism is summarized in Fig. 7, which illustrates how sorafenib-conditioned medium drives pro-inflammatory activation specifically through the ERK signaling axis.

Fig. 7.

Schematic diagrams of the signaling pathways involved in macrophage reprogramming induced by chemotherapy-derived DAMPs. Treatment of HepG2 cells with sorafenib or oxaliplatin induces immunogenic cell death, resulting in the release of DAMPs (HMGB1, HSP90, and S100A9). These extracellular mediators bind to pattern recognition receptors (RAGE and TLR4) on the surface of macrophages. The solid red arrows indicate the primary signaling axis identified in this investigation: DAMP-mediated activation of the MyD88-dependent pathway and DIA-1/TIRAP recruitment, which converges on the MEK/MAPK and ERK signaling cascade. This leads to the activation of the transcription factor AP-1 and subsequent pro-inflammatory gene transcription. Dashed grey lines represent canonical signaling pathways (including PI3K/AKT, JAK/STAT, and NF-κB) previously reported in literature but observed to be distinct from the primary ERK-driven mechanism in this model.

The proteomic landscape observed in this study provides a functional snapshot that mirrors the metabolic dichotomy between M1 and M2 macrophages. In the control tumor medium group (M2-like), the upregulation of nucleolar and ribosomal proteins (e.g., nucleophosmin, ribosomal subunits) and histones reflects the high biosynthetic activity required for the trophic functions of TAMs. M2 macrophages are characteristically involved in tissue remodeling and cell proliferation, processes that demand robust protein synthesis machinery²⁸. Furthermore, the specific upregulation of mitochondrial stress-70 protein (Mortalin/GRP75) in this group aligns with the established metabolic profile of M2 macrophages, which rely primarily on oxidative phosphorylation (OXPHOS) for energy production²⁹. The presence of CD44 in this group is also consistent with its role as a marker for aggressive, pro-tumoral TAMs involved in tumor cell invasion and metastasis³⁰. Conversely, sorafenib-conditioned medium treatment triggered a proteomic shift indicative of acute cellular stress and inflammatory activation. The significant upregulation of Endoplasmin (GRP94) suggests a state of ER stress. As macrophages polarize to an M1 phenotype, the massive demand for cytokine synthesis and secretion overloads the ER folding machinery. Crucially, GRP94 is a master chaperone for TLRs and integrins, and its upregulation has been directly linked to M1 polarization and inflammation-induced insulin resistance³¹. This ER stress response is likely a key driver of the NLRP3 activation we observed, as the Unfolded Protein Response (UPR) is a known trigger for inflammasome assembly. Moreover, the upregulation of calcium-binding proteins like Calumenin and EF-hand domain-containing protein D2 (EFHD2/Swiprosin-1) provides the mechanistic second signal for our observed pathway. EFHD2 is critical for macrophage migration and cytokine release³², while intracellular calcium flux is an essential prerequisite for NLRP3 inflammasome activation³³. The increase in Y-box-binding protein (YB-1) reflects the profound transcriptional and translational reprogramming occurring in these cells. HMGA1, which closely related with YB-1, acts as an architectural transcription factor that remodels chromatin to enhance the expression of pro-inflammatory genes, often cooperating with NF- κB and STAT3³⁴. In addition, the HMGB1, the co-family of HMG proteins, is highly expressed in macrophages in several inflammatory conditions including sepsis, COPD, gingivitis and intratumor mass. Overexpression of HMGB1 in macrophages cell drives M1-polarization^35–37. Collectively, these proteomic data suggest that sorafenib-derived DAMPs forces TAMs out of their homeostatic, oxidative (M2) state and into a high-stress, calcium-active state that fuels the NLRP3-mediated inflammatory burst in M1 like macrophages (Fig. 7).

Despite the novel insights provided into the immunomodulatory effects of sorafenib and oxaliplatin, several limitations of this study must be acknowledged. First, as an in vitro investigation, our model isolates specific molecular interactions but lacks the multicellular complexity of the in vivo tumor microenvironment, potentially oversimplifying the dynamic spectrum of macrophage polarization and future studies utilizing a broader panel of HCC cell lines (e.g., Huh7, PLC/PRF/5) will help further delineate the generalizability of these findings. Second, the use of murine macrophages to assess human tumor-derived signals, while supported by the evolutionary conservation of DAMPs, requires validation in primary human TAMs to confirm species-specific signaling kinetics. Finally, while we characterized the molecular and proteomic phenotype of these macrophages, future work is needed to assess their net functional impact on tumor progression and T-cell activation in immunocompetent animal models.

Conclusion

In conclusion, this study establishes that sorafenib and oxaliplatin induce immunogenic cell death characterized by the release of HMGB1, S100A9, and HSP90, which actively drives the reprogramming of M2-like tumor-associated macrophages toward a pro-inflammatory M1 phenotype. Mechanistically, this polarization is mediated by a distinct non-canonical ERK-NLRP3 inflammasome axis that functions independently of the classical NF- κB pathway. Clinically, these findings characterize sorafenib not merely as a cytotoxic agent but as an immunological primer capable of converting cold tumor microenvironments into hot ones, thereby providing a strong rationale for its combination with immune checkpoint inhibitors. This alignment with the shifting landscape of cancer drivers, where inflammatory and metabolic environments are increasingly recognized as primary drivers of malignancy, underscores the importance of targeting the tumor immune niche³⁸.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Songjang W: Writing – original draft, Supervision, Investigation, Conceptualization, Writing – review & editing. Paiyabhroma N: Supervision, Conceptualization, Writing – review & editing. Srisuwan J: Investigation, Conceptualization, Writing – review & editing. Promchai S: Investigation, Conceptualization, Writing – review & editing. Nensat C: Investigation, Conceptualization, Writing – review & editing. Pankhong P: Investigation, Conceptualization, Writing – review & editing. Kumphune S: Supervision, Investigation, Conceptualization, Writing – review & editing. Jiraviriyakul A: Supervision, Investigation, Conceptualization, Writing – review & editing.

Funding

This work was supported by Office of the Permanent Secretary, Ministry of Higher Education, Science, Research and Innovation (OPSMHESI), Thailand Science Research and Innovation (TSRI) and Naresuan University (NU) (Grant numbers RGNS 65–118) to Songjang W. The Naresuan University (NU) and the National Science, Research, and Innovation Fund (NSRF) (Grant numbers R2568B087) to Songjang W, and Grant number R2568B083 to Jiraviriyakul A.

Data availability

The datasets during the current study are available from the first author, Dr. Songjang, upon reasonable request at worawats@nu.ac.th.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval

This study was conducted in strict adherence to the ethical principles outlined in the Declaration of Helsinki. The study protocol was reviewed and approved by the Ethics Committee of Naresuan University (Approval No. P1-0116/2565). As the study utilized unidentified buffy coats, the requirement for informed consent was waived by the committee.