ANP32A-mediated histone 3 K27 acetylation is essential for sotorasib activity in KRAS-mutant non-small cell lung cancer.

Results

ANP32A is an unfavorable prognosis factor in KRAS-mutant NSCLC
We investigated the potential function of ANP32A in lung cancer by analyzing The Cancer Genome Atlas dataset. ANP32A was upregulated in both lung adenocarcinoma (LUAD) and lung squamous cell carcinoma (LUSC) (Fig. 1A). Interestingly, the expression of ANP32A was negatively correlated to the overall survival (OS) rate in LUAD but not in LUSC (Figs. 1B and S1A). ANP32A protein expression was upregulated in almost all lung cancer cell lines (Figs. 1C and S1B). Analysis on tissue array confirmed that LUAD tumor tissues (T) expressed higher levels of ANP32A protein than the paired nontumor tissues (Fig. 1D). Although the correlation of ANP32A protein expression level and the OS rate was not significant, LUAD patients with high ANP32A protein expression (high) exhibited shorter survival months than LUAD patients with low ANP32A protein expression (low), and ANP32A protein expression was positively correlated to the World Health Organization grade (Figs. 1, E, F, S1C, and Table 1). Upregulation of ANP32A at both mRNA and protein levels was verified in our LUAD but not LUSC patient cohort (Figs. 1, G, H and S1D). Moreover, ANP32A upregulation was correlated to KRAS mutations (KRAS-Mut) but not other driver mutations, and ANP32A expression level was negatively correlated to the OS rate in these KRAS-mutant lung cancer patients (Figs. 1, I, J and S1, E–I). These findings suggest that ANP32A is an unfavorable prognosis factor in LUAD and may play important roles in KRAS-mutant lung cancer.

ANP32A promotes the proliferation of KRAS-mutant lung cancer cells in vitro
The function of ANP32A in lung cancer was tested by knocking out ANP32A in H2030 and H358 cells (Figs. 2A and S2, A–B). Apparently, ANP32A knockout cells (sgANP) exhibited impaired cell proliferation, colony formation, migration and invasion, and arrested the cell cycle at G2/M phase compared with control cells (sgCtrl) (Figs. 2, B–F and S2, C–F). To verify the specific effect of ANP32A, ANP32A knockdown cells (shANP) and ANP32A knockdown with ANP32A reoverexpression cells (shANP + ANP) were established (Figs. 2G and S2G). As expected, shANP cells showed similar phenotypes to sgANP cells, and these phenotypes were reversed in shANP + ANP cells (Figs. 2, H–L and S2, H–J). In addition, ANP32A deficiency did not significantly induce apoptosis (Fig. S2, K and L). These results suggest that ANP32A upregulation promotes the proliferation, migration, and invasion of lung cancer cells in vitro.

ANP32A is essential for the sensitivity of KRAS-mutant cells to Sot
Given that KRAS-mutant lung cancer expressed higher levels of ANP32A than wildtype lung cancer, we further tested whether ANP32A may play a role in target therapy of KRAS-mutant lung cancer. Notably, ANP32A knockout (sgANP) and knockdown (shANP) H2030 cells showed much higher IC50 of Sot than control cells (sgCrtl and shCtrl), and reoverexpression of ANP32A in shANP cells reduced the IC50 (Fig. 3, A and B). Although Sot treatment (1 μM) significantly caused cell cycle arrest at G1/G0 phase in sgCtrl H2030 cells, it barely affected cell cycle profile in sgANP H2030 cells (Fig. 3C). Moreover, sgANP cells exhibited less Sot-induced apoptosis than sgCtrl cells (Fig. 3D). Interestingly, Sot-resistant H2030 cells (H2030/SR) showed ANP32A downregulation compared with the parental H2030 cells (Fig. 3, E and F). These observations demonstrate that ANP32A may play an essential role in the sensitivity of KRAS-mutant cells to Sot.

ANP32A enhances H3K27 acetylation and interacts with YEATS4
ANP32A has been shown to regulate H3 acetylation (10). Indeed, sgANP cells displayed reduced histone H3 acetylation, particularly at lysine 4 and 27 sites (H3K4Ac and H3K27Ac) (Figs. 4A and S3A). The reduced H3K27Ac was confirmed in shANP H2030 cells, and ANP32A reoverexpression in these cells (shANP + ANP) reversed the effect, whereas ANP32A knockdown only mildly affected H3K4Ac (Figs. 4B and S3B). Since ANP32A is not an acetyltransferase, ANP32A may mediate H3K27Ac by recruiting partners. Analysis on potential ANP32A-interacting proteins (biogrid.org) suggests that BRD4 and YEATS4 involving histone acetylation may interact with ANP32A. FLAG-YEATS4 but not FLAG-BRD4 was coimmunoprecipitated with His-ANP32A (Figs. 4C and S3C). Immunofluorescent experiment demonstrated that merging of ANP32A and YEATS4 or H3K27Ac gave a yellow signal, confirming ANP32A colocalization with YEATS4 and H3K27Ac (Fig. 4D). ANP32A deficiency downregulated YEATS4 rather than BRD4 at protein and mRNA levels, which was rescued by reintroduction of ANP32A (Figs. 4, E–H and S3, D–G). MG132 treatment did not normalize YEATS4 expression in these cells, excluding possible involvement of the proteosome pathway (Fig. S3H). Moreover, H2030/SR cells exhibited downregulation of YEATS4 and H3K27Ac compared with the parental H2030 cells (Figs. 4I and S3I). In support, we found that YEATS4 served as a prognostic factor in KRAS-mutant NSCLC (Fig. 4J). Moreover, YEATS4 expression levels in the high ANP32A expression group (ANP-high) were higher than those in the low ANP32A expression group (ANP-low), and YEATS4 expression was positively correlated with ANP32A expression in KRAS-mutant NSCLC (Fig. 4, K and L). Consistent with the findings for ANP32A, upregulation of YEATS4 at both the mRNA and protein levels was confirmed in our LUAD patient cohort (Fig. 4, M and N). Considering the function of YEATS4 in assembling SRCAP and TIP60–p400–TRRAP complexes for histone acetylation, ANP32A may promote Sot sensitivity by enhancing H3K27Ac via upregulating and interacting with YEATS4.

ANP32A promotes H3K27Ac by upregulating and interacting with YEATS4
We further tested whether H3K27Ac mediated ANP32A function. Chromatin immunoprecipitation (ChIP)–quantitative PCR (qPCR) revealed a significant enrichment of H3K27Ac on the upstream of YEATS4 gene and YEATS4 target gene LMO4 gene promoters, particularly YEATS4 gene in the region between −1200 bp and −1000 bp and LMO4 gene in the region between −800 bp and −200 bp (Figs. 5A and S4A). Furthermore, H3K27Ac in these regions was impaired in shANP cells but restored in shANP + ANP cells (Figs. 5B and S4B). In shANP cells, much less of H3K27Ac was coimmunoprecipitated with FLAG-YEATS4 than that in shCtrl cells, suggesting that ANP32A deficiency impaired the interaction of YEATS4 with H3K27Ac (Figs. 5C and S4C). Restoration of YEATS4 expression in ANP32A knockout cells (sgANP + YEATS4) did not affect ANP32A expression but partially rescued the LMO4 expression (Figs. 5, D, E and S4, D, E). Notably, YEATS4 overexpression in ANP32A-knockout cells (sgANP + YEATS4) restored ANP32A knockout–impaired H3K27Ac, cell proliferation, and cell cycle profile (Figs. 5, E–G and S4, F, G). YEATS4 also potently antagonized Sot-induced cell cycle arrest and apoptosis (Figs. 5, G, H and S4, G, H). Most importantly, shANP + YEATS4 cells exhibited less IC50 of Sot by twofolds than shANP cells (Fig. 5I). These results suggest that ANP32A promotes the proliferation and Sot sensitivity of lung cancer cells by enhancing H3K27Ac to upregulate YEATS4 and interacting with YEATS4.

Inhibition of histone deacetylation enhances Sot efficacy on KRAS-mutant lung cancer in vitro and in vivo
The promoting effect of ANP32A on H3K27Ac and Sot sensitivity in KRAS-mutant lung cancer cells implied that histone acetylation might improve Sot activity. To test this possibility, TSA, as an inhibitor of HDAC, was used to mimic the effect of ANP32A on histone acetylation and treat Sot-resistant H2030 cells. HDAC inhibitors enhance histone acetylation and have been widely used for cancer treatment. H2030/SR cells were relatively resistant to HDAC inhibitor TSA with an IC50 of 203.3 nM compared with parental H2030 cells (Fig. 6A). The presence of TSA (100 nM) decreased IC50 of Sot to less than one-fourth in H2030/SR cells, suggesting that TSA treatment reduced Sot resistance (Fig. 6B). Furthermore, TSA (100 nM) or Sot (Sot, 1 μM) alone caused cell cycle arrest at G1/G0 phase and mildly affected apoptosis in H2030/SR cells, whereas the combination of Sot and TSA (ST) significantly worsened phenotypes (Figs. 6, C, D and S5, A, B). TSA also potently improved the sensitivity of sgANP cells to Sot (Fig. 6E). Sot promoted cell cycle arrest at G1/G0 phase and apoptosis in sgCtrl cells but was less effective in sgANP cells (Figs. 6, F, G and S5, C, D). Coadministration of TSA (ST) enhanced Sot-induced cell cycle arrest at G1/G0 phase in both sgCtrl and sgANP cells, whereas TSA only increased Sot-induced apoptosis in sgANP cells but not sgCtrl cells (Figs. 6, F, G and S5, C, D). Interestingly, Sot treatment alone decreased the expression of ANP32A and YEATS4, whereas coadministration of TSA (ST) restored the expression of both genes in H2030 cells (Fig. 6, H and I). In vivo, ANP32A knockout delayed H2030 cells to form subcutaneous tumors and suppressed cell proliferation as expected (Fig. 7, A–D). Sot treatment alone significantly suppressed tumor formation and cell proliferation in sgCtrl cells but failed to do so in sgANP cells (Fig. 7, A–D). Most importantly, ST further repressed tumor formation and cell proliferation in sgANP cells, which was more effective than that in sgCtrl cells (Fig. 7, A–D). These results demonstrate that inhibition of histone deacetylation by TSA enhances the sensitivity of KRAS-mutant lung cancer cells to Sot in vitro and in vivo. TSA treatment partially mimics the effect of ANP32A and promotes Sot efficacy.

Activation of the p53 signaling pathway via ANP32A-mediated histone acetylation
To dissect how TSA may increase Sot sensitivity, H2030/SR cells were treated with TSA, Sot, or combination (ST), and used for RNA-Seq. As expected, few genes were altered by Sot treatment, whereas amounts of genes were altered by TSA alone or ST (Figs. 8, A, B and S6, A, B). Of note, mitogen-activated protein kinase and p53 pathways were among the top-ranked signatures commonly enriched in Kyoto Encyclopedia of Genes and Genomes analysis (Fig. 8, C and D). p53 and its multiple p53 target genes were activated by TSA (Fig. 8E). In support, downregulation of p53 and Bax and upregulation of Bcl2 were observed in H2030/SR cells, whereas the Erk phosphorylation was unchanged (Figs. 8F and S6C). Moreover, p53 and its target genes were reduced in ANP32A-deficient cells, which were reversed by reintroduction of ANP32A or YEATS4 (Figs. 8, G–I and S6, D–F). Notably, Sot treatment downregulated p53, whereas ST reversed the effect of Sot (Figs. 8J and S6, G–H). ChIP–qPCR verified that ANP32A knockdown impaired H3K27Ac in the promoter region of TP53 (Fig. 8, K and L). Taken together, these observations suggest that activation of the p53 pathway may play an important role in ANP32A-mediated Sot activity in KRAS-mutant lung cancer cells.

Discussion
KRAS G12C–selective inhibitors mark a significant milestone in cancer drug discovery and greatly improve the treatment of lung cancer (18, 19). However, acquired resistance is observed in most patients who initially respond to KRAS G12C inhibition but then develop resistance after 4 to 6 months (20). Despite the eventually developed resistance, multiple combinatorial treatments have been proposed to enhance the efficacy of KRAS G12C inhibitors, including genetic mechanisms and adaptive nongenetic mechanisms (5). There is no doubt that biomarkers are essential for the selection of combinatorial treatments to overcome KRAS G12C inhibitor resistance.
Our study demonstrates that ANP32A is an unfavorable prognosis factor for KRAS-mutant lung cancer. ANP32A promoted KRAS-mutant lung cancer but caused KRAS-mutant lung cancer cells to be sensitive to Sot in vitro and in vivo. These seemingly controversial observations may reflect the inseparable duality of uncontrolled proliferation in cancer. It not only confers cancer cells of growth advantage but also causes multiple stresses, such as metabolic and replicative stress, redox stress, and genomic instability, and so on, which are potential vulnerable points for cancer therapy. Cancer cells may maintain the balance to favor tumor development, whereas drug intervention may break the balance and push it to the opposite. In our study, ANP32A expression not only promoted cell proliferation but also caused upregulation of apoptosis genes but downregulation of survival genes, which were elegantly balanced. Moreover, Sot treatment or Sot resistance potently suppressed ANP32A and tumor suppressor p53 but upregulated the expression of survival gene, which explained the rapid development of drug resistance. However, TSA treatment reversed p53 expression because of ANP32A deficiency or Sot treatment through regulating H3K27Ac, which increased apoptosis, promoting Sot activity. We also noticed and confirmed that TSA treatment also upregulated p63 and p73 (Fig. S6, F–H). Thus, we speculate that the ability of ANP32A to activate these apoptosis pathways may determine its role in promoting Sot activity. ANP32A possibly serves as a biomarker for early selection of combinatorial treatment to overcome drug resistance.
Histone acetylation appears to be important for ANP32A-mediated Sot sensitivity. ANP32A has been suggested to enhance or reduce histone acetylation dependent on partners (10). However, the underlying mechanisms have never been addressed. In this study, we demonstrated that YEATS4 was an ANP32A partner that enhanced H3K27Ac. ANP32A upregulated YEATS4 by enhancing H3K27Ac in the YEATS4 gene promoter region. ANP32A also interacted with YEATS4 and increased YEATS4 interaction with H3K27Ac. Given that ANP32A preferentially binds to unmodified H3 and YEATS4 binds to H3K27Ac, ANP32A likely binds to unmodified H3 and promotes H3K27Ac by recruiting YEATS4, which is a component of SRCAP and the TIP60–p400–TRRAP complexes promoting histone acetylation (17). In turn, H3K27Ac further promotes YEATS4 expression forming a positive feedback loop. This self-reinforcing cycle may ensure the stability of the ANP32A–YEATS4–H3K27Ac axis. ANP32A suppression disrupts this loop contributing to the acquisition of Sot resistance. It should be noted that YEATS4 also functions through other mechanisms such as LncAKHE (21), which is independent of ANP32A. We could not exclude that other factors may also mediate the effect of ANP32A.
Multiple genomic mechanisms and adaptive nongenomic mechanisms underlying drug resistance to KRAS G12C–selective inhibitors have been dissected (22). In our study, we found that the p53 pathway was notably suppressed in Sot-resistant cells and downregulated by ANP32A deficiency and Sot treatment in H2030 cells. ANP32A knockdown and Sot treatment also suppressed the expression of apoptosis gene but upregulated survival protein Bcl2. These observations may explain why ANP32A deficiency causes significant cell cycle arrest but mild apoptosis. Moreover, TSA treatment reactivated Sot-suppressed p53 expression in H2030 cells, which was similar to p53 reactivation by re-expression of ANP32A in shANP cells. It appears that both TSA treatment and ANP32A target the p53 pathway to promote Sot activity. This may explain why TSA treatment partially mimics the effect of ANP32A promoting Sot efficacy. Previous studies showed that TSA promoted p53 acetylation at K382 and activated the p53–Bax apoptotic pathway (23). TSA also suppressed tumors through distinct mechanisms: inducing p21-mediated G1/S arrest in wildtype p53 cells or triggering mitotic catastrophe in p53-mutant cells (24). Our findings are consistent to these observations and suggest that the p53 pathway may play a critical role in Sot resistance. Reactivation of p53 pathway may be an alternative strategy to improve Sot therapy.
Our study also proposes novel strategies for combinatory treatment. Previously, we showed that direct targeting of ANP32A with a small peptide by blocking ANP32A binding to H3 potently repressed leukemia (25). Considering the effect of ANP32A in KRAS-mutant lung cancer, such a peptide is likely to suppress lung cancer and enhance Sot efficacy. Moreover, we showed that TSA potently reversed the effect of ANP32A deficiency and potently sensitized KRAS-mutant lung cancer cells to Sot in vitro and in vivo. Since HDAC inhibitors have been developed and widely used in clinics (26), it is worthy of testing the combinatorial treatment with KRAS G12C–selective inhibitors and HDAC inhibitors.
In summary, mutant KRAS signaling activates the ANP32A–YEATS4–H3K27Ac axis to retain the expression of apoptosis genes in lung cancer. Mutant KRAS signaling also promotes survival and proliferation. Without drug intervention, it is well balanced to favor proliferation and retain the drug sensitivity. Sot treatment suppresses proliferation and impairs the ANP32A–YEATS4–H3K27Ac axis, which causes reduced proliferation and eventually develops drug resistance. TSA coadministration restores ANP32A–YEATS4–H3K27Ac axis and activates apoptosis pathway to reverse drug resistance (Fig. S6I).

Experimental procedures

Cell culture
Cell lines Beas-2B (Research Resource Identifier [RRID]: CVCL_0168), Calu-1 (RRID: CVCL_0608), A549 (RRID: CVCL_0023), H157 (RRID: CVCL_0463), H1299 (RRID: CVCL_0060), PC9 (RRID: CVCL_B260), HCC827 (RRID: CVCL_2063), H1975 (RRID: CVCL_1511), H2228 (RRID: CVCL_1543), H2030 (RRID: CVCL_1517), H358 (RRID: CVCL_1559), and 293T (RRID: CVCL_0063) (passage number = 4) were obtained from the Cell Bank of the Chinese Academy of Sciences. Calu-1 cells were cultured in McCoy’s 5A medium, Beas-2B and 293T in Dulbecco's modified Eagle's medium, and the rest cells in RPMI1640, supplemented with 10% fetal bovine serum (Gibco) and 1% penicillin–streptomycin (Sbjbio). Authentications were obtained by a Human STR Profiling Cell Authentication Service (American Type Culture Collection), and mycoplasma was tested in all cell lines.
To create Sot-resistant H2030/SR cells, H2030 cells were first exposed to 2 nM Sot for 24 h and then cultured without the inhibitor until normal growth resumed. Cells underwent 6 to 10 cycles of drug exposure, followed by doubling the Sot concentration for another 6 to 10 cycles, until the IC50 increased more than 10-folds.
To generate a stable cell line, lentivirus-based plasmids pLKO.1 (RRID: Addgene_8453), pLKO.1-shANP were constructed as described previously (10). The lentiviral plasmids used in this study included pU6-Cas9 (P88382, RRID: Addgene_62988), pU6-sgRNA-ANP32A (P65521), pLV-CMV-puro (P80484, RRID: Addgene_123223), pLV-ANP32A (P65538), and pLV-YEATS4-FLAG (P76387), were synthesized by Miaolingbio. Lentiviral plasmids along with packaging plasmids psPAX2 (RRID: Addgene_12260) and pMD2.G (RRID: Addgene_12259) were cotransfected into 293T cells. Viral supernatants were collected 48 h post-transfection and used to infect H2030 and H358 cells with polybrene. Stably transduced cells were selected in puromycin or G418 for 5 days. Full-length details of the plasmids used are provided in the supporting information.

Xenograft mouse lung cancer model
The Animal Ethics Committee of Jinhua Central Hospital approved the mouse study (license: AL-JHYY2023-39). NOD–SCID mice were obtained from Shanghai Ziyuan Biotechnology. After cell counting, cells were mixed with Matrigel, and a 100 μl suspension containing 1 × 107 cells was injected into the right axillary region of each mouse. Once tumors reached 50 to 100 mm3, mice were divided into three groups: control, Sot, and Sot with TSA. TSA was given subcutaneously at 0.5 mg/kg/day, whereas Sot was given orally at 30 mg/kg/day in a 0.5% CMC–Na solution. Mice were weighed daily, and tumor volumes were tracked. Tumor volume = length × (width)2/2. On day 30, mice were euthanized, and tumor weights were recorded.

Online database analysis and verification in lung cancer patient samples
Prognostic and mRNA expression data for lung cancer patients were sourced from The Cancer Genome Atlas database (https://tcga-data.nci.nih.gov/tcga/). Analyses of OS and expression data for ANP32A were conducted via the Home for Researchers platform (https://www.home-for-researchers.com/#/). Potential interacting proteins of ANP32A were predicted using the BioGRID database (https://thebiogrid.org/).
To detect ANP32A at the protein and mRNA levels, 66 human lung cancer samples were collected from Jinhua Central Hospital, with histological confirmation and patient consent. The study adhered to The Declaration of Helsinki and received ethics approval (no.: AL-JHYY202437). A tissue microarray from Shanghai Outdo Biotech was also used, and two pathologists independently evaluated stained slides. Staining intensity was scored from 1 to 4, and the final score was calculated by multiplying this intensity by the percentage of positive cells. The characteristics of LUAD patients are listed in Table S1.

Quantitative RT–PCR
Total RNA was extracted utilizing TRIzon Reagent (CW0580S; Cwbio). Complementary DNA synthesis was conducted employing the PrimeScript RT Master Mix (RR036A; Takara). qPCR was subsequently performed using the TB Green Fast qPCR Mix (RR430A; Takara). The primers used for qPCR are detailed in Table S2.

Immunoprecipitation and Western blotting
For coimmunoprecipitation (IP), cells were lysed on ice for 30 min with IP lysis buffer containing 1% PMSF. After centrifugation, 100 μl of the supernatant was saved as input. The rest was incubated overnight at 4 °C with the FLAG antibody, followed by a 4-h incubation with 20 μl magnetic beads. Beads were washed 3 to -4 times with IP lysis buffer, and proteins were eluted by boiling in loading buffer (LT101; Epizyme) at 95 °C for 10 min and used for Western blotting analysis.
To perform a Western blot, proteins were extracted from lysed cells using radioimmunoprecipitation assay buffer with 1% PMSF. Loading buffer was added to the samples, which were then boiled for 10 min. After cooling, 10 μl of each sample was loaded onto SDS-PAGE gels for electrophoresis. Separated proteins were transferred to polyvinylidene difluoride membranes (IPVH00010; Millipore), blocked with 5% skim milk in Tris-buffered saline with Tween-20 for 1 h, and incubated overnight with primary antibodies at 4 °C. The next day, membranes were washed with Tris-buffered saline with Tween-20, incubated with secondary antibodies for 1 h, and protein bands were visualized using an ECL detection kit (36208ES60; Yeasen Biotechnology) on a Bio-Rad ChemiDoc XRS system. Chemical reagents and antibodies used in cell culture and Western blot, all of which are commercially available products, are listed in Table S3 along with their corresponding RRID identifiers.

Cell Counting Kit-8 assay
Cells (800/well) were seeded in 96-well plates with 100 μl of culture medium. Viability was measured on days 0 to 6 with the Enhanced Cell Counting Kit-8 (BL1055E; Biosharp). For drug sensitivity tests, cells were seeded at 1500 cells per well under the same conditions and exposed to varying drug concentrations for 24 or 72 h. Viability was measured by Cell Counting Kit-8, and IC50 was calculated.

Cell migration and invasion assay
A total of 2 × 104 cells were suspended in a serum-free medium and placed in the upper chamber of a Transwell apparatus (84052ES12; Yeasen Biotechnology). The lower chamber contained medium with 20% fetal bovine serum. After 24 h, cells migrating into the lower chamber were fixed and stained with crystal violet. For the invasion assay, the upper chamber’s polycarbonate membrane was precoated with Matrigel (356230; Corning) and diluted 1:10 with serum-free medium.

Apoptosis analysis
Cells (1 × 105/well) were seeded in 6-well plates with 2 ml of culture medium and incubated for 12 h. Cells were treated with different drug concentrations for 24 or 72 h and stained with an Annexin V-PE/7-AAD kit (Vazyme), following the guidelines of the manufacturer. Data were collected by flow cytometry (CytoFLEX; Beckman Coulter), and cell apoptosis was analyzed with FlowJo (version 10.8.1; BD [Becton, Dickinson & Company] Life Sciences).

5-Ethynyl-2′-deoxyuridine assay
The BeyoClick EdU-555 Cell Proliferation Detection Kit (C0075S; Beyotime) was used to treat cells (2 × 104) in a 24-well plate with 500 μl of culture medium per well following the manufacturer’s instructions. Microscopic images were acquired with an Olympus BX43 microscope.

Cell cycle assay
Cells (1 × 105) were seeded into 6-well plates with 2 ml of culture medium and incubated for 12 h. After incubation, cells were collected and stained with the Cell Cycle Detection Kit (KeyGEN BioTECH) following the guidelines of the manufacturer. Data were collected by flow cytometry (CytoFLEX), and the cell cycle was analyzed with ModFit software (Verity Software House).

H&E staining
H&E Staining Kit (C0105M; Beyotime) was used. Briefly, sections were stained with hematoxylin, followed by a 30-s incubation of eosin solution. Subsequently, sections were dehydrated and mounted using neutral balsam. Microscopic images were acquired with an Olympus BX43 microscope.

Immunohistochemical staining
Antigen retrieval was performed by utilizing the One-Step Dewaxing/Antigen Retrieval Buffer (pH 6.0) (20×) (Elabscience). The ready-to-use SABC-POD(F) (Rabbit IgG) Kit (catalog no.: SA1028) and the SABC-POD(F) (Mouse IgG) Kit (catalog no.: SA1027) were used (Boster Biological Technology) for the subsequent staining of mouse tumor tissue sections following the manual provided by the manufacturer. Microscopic images were acquired with an Olympus BX43 microscope.

Chromatin IP assay
Cells were harvested, and ChIP was conducted utilizing the BeyoChIP Enzymatic ChIP Assay Kit (P2083S; Beyotime), following the manufacturer’s protocol. The primers employed for ChIP–qPCR are detailed in Table S2.

Immunofluorescence assay
A cell climb slip was placed in each well of a 24-well plate. About 2 × 104 cells were seeded and cultured for 12 h. Slides were washed with PBS and fixed with 4% paraformaldehyde for 15 min at room temperature. After washing with PBS containing 3% bovine serum albumin (BSA) (9048-46-8; Yeasen Biotechnology), cells were permeabilized with Immunostaining Strong Permeabilization Buffer (P0097-500 ml; Beyotime) for 20 min at room temperature and washed three times for 5 min each with PBS containing 3% BSA. After blocking with PBS containing 5% BSA for 1 h at room temperature, slides were incubated with primary antibodies overnight at 4 °C, followed by incubation with fluorescent-conjugated secondary antibodies for 2 h at 37 °C in the dark. Coverslips were finally mounted with AntiFade Mounting Medium (with 4',6-diamidino-2-phenylindole) (HY-K1047; MedChemExpress) and stored in a protected environment. Microscopic images were acquired with an NIS microscope (Nikon).

Transcriptomic sequencing and data analysis
Sot-resistant H2030 cells (H2030/SR) were grown to 70% confluence in 10 cm dishes and treated with 1 μM Sot, 100 nM TSA, or both for 24 h. Cells were harvested, and total RNA was extracted with TRIzon Reagent (CW0580S; Cwbio) for transcriptomic sequencing (Tsingke Biotechnology). The obtained sequencing data were analyzed using the online analysis platform provided by the company (https://cloud.tsingke.com.cn/main/analysisTool). For differential gene expression analysis, genes with |log2 fold change| >1 and p < 0.05 were identified as differentially expressed between groups. All identified differentially expressed genes were subsequently subjected to Kyoto Encyclopedia of Genes and Genomes functional enrichment analysis.

Statistical analysis
For quantification of Western blot analysis, intensity values were obtained using ImageJ (National Institutes of Health). All signals were normalized to the total protein loaded per lane, with histone acetylation levels and phosphorylated Erk further normalized to total histone and total Erk levels, respectively. The data were processed and visualized with GraphPad Prism (version 9.0; GraphPad Software, Inc). Data are presented as the mean ± standard deviation (x ± s). For statistical analysis, an unpaired Student’s t test was employed to compare two groups, whereas a one-way ANOVA was conducted for comparisons across multiple groups. Individual data points are overlaid on the bar graphs, with each point representing a biological replicate. A p value of less than 0.05 was deemed to indicate statistical significance.

Data availability
Raw RNA-Seq data have been deposited in the Sequence Read Archive under BioProject accession number PRJNA1337559.

Supporting information
This article contains supporting information.

Conflict of interest
The authors declare that they have no conflicts of interest with the contents of this article.