G3BP2-K76 acetylation promotes tumor immunoescape by stabilizing PD-L1 expression in colorectal cancer

Background

Colorectal cancer (CRC) often evades immune surveillance, leading to poor responses to PD-1/PD-L1 blockade immunotherapy. The underlying mechanisms regulating PD-L1 expression remain incompletely understood.

Methods

We utilized co-immunoprecipitation (Co-IP) and RNA immunoprecipitation (RIP) to investigate protein-protein and protein–RNA interactions. Western blot (WB) and quantitative PCR (qPCR) were applied to assess protein and mRNA expression levels, respectively. Immunocompetent mouse models and patient-derived CRC samples were used to evaluate clinical relevance and therapeutic targeting.

Results

We identified acetylation of G3BP2 at lysine 76 (K76) as a key regulator of PD-L1-mediated immune evasion. This modification is catalyzed by p300 and reversed by HDAC6. Acetylated G3BP2 enhances binding to PABPC1 and PD-L1 mRNA, increasing its stability and upregulating PD-L1 expression. Consequently, tumors exhibit reduced cytotoxic T lymphocyte infiltration and acquire resistance to anti-PD-L1 therapy. Elevated G3BP2-K76 acetylation was observed in immunotherapy-resistant CRC tissues and promoted tumor growth in mice. A peptide inhibitor targeting K76 acetylation (K76-pe) effectively suppressed PD-L1 expression and synergized with anti-PD-L1 treatment in vivo.

Conclusion

G3BP2-K76 acetylation represents a critical post-translational mechanism driving immune evasion in CRC. Targeting this pathway may provide a promising strategy to overcome resistance to immunotherapy.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12964-026-02804-4.

Highlights

G3BP2 acetylation at lysine 76 promotes immune evasion in colorectal cancer by enhancing its binding to PABPC1 and PD-L1 mRNA, thereby stabilizing PD-L1 expression and reducing cytotoxic T cell infiltration. This modification is mediated by p300 and reversed by HDAC6. Targeting G3BP2-K76 acetylation with a specific peptide restores anti-tumor immunity and improves response to PD-L1 blockade, highlighting its critical role in regulating immune escape and its potential as a therapeutic target.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12964-026-02804-4.

Introduction

Colorectal cancer (CRC) is a major global health issue, with 1,926,118 new cases and 903,859 deaths reported annually [1]. Under normal circumstances, the adaptive immune system is capable of recognizing and eliminating nascent tumors [2, 3]. However, CRC encompasses a heterogeneous group of diseases, each characterized by distinct genetic alterations. CRC tumors are typically associated with a highly immunosuppressive microenvironment, which enables cancer cells to evade immune surveillance and facilitates tumor progression [4, 5]. Thus, understanding the mechanisms behind CRC immune evasion is crucial for developing effective therapeutic strategies.

Post-translational modifications (PTMs), such as phosphorylation, ubiquitination, acetylation, and methylation, are known to dynamically regulate protein stability, localization, and catalytic activity. These modifications play a pivotal role in organizing and controlling biological systems [6, 7]. Among them, protein lysine acetylation is a widespread and dynamic PTM, with studies showing extensive acetylation of non-histone proteins [8, 9]. Protein lysine acetylation occurs through two distinct mechanisms: nonenzymatic acetylation, where acetyl phosphate (AcP) or acetyl-coenzyme A (Ac-CoA) donates the acetyl group, and enzymatic acetylation, which depends on lysine acetyltransferases to transfer the acetyl group from Ac-CoA [10, 11]. As a key regulator of various cellular processes, acetylation has significant implications in cancer biology. Understanding the acetylation of key oncogenes is crucial for elucidating the mechanisms of cancer progression and developing novel therapeutic strategies.

RasGAP SH3 domain-binding protein 2 (G3BP2) is an RNA-binding protein that plays a critical role in stress granule (SG) assembly [12]. Recent studies have underscored G3BP2’s pivotal functions in regulating key cellular processes, including growth, proliferation, and apoptosis [13, 14]. As an RNA-binding protein, G3BP2 contributes to post-transcriptional regulation by modulating mRNA stability, localization, and translation through interactions with both coding and noncoding RNAs [15–17]. Dysregulated expression of RNA-binding proteins such as G3BP2 disrupts gene expression homeostasis and contributes to various human diseases, including cancer. G3BP2 has been identified as an oncogene in prostate cancer, head and neck squamous cell carcinoma, and esophageal carcinoma. A study by Chen et al. demonstrated that phosphorylated G3BP2 promotes p53 degradation, leading to radiotherapy resistance in colorectal cancer (CRC) [18]. Despite these insights, it remains unclear whether G3BP2 undergoes acetylation and how acetylation may modulate its function, particularly in the context of CRC immune evasion. Further investigation into G3BP2 acetylation is critical to address this gap in our understanding.

In this study, we identified an acetylation modification at lysine 76 in the G3BP2 amino acid sequence, with significantly elevated levels of G3BP2 K76 acetylation in immunotherapy resistance CRC tissues compared to sensitive tissues. Furthermore, we found that the acetylation of G3BP2 K76 is regulated by the acetyltransferase p300 and the deacetylase HDAC6. Mechanistically, we discovered that G3BP2 K76 acetylation enhances G3BP2’s interaction with PABPC1 and its binding to PD-L1, thereby promoting the stability of PD-L1 mRNA and increasing PD-L1 protein levels. This results in resistance to immunotherapy. Functionally, in immunocompetent mice, G3BP2 K76 acetylation significantly increased tumor size. These findings reveal the complex post-translational regulatory mechanisms governing G3BP2, offering valuable insights for developing targeted therapeutic strategies in CRC treatment.

Methods

Tissue specimens and cell lines

Twelve pairs of fresh-frozen primary CRC tissues were meticulously collected from the Sixth Affiliated Hospital of Sun Yat-sen University (Guangzhou, China), with approval from the Institutional Review Board and Ethics Committee. Tissue samples were exclusively obtained from patients with a confirmed diagnosis of colorectal adenocarcinoma, verified by colonoscopic pathology biopsy, and who had not undergone neoadjuvant therapy. Human embryonic kidney (HEK293) and colorectal cancer (HCT116) cell lines were sourced from the American Type Culture Collection (ATCC). Cells were cultured in RPMI-1640 or Dulbecco’s Modified Eagle Medium (DMEM), both from Gibco, and supplemented with 10% fetal bovine serum (FBS) in a humidified incubator at 37 °C with 5% CO2. All cell lines were authenticated via short tandem repeat (STR) fingerprinting and tested negative for mycoplasma contamination before experimentation.

Patients with PD-L1 treatment

We established a cohort of 36 patients with colorectal cancer (CRC) who received pembrolizumab treatment. All participants met the following eligibility criteria: pathological confirmation of CRC, prior anti-PD-L1 treatment for at least four weeks, availability of adequate tumor biopsy samples, baseline patient and disease information, and sufficient archival tissue for analysis. Endoscopic biopsy samples were collected from each patient, fixed in formalin, and paraffin-embedded. Clinical, radiographic, and treatment-related data were retrospectively reviewed. Written informed consent was obtained from all donors, and the study procedures complied with the Declaration of Helsinki. Based on objective clinical response evaluated using the Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1, the cohort was equally divided into responder (sensitivity) and non-responder (resistance) groups, with 18 patients in each.

Immunoblotting analysis and antibodies

The indicated cells were lysed using RIPA buffer (10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Triton X-100, 0.1% SDS, 1 mM EDTA), supplemented with a protease inhibitor cocktail. Protein samples were separated by polyacrylamide gel electrophoresis (PAGE) and subsequently transferred to polyvinylidene fluoride (PVDF) membranes (Millipore). The membranes were blocked with 5% bovine serum albumin (BSA) at room temperature (RT) for 1 h and incubated with primary antibodies overnight at 4 °C. The membranes were then incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies at RT for 1 h. The signal was detected using the Azure 400 biosystem (Azure, USA). Band intensities were quantified using ImageJ software. The following antibodies were used: anti-G3BP2 antibody (1:1000, ab86135, Abcam), anti-acetyllysine (Pan-Ac) antibody (1:1000, PTM-101, PTM), anti-β-actin (1:10000, 81115-1-RR, Proteintech), DYKDDDDK Tag anti-flag antibody (1:1000, 14793, CST), anti-HA-Tag antibody (1:1000, 3724, CST), anti-myc tag antibody (1:1000, 2276 S, CST), anti-p300 antibody (1:1000, ab275378, Abcam), anti-HDAC6 antibody (1:1000, ab239362, Abcam), anti-Histone H3 antibody (1:5000, 68345-1-Ig, Proteintech), and anti-PD-L1 antibody (1:1000, ab228415, Abcam).

Immunoprecipitation

Cells were lysed using Immunoprecipitation Lysis Buffer (P0013, Beyotime), supplemented with a protease inhibitor cocktail, for 30 min at 4 °C. Protein A/G magnetic beads (P2108, Beyotime) were used to preclear the cell extracts, which were then incubated overnight at 4 °C with either IgG (1:50, 30000-0-AP, Proteintech) or the indicated antibodies (1:50). The immune complexes were subsequently incubated with protein A/G beads for 2 h. The immunoprecipitates were washed and analyzed via immunoblotting (IB).

Construction of plasmids and cell transfection

The pLKO-G3BP2 shRNA (5’-CGGGAGTTTGTGAGGCAATAT-3’), pLKO-p300 shRNA (5’-CAATTCCGAGACATCTTGAGA-3’), pLKO-HDAC6 shRNA (5′-CGGTAATGGAACTCAGCACAT-3′), and pLKO-PABPC1 shRNA (5’-AGCTGTTCCCAACCCTGTAAT-3’) were designed and cloned into the pLKO-puro vector. The pLKO shRNAs and the control pLKO.1 vector were purchased from Guangzhou Boya Biotechnology Development Co., Ltd. Plasmids encoding myc-G3BP2 wild-type (WT), myc-G3BP2 K76R, myc-G3BP2 K76Q, myc-PABPC1, the eleven deacetylases (HDAC1-11), and the six acetyltransferases (PCAF, KAT8, p300, GCN5, Tip60, and CBP) were obtained from Guangzhou Boya Biotechnology Development Co., Ltd. All plasmids were verified by sequencing. Plasmids were transfected into cells via lentiviral infection as described previously [19]. To establish stable cell lines expressing G3BP2, G3BP2 K76R, or G3BP2 K76Q, the corresponding plasmids were transfected into G3BP2 knockout (KO) cells. Transfection efficiencies were confirmed by qRT-PCR or immunoblotting.

In vitro acetylation and deacetylation assay

Plasmids expressing wild-type myc-G3BP2, its mutant myc-K76R, or p300-FLAG were transfected into the indicated cells. The proteins myc-G3BP2, myc-G3BP2 K76R, and p300-FLAG were then purified according to the instructions provided in the FLAGIPT1 or IP0010 kit (Sigma). The in vitro deacetylation assay was performed as previously described with minor modifications [20]. Briefly, immunopurified wild-type G3BP2 or K76R was incubated with immunopurified p300 in HAT buffer (Millipore) in a shaking incubator at 30 °C for 1 h. The acetylation of G3BP2 was assessed via immunoblotting (IB) using the Pan-Ac antibody.

The myc-G3BP2 plasmid was co-transfected into the indicated cells with the p300 vector to induce K76 acetylation of G3BP2. Simultaneously, the FLAG-HDAC6 plasmid was introduced into the indicated cells. The proteins myc-G3BP2 and FLAG-HDAC6 were then purified according to the instructions provided in the FLAGIPT1 or IP0010 kit (Sigma). The in vitro deacetylation assay was performed as previously described [20]. Immunopurified myc-G3BP2 containing Ac-K76-G3BP2 was incubated with immunopurified HDAC6 in HEPES buffer (40 mM HEPES, 1 mM MgCl2, 1 mM DTT, and 5 mM NAD+) at 37 °C for 1 h. The acetylation status of G3BP2 was evaluated via immunoblotting (IB) using the Pan-Ac antibody.

RNA immunoprecipitation assay

RNA immunoprecipitation (RIP) assays were performed using 5 µg of anti-myc tag antibody and the Magna RIP™ Quad RNA-Binding Protein Immunoprecipitation Kit (17–704, Millipore) following the manufacturer’s instructions. RT-qPCR was used to analyze the enrichment of RNAs, including GAPDH.

mRNA half-life analysis

Cells were seeded in six-well plates and cultured for one day before being treated with 5 µg/mL actinomycin D for various durations. Total RNA was then extracted using TRIzol (15596018CN, Thermo Fisher) and analyzed by RT-qPCR. An mRNA decay curve was generated, with the value of untreated cells set to 100%. The time point at which 50% of the mRNA remained was defined as the mRNA half-life.

PD-L1 forward primer: 5’-TTGCTGAACGCCCCATACAA-3’

PD-L1 reverse primer: 5’-GGAATTGGTGGTGGTGGTCT-3’

Pd-l1 forward primer: 5’-CGCCTGCAGATAGTTCCCAA-3’

Pd-l1 reverse primer: 5’-ATCGTGACGTTGCTGCCATA-3’

Xenograft mouse model

All animal experiment procedures were conducted in accordance with the guidelines of the Public Health Service Policy on Humane Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee (IACUC). Male C57BL/6 or nude mice (18–20 g), aged 4–6 weeks, were subcutaneously injected with approximately 2.5 × 10⁶ indicated cells per mouse into the flank, with 5 mice per group. Tumor diameters were measured using a caliper every 2 days. After 3 weeks, the mice were sacrificed, and the xenografts were harvested. For the anti-PD-L1 antibody experiment, mice were randomized into two groups: control (DMSO) and anti-PD-L1 antibody treatment. When tumors reached a size of 3 mm, mice were injected with either IgG or anti-PD-L1 antibody (Invitrogen CD274 (PD-L1, B7-H1) Monoclonal Antibody (MIH5), Functional Grade, ) at a dose of 200 µg per mouse every 2 days. After 3 weeks, the mice were sacrificed, and the xenografts were harvested.

Statistical analysis

Statistical analyses were conducted using the Statistical Package for the Social Sciences software (v26.0; SPSS Inc., Chicago, IL, USA). Variables were summarized as mean ± standard deviation (SD). An unpaired Student’s t-test was used to evaluate statistical significance between two groups. The number of technical and experimental replicates is provided in the figure legends for each experiment. P values less than 0.05 were considered significant and were represented graphically as follows: *P < 0.05; **P < 0.01; ***P < 0.001, unless otherwise specified.

Results

G3BP2 is acetylated at K76

Given the limited literature on G3BP2 acetylation, this study aims to determine whether G3BP2 is acetylated and to investigate the functional impact of this modification. To address this, the acetylation of G3BP2 was examined using a pan-anti-acetylated antibody (Pan-Ac) following immunoprecipitation of G3BP2 from lysates of HEK293T, HCT116, and MC38 cells (Fig. 1A-B and Supplementary Figure S1A). Similarly, when immunoprecipitation was performed using the Pan-Ac antibody, the presence of G3BP2 was detected using an anti-G3BP2 antibody (Fig. 1C-D and Supplementary Figure S1B). These results suggest that endogenous G3BP2 is acetylated. Based on this observation, lysates from G3BP2 knockout (KO) cells transfected with Myc-G3BP2 were immunoprecipitated using a Myc antibody, and acetylation of exogenous G3BP2 was detected using the Pan-Ac antibody (Fig. 1E-F and Supplementary Figure S1C-F). The results indicate that exogenous G3BP2 is also acetylated.

Fig. 1.

G3BP2 is acetylated at K76. A-B Immunoblotting (IB) analysis of endogenous G3BP2 acetylation, using a Pan-Ac antibody, in whole-cell lysates (WCL) from HEK293T (A) and HCT116 (B) cells. Acetylation was evaluated by immunoprecipitation (IP) with a G3BP2-specific antibody. C, D IB analysis of endogenous G3BP2 acetylation, detected with a G3BP2 antibody, following IP with a Pan-Ac antibody, in WCL from HEK293T (C) and HCT116 (D) cells. E, F After transfection of Myc-G3BP2 plasmid into G3BP2 knockout (KO) HEK293T (E) and HCT116 (F) cells, IP was performed with a Myc antibody, and acetylation of exogenous G3BP2 was assessed using a Pan-Ac antibody. G Mass spectrometry analysis was employed to identify the acetylation site of purified G3BP2 from HEK293T cells. H After the transfection of Myc-G3BP2, Myc-G3BP2 K76R, or Myc-G3BP2 K76Q plasmids into G3BP2 KO HEK293T cells, IP was carried out using a Myc antibody, and acetylation of exogenous G3BP2 was analyzed by IB with a Pan-Ac antibody

To identify acetylation sites on G3BP2, the protein was purified from cells and analyzed using liquid chromatography-tandem mass spectrometry (LC-MS). This analysis identified lysine 76 (K76) as a potential acetylation site on G3BP2 (Fig. 1G and Supplementary Figure S1G-H). To confirm acetylation at K76, the lysine residue was mutated to arginine (K76R) or glutamine (K76Q) to mimic the absence or presence of acetylation, respectively. Upon re-expression of wild-type (WT) G3BP2 and its mutants (K76R and K76Q) in G3BP2 KO cells, acetylation detected by the Pan-Ac antibody was significantly reduced in the K76R and K76Q mutants compared to G3BP2-WT (Fig. 1H and Supplementary Figure S1I). Additionally, the K76 residue and its surrounding sequence were highly conserved across different species (Figure S1J). These findings suggest that K76 acetylation represents the predominant acetylation modification of G3BP2.

Acetyltransferase p300 acetylates G3BP2 at K76

To confirm the acetyltransferase responsible for regulating G3BP2 K76 acetylation, six acetyltransferases—p300, CBP, GCN5, Tip60, MOB1, and KAT8—were examined. G3BP2 knockout (KO) cell lines with ectopic expression of myc-G3BP2 were transfected with these six acetyltransferases. Overexpression of p300, but not the other five acetyltransferases, significantly enhanced the K76 acetylation level of G3BP2 (Fig. 2A). To further investigate, interactions between G3BP2 and p300 were analyzed at both endogenous and exogenous levels using co-immunoprecipitation assays (Fig. 2B, C). Additionally, increased expression of p300 via an overexpression plasmid resulted in a dose-dependent increase in the acetylation level of endogenous G3BP2 (Fig. 2D and Supplementary Figure S2A-B).

Fig. 2.

p300 is the acetyltransferase for G3BP2 K76 acetylation. A After overexpression of critical acetyltransferases in Myc-G3BP2 HEK293T cells, immunoprecipitation (IP) was performed using a Myc antibody, and acetylation of exogenous G3BP2 at K76 was assessed with a Pan-Ac antibody. B IP was performed using a G3BP2 antibody, and the complexes of endogenous G3BP2 and p300 were assessed with G3BP2 and p300 antibodies. C Following the transfection of Myc-G3BP2 and HA-p300 plasmids into HEK293T cells, IP was conducted using a Myc antibody, and the Myc-G3BP2 and HA-p300 complexes were assessed using Myc and HA antibodies. D After transfection of HEK293T cells with the indicated dose of HA-p300 plasmid, IP was performed using a G3BP2 antibody, and acetylation of endogenous G3BP2 at K76 was assessed with a Pan-Ac antibody. E After transfection of Myc-G3BP2-WT, Myc-G3BP2 K76R, or Myc-G3BP2 K76Q plasmids with HA-p300 into G3BP2 KO HEK293T cells, IP was conducted using a Myc antibody, and acetylation of exogenous G3BP2 was assessed with a Pan-Ac antibody. F Purified Myc-G3BP2 was incubated with or without purified HA-p300 and acetyl-CoA, and the acetylation of G3BP2 was assessed using the Pan-Ac antibody

Conversely, in G3BP2 KO cells transfected with various exogenous G3BP2 constructs, p300 failed to induce acetylation of G3BP2 at the K76 site when it was mutated to K76R or K76Q (Fig. 2E and Supplementary Figure S2C-D). Furthermore, the vitro acetylation assays demonstrated that immunopurified p300 directly and efficiently acetylated immunopurified wild-type (WT) G3BP2, whereas this activity was absent in the K76R and K76Q mutants (Fig. 2F). Collectively, these findings identify p300 as the acetyltransferase responsible for G3BP2 K76 acetylation.

HDAC6 is the G3BP2 deacetylase

Reversible acetylation is catalyzed by lysine acetyltransferases and reversed by lysine deacetylases, which are classified into two groups: classical zinc-dependent histone deacetylases (HDACs) and NAD⁺-dependent sirtuins (SIRTs). To identify the deacetylase responsible for G3BP2 regulation, the deacetylase inhibitors nicotinamide (NAM) and trichostatin A (TSA) were used to target the SIRT family and HDAC family of deacetylases, respectively. The results showed that G3BP2 acetylation significantly increased following TSA treatment, whereas NAM had no significant effect (Fig. 3A). Based on this observation, G3BP2 knockout (KO) cell lines with ectopic expression of myc-G3BP2 were transfected with HDAC1-11. Overexpression of HDAC6, but not the other ten HDACs, decreased the K76 acetylation level of G3BP2 (Fig. 3B). Co-immunoprecipitation assays revealed that HDAC6 interacts with G3BP2, indicating that HDAC6 could be co-immunoprecipitated with G3BP2 (Fig. 3C, D). To further confirm this, increased expression of HDAC6 via an overexpression plasmid resulted in a dose-dependent decrease in the acetylation level of endogenous G3BP2 (Fig. 3E and Supplementary Figure S3A-B).

Fig. 3.

HDAC6 deacetylated G3BP2 K76 acetylation. A After HEK293T cells were treated with NAM or TSA, immunoprecipitation (IP) was performed using a G3BP2 antibody, and the acetylation of endogenous G3BP2 at K76 was assessed with a Pan-Ac antibody. B After overexpression of HDAC1-11 in Myc-G3BP2 HEK293T cells, IP was performed using a Myc antibody, and the acetylation of exogenous G3BP2 at K76 was assessed with a Pan-Ac antibody. C IP was conducted using a G3BP2 antibody, and the complexes between endogenous G3BP2 and HDAC6 were assessed with G3BP2 and HDAC6 antibodies. D After transfection of Myc-G3BP2 and Flag-HDAC6 plasmids into HEK293T cells, IP was performed using a Myc antibody, and the complexes between Myc-G3BP2 and Flag-HDAC6 were assessed with Myc and Flag antibodies. E After HEK293T cells were transfected with the indicated dose of Flag-HDAC6 plasmid, IP was performed using a G3BP2 antibody, and acetylation of endogenous G3BP2 at K76 was assessed with a Pan-Ac antibody. F Purified Myc-G3BP2 was incubated with purified Flag-HDAC6 and acetyl-CoA, and acetylation of G3BP2 was assessed with a Pan-Ac antibody. G After downregulation of HDAC6 in Myc-G3BP2-WT, Myc-G3BP2 K76R, or Myc-G3BP2 K76Q HEK293T cells, IP was conducted using a Myc antibody, and the acetylation of exogenous G3BP2 was assessed with a Pan-Ac antibody

Furthermore, immunopurified HDAC6 directly and efficiently reduced the acetylation level of immunopurified G3BP2 in a dose-dependent manner (Fig. 3F). Knockdown of HDAC6 markedly enhanced the acetylation levels of G3BP2; however, this effect was not observed in the K76R and K76Q mutants (Fig. 3G and Supplementary Figure S3C-D). Collectively, these findings demonstrate that HDAC6 functions as the deacetylase for G3BP2 at the K76 residue.

K76 acetylated G3BP2 enhanced binds to the PD-L1 mRNA and promote its stability

Protein acetylation can influence the intracellular localization, stability, and binding affinity of proteins to other molecules. To determine whether G3BP2 K76 acetylation affects its intracellular localization, cellular fractionation assays were performed. Previous studies have reported that G3BP2 is primarily localized in the cytoplasm but is also detectable in the nucleus. Our findings confirm that G3BP2 is present in both the cytoplasm and nucleus in WT, K76R, and K76Q cells, with the cytoplasm-to-nucleus ratio remaining consistent across these groups (Fig. 4A). These results indicate that K76 acetylation does not affect intracellular localization. We next analyzed whether K76 acetylation impacts the stability of G3BP2. Following treatment with cycloheximide (CHX) for 0, 2, 4, and 8 h, intracellular G3BP2 levels in WT, K76R, and K76Q cells were assessed via immunoblotting (IB). The results revealed no significant differences in G3BP2 expression levels at any time point (Fig. 4B), suggesting that the protein half-life of G3BP2 remains unchanged regardless of K76 acetylation status.

Fig. 4.

K76 acetylated G3BP2 bindings and stabilized to PD-L1 mRNA. A IB depicting G3BP2 in cytoplasm, nucleus, and cellular membrane fractions derived from G3BP2 KO HEK293T cells with stable reexpression of WT G3BP2 or its mutants K76R or K76Q. B IB depicting G3BP2 after indicated HEK293T cells treated by cycloheximide (CHX) for 0, 2, 4, and 8 h. C-D RIP lysate prepared from indicated HEK293T (C) and HCT116 (D) cells were immunoprecipitated using G3BP2 antibody, and associated RNA level was validated by qRT-PCR. E, F After overexpression p300 in indicated HEK293T (E) and HCT116 (F) cells, RIP lysates were immunoprecipitated using either a normal rabbit IgG or anti-G3BP2 antibody, and associated RNA was validated by qRT-PCR. G Indicated HEK293T cells were treated with 5 μm actinomycin D for 0, 2, 4, and 6 h. Stability of PD-L1 mRNA was assessed by qRT-PCR. H After overexpression HDAC6 in G3BP2-WT HCT116 cell, qRT-PCR was performed to detect stability of PD-L1 mRNA. I IB depicting the expression of PD-L1 in G3BP2-WT HCT116 cells after overexpression HDAC6

Since K76 acetylation does not affect subcellular localization or protein stability, we investigated its role in RNA-binding ability. Published studies have shown that G3BP2 binds to PD-L1, CDK7, CDK9, and c-Myc mRNAs to regulate mRNA synthesis and degradation, thereby modulating cellular functions [21–23]. To explore whether K76 acetylation affects the interaction between G3BP2 and PD-L1 mRNA, RNA immunoprecipitation (RIP) assays were conducted. Total lysates were immunoprecipitated with either normal rabbit IgG or anti-myc antibody, and the immunoprecipitated RNAs were analyzed via RT-qPCR. The results showed that K76 deacetylation (K76R) significantly impaired the binding between G3BP2 and PD-L1 mRNA compared to WT G3BP2, while the acetylation-mimicking K76Q mutant retained its binding ability (Fig. 4C-D and Supplementary Figure S4A). This effect was specific to PD-L1 mRNA, as no differences were observed with other mRNAs (Supplementary Figure S4B-D). Subsequent experiments demonstrated that increasing the acetylation level of G3BP2 via p300 overexpression markedly enhanced its interaction with PD-L1 mRNA, whereas this enhancement was not observed for the K76R mutant (Fig. 4E-F and Supplementary Figure S4E).

Next, we investigated how acetylated G3BP2 affects PD-L1 expression. G3BP2 is known to regulate mRNA stability, localization, and translation, and previous studies have shown that G3BP2 stabilizes PD-L1 mRNA [21]. Following treatment with actinomycin D (5 µM) for different durations, PD-L1 mRNA decay was analyzed. The results revealed a substantial decrease in PD-L1 mRNA stability in K76R cells compared to WT cells, whereas PD-L1 mRNA stability was maintained in K76Q cells (Fig. 4G). Similarly, PD-L1 protein levels were reduced in K76R cells compared to WT and K76Q cells (Supplementary Figure S4F). Furthermore, reducing the acetylation level of G3BP2 via HDAC6 overexpression significantly decreased the stability of PD-L1 mRNA (Fig. 4H and Supplementary Figure S4G) and PD-L1 protein levels (Fig. 4I and Supplementary Figure S4H). Collectively, these findings demonstrate that K76 acetylation plays a critical role in promoting the binding of G3BP2 to PD-L1 mRNA and enhancing PD-L1 mRNA stability.

K76 acetylated G3BP2 interacts with PABPC1 to bind PD-L1 mRNA and promote its stability

G3BP2 and poly-A binding protein (PABP) are key components of stress granules. Previous studies have reported that G3BP2 interacts with PABPC1 to promote mRNA stability [24, 25]. Therefore, we hypothesized that K76 acetylation of G3BP2 might influence its interaction with PABPC1, thereby modulating G3BP2’s regulation of PD-L1. Using co-immunoprecipitation (co-IP), we confirmed the interaction between G3BP2 and PABPC1 (Fig. 5A-B). Notably, the co-IP results demonstrated that K76 deacetylation (K76R) significantly impaired the interaction between G3BP2 and PABPC1 compared to WT G3BP2, while the acetylation-mimicking K76Q mutant retained its ability to interact with PABPC1 (Fig. 5C). Increasing the acetylation level of endogenous G3BP2 via p300 overexpression markedly enhanced the interaction between G3BP2 and PABPC1 (Fig. 5D and Supplementary Figure S5A-B). However, when PABPC1 was knocked out, the binding ability of G3BP2 to PD-L1 mRNA did not decrease (Fig. 5E and Supplementary Figure S5C-D), but the stability of PD-L1 mRNA was significantly reduced (Fig. 5F and Supplementary Figure S5E-F). Collectively, these findings suggest that K76-acetylated G3BP2 enhances its interaction with PABPC1, thereby promoting PD-L1 mRNA stability.

Fig. 5.

G3BP2 K76 acetylation promotes PD-L1 stabilization through interact with PABPC1. A, B Immunoprecipitation (IP) was performed using G3BP2 and PABPC1 antibodies in whole-cell lysates (WCL) of HEK293T cells, and the G3BP2-PABPC1 complexes were assessed with G3BP2 and PABPC1 antibodies. C IP was conducted using a G3BP2 antibody in WCL of the indicated HEK293T cells, and the G3BP2-PABPC1 complexes were assessed with G3BP2 and PABPC1 antibodies. D After overexpression of p300 in HEK293T cells, IP was performed using a G3BP2 antibody, and the G3BP2-PABPC1 complexes were assessed with G3BP2 and PABPC1 antibodies. E After knock-out of PABPC1 in HEK293 cells, RIP lysates were immunoprecipitated using a G3BP2 antibody, and associated RNA was validated by qRT-PCR. F After knock-out of PABPC1 in HEK293 cells, qRT-PCR was performed to detect the stability of PD-L1 mRNA

K76 acetylated G3BP2 correlates with CRC progression and immune checkpoint inhibitors treatment resistance

Cancer cells often exploit immune checkpoint molecules, including PD-L1, to suppress and evade the immune system [26, 27]. Immunotherapy, particularly anti-PD-1 and anti-PD-L1 therapies, has shown promising potential in colorectal cancer (CRC) [28]. However, due to heterogeneous PD-L1 expression and the presence of other immunosuppressive mechanisms, some CRC patients do not respond to anti-PD-1/PD-L1 treatment [29, 30]. This led us to hypothesize that G3BP2 K76 acetylation may contribute to immunotherapy resistance.

First, we assessed the acetylation level of G3BP2 and the expression of PD-L1 in CRC tissues from 36 patients (supplement Table 1). We assessed that PD-L1 expression and the acetylation level of G3BP2 in CRC tissues (Fig. 6A). Additionally, the density of cytotoxic T lymphocytes (CTLs) was also evaluated (Fig. 6B). The tissues with higher acetylation level of G3BP2 may correlate with the higher expression of PD-L1 and lower infiltration of T lymphocytes. To further investigate the correlation between the acetylation level of G3BP2 with density of cytotoxic T lymphocytes and the PD-L1 expression, we divided the 36 CRC patients into a resistance group and a sensitive group according to their respondence to anti-PD-L1 treatment. We next examined whether G3BP2 acetylation contributes to CRC resistance to PD-L1. The acetylation level of G3BP2 was higher in those who developed resistance to treatment compared to those who were sensitive to immunotherapy (Fig. 6C, Supplementary Figure S6A-C). We assessed and compared the band density in resistance group and sensitive group, the acetylation level of G3BP2 and PD-L1 expression was significantly higher in resistance group (Supplementary Figure S6D-E). However, the mRNA level of PD-L1 had no significant difference between resistance group and sensitive group (Fig. 6D). And the PD-L1expression was associated with acetylation level of G3BP2 (Supplementary Figure S6F). We further explored the CTLs in CRC tissues, the results showed the density of CTLs were much higher in sensitivity group compared with resistance (Fig. 6E-F). These results suggest that G3BP2 acetylation level was correlated with the PD-L1 expression and infiltration of CTLs.

Fig. 6.

G3BP2 K76 acetylation promotes PD-L1 expression and reduces T lymphocyte infiltration in colorectal cancer. A Tumor tissues from 36 colorectal cancer (CRC) patients were lysed, and the expression levels of G3BP2 and PD-L1 were analyzed using the indicated antibodies. The acetylation of endogenous G3BP2 was assessed with a Pan-Ac antibody by immunoprecipitation (IP) using a G3BP2 antibody. B Corresponding paraffin-embedded sections of colorectal cancer (CRC) patients’ tumor tissues were subjected to immunohistochemistry (IHC) to analyze the density of T cells using CD3 + and CD8 + antibodies and the PD-L1 expression level by PD-L1 antibody. C According to their sensitivity to immunotherapy, 36 patients were classified into a sensitive group and a resistant group. Tumor tissues from two groups were lysed, and the expression levels of p300 and PD-L1 were analyzed using the indicated antibodies. The acetylation of endogenous G3BP2 was assessed with a Pan-Ac antibody by immunoprecipitation (IP) using a G3BP2 antibody. D qRT-PCR analysis of PD-L1 in 36 tumor tissues. E Examples of manual gating for CD8 + immune cells with in the CD45 + population from the tissues of sensitive group and resistant group. F Quantification of CD3 + and CD8 + T lymphocytes from immunohistochemistry (IHC)

To further investigate the association between acetylation level of G3BP2 and the expression of PD-L1 which promotes tumor growth by reducing CTL infiltration, G3BP2-WT and G3BP2-K76R MC38 cells were used in a subcutaneous tumor model in C57BL/6 mice. Tumors in the G3BP2-K76R group were significantly smaller than those in the G3BP2-WT group (Fig. 7A-C).

Fig. 7.

G3BP2 K76 acetylation promotes immunotherapy resistance and reduces T lymphocyte infiltration in vivo. A-C Indicated MC38 cells were used to establish a subcutaneous tumor model, Tumor volumes were measured every two days from 12 days (A), and tumor weight was analyzed on day 22 (B), representative pictures of xenograft tumors in mice after inoculation of MC38 cells with G3bp2-WT or with G3bp2-K76R lentivirus (C). D Examples of manual gating for CD8 + immune cells within the CD45 + population from the subcutaneous tumor tissues. E Quantification of CD3 + and CD8 + T lymphocytes from immunohistochemistry (IHC). F Corresponding paraffin-embedded sections of subcutaneous tumor tissues were subjected to immunohistochemistry (IHC) to analyze the density of T cells using CD3 and CD8 antibodies. F-H After injection with anti-PD-1, tumor volumes were calculated every two days, and tumor weight was measured on day 10

Immunohistochemistry and flow cytometry further revealed that the density of CTLs in tumor tissues from the G3BP2-K76R group was significantly higher than that in the G3BP2-WT group (Fig. 7D-F). These results suggest that G3BP2 acetylation promotes tumor growth by decreasing CTL infiltration in vivo. We next examined the efficiency of anti-PD-L1 in C57BL/6 mice injected G3BP2-WT and G3BP2-K76R MC38 cells which were used in a subcutaneous tumor model, which injected with anti-PD-L1 antibody or control IgG. Anti-PD-L1 treatment significantly inhibited tumor growth in the G3BP2-K76R group compared to the IgG control (Fig. 7G), whereas anti-PD-L1 treatment had a limited effect in the G3BP2-WT group (Fig. 7H). These findings indicate that elevated acetylation levels of G3BP2 contribute to resistance against anti-PD-L1 treatment in CRC by reducing CTL infiltration and promoting immunotherapy resistance.

Inhibiting G3BP2 K76 acetylation with a peptide inhibitor enhances anti-PD-1 sensitivity

Since targeting p300 affects multiple signaling pathways, a strategy that specifically targets G3BP2 K76 acetylation is needed. Based on motifs involved in protein post-translational modifications (PTMs), specifically designed peptides have shown promise as an effective strategy to inhibit protein PTMs [31, 32]. After analyzing the sequence around the G3BP2 K76 site, we synthesized five short peptides linked to cell-penetrating peptides (Fig. 8A). Among these, K76-peptide-3# (referred to as K76-pe) exhibited the strongest inhibitory effect on G3BP2 acetylation (Fig. 8B) and was selected for further experiments. Additionally, we designed K76R-pe, a control peptide. The level of G3BP2 acetylation was significantly decreased in cells treated with K76-pe but not in those treated with K76R-pe (Fig. 8C). Furthermore, in cells overexpressing p300, the acetylation level of G3BP2 was also significantly lower in the K76-pe group compared to the K76R-pe group (Fig. 8D). Next, we assessed the stability of PD-L1 mRNA in cells treated with K76-pe and K76R-pe. The results showed that PD-L1 mRNA stability and protein levels were significantly reduced following K76-pe treatment (Fig. 8E-F), whereas no such effect was observed in cells treated with K76R-pe.

Fig. 8.

Targeting G3BP2 K76 acetylation with synthetic peptides modulates PD-L1 expression and tumor progression. A Schematic illustration of the designed peptides. Green: cell-penetrating peptide (CPP); Red: site of G3BP2 acetylation. B After treatment with K76-peptide 1–5#, immunoprecipitation (IP) was conducted using G3BP2 antibody, and G3BP2 acetylation at K76 was assessed using Pan-Ac antibody in MC38 cells. C After treatment with K76-pe or K76R-pe, IP was performed using G3BP2 antibody and Pan-Ac antibody, and G3BP2 acetylation at K76 was assessed with the Pan-Ac antibody in MC38 cells. D After overexpression of p300 in MC38 cells and treatment with K76-pe or K76R-pe, IP was conducted using G3BP2 antibody and Pan-Ac antibody, and the acetylation of G3BP2 at K76 was assessed with the Pan-Ac antibody. E-F Indicated MC38 cells were treated with 5 µM actinomycin D for 0, 2, 4, and 6 h, and PD-L1 mRNA stability was assessed by qRT-PCR (E) and the protein level of PD-1 was depicting by immunoblotting (IB). G-H MC38 cells were used in a subcutaneous tumor model. Tumor weight of mice injected with K76-pe or K76R-pe for 2 weeks. I-J Tumor weight of mice injected with anti-PD-1 and K76-pe or K76R-pe for 2 weeks

We then investigated whether K76-pe could inhibit tumor growth and enhance the efficacy of anti-PD-L1 treatment. In the G3BP2-WT MC38 subcutaneous tumor model, we found that K76-pe did not exhibit obvious toxic effects in mice, as evidenced by the survival rate and normal renal and liver function (Supplementary Figure S6G-I). As shown in Fig. 7G, K76-pe significantly inhibited tumor growth compared to control DMSO treatment. Additionally, K76-pe reduced G3BP2 acetylation levels (Supplementary Figure S6J). Importantly, we found that the combination of K76-pe and anti-PD-L1 treatment enhanced the efficacy of anti-PD-L1 therapy compared to anti-PD-L1 treatment alone (Fig. 8I-J). Taken together, these results suggest that targeting K76 acetylated G3BP2 with K76-pe may enhance the therapeutic efficacy of anti-PD-L1 treatment in cancers with high G3BP2 K76 acetylation levels.

Discussion

Cancer immune evasion is a multistep process orchestrated by dynamic interactions between tumor cells, the tumor microenvironment (TME), and various immune cells [33]. Tumor cells employ multiple strategies to evade immune recognition and elimination, including defective antigen presentation, establishment of an immunosuppressive TME, and direct inhibition of immune effector cells [34, 35]. These mechanisms collectively enable tumor persistence and progression despite ongoing immune surveillance.

Though recent advances in immunology and oncology have uncovered a spectrum of immune evasion pathways, providing deeper insights into how tumors subvert host immunity. Central to these processes are still several key signaling axes, such as the programmed cell death protein 1/programmed cell death ligand 1 (PD-1/PD-L1) pathway, cytotoxic T lymphocyte-associated antigen 4 (CTLA-4), transforming growth factor-β (TGF-β), nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB), and the cyclic GMP-AMP synthase-stimulator of interferon genes (cGAS-STING) pathways [36–42]. These pathways play pivotal roles in mediating immunosuppression within the TME.

Among these, CD274 (commonly known as PD-L1) is the most extensively characterized surface molecule exploited by cancer cells to facilitate immune evasion. PD-L1 exerts its immunosuppressive effects primarily by engaging the co-inhibitory receptor PD-1—typically upregulated during immune activation—on CD8 + cytotoxic T lymphocytes (CTLs), natural killer (NK) cells, and certain myeloid subsets, thereby suppressing their effector functions [41, 43]. Additionally, PD-L1 is abundantly expressed on tumor-infiltrating myeloid cells, further contributing to localized immunosuppression [44]. Similar to normal cells, cancer cells upregulate PD-L1 expression in response to interferon-γ (IFN-γ), predominantly derived from CTLs, as part of a physiological negative feedback loop designed to terminate immune responses. This mechanism explains why malignancies with robust CD8 + CTL infiltration often respond favorably to immune checkpoint inhibitors targeting PD-1 or PD-L1. Furthermore, genetic alterations disrupting the 3’-untranslated region (UTR) of CD274 occur frequently in hematological malignancies, resulting in elevated CD274 mRNA levels that promote immune evasion [45]. Likewise, CD274 gene amplification is commonly observed across various solid tumor types and correlates with increased PD-L1 surface expression on malignant cells [46], underscoring its critical role in tumor immune escape.

Recently, immune checkpoint inhibitors, such as anti-PD-1 and anti-PD-L1 monoclonal antibodies, have shown promising potential in the treatment of colorectal cancer (CRC). However, immunotherapy is currently effective only in CRCs with microsatellite instability-high (MSI-H) status and not in those with microsatellite stable (MSS) status. Moreover, some MSI-H CRC patients develop resistance to immunotherapy. Understanding the mechanisms underlying immunotherapy resistance in CRC and improving the efficacy of immunotherapy are crucial for enhancing patient outcomes. Our findings reveal that G3BP2 K76 acetylation is significantly increased in patients resistant to PD-1 therapy compared to those sensitive to treatment. Mechanistically, acetylation of G3BP2 at K76 enhances its binding to PD-L1 mRNA and interaction with PABPC1, thereby stabilizing PD-L1 mRNA. This results in elevated PD-L1 expression, which promotes resistance to immunotherapy. The use of small-molecule inhibitors specifically targeting G3BP2 K76 acetylation reduces PD-L1 expression and enhances the therapeutic efficacy of immunotherapy.

Post-translational protein modifications, including acetylation, are typically maintained in a dynamic equilibrium under normal physiological conditions, regulated by acetyltransferases and deacetylases. Zhang et al. reported that KAT8 acetylates ALKBH5 at K235, while HDAC7 deacetylates ALKBH5 at the same site. Similarly, Qiu et al. identified GCN5 as the acetyltransferase and SIRT6 as the deacetylase for KAT8 acetylation. Notably, the same enzymes can modify multiple targets. For instance, p300 has been reported to acetylate SF3A2 [47], fascin [48], and Nrf2 [49]. In this study, we found that p300 acetylates G3BP2 at K76. HDAC6 has been shown to deacetylate alpha-tubulin [50], p62 [51], and MOB1 [52]. Here, we demonstrated that HDAC6 is responsible for G3BP2 K76 deacetylation. Our findings further highlight the dynamic balance between acetyltransferases and deacetylases in regulating protein acetylation.

Emerging studies have demonstrated that acetylation acts as a molecular switch that regulates protein function in the progression of colorectal cancer (CRC). Li et al. reported that K505 acetylation of ELMO1 promotes CRC progression by preferentially binding to Dock18 and activating Rac1 [19]. Wang et al. found that K29-acetylated PHF5A induces alternative splicing, stabilizing KDM3A mRNA and promoting its protein expression, which enhances the stress resistance of CRC cells and contributes to colorectal carcinogenesis [53]. Furthermore, KAT8 can also undergo acetylation, and acetylated KAT8 effectively inhibits the migration and invasion of CRC cells without affecting their proliferation [54]. In this study, we found that p300-mediated G3BP2 K76 acetylation upregulates PD-L1 expression, inducing CRC patients’ resistance to immunotherapy. These findings enhance our understanding of the epigenetic regulation of proteins in immunotherapy resistance of CRC and highlight a potential therapeutic target for CRC treatment.

G3BP2 is an RNA-binding protein known for its role in stress granule (SG) assembly. As an RNA-binding protein, G3BP2 participates in post-transcriptional regulation by modulating mRNA stability, localization, and translation through interactions with coding and noncoding RNAs [15–17]. Published studies have shown that G3BP2 binds to PD-L1 mRNA, CDK7 mRNA, CDK9 mRNA, and c-Myc mRNA to regulate mRNA synthesis and degradation, thereby modulating cellular functions [21–23]. In colorectal cancer (CRC), K76-acetylated G3BP2 enhances its binding to PD-L1 mRNA and interaction with PABPC1, thereby increasing PD-L1 mRNA stability. This leads to elevated PD-L1 expression, promoting CRC progression and resistance to immunotherapy. Targeting p300-mediated G3BP2 K76 acetylation using specific peptides could inhibit tumor growth and enhance the efficacy of immunotherapy. Our study contributes to this understanding by revealing that K76 acetylation of G3BP2 promotes CRC progression and immunotherapy resistance by preferentially binding to PD-L1 mRNA and PABPC1. These findings suggest that treatment with specific inhibitors targeting G3BP2 K76 acetylation could serve as a viable strategy to overcome CRC immunotherapy resistance.

In summary, this study identifies elevated acetylation of G3BP2 at K76 as a critical driver of colorectal cancer (CRC) resistance to immunotherapy. Targeting G3BP2 K76 acetylation effectively enhances the efficacy of immunotherapy, providing a promising therapeutic strategy to overcome immunotherapy resistance.

Authors’ contributions

Haiqing Jie: Writing – conceptualization, data curation, methodology, resources, writing – original draft. Writing – review & editing. Xuanhua Yang: data curation, formal analysis, methodology, resources, software. Ziwei Zeng：data curation, formal analysis, investigation, software. Wenfeng Liang: data curation, formal analysis, investigation, software. Wenxin Li: data curation, formal analysis, methodology. Ze Li: data curation, investigation, methodology. Dongxu Lei: formal analysis, methodology. Hao Xie: formal analysis, methodology. Tanxing Cai: data curation, methodology. Zhiping Chen: data curation, formal analysis, software. Yang Li: conceptualization, formal analysis, funding acquisition, project administration. Huashan Liu: conceptualization, data curation, formal analysis, methodology, supervision, writing – review & editing. Liang Kang: funding acquisition, investigation, resources, supervision, writing – review & editing.

Funding

This study was supported by grants from National Natural Science Foundation of China [82403169], Basic and Applied Basic Research Foundation of Guangdong Province [2023A1515111059].

Data availability

All data generated in the present study may be requested from the corresponding authors.

Declarations

Ethics approval and consent to participate

This study was conducted in compliance with the Declaration of Helsinki and received informed consent from all human subjects. The Ethics Committee of The Sixth Affiliated Hospital of Sun Yat-sen University approved the ethics statements for human subjects in this study (No: 2022ZSLYEC-116).

All animal experiments were performed according to the Health Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee (IACUC) of The Sixth Affiliated Hospital of Sun Yat-sen University (No: SYSU-IACUC-2023-000085).

Competing interests

The authors declare no competing interests.

Consent for publication

All authors agree to publication of the article.