Blocking the TCA cycle in Cancer Cells Potentiates CD36⁺ T Cell-Mediated Antitumor Immunity by Suppressing ER Stress-Associated THBS2 Signaling

Abstract

The TCA cycle is often rewired or dysregulated to meet the increased energy and biosynthetic demands of rapidly dividing cancer cells, and targeting the TCA cycle is a potential therapeutic strategy for treating cancer. However, tumor cell metabolism can impact other cells in the tumor microenvironment, and disrupting the TCA cycle in cancer cells could impact the antitumor immune response. Here, using CPI-613 as a model drug for TCA cycle inhibition, we identified a molecular mechanism by which blocking the TCA cycle enhances T cell-mediated antitumor immunity in the context of head and neck squamous cell carcinoma (HNSCC). Impairment of mitochondrial metabolism by CPI-613 induced endoplasmic reticulum (ER) stress in HNSCC cells, leading to increased expression of spliced X-box binding protein 1 (XBP1s). This, in turn, directly suppressed the transcriptional activity of the thrombospondin-2 (THBS2) gene. Correspondingly, CPI-613 reduced the secretion of THBS2 from HNSCC cells, enhancing the proliferation and cytotoxic potential of tumor-infiltrating CD36⁺CD8⁺ T cells by upregulating AKT-mTOR signaling. This mechanism ultimately enhanced antitumor immunity in a syngeneic mouse model of orthotopic HNSCC following CPI-613 treatment. These findings uncover the immunomodulatory role of the TCA cycle in cancer cells and suggest that targeting it is a promising approach to harness tumor-reactive immune cells.

Significance:

The immunomodulatory role of the TCA cycle in cancer cells provides a therapeutic opportunity to enhance antitumor immunity by targeting tumor cell metabolism.

Introduction

Cancer cells have the potential to manipulate the tricarboxylic acid (TCA) cycle, a critical component of mitochondrial respiration, to support their rapid proliferation and survival (1–3). In cancer, the TCA cycle is often rewired or dysregulated to meet the increased energy and biosynthetic demands of rapidly dividing cells (2–4). Targeting enzymes that are overexpressed or mutated in the TCA cycle, such as pyruvate dehydrogenase (PDH) and α-ketoglutarate dehydrogenase (α-KGDH), can disrupt the metabolic homeostasis of cancer cells, inhibit tumor growth and survival, and thus offer a promising therapeutic strategy (3,5). However, it remains to be determined whether interference with the TCA cycle in cancer cells affects the antitumor response and the underlying molecular mechanisms.

Tumor secretome and antitumor immunity are intertwined in the complex dynamics of cancer progression, immune evasion, and therapeutic response (6,7). The tumor secretome is the collection of molecules secreted by tumor cells into the tumor microenvironment (TME). This intricate mixture comprises proteins, lipids, nucleic acids, metabolites, and extracellular vesicles, all of which are pivotal in modulating the immune response by regulating the recruitment, function, and activation of immune cells in the TME (8). Gaining insight into the complexity of the tumor secretome and its role in reshaping the TME is essential for identifying novel biomarkers for cancer diagnosis and prognosis. Additionally, it supports the development of targeted therapies aimed at disrupting the interactions between tumor cells and immune cells (9,10).

Our previous study has shown the therapeutic potential of CPI-613, a lipoic acid analog selectively inhibiting PDH and KGDH function, in pancreatic cancer and head and neck squamous cell carcinoma (HNSCC) (11,12). Here we report how inhibition of the TCA cycle with CPI-613 enhances antitumor immunity against HNSCC. Specifically, CPI-613 promotes an increase in the expression of spliced X-box binding protein 1 (XBP1s) in HNSCC cells via endoplasmic reticulum (ER) stress, which in turn represses thrombospondin-2 (THBS2) gene transcription. As a result, CPI-613 decreases THBS2 protein secretion from HNSCC cells and promotes proliferation and cytotoxic potential of tumor-infiltrating CD36⁺CD8⁺ T cells via AKT-mTOR signaling. In a syngeneic mouse model of orthotopic HNSCC, this mechanism ultimately enhances T cell-mediated antitumor immunity following CPI-613 treatment. These findings underscore the critical role of the TCA cycle in cancer cell-mediated immunomodulation of immune cells and provide a basis for advancing CPI-613 as a potential treatment option for HNSCC.

Methods and materials

Reagents, plasmids, antibodies, and standard assays

Thapsigargin (TG) and doxycycline (Dox) were purchased from Sigma-Aldrich (St Louis, MO). Malonate, lonidamine, MKC8866, sulfo-N-succinimidyl oleate (SSO), and AZD5363 were purchased from MCE (Monmouth Junction, NJ). CPI-613 was purchased from Selleckchem (Houston, TX). ER-BioID^HA was a kind gift from Dr. Toren Finkel at University of Pittsburgh (13). Human full-length THBS2 overexpression plasmid and mouse full-length Thbs2 overexpression plasmid were obtained from Sinobiological Inc (Beijing, China). pLKO.1-puro TRC control shRNA targeting the gene encoding green fluorescent protein (shGFP) and gene specific shRNAs targeting GLS1, Thbs2, or GRP78 were purchased from Horizon Discovery (Waterbeach, UK). Antibodies used for Western blot and immunofluorescence (IF) are listed in Supplementary Table S1. Plasmid transfection and lentivirus infection, Western blot, and immunoprecipitation (IP) were carried out as we previously described (14–16).

Cell lines and primary tissue specimens

Human HNSCC cell lines HN6 (RRID: CVCL_8129) and HN12 (RRID: CVCL_5518) were a kind gift from Dr. Andrew Yeudall in 2016. Mouse HNSCC cell line MOC2 (RRID: CVCL_ZD33) was purchased from Kerafast (Boston, MA). Cells were cultured in DMEM medium containing 10% fetal bovine serum (FBS) at 37°C in a humidified incubator supplied with 5% CO2. The GenePrint^® 10 System was used to authenticate the cell lines. Routine mycoplasma screening was performed using the MycoAlert Mycoplasma Detection Kit (Lonza), and all cell lines were used for experiments before passage 10. Formalin-fixed, paraffin-embedded (FFPE) tissues from HNSCC patients were obtained from the Head and Neck Satellite Tissue Bank of Emory University. All clinical specimens were obtained with written informed consent from the patients. The studies were conducted in accordance with recognized ethical guidelines (Declaration of Helsinki) and were approved by the Emory Institutional Review Board (IRB).

Liquid chromatography-mass spectrometry (LC-MS)

HN12 cells were infected with ER-BioID^HA lentiviral plasmid and selected by 1.5 μg/ml puromycin for two weeks to generate ER-BioID^HA stably expressing HN12 (HN12-ER) cells. HN12-ER cells were pulsed with 50 μM biotin for 12 hrs, followed by treatment with or without 50 μM CPI-613 for 24 hrs. Conditioned medium (CM) was collected and incubated with High Capacity Magne Streptavidin (SA) Beads (Promega, Cat#PRV7820) overnight. Beads were collected using a magnetic stand, washed five times with 50 mM ammonium bicarbonate, and resuspended for mass spectrometric detection at the Taplin Mass Spectrometry Facility (Harvard Medical School). LC-MS data were analyzed using MaxQuant v2.5.2.0 software (RRID: SCR_014485). The peak list was generated for extensive database searching, and differentially expressed proteins (DEPs) (FC ≥ 2, p < 0.05) were identified. The resulting peaks were plotted on a two-dimensional graph with retention time on the x-axis and m/z ratio on the y-axis, providing a clear visual representation of the data.

Measurement of mitochondrial respiration

To determine mitochondrial respiration profiles, 5 × 10⁵ HN12 cells treated with or without the TCA inhibitors were seeded directly onto Seahorse XFPS plates. OCR was acquired in cells by sequential treatment with 1.5 μM oligomycin, 500 nM FCCP, and 0.5 μM antimycin A/rotenone and analyzed using a Seahorse XFe24 flux bioanalyzer (Agilent Technologies).

Chromatin immunoprecipitation followed by quantitative PCR analysis (ChIP-qPCR)

ChIP assay was performed using a ChIP Assay Kit (ThermoFisher Scientific) as we previously described (17). Briefly, cells were crosslinked with 1% formaldehyde for 10 mins, followed by quenching with glycine for 10 mins and lysed to generate soluble chromatin, which was sonicated with a Sonifier F-100 (ThermoFisher Scientific). Immunoprecipitations of soluble chromatin were incubated with anti-XBP1s antibody and protein A/G beads overnight at 4°C. The immunoprecipitates were washed and eluted from the beads, and reverse crosslinked with 0.5M NaCl for 1.5 hrs at 65°C. Crosslinked chromatin was treated with RNAse A for 1 hour at 37°C, followed by phenol extraction. Relative enrichment of specific DNA sequences was measured using the StepOne Plus Real-Time PCR System (Applied Biosystems). Primers used are listed in Supplementary Table S2.

Specific transcription factor (TF)-DNA interaction

JASPAR (RRID: SCR_003030) was utilized to pinpoint the potential location of the XBP1s binding site on the THBS2 gene promoter. Dox-inducible dCas9 vector p-LV-TRE3G-dCas9-DsRed-Zeo with an HA tag was a kind gift from Dr. S. Ali Shariati at UC Santa Cruz (18), which was designed to identify specific TF-DNA interactions. An sgRNA targeting the THBS2 gene promoter (sgTarget) was designed based on available PAM sites (NGG) near the XBP1s binding site to fully cover the binding site, which was co-transfected with p-LV-TRE3G-dCas9-DsRed-Zeo into HNSCC cells with 1 mg/ml Dox to the cell medium (18). An sgRNA against the GFP gene (sgGFP) was used as a negative control. All sgRNAs used were synthesized by Synthego (Redwood City, CA).

IF

HN12-ER cells were fixed with 3.7% formaldehyde in PBS for 15 mins and permeabilized with 0.1% Triton X-100 in PBS for 10 mins. Cells were blocked with IF buffer (PBS plus 1% BSA and 2% FBS) for 30 mins and were then incubated with mouse anti-human HA antibodies and rabbit anti-human calnexin overnight at 4°C, followed by incubation with goat anti-mouse Alexa Flour 488-conjugated secondary antibody and goat anti-rabbit Alexa Flour 594-conjugated secondary antibody in the dark for 1 h. For tissue IF, FFPE tissue sections from HNSCC patients were deparaffinized in xylene, rehydrated through a graded series of alcohol, and incubated in 3% hydrogen peroxide. After antigen retrieval and blocking, tissues were incubated with the indicated primary antibodies overnight at 4 °C, followed by incubation with fluorescence-conjugated secondary antibodies in the dark for 1 h. Slides were mounted with Vectashield mounting medium (Vector Laboratories, Burlingame, CA) containing the DAPI nuclear stain before examination under an inverted fluorescence microscope (BZ-X710 All-in-one, Keyence). The quantitation of positive CD8 or Ki67 cells in the tumor sections was counted in 10 random fields, and the average number of positive cells per reported field was calculated based on the results provided by three investigators.

RNA-seq analysis

Total RNA was extracted from HN12 cells treated with or without CPI-613 using TRIzol reagent (Invitrogen, Carlsbad, CA). The purified RNA samples were sent to Novogene Corporation (Sacramento, CA) for library construction and sequencing using the Illumina HiSeq^™ 2000 platform (RRID: SCR_020130) to obtain expression libraries of 50-nt read length. Independent duplicate cultures were sampled to avoid random error. Differentially expressed genes (DEGs) were identified using the DESeq R package (version 4.1.2) functions estimateSizeFactors and nbinomTest. Enrichment analysis on the sets of differentially expressed genes (LogFC ≥ |±0.25|, p < 0.05) was performed via Gene Ontology (GO).

Flow cytometry

Mouse tumor tissue, spleen and lymph nodes were mechanically digested by mincing the tissue and filtered through a 70 μm filter. Cell suspension was chemically digested by incubation with collagenase (Roche, Cat #11088866001), dispase (Stemcell Technologies, Cat#07923), and liberase enzymes (Sigma-Aldrich, Cat #5401119001) at 37°C for 30 mins. Single cell suspensions were then incubated with fluorochrome-conjugated antibodies (Supplementary Table S1) for 10 mins, washed, and fixed in FACS buffer containing 2% FBS and 4% paraformaldehyde (Sigma-Aldrich, Cat#95-30525-89-4). Zombie Aqua Fixable (BioLegend, Cat#423101) or Zombie UV Viability Dye (BioLegend, Cat#423107) was used to detect live cells. Samples were analyzed on a BD Symphony A3 cytometer and further analyzed using FlowJo (V.10.8.1) software (BD Biosciences) (RRID: SCR_008520). ‘Fluorescence minus one’ controls were tested for each multicolor flow panel.

T cell assays

Human peripheral blood mononuclear cells (PBMCs) were collected from healthy donors at Emory University with Ficoll-Paque Premium density gradient centrifugation (Cytiva) as we previously described (15). CD8⁺ T cells were isolated from the fresh PBMCs using an EasySep^™ Human CD8 Positive Selection Kit II (Stemcell Technologies, Cat#17853). Mouse CD8⁺ T cells were isolated from the spleen of six-week-old male C57BL/6 mice using an EasySep^™ Mouse CD8a Positive Selection Kit II (Stemcell Technologies, Cat#18953). Human CD8⁺ T cells from PBMCs were stimulated Dynabeads Human T-Activator αCD3/αCD28 (ThermoFisher Scientific, Cat#11161D) and mouse splenic CD8⁺ T cells were stimulated with Dynabeads mouse T-Activator αCD3/αCD28 (ThermoFisher Scientific, Cat#11452D) in complete RPMI 1640 medium for 24 hrs. Mouse splenic CD8⁺ T cells were further resuspended in the media containing 100 U/ml murine IL-2 (Biolegend) and incubated with CM collected from MOC2 cells treated with or without CPI-613. After 48 hrs of incubation, CD8⁺ T cells were harvested, and their proliferation and cytokine production were analyzed by flow cytometry.

Animal studies

Six-week-old male wild-type and Rag1 knockout (KO) C57BL/6 mice were purchased from the Jackson Laboratory (Bar Harbor, ME). All animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of Emory University. To generate buccal mucosal tumor models, 1×10⁶ gene-modified or control MOC2 cells were suspended in 100 μl of PBS/Matrigel (3:1) and implanted into mice by intramucosal injection. On day 6 after cell inoculation, mice were randomized to receive treatment with vehicle or CPI-613. CPI-613 was dissolved in 5% DMSO of corn oil and given by intraperitoneal (i.p.) injection at 50 mg/kg body weight once daily consecutively for 12 days. Tumor dimensions were measured twice a week with electronic calipers, and tumor volume was calculated by the formula of V = length × width² × 1/2. Mice were sacrificed when the experiment was terminated, and tumors and major organs were excised for HE staining, IF and flow cytometry analysis. Blood was collected via the ocular vein, and serum alanine transaminase (ALT), aspartate transaminase (AST) and creatinine levels were measured using the EnzyChrom^™ Alanine Transaminase Assay Kit, Aspartate Transaminase Assay Kit (BioAssay Systems, Hayward, CA) and Creatinine Assay Kit (Cayman Chemical, Ann Arbor, MI), respectively.

Bioinformatics and statistical analysis

Overall survival associated with the THBS2 gene or protein with high and low expression (optimal proportion) was calculated using the ‘survival’ and ‘surviminer’ package in R software. The ‘TIMER2.0’ was performed for pan-cancer analysis of THBS2 expression across 33 cancer types. Differences in tumor immune infiltration between the THBS2 high group and low group were analyzed using ssGSEA package in R software. For comparison between two groups, statistical analysis was performed using unpaired Student’s t-test. One-way analysis of variance (ANOVA) followed by Tukey’s multiple testing correction was used for comparison of more than two groups. Unless otherwise specified in the figure legends, all experiments reported in this study were performed using at least two independent experiments or biological replicates. p < 0.05 was considered statistically significant. Data were analyzed using statistical software GraphPad Prism 9 (San Diego, CA). Experimental values were expressed as mean ± standard deviation (SD).

Data availability

The RNA-seq data generated in this study is available in Gene Expression Omnibus (GEO) at GSE243460. Gene expression data were downloaded from the Cancer Genome Atlas (TCGA) (RRID: SCR_003193) (https://portal.gdc.cancer.gov/repository). TCGA HNSCC cohort includes 415 HPV-negative (HPV-) cases and 70 HPV-positive (HPV+) cases. Clinical data associated with TCGA HNSCC cohort were obtained from cBioPortal (RRID: SCR_014555) (http://www.cbioportal.org/datase). Protein level and clinical data were downloaded from Clinical Proteomic Tumor Analysis Consortium (CPTAC) (RRID: SCR_017135) (https://proteomics.cancer.gov/data-portal). All other raw data generated in this study are available upon request from the corresponding author.

Results

CPI-613 potentiates CD8⁺ T cell-mediated antitumor immunity in the TME

CPI-613, malonate, and lonidamine are effective TCA cycle inhibitors (12,19). To assess and compare the effectiveness of these inhibitors in mitochondrial respiration, oxygen consumption rate (OCR) was measured in HN12 cells with or without these treatments. This analysis revealed that three inhibitors suppressed mitochondrial respiration, with CPI-613 being the most effective (Supplementary Fig. S1). Consequently, CPI-613 was chosen as the model drug for TCA cycle inhibition.

Tumors originating from murine MOC2 cells display a low prevalence of CD8⁺ T cells and an immunosuppressive TME (9). To understand the potential of CPI-613 in modulating tumor immunity, we treated MOC2 tumor-bearing C57BL/6 mice with CPI-613 for 12 days. This treatment led to a significant reduction in tumor growth and weight without affecting the mouse body weight (Fig. 1A and B). There were no notable differences in serum AST, ALT, and creatinine levels (Supplementary Fig. S2A–C) or in the histopathological examination of major organs (Supplementary Fig. S2D) between tumor-bearing mice treated with or without CPI-613, suggesting no side effect produced by CPI-613 at this dosage. Moreover, CPI-613 treatment in Rag1 KO C57BL/6 mice lacking functional lymphocytes also reduced MOC2 tumor burden (Fig. 1C and D), which was consistent with our previous findings in immunodeficient NSG mouse (12), indicating the direct antitumor effect of CPI-613. However, the treatment efficacy of CPI-613 in Rag1 KO C57BL/6 mice was much lower compared to wild-type C57BL/6 mice (Fig. 1A–D). IF analysis showed an increase not only in the number of CD8⁺ T cells but also in CD8⁺Ki67⁺ T cells within MOC2 tumors following CPI-613 treatment (Fig. 1E and F), suggesting that CPI-613 not only enhances tumor-infiltration of CD8⁺ T cells but also stimulates their proliferation.

Figure 1. CPI-613 enhances antitumor immunity of tumor-infiltrating effector T cells in syngeneic tumor models of orthotopic head and neck tumors.

(A) Effect of CPI-613 treatment on MOC2 tumor growth and weight in wild-type C57BL/6 mice (n = 5 mice/group). (B) Effect of CPI-613 treatment on body weight of wild-type C57BL/6 mice bearing MOC2 tumors (n = 5 mice/group). (C) Effect of CPI-613 treatment on MOC2 tumor growth and weight in Rag1 KO C57BL/6 mice (n = 5 mice/group). (D) Effect of CPI-613 treatment on body weight of Rag1 KO C57BL/6 mice bearing MOC2 tumors (n = 5 mice/group). In (A-D), CPI-613 was administered at a dose of 50 mg/kg by i.p. injection daily for 12 days when buccal mucosal tumors were established on day 8. (E, F) Immunofluorescent staining with anti-pan-Keratin/anti-CD8α antibodies and anti-Ki67/anti-CD8α antibodies in MOC2 tumors derived from wild-type C57BL/6 mice with or without CPI-613 treatment for 12 days. (G, H) Percent of cytotoxic (GzmB⁺) and proliferative (Ki67⁺) CD8⁺ T cell subsets in MOC2 tumors and spleens derived from wild-type C57BL/6 mice treated with CPI-613 for 3 days or 12 days. In (E-H), representative images and quantitative data (n=5) are shown in the left and right panels, respectively. *p<0.05; **p<0.01.

Next, we examined the dynamic alterations in CD8⁺ T cell proliferation and cytotoxicity in the presence or absence of CPI-613. A significant increase in the number of proliferative and cytotoxic CD8⁺ T cells was observed in MOC2 tumors following either short-term (3 days) or long-term (12 days) treatment with CPI-613 (Fig. 1G). Notably, long-term treatment resulted in a higher number of proliferative and cytotoxic CD8⁺ T cells compared to short-term treatment (Fig. 1G). There were no changes in cell proliferation and cytotoxicity of splenic CD8⁺ T cells with or without CPI-613 treatment regardless of treatment duration (Fig. 1H). The same tendency was seen in CD8⁺ T cells in lymph nodes treated with or without CPI-613 (Supplementary Fig. S3), suggesting that CPI-613 only has the potential to modulate tumor-infiltrating CD8⁺ T cells. No significant changes were observed in regulatory T cells (Tregs) (Supplementary Fig. S4A) or monocytic and granulocytic myeloid-derived suppressor cells (MDSCs) (Supplementary Fig. S4B) in MOC2 tumors regardless of CPI-613 treatment. These findings support the notion that CPI-613 represses HNSCC partially through inducing antitumor immunity mediated by tumor-infiltrating CD8⁺ T cells.

CPI-613 promotes the proliferation and cytotoxic potential of CD8⁺ T cells by acting on HNSCC cells

Next, we assessed the proliferation of either HNSCC cells or CD8⁺ T cells isolated from human PBMCs after 48 hrs of CPI-613 treatment. CPI-613 significantly inhibited the proliferation of both HN6 and HN12 cells but did not affect the proliferation of human CD8⁺ T cells (Fig. 2A). Similar tendencies in cell proliferation were observed in CPI-613-treated MOC2 cells and splenic CD8⁺ T cells isolated from wild-type C57BL/6 mice (Fig. 2B). Consistently, there were no changes in the proliferative (Ki67⁺) and cytotoxic (GzmB⁺) subsets of mouse splenic CD8⁺ T cells between CPI-613 treatment and control groups (Fig. 2C), indicating no direct impact of CPI-613 on CD8⁺ T cells. Intriguingly, mouse splenic CD8⁺ T cells showed increased proliferation when incubated with CM collected from CPI-613-treated MOC2 cells (Fig. 2D). At the molecular level, increased Ki67⁺ and GzmB⁺ cell populations were seen in mouse splenic CD8⁺ T cells incubated with CM collected from CPI-613-treated MOC2 cells compared to control group (Fig. 2E). These findings demonstrate that CPI-613 enhances the proliferation and cytotoxic potential of effector T cells through its action on HNSCC cells.

Figure 2. I-613 promotes the proliferation and cytotoxic potential of effector T cells by acting on HNSCC cells.

CP (A) Effect of CPI-613 on proliferation of HN6 cells, HN12 cells, and CD8⁺ T cells isolated from healthy human peripheral blood. (B) Effect of CPI-613 on proliferation of MOC2 cells and CD8⁺ T cells isolated from C57BL/6 mouse spleen. (C) Percent of proliferative (Ki67⁺) and cytotoxic (GzmB⁺) subsets in mouse splenic CD8⁺ T cells treated with or without CPI-613. (D) Effect of CPI-613 on proliferation of mouse splenic CD8⁺ T cells incubated with CM collected from DMSO- or CPI-613-treated MOC2 cells. (E) Percent of proliferative (Ki67⁺) and cytotoxic (GzmB⁺) subsets in mouse splenic CD8⁺ T cells incubated with CM collected from DMSO- or CPI-613-treated MOC2 cells. In (A-E), HNSCC cells or CD8⁺ T cells were treated with 50 μM CPI-613 for 48 hrs. In (C) and (E), representative flow images and quantitative data (n = 3) are shown in the left and right panels, respectively. CM, conditioned medium. *p<0.05; **p<0.01.

Blocking the TCA cycle with CPI-613 reduces THBS2 secretion from HNSCC cells

To create a molecular platform for capturing the tumor-specific secretome while excluding the influence of serum proteins, we exploited the properties of proximity biotinylation. Since most membrane and secreted proteins transit through the ER, expression of an ER lumen-resident BioID expression plasmid (termed ER-BioID^HA) provides a method to biotinylate proteins destined for conventional secretion (13). We generated ER-BioID^HA stably expressing HN12-ER cells, in which ER-BioID^HA was exclusively localized to the ER (Fig. 3A). The bands visualized in the presence of biotin using an SA antibody confirmed the response of ER-BioID^HA to biotin (Fig. 3B).

Figure 3. Inhibiting the TCA cycle with CPI-613 significantly reduces both intracellular levels and secretion of THBS2 in HNSCC cells.

(A) Colocalization of ER-BioID^HA and ER-resident protein calnexin in HN12-ER cells determined by IF. (B) Detection of secreted biotinylated proteins in HN12-ER cell supernatant in the presence or absence of exogenous biotin. Cell supernatant collected from HN12-ER cells with or without addition of 50 μM biotin was purified by streptavidin (SA) and determined by SA-based Western blot. Initial cell lysate was assessed for ER-BioID^HA expression (HA epitope) and β-Actin as a loading control. (C) Schematic of the process of assessing differential secretome of HN12-ER cells in the presence or absence of CPI-613. HN12-ER cells stably expressing ER-BioID^HA construct were treated with DMSO or 50 μM CPI-613 for 24 hrs, and supernatant was collected and incubated with SA beads, followed by liquid chromatography-mass spectrometry (LC-MS) analysis (Created in BioRender. Teng, Y. (2025) https://BioRender.com/ t49t849). (D) Top ten most downregulated proteins in HN12-ER cells following CPI-613 treatment determined by LC-MS analysis. (E) A representative spectrogram of tryptic peptides derived from THBS2 protein. (F) Effect of CPI-613 treatment on secreted or cellular THBS2 protein levels in HN6 and HN12 cells determined by Western blot. HN12-ER cells, ER-BioID^HA stably expressing HN12 cells; DEPs, differentially expressed proteins; Sup, supernatant; Lys, lysate.

Next, the supernatants from HN12-ER cells treated with DMSO or 50 μM CPI-613 were collected and incubated with SA beads, followed by LC-MS analysis (Fig. 3C). This study identified 148 DEPs in CPI-613-treated HN12-ER cells compared to control cells. Among them, THBS2 was the most downregulated protein identified in the supernatants from CPI-613-treated HN12-ER cells (Fig. 3D and E). CPI-613-induced THBS2 reduction was confirmed by Western blot using HN6 and HN12 cell supernatants (Fig. 3F). Interestingly, reduced THBS2 protein levels were also observed in the lysates of HN6 and HN12 cells treated with CPI-613 (Fig. 3F), suggesting that blocking the TCA cycle with CPI-613 has the potential to reduce both intracellular levels and secretion of THBS2 in HNSCC cells.

CPI-613 represses THBS2 gene expression in HNSCC cells through inducing the expression of ER stress-associated XBP1s

To investigate the mechanism by which CPI-613 suppressed THSB2, RNA extracted from CPI-613- and DMSO-treated HN12 cells was subjected to RNA-seq. Pathway enrichment analysis showed two signaling pathways, “Endoplasmic reticulum unfolded protein response” and “Response to endoplasmic reticulum stress”, were upregulated in HN12 cells treated with CPI-613, suggesting that CPI-613 may trigger ER stress in HNSCC cells (Fig. 4A). Moreover, XBP1 was identified as one of the most upregulated ER stress-enriched genes in CPI-613-treated HN12 cells compared to control cells (Fig. 4B). Following CPI-613 treatment, levels of other ER stress-related proteins, including ATF6 and GRP78, also increased in HNSCC cells (Supplementary Fig. S5A), further supporting its potential to induce ER stress. Splicing of XBP1 mRNA generates the stress-responsive XBP1s (20), we thus determined XBP1s levels in HNSCC cells in the presence or absence of CPI-613. In HN12 and HN6 cells, XBP1s was undetectable but dramatically increased upon CPI-613 treatment (Fig. 4C). No XBP1s was detected in the cell supernatants regardless of CPI-613 treatment (Fig. 4C), indicating that XBP1s is not a secreted protein. We also found that the ER stress inducer TG increased XBP1s levels while decreased THBS2 levels in HN6 and HN12 cells (Fig. 4D). Moreover, treatment with the ER stress deactivator MKC8866 remarkably attenuated XBP1s levels induced by TG or CPI-613, which was accompanied by increased THBS2 levels (Fig. 4E). These observations suggest the involvement of ER stress in the regulation between XBP1s and THBS2 in the context of CPI-613 treatment.

Figure 4. CPI-613 induces the expression of ER stress-associated XBP1s to inhibit THBS2 transcriptional activity in HNSCC cells.

(A) GO pathway enrichment analysis of RNA-seq data from HN12 cells treated with or without CPI-613. Significant alterations in ER stress-related pathways are highlighted in red. (B) Heatmap analysis of RNA-seq data showing differential expression of ER stress-associated genes in HN12 cells upon CPI-613 treatment. (C) Effect of CPI-613 treatment on secreted or cellular XBP1s protein levels in HN6 and HN12 cells determined by Western blot. Sup, supernatant; Lys, lysate. (D) Effect of CPI-613 or ER stress inducer TG on protein levels of XBP1s and THBS2 in HN6 and HN12 cells. (E) Effect of ER stress deactivator MKC8866 on TG or CPI-613-induced inhibition of XBP1s and THBS2 expression in HN6 and HN12 cells. Cells were pretreated with 10 μM MKC8866 for 4 hrs before TG or CPI-613 treatment. (F) Effect of CPI-613, malonate, or lonidamine on protein levels of XBP1s and THBS2 in HN6 and HN12 cells. (G) Schematic of the DNA binding site of XBP1s on the upstream promoter of the THBS2 gene. (H) Binding of XBP1s to the THBS2 gene promoter in HN6 and HN12 cells following CPI-613 or TG treatment determined by ChIP assay. (I) Schematic showing the designed sgRNA that can specifically target XBP1s binding site at the THBS2 gene promoter (Created in BioRender. Teng, Y. (2025) https://BioRender.com/ q44w604). (J, K) Effect of sgTarget on THBS2 expression (J) and protein (K) levels in HN6 and HN12 cells in the presence or absence of TG or CPI-613. HN6 and HN12 cells co-transfected with dCas9 and sgTarget or sgGFP were treated with or without TG or CPI-613, followed by qRT-PCR and western blot. In (C-D, (F), (H) and (J-K), HNSCC cells were treated with 0.5 μM TG or 50 μM CPI-613 for 24 hrs. sgGFP, sgRNA targeting the GFP gene; sgTarget, sgRNA targeting XBP1s binding site on the THBS2 gene promoter. ns, not significant; *p<0.05; **p<0.01.

Next, we treated HN12 and HN6 cells with two additional TCA cycle inhibitors, malonate and lonidamine. Similar to the effects seen with CPI-613, treatment with both malonate and lonidamine led to increased levels of XBP1s, alongside decreased levels of THBS2 (Fig. 4F). This finding implies that the changes in the XBP1s-THBS2 signaling axis are triggered by TCA cycle inhibition, rather than by other effects of CPI-613. Interestingly, we identified a single consensus XBP1s binding site spanning −1240 bp and −1253 bp on the THBS2 gene promoter (Fig. 4G). ChIP assays revealed a distinct XBP1s occupancy on the THBS2 gene promoter in both HN6 and HN12 cells following either CPI-613 or TG treatment (Fig. 4H). Knockdown of GRP78 in these two cell lines did not affect THBS2 expression, either in the presence or absence of CPI-613, excluding the involvement of GRP78 in the regulation of THBS2 (Supplementary Fig. S5B–S5D). As CPI-613 upregulates GLS1-dependent glutaminolysis in certain HNSCC cells (12), we also assessed THBS2 levels in GLS1 knockdown and control cells with or without CPI-613 treatment. Knockdown of GLS1 in HN6 and HN12 cells did not influence THBS2 expression, regardless of whether CPI-613 was present or not (Supplementary Fig. S6), suggesting that GLS1 plays no role in CPI-613-induced THBS2 suppression.

To determine the essential function of the single XBP1s binding site within transcription networks, CRISPRd was used to disrupt XBP1s binding to the THBS2 gene promoter (18). In this study, sgTarget or sgGFP was co-transfected with Dox-inducible deactivated Cas9 (dCas9) vector p-LV-TRE3G-dCas9-DsRed-Zeo (Fig. 4I). In sgTarget-transfected HNSCC cells, THBS2 expression was restored at both the transcriptional and protein levels in the presence of CPI-613 or TG (Fig. 4J and 4K), indicating that THBS2 expression depends on XBP1s activity.

High levels of THBS2 correlate with a low presence of tumor-infiltrating CD8⁺ T cells in HNSCC

To explore the clinical significance of THBS2 expression, we conducted a bioinformatics analysis using TCGA pan-cancer database. We found elevated THBS2 expression in HNSCC and 11 other cancer types compared to their respective normal tissues (Supplementary Fig. S7). Within HNSCC, THBS2 expression was significantly higher in the HPV- subtype compared to the HPV+ subtype (Supplementary Fig. S7). Survival analysis revealed a significant inverse correlation between high THBS2 expression and poor overall survival in HPV- patients included in TCGA database, and this correlation was also observed in HPV- patients from CPTAC database (Fig. 5A). In contrast, THBS2 expression did not affect survival in HPV+ patients (Supplementary Fig. S8), suggesting that THBS2 plays a critical role in the development and progression of HPV- HNSCC. Bioinformatics analysis also showed that THBS2 gene expression was tightly associated with the presence of several immune cell types, including T cells, B cells, NK cells, macrophages, and neutrophils in HPV- HNSCC (Fig. 5B and C). More specifically, higher THBS2 gene expression was significantly correlated with a lower presence of tumor-infiltrating activated CD8⁺ T cells in HPV- HNSCC (Fig. 5B–D), which was confirmed by IF using the primary tumor samples from untreated HPV- HNSCC patients (Fig. 5E). These results suggest that THBS2 may promote a less favorable immune environment in HNSCC.

Figure 5. High levels of THBS2 expression are negatively correlated with poor survival in HNSCC patients and reduced tumor-infiltrating activated CD8+ T cells.

(A) Correlation between THBS2 gene expression and overall survival in TCGA or CTPAC HPV- HNSCC patient cohort. (B) Immune infiltration analysis of 31 immune cell types in THBS2 high (205 cases) vs. low (205 cases) expression group using TCGA HPV- HNSCC cohort dataset. (C) Identification of immune infiltration of ‘activated CD8 T cells’ as the most downregulated signature in THBS2 high vs. low expression group in TCGA HPV- HNSCC cohort. (D) Correlation between immune infiltration of activated CD8⁺ T cells and THBS2 expression in TCGA HPV- HNSCC cohort (R = −0.17, P = 0.012). (E) Representative IF images showing the correlation between the presence of CD8-positive cells and THBS2 protein levels in primary HNSCC tumor specimens.

CPI-613 enhances antitumor immunity by disrupting the THBS2-CD36 axis between HNSCC cells and CD8⁺ T cells

To investigate whether CPI-613 impacts CD8⁺ T cells by inhibiting tumoral THBS2, we developed THBS2-overexpressing HN12 cells and Thbs2-overexpressing MOC2 cells (Fig. 6A). The gene overexpression led to increased protein secretion from HNSCC cells (Fig. 6B), but it did not affect cell proliferation regardless of CPI-613 treatment (Fig. 6C). CD36, a scavenger receptor involved in immunity, is widely expressed in immune cells, which can pump fat to defang effector T cells in tumors (21). Interestingly, THBS2 has been reported to regulate a diverse range of activities through binding to CD36 on the cell surface membrane (22,23). To determine the THBS2-CD36 interaction, CD8⁺ T cells isolated from C57BL/6 mouse spleen were collected for IP with a CD36 antibody. THBS2 was identified in the CD36 immunocomplex in CD8⁺ T cells (Fig. 6D). IP using a THBS2 antibody followed by Western blot analysis with a CD36 antibody further confirmed this protein-protein interaction (Supplementary Fig. S9). There was no difference in the amount of THBS2 bound to CD36 in CD8⁺ T cells with or without CPI-613 treatment (Fig. 6E). An increased amount of THBS2 protein was detected in the CD36 immunocomplex in CD8⁺ T cells when incubated with CM collected from MOC2 cells (Fig. 6E). However, a remarkably reduced amount of THBS2 was found to interact with CD36 in CD8⁺ T cells when incubated with CM collected from CPI-613-treated MOC2 cells (Fig. 6E).

Figure 6. CPI-613 enhances antitumor immunity through blocking the THBS2-CD36 axis between HNSCC cells and CD8+ T cells.

(A) Overexpression of THBS2/Thbs2 in HN12 and MOC2 cells. (B) Effect of THBS2/Thbs2 overexpression on THBS2 secretion in HN12 and MOC2 cells. (C) Effect of THBS2/Thbs2 overexpression on cell proliferation in HN12 and MOC2 cells in the presence or absence of CPI-613 for 72 hrs. (D) IP analysis of the THBS2-CD36 interaction in mouse splenic CD8⁺ T cells. (E) IP analysis of THBS2-CD36 interaction in mouse splenic CD8⁺ T cells incubated with CM collected from DMSO- or CPI-613-treated MOC2 cells. In (D) and (E), the immunoprecipitates from mouse splenic CD8⁺ T cells were pulled down with a CD36 antibody, followed by Western blot with a THBS2 antibody. Pre-immune IgG was used as a negative control. (F) Alterations in phosphorylation levels of AKT and mTOR in mouse splenic CD8⁺ T cells incubated with or without CM collected from CPI-613-treated MOC2 cells. (G) Alterations in the phosphorylation levels of AKT and mTOR in mouse splenic CD8⁺ T cells incubated with CM collected from CPI-613-treated Thbs2-overexpressing or control MOC2 cells. (H) Effect of AZD5363 on mTOR phosphorylation in mouse splenic CD8⁺ T cells incubated with CM collected from DMSO- or CPI-613-treated MOC2 cells. (I) Effect of AZD5363 on cell proliferation in mouse splenic CD8⁺ T cells incubated with CM collected from DMSO- or CPI-613-treated MOC2 cells. (J) Alterations in the phosphorylation of AKT and mTOR in mouse splenic CD8⁺ T cells treated with CPI-613 or SSO. (K) Effect of SSO on the phosphorylation of AKT and mTOR in mouse splenic CD8⁺ T cells incubated with CM collected from MOC2 cells or not. (L) Effect of SSO on the phosphorylation of AKT and mTOR in mouse splenic CD8⁺ T cells incubated with CM collected from Thbs2-overexpressing or control MOC2 cells. (M) Effect of CPI-613 treatment on XBP1s and Thbs2 levels in mouse splenic CD8⁺ T cells. MOC2 cells or CD8⁺ T cells were treated with 50 μM CPI-613 in (C, E-J, and M), 20 μM SSO in (J-L) and 5 μM AZD5363 in (H-I), respectively. Mouse splenic CD8⁺ T cells were CD8⁺ T cells isolated from 8-week-old C57BL/6 mouse spleen. CM, conditioned medium; EV, empty vector; Thbs2 O/E, Thbs2 overexpression vector. *p<0.05; **p<0.01.

The AKT-mTOR and ERK1/2 signaling pathways play crucial roles downstream of CD36 in regulating cell growth and other cellular activities (24,25). No changes in ERK1/2 phosphorylation were observed in mouse splenic CD8+ T cells, regardless of whether they were treated with CPI-613 (Supplementary Fig. S10A) or incubated with CM (Supplementary Fig. S10B) collected from CPI-613-treated MOC2 cells. Strikingly, there was an increase in the phosphorylation levels of AKT and mTOR in CD8⁺ T cells incubated with CM collected from CPI-613-treated MOC2 cells, an effect that was not observed in CD8⁺ T cells treated with CPI-613 (Fig. 6F). When cultured with CM collected from CPI-613-treated Thbs2-overexpressing MOC2 cells, CD8⁺ T cells showed reduced levels of p-AKT and p-mTOR compared to CD8⁺ T cells incubated with CM from CPI-613-treated MOC2 cells (Fig. 6G), suggesting that CPI-613-induced activation of the AKT-mTOR signaling in effector T cells is mediated by THBS2 secreted from HNSCC cells. Furthermore, when CD8⁺ T cells were cultured with CM collected from MOC2 cells, the AKT inhibitor AZD5363 effectively inhibited mTOR phosphorylation, resulting in decreased CD8⁺ T cell proliferation (Fig. 6H and I). A similar trend was observed in CD8⁺ T cells incubated with CM collected from CPI-613-treated MOC2 cells (Fig. 6H and I), supporting the critical role of AKT-mTOR signaling in CPI-613-induced CD8⁺ T cell proliferation.

We further examined the changes in AKT and mTOR phosphorylation in CD8⁺ T cells treated with or without the CD36 inhibitor SSO. Blocking CD36 with SSO resulted in increased levels of p-AKT and p-mTOR in CD8⁺ T cells (Fig. 6J), regardless of whether they were incubated with CM collected from MOC2 cells or not (Fig. 6K). Additionally, the phosphorylation of AKT and mTOR was reduced in CD8⁺ T cells when incubated with CM collected from Thbs2-overexpressing MOC2 cells (Fig. 6L). However, the decreased levels of p-AKT and p-mTOR in CD8⁺ T cells were restored following SSO treatment (Fig. 6L). Unlike in HNSCC cells, CPI-613 did not alter the protein levels of XBP1s and THBS2 in CD8⁺ T cells (Fig. 6M). Loss of Thbs2 did not directly affect MOC2 cell proliferation (Supplementary Fig. S11A–B), but this loss repressed MOC2 tumors through activating CD36-AKT/mTOR pathway in CD8⁺ T cells (Supplementary Fig. S11C–I), which mirrors the effects observed in CPI-613 treatment. These findings indicate that CPI-613 potentiates antitumor immunity by disrupting the THBS2-CD36 axis between tumor and tumor-infiltrating CD8⁺ T cells.

CPI-613 potentiates the antitumor response of CD36⁺ effector T cells by suppressing tumor-secreted THBS2

Next, we evaluated the importance of THBS2 signaling in CPI-613-mediated antitumor immune response. In wild-type C57BL/6 mice, overexpression of Thbs2 increased MOC2 tumor burden, as evidenced by larger and heavier tumors (Fig. 7A and B). While in Rag1 KO C57BL/6 mice, there was no difference in tumor size or weight between Thbs2-overexpressing and control MOC2 tumors (Fig. 7C and 7D). Consistent with our previous data (Fig. 1A), CPI-613 treatment induced the regression of MOC2 tumors in both wild-type and Rag1 KO C57BL/6 mice (Fig. 7A–D). However, MOC2 tumors with Thbs2 overexpression were less sensitive to CPI-613 treatment in wild-type C57BL/6 mice, but not in Rag1 KO C57BL/6 mice (Fig. 7A–D). Moreover, wild-type mice with MOC2 tumors survived longer than those with Thbs2-overexpressing MOC2 tumors after CPI-613 treatment (Fig. 7E). These findings support the notion that CPI-613 exhibits antitumor potency by enhancing antitumor immunity through suppressing tumoral THBS2.

Figure 7. CPI-613 specifically potentiates the antitumor response of CD36+ effector T cells by suppressing tumor-derived THBS2 in orthotopic syngeneic tumor models.

(A, B) Effect of Thbs2 overexpression on MOC2 tumor growth and weight in wild-type C57BL/6 mice treated with or without CPI-613 (n = 5 mice/group). (C, D) Effect of Thbs2 overexpression on MOC2 tumor growth and weight in Rag1 KO C57BL/6 mice treated with or without CPI-613 (n = 5 mice/group). (E) Effect of Thbs2 overexpression on survival of MOC2 tumor-bearing mice treated with or without CPI-613 (n = 5 mice/group, log-rank test). (F) Effect of Thbs2 overexpression on proliferative (Ki67⁺) and cytotoxic (GzmB⁺ or IFN-γ⁺) CD36⁺CD8⁺ T cell subsets in MOC2 tumors from wild-type C57BL/6 mice treated with or without CPI-613. (G) Effect of Thbs2 overexpression on proliferative (Ki67⁺) and cytotoxic (GzmB⁺ or IFN-γ⁺) of CD36⁻CD8⁺ T cell subsets in MOC2 tumors from wild-type C57BL/6 mice treated with or without CPI-613. Quantitative data from 5 mice/group are shown in (F-G). (H) A proposed model for CPI-613-mediated antitumor immunity involves disrupting the THBS2-CD36 axis between HNSCC cells and CD36⁺CD8⁺ T cells in the TME (Created in BioRender. Teng, Y. (2025) https://BioRender.com/e64w812). EV, empty vector; Thbs2 O/E, Thbs2 overexpression vector. ns, not significant; *p<0.05; **p<0.01.

At the molecular level, CPI-613 significantly increased the proliferation and cytotoxicity of CD36⁺CD8⁺ T cells in MOC2 tumors (Supplementary Fig. S12 and Fig. 7F), but this treatment had no noticeable effect on the population of tumor-infiltrating CD36⁻CD8⁺ T cells (Supplementary Fig. S13 and Fig. 7G). Overexpression of Thbs2 in MOC2 tumors remarkably attenuated the increase in proliferation and cytotoxicity of CD36⁺CD8⁺ T cells induced by CPI-613, without affecting the CD36⁻CD8⁺ T cell subset. These observations demonstrate that CPI-613 enhances antitumor immunity through disrupting the THBS2-CD36 axis in the TME.

Discussion

Targeting TCA cycle impairs mitochondrial respiration, resulting in decreased ATP production and altered cellular metabolism(3,19). Using CPI-613 as a model drug for TCA cycle inhibition and combining findings from both in vitro and in vivo models, we show that CPI-613 induces ER stress in HNSCC cells, leading to an increase in XBP1s expression. This, in turn, negatively regulates the transcription of the THBS2 gene. Correspondingly, CPI-613 reduces secretion of the THBS2 protein from HNSCC cells, which enhances the proliferation and cytotoxic potential of intratumoral CD8⁺ T cells by upregulating THBS2-dependent CD36-AKT-mTOR signaling and ultimately results in the potentiation of antitumor immunity (Fig. 7H). These findings suggest a novel strategy for enhancing antitumor immunity by targeting TCA cycle in cancer cells.

CPI-613 is a novel therapeutic agent targeting TCA cycle through inhibition of PDH and α-KGDH(26). CPI-613 is freely soluble (~140–174 mg/mL) in acetone, chloroform, ethyl acetate, tetrahydrofuran, and toluene and practically insoluble in water (~ 0.014 mg/mL), and it has been mainly used in treating pancreatic cancer(27–29). In addition to its potent anticancer activity, CPI-613 has been well-tolerated in clinical trials and is now under clinical investigation in combination with different chemotherapeutics in treating metastatic pancreatic cancer (NCT03504423), biliary tract cancer (NCT04203160), clear cell sarcoma (NCT04593758) and relapsed/refractory acute myeloid leukemia (NCT03504410). One of our mechanistic studies shows that CPI-613 exhibits anticancer activity in pancreatic cancer cells by inducing apoptosis, increasing autophagy, and suppressing lipid metabolism. This is achieved through the activation of the AMPK-acetyl-carboxylase (ACC) signaling pathway (11). In HNSCC cells, CPI-613 treatment leads to an increase in GLS1-mediated glutaminolysis (12). The present study is the first to demonstrate that CPI-613 exerts an immunomodulatory effect by disrupting the THBS2-CD36 axis between HNSCC cells and CD8⁺ T cells. This mechanism may differ among different cancer types; thus, further studies in a broader range of cancers are needed to determine the general applicability and impact of our findings.

XBP1 has two isoforms: XBP1u (unspliced) is the inactive form, while XBP1s (spliced) is the active form when unfolded or misfolded proteins accumulate in the ER. XBP1s is a critical TF that plays a vital role in the cellular response to ER stress by regulating gene expression involved in protein folding and ER-associated degradation (30). Our RNA-seq data revealed that CPI-613 activates the ER stress pathway, as evidenced by the enrichment of ER stress-related genes, including XBP1s. CPI-613-induced ER stress may be attributed to its impact on mitochondria. When mitochondrial function is compromised, it can lead to an accumulation of unfolded or misfolded proteins, disrupting standard protein folding and trafficking processes in the ER, and resulting in ER stress (31). However, the communication between the ER and mitochondria in the context of CPI-613 remains to be further elucidated. ER stress has been associated with cancer progression, and recent evidence increasingly supports the role of ER stress in the TME and treatment responses (32–34). ER stress signals not only influence the presentation of tumor antigens on major histocompatibility complex (MHC) molecules, which affects how effectively the immune system can recognize and target cancer cells, but also lead to the production of cytokines and chemokines that recruit immune cells to the TME, potentially enhancing antitumor immunity. Here, we demonstrate that CPI-613 suppresses THBS2 expression and secretion in HNSCC cells by inducing the expression of ER stress-associated XBP1s, which in turn enhances the proliferation and cytotoxicity of tumor-infiltrating CD36⁺CD8⁺ T cells. This new mechanism adds another layer to our understanding of the interconnection between ER stress and tumor immunity. At this stage, we cannot exclude the involvement of other genes related to the increase in XBP1s. Further experiments are necessary to explore additional effects associated with CPI-613-induced XBP1s signaling.

The THBS family comprises a group of extracellular matrix (ECM) proteins, including THBS1, THBS2, THBS3, THBS4, and THBS5. Based on their molecular structure, these five THBS proteins are divided into two groups: THBS1 and THBS2 belong to the first group, while THBS3, THBS4, and THBS5 belong to the second group (35). THBS1 has been extensively researched, with diverse functions in tumor progress and immunity. Although THBS2 is known to play roles in tissue repair, angiogenesis, and tumorigenesis, research on its involvement in immunity is particularly limited. A recent report showed that THBS2-high lung adenocarcinoma (LUAD) was associated with decreased immune cell infiltrates and increased immune exhaustion markers (36). Moreover, treatment with THBS2 recombinant protein suppressed in vitro T cell proliferation and promoted in vivo LUAD growth and distant micro-metastasis. Consistently, we found that high THBS2 expression negatively correlates with the presence of CD8⁺ cells in HPV- HNSCC. We also identified THBS2 as the most downregulated protein secreted from HNSCC cells following CPI-613 treatment. Importantly, restoring THBS2 expression in HNSCC cells significantly counteracted the CPI-613-induced increase in proliferation and cytotoxicity of tumor-infiltrating CD8⁺ T cells. It’s crucial to recognize the disparities in immune cell infiltration and composition between HPV+ and HPV- tumors, which partially reflect their intrinsic immunogenicity. The TME of HPV- HNSCC is immunologically “cold” compared to that of HPV+ HNSCC, exhibiting a reduced capacity for immune activation. Our findings suggest that the inverse correlation between THBS2 levels and CD8⁺ T cell infiltration in HPV- HNSCC indicates that THBS2 may play a pivotal role in maintaining a “cold” TME. Elevated THBS2 levels likely contribute to an immunosuppressive or immune-excluded TME by promoting a “cold” tumor phenotype. This study underscores the potential of targeting THBS2 to reverse the immunosuppressive TME and suggests TCA inhibitors like CPI-613 could demonstrate synergistic effects when combined with other therapeutic strategies due to their role in suppressing THBS2. Notably, secretome profiling of primary cells derived from oral cavity squamous cell carcinoma (OCSCC) and pericancerous normal epithelial tissue identified THBS2 as a potentially useful salivary marker for OCSCC detection(37). Undoubtedly, THBS2 could also be developed as a biomarker to predict the development of HNSCC.

CD36 plays a role in lipid metabolism and lipid peroxidation, impairing the production of cytotoxic cytokines and weakening the antitumor activity of CD8⁺ T cells(38,39). Here, we show that CPI-613 effectively reduces THBS2 secretion by inhibiting its expression in HNSCC cells. The reduction in THBS2 decreases its binding to CD36 on the surface of tumor-infiltrating CD8⁺ T cells, subsequently activating the AKT-mTOR signaling pathway to promote T cell proliferation and the potential production of cytotoxic cytokines. Although CPI-613 does not affect the proliferation of intratumoral CD36⁻CD8⁺ T cells, it enhances the cytotoxicity of this T cell subset (Fig. 7G). We were aware that CPI-613 may induce pyroptosis, a form of immunogenic cell death, to activate CD8⁺ T cells (40). However, only CPI-613 at the higher dosage than that used in this study (200 μM vs. 50 μM) has the potential to induce pyroptosis in HNSCC cells (Supplementary Fig. S14). Therefore, further investigations are warranted to explore CD36-independent mechanisms involved in CPI-613-induced T cell immunity.

In summary, we demonstrated for the first time that blocking the TCA cycle leads to an enhanced T cell-mediated antitumor response against HNSCC. This effect is due, at least in part, to the disruption of the ER stress-associated THBS2-CD36 axis between tumor and effector T cells. These novel findings provide valuable insights into the complex cell interactions within the TME. Our study also highlights the potential of CPI-613 as an immunomodulatory agent in cancer therapy and provides a basis for further exploration of therapeutic strategies using CPI-613 in combination with immune checkpoint inhibitors or other immunotherapeutic approaches to achieve durable and curative outcomes.

Abbreviation

α-KGDH

α-ketoglutarate dehydrogenase

ACC

AMPK-acetyl-carboxylase

ChIP

Chromatin immunoprecipitation

CM

Conditioned medium

CPTAC

Clinical Proteomic Tumor Analysis Consortium

DEPs

Differentially expressed proteins

Dox

Doxycycline

ECM

Extracellular matrix

ER

Endoplasmic reticulum

FBS

Fetal bovine serum

HN12-ER

ER-BioID^HA stably expressing HN12 cells

HNSCC

Head and neck squamous cell carcinoma

IACUC

Institutional Animal Care and Use Committee

KO

knockout

IF

Immunofluorescence

IP

Immunoprecipitation

IRB

Institutional Review Board

LC-MS

Liquid chromatography-mass spectrometry

LUAD

Adenocarcinoma

MDSCs

Myeloid-derived suppressor cells

MHC

Major histocompatibility complex

OCSCC

Oral cavity squamous cell carcinoma

OCR

Oxygen consumption rate

PBMC

Human peripheral blood mononuclear cell

PDH

Pyruvate dehydrogenase

RNA-seq

RNA sequencing

SA

Streptavidin

SD

Standard deviation

SSO

Sulfo-N-succinimidyl oleate

TCA

Tricarboxylic acid

TCGA

The Cancer Genome Atlas

TG

Thapsigargin

THBS2

Thrombospondin-2

TME

Tumor microenvironment

Tregs

Regulatory T cells

XBP1s

Sliced X-box binding protein 1