Targeting super-enhancer-driven SKIL transcription by CDK7 inhibitor THZ1 to suppress gastric cancer progression

Abstract

Background

Gastric cancer (GC) is a lethal malignancy characterized by high incidence, mortality, and limited treatment options. Transcriptional addiction is a key cancer hallmark that drives tumor pathogenesis, making its inhibition a promising therapeutic strategy for GC. The study aims to investigate the roles and mechanisms of super-enhancer (SE)-driven oncogenic transcriptional addiction in GC progression and to identify novel targetable vulnerabilities.

Methods

We utilized cellular and animal models to assess the effects of THZ1 treatment and CDK7 knockdown on GC progression. RNA sequencing was employed to elucidate the potential molecular mechanism of THZ1 treatment. ChIP-seq was performed to establish SE landscape in GC. Integrative analysis of transcriptomic and SE profiling was used to identify THZ1-targeted oncogenic genes. Rescue experiments were conducted to confirm that THZ1 treatment suppresses GC malignant progression by targeting SE-driven SKIL transcription.

Results

GC cells exhibited pronounced sensitivity to THZ1 compared to normal gastric mucosa cells, and the treatment potently suppressed tumor growth and migration in both cellular and animal models. CDK7 was significantly upregulated in GC tissues, and its knockdown inhibited malignant progression in vitro and in vivo, whereas its overexpression accelerated tumor progression. Mechanistically, SE-driven oncogenic transcriptional amplification underlies GC cell susceptibility to THZ1, supported by the identification of novel oncogenic genes such as SKIL. SKIL, a key Hippo pathway regulator, was highly expressed in GC cells, and its elevated expression predicted poor patient prognosis. SKIL silencing attenuated malignant phenotypes, while its overexpression diminished THZ1’s suppression of GC cell proliferation and migration.

Conclusion

Our findings demonstrate that THZ1 inhibits GC progression by disrupting SE-driven oncogenic transcription, thereby offering CDK7 inhibition as a promising therapeutic intervention for GC.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12967-026-07721-1.

Introduction

Gastric cancer (GC) is a highly lethal malignancy with a significant global burden. It accounts for over 968,350 new cases and 659,853 deaths annually worldwide, ranking fifth in incidence and mortality [5, 29, 36]. Despite the availability of multiple primary treatments for GC, including surgery, chemotherapy, radiotherapy, immunotherapy, and targeted therapy, the overall prognosis remains dismal, with a 5-year survival rate of just 25% across all stages and less than 5% for advanced metastatic GC [20, 45]. Thus, urgent action is required to develop innovative and more efficient treatment strategies.

Transcriptional addiction is a key molecular signature of cancer driven by super-enhancers (SEs), which are densely packed enhancer regions that sustain high levels of oncogenic transcription to drive tumorigenesis [2, 4, 34, 48]. Thus, targeting SE-driven oncogenic transcription offers a promising treatment option. The transcription process catalyzed by RNA polymerase II (Pol II) comprises initiation, elongation, and termination [13]. Carboxy-terminal domain (CTD) hyperphosphorylation of RNA Pol II, which is precisely regulated by Cyclin-dependent kinases (CDKs), is essential for transcription [16, 30, 42], making small molecules targeting transcriptional CDKs promising for treatment of cancers.

CDK7, a component of transcription factor IIH (TFIIH), is crucial for transcription initiation and elongation by phosphorylating RNA Pol II’s CTD [38]. Studies confirmed that CDK7 is vital for the assembly and maintenance of SEs, and that SE-driven oncogenes are highly responsive to CDK7 inhibition [8, 34]. THZ1 is a covalent CDK7 inhibitor that selectively blocks its kinase activity by binding to a unique cysteine outside the kinase domain [27]. The role of THZ1 was identified in multiple cancers, including glioblastoma [3], pancreatic ductal adenocarcinoma [44], gallbladder cancer [19] and oesophageal squamous cell carcinoma [23]. The effect of THZ1 on GC still remains uncharacterized, prompting us to assess its therapeutic potential and elucidate its underlying mechanisms of action.

Studies have revealed that SE-driven transcription of oncogenes mediates the vulnerability of malignant tumors to THZ1 [10, 27, 43]. SKIL (also known as SnoN) was initially identified as an oncogene and has been reported to be overexpressed in breast cancer [46], NSCLC [31], and colorectal carcinoma [6]. It exerts pro-oncogenic effects by regulating pathways like Hippo to support tumorigenesis and epithelial to mesenchymal transition (EMT) [31, 50]. While some evidence, such as the boosted ESCC cell proliferation following SKIL silencing, suggests anti-oncogenic activities [21, 35], its function in GC progression is nevertheless unknown.

In this study, we demonstrated the covalent CDK7 inhibitor THZ1 exhibits significant anti-tumor activity in GC. Combining transcriptome and genomic analysis, we revealed THZ1 treatment suppresses GC malignant progression by targeting SE-driven SKIL transcription. Overall, our findings elucidate a mechanism through which THZ1 counteracts GC proliferation and migration by targeting transcriptional addiction, offering a promising therapeutic intervention for GC.

Methods

Cell culture

Cell lines used in this study include the normal gastric mucosa cell GES-1, three GC cell lines (AGS, HGC-27, MKN45), and human embryonic kidney cells (HEK293T), all of which were sourced from the Chinese Academy of Sciences (CAS) Cell Research Center, Shanghai, China. The cell lines were cultured in RPMI-1640 medium (Gibco, NY, USA) containing 10% fetal bovine serum (FBS; Biological Industries, KBH, IL) and 1% penicillin-streptomycin (Gibco, NY, USA), except HEK293T cell lines, which were maintained in DMEM medium (Gibco, NY, USA) supplemented with 10% FBS (Biological Industries, KBH, IL). Stable GC cell lines overexpressing CDK7 or SKIL were constructed using a lentiviral packaging system, with the corresponding overexpression plasmids procured from Beijing Tsingke Biotech Co., Ltd. and Guangzhou IGE Biotechnology LTD, respectively. The conditions of cell culture were strictly controlled at 5% CO₂ and 37℃.

Chemicals and reagents

THZ1 (Cat. 7549) was first obtained from Selleck (Houston, TX, USA) and then solubilized in dimethyl sulfoxide (DMSO; Solarbio, Cat. 8371) to prepare a working solution at a concentration of 10 mM for storage at -80 °C.

CRISPR–Cas9 mediated gene knockdown of CDK7

To obtain stable GC cell lines expressing Cas9, lentiCas9-Blast (Addgene, Cat. 52962) was co-transfected with packaging plasmids (PMD2G and PSPAX2) into HEK293T cell lines. After 8 h of transfection, the fresh complete DMEM medium was replaced. The medium containing lentivirus particles was collected and concentrated from HEK293T cell culture dishes following 48 and 72 h of incubation, respectively. Then GC cell lines were infected with a certain amount of lentivirus particles, and the fresh medium was replaced after 12 h and continued to be cultured for 48 h. Finally, blasticidin (TargetMol, Cat. T64911) was added for screening. CDK7-KD GC cell lines were produced by cloning gRNA targeting the CDK7 exon into the lentiGuide-Puro vector (Addgene, Cat. 52963). The sgRNA Designer tool from the Broad Institute was utilized to design sgRNAs. GC-Cas9 cell lines were infected and selected with puromycin (Solarbio, Cat. P8230). The sgCDK7 sequences used are given as: Sense 5’-TTTCCATAAAATCAAAGACA-3’, Antisense (5’- TGTCTTTGATTTTATGGAAA − 3’).

SiRNA knockdown assay

Lipofectamine RNAiMAX (Invitrogen) was used to transfect siRNAs. The final concentration of the siRNA is 20 nM. GC cells were seeded. After 12 h, the transfection procedure was executed at a cellular confluence level of 30%. The siRNA sequences were: siCDK7 (5’-GCUGUAGAAGUGAGUUUGUAA-3’), siSKIL1 (5’- GCAAGUAAGUCCAUAUCAA-3’) and siSKIL2 (5’- GGCUCACAGUAGUGGUAAA − 3’).

Cell viability assay

Cell cultures were prepared by seeding 3 × 10³ cells per well into 96-well microplates. Cells were incubated for 12 h before being treated with specified concentrations of THZ1 and DMSO for durations of 0, 24, 48, and 72 h. Viability assays were subsequently performed using the Cell Counting Kit-8 (CCK8) as per the protocol outlined by the manufacturer (MCE, Cat. HY-K0301). Data are expressed as the percentage (%) of treatment group absorbance relative to the control group. CDK7-KD, CDK7-OE, SKIL-OE or SKIL-KD GC cell lines were performed as described above, with the difference that THZ1 treatment were not required.

Colony formation assay

A total of 1000 cells per well were plated in 6-well plates, with medium changes every three days and treatments applied during culture. The culture was terminated after 12–14 days. Colonies were preserved with 4% paraformaldehyde, stained using a 0.1% crystal violet solution, and imaged for analysis. ImageJ software was employed to quantify the clones. CDK7-KD, SKIL-OE or SKIL-KD GC cell lines were performed as described above, with the difference that THZ1 treatment were not required.

Wound healing assay

Cells were seeded into 6-well plates at approximately 90% confluence. Cells were cultured for 10–12 h for adhesion, scratches were made with a 200 µl tip, and 1% FBS-supplemented medium was added to continue culturing. THZ1 and DMSO were added according to the experimental requirements, and photographs were taken to observe the healing of scratches respectively. Image J software analyzed the scratch areas. CDK7-KD, CDK7-OE, SKIL-OE or SKIL-KD GC cell lines were performed as described above, with the difference that THZ1 treatment were not required.

Cell apoptosis assay

Cells were plated in 6-well plates at 30% confluence and incubated for 12 h. Following incubation, cells were treated with THZ1, and apoptosis was assessed using a FITC Annexin V/PI kit (7sea biotech, Cat. A005-3) and flow cytometry (Agilent, CA, USA).

Cell cycle analysis

At 30% confluence, cells were seeded into a 6-well plate and administered with THZ1 after 12 h. Following fixation in ethanol at -20 °C for 15 min, cells were reconstituted in phosphate-buffered saline (PBS), and evaluated with the Cell-cycle Analysis Kit (7sea biotech, Cat. C001-050). Flow cytometry (Agilent, CA, USA) was employed to assess cell-cycle phases.

Western blotting

Cells were lysed with highly efficient RIPA buffer (Solarbio, Cat. R0010) containing 1% Protein Phosphatase Inhibitor (Solarbio, Cat. P1260) and 1% PMSF (Solarbio, Cat. P0100). Total protein was quantified using the BCA Protein Assay Kit (Solarbio, Cat. PC0020). The protein extracts were separated by SDS-PAGE and electrophoretically transferred onto nitrocellulose membranes (Cytiva, Cat. 66485). Membranes were blocked using 5% skim milk (BD, Cat. 232100) for 2 h, then incubated at 4 ℃ overnight with primary antibodies to ensure specific binding, and treated with secondary antibodies for 2 h the next day. Protein band intensities from three independent biological replicates were quantified using ImageJ software. Detailed information on the primary antibodies used is documented in Supplementary Table S1.

RNA extraction and RT-qPCR

Total cellular RNA was extracted using TRIzol reagent, and complementary DNA was synthesized using the PrimeScript™ RT reagent Kit with gDNA Eraser (TaKaRa, Cat. RR047A), following the manufacturer’s instructions. The reaction mixture was prepared as per the guidelines provided with the ChamQ Universal SYBR qPCR Master Mix (Vazyme, Cat. Q711-02) and subsequently quantified using the ABI 7500 real-time fluorescence qPCR platform (MA, USA). GAPDH served as the reference gene, with 2^−△△Ct employed for calculating relative gene expression. Primer sequences for RT-qPCR are detailed in Supplementary Table S2.

RNA-seq

GES-1 and AGS cell lines underwent incubation with 100 nM THZ1 or DMSO for 6 h. The cellular samples were then harvested, lysed thoroughly using TRIzol (Invitrogen, Cat. 10296010) and sent to Novogene (Beijing, China) for RNA-seq analysis. Differentially expressed genes (DEGs) between sample groups were identified by DESeq2 algorithms. DEGs were screened according to a fold change criterion of ≥ 1.5 and an adjusted P-value threshold of < 0.05. Enrichment analyses for Gene Ontology (GO) terms and KEGG pathways were carried out using the DAVID bioinformatics resource. Data visualization was performed using the online platform https://www.bioinformatics.com.cn.

ChIP-seq

AGS H3K27ac ChIP-seq data (GSE162420) were obtained from Gene Expression Omnibus (GEO), and super-enhancer analysis in AGS cell lines was conducted using the Rank Ordering of Super-Enhancers (ROSE) algorithm. The data screening followed the principle of a P-value<0.05.

Immunohistochemistry (IHC) and Hematoxylin–Eosin (HE) staining

IHC analysis involved heating paraffin sections at 60℃ (2–4 h), dewaxing in xylene, and rehydrating with graded alcohols. Antigen retrieval was followed by incubation in 3% H₂O₂ for 15 min to inactivate peroxidase. Sections were permeabilized with 0.3% Triton X-100, and 10% BSA was used to block non-specific binding. Primary antibodies were incubated at 4℃ overnight, followed by the addition of secondary antibodies, which were incubated for 1 h at room temperature. Sections were incubated with diaminobenzidine (DAB), stained with hematoxylin, dehydrated, mounted with resin, and observed under a microscope. For HE analysis, paraffin sections were heated at 60℃ for 2–4 h, dewaxed in xylene and hydrated by graded alcohols. The sections were treated with hematoxylin and eosin staining and dehydrated. Ethical clearance for the study protocol was secured from the Animal Research Ethics Committee of Guizhou Medical University (approval number 2403428) and the Ethics Committee of the Affiliated Hospital [approval number 2024 (343)].

Xenograft model

All experimental procedures involving animals adhered to the National Institutes of Health guidelines and were authorized by the Animal Research Ethics Committee of Guizhou Medical University (approval number 2403428). The THZ1 xenograft model was established by subcutaneous inoculation of HGC-27 cell suspension at a density of 1 × 10⁷ cells / 200 µl on the right side of the back of six-week-old female BALB/c nude mice, a commonly used immunodeficient murine model (SLRC Laboratory Animal Center, Shanghai, China). Seven days after inoculation, mice were divided into control and experimental groups, with five subjects per group (n = 5/group). For 14 days, the animals were given intraperitoneal injections of either vehicle or THZ1 (10 mg/kg) twice daily. Nude mice body weight and tumor volume were assessed and recorded every 2–3 days. Tumor tissues and vital organs (hearts, livers, spleen, lungs, kidneys) were collected. The sgCDK7 xenograft model was established as described above, with the difference that THZ1 and DMSO injections were not required.

Statistical analysis

Data analysis was carried out using SPSS 25.0 and GraphPad Prism 8.1, and results were presented as mean ± SD. Expression and survival were analyzed through the Gene Expression Profiling Interactive Analysis (GEPIA)/UALCAN database and Kaplan-Meier curves. Two-group comparisons were analyzed with a Student’s t-test, while one-way ANOVA was used for multiple-group comparisons. Statistical significance was set at a P-value<0.05.

Results

THZ1 exhibits high potency and selectivity for GC

The impact of THZ1 was evaluated on three GC cell lines (AGS, HGC-27, MKN45) and a normal gastric mucosa cell line (GES-1). CCK8 assays demonstrated that GC cells are more sensitive to low-dose THZ1 than GES-1 (Fig. 1A). Consistently, colony formation was significantly suppressed in GC cells but not in GES-1 (Fig. 1B and Fig. S1A), and Western blot analysis showed downregulation of the proliferation marker PCNA in THZ1-treated GC cells (Fig. 1C, Fig. S1B and Fig. S6A). Cell cycle analysis revealed that low-dose THZ1 specifically induces G2/M phase arrest in GC cells, with no discernible effect on GES-1 (Fig. 1D and Fig. S1C). Furthermore, THZ1 triggered apoptosis in GC cells, as shown by Annexin V staining (Fig. 1E and Fig. S1D) and concomitant decrease in Bcl2 and increase in Bax protein levels (Fig. 1F, Fig. S1E and Fig. S6B). The wound healing assay indicated that THZ1 strongly inhibits GC cell migration (Fig. 1G and Fig. S1F). These findings reveal THZ1 attenuate GC cell growth and migration with notable potency and selectivity.

Fig. 1.

THZ1 exhibits high potency and selectivity for GC. A Cell viability of GES-1, the AGS, HGC-27, and MKN45 cell lines were subjected to analysis using CCK8 assay after THZ1 treatment. B Proliferation of GES-1, the AGS and HGC-27 cell populations were evaluated by colony formation assay after THZ1 treatment. C The expression of the proliferation marker (PCNA) after THZ1 treatment was analyzed via Western blotting. D Flow cytometry with propidium iodide (PI) staining was applied to evaluate the cell cycle in cells exposed to THZ1. E Apoptosis in THZ1-treated cells was analyzed via Annexin V-FITC/PI staining. F The relative expression of apoptosis-related proteins Bax and Bcl2 after THZ1 treatment was assessed via Western blot analysis. G A wound healing assay was employed to evaluate the influence of THZ1 on the migration of GES-1, AGS, and HGC-27 cells. The scale bar measures 500 μm, and the results are expressed as mean ± SD. Statistical significance is denoted by *P < 0.05, **P < 0.01, ***P < 0.001

CDK7 is highly expressed in GC and essential for GC malignant phenotypes

As THZ1 inhibits CDK7 covalently, we further studied CDK7’s role in GC. Consistent with the bioinformatics analysis of GEPIA and UALCAN database (Fig. 2A, B), immunohistochemical staining confirmed that CDK7 is significantly upregulated in GC tissues (Fig. 2C). To assess its function, we successfully generated CDK7-knockdown GC cell lines (Fig. 2D, Fig. S2A, S2B and Fig. S6C), with knockdown efficiency confirmed by Western blotting through the marked reduction in both CDK7 protein abundance and Ser5 phosphorylation of the RNA Pol II CTD. Functional assays revealed that CDK7 depletion inhibits GC cell growth (Fig. 2E, F and Fig. S2C, D) and migration (Fig. 2G and Fig. S2E). Conversely, complementary gain-of-function experiments confirmed that CDK7 overexpression promotes these malignant phenotypes (Fig. S3). Collectively, these results establish CDK7 as a driver of GC, whose upregulation promotes proliferative and migratory capacities.

Fig. 2.

CDK7 is highly expressed in GC and essential for GC malignant phenotypes. A-B CDK7 expression in GC tissues was analyzed using GEPIA and UALCAN. C CDK7 expression in GC tissues was evaluated via IHC; scale bar = 50 μm, n = 3/group. D Western blot detected CDK7 and Ser5 phosphorylation of RNA Pol II CTD following CDK7 knockdown. E Viability of GC cells after CDK7 knockdown was evaluated by CCK8 assays. F Colony formation assay assessed GC cell proliferation after CDK7 knockdown. G GC cell migration after CDK7 knockdown was assessed via wound healing assay (scale bar: 500 μm). Data are given as mean ± SD, with significance marked as *P < 0.05, **P < 0.01, ***P < 0.001

THZ1 treatment effectively suppressed GC growth in vivo

We next examine THZ1’s impact on GC growth utilizing in vivo HGC-27 xenograft systems. THZ1 treatment significantly suppressed tumor growth (Fig. 3A, B) and reduced final tumor weight (Fig. 3C) compared to DMSO control. No significant body weight loss (Fig. 3D) or tissue damage in the heart, liver, or kidney (Fig. 3E) was observed, indicating that the treatment was well-tolerated. Consistent with the growth inhibition, THZ1-treated xenografts have fewer proliferating cells than DMSO-treated ones, as indicated by Ki67 staining (Fig. 3F). Notably, CDK7 expression had no significant differences between the two groups (Fig. 3G). These findings demonstrate that THZ1 effectively inhibits GC growth in vivo with limited toxicity.

Fig. 3.

THZ1 treatment effectively suppressed GC proliferation in vivo. A-B Tumor images and growth curves for HGC-27 xenografts in nude mice treated with DMSO (n = 5) or THZ1 (n = 5). C Tumor weights for HGC-27 xenografts in nude mice treated with DMSO (n = 5) or THZ1 (n = 5). D Body weight changes in HGC-27 xenograft-bearing nude mice treated with DMSO (n = 5) or THZ1 (n = 5). E HE staining of the heart, liver, and kidney tissues in nude mice with HGC-27 xenografts treated with DMSO or THZ1, scale bar: 100 μm. F Ki67 expression in subcutaneous tumor tissues was identified via IHC, scale bar: 100 μm. G CDK7 expression was analyzed in subcutaneous tumor tissue samples using IHC (scale bar: 100 μm), with statistical significance at **P < 0.01, ***P < 0.001

CDK7 knockdown significantly decreased tumor growth in vivo

The effects of CDK7 knockdown on GC growth was evaluated using in vivo xenograft models. CDK7 knockdown markedly suppressed tumor growth without affecting body weight, as no significant difference was observed between the control and knockdown groups (Fig. 4A-D). IHC staining confirmed successful CDK7 knockdown (Fig. 4E) and revealed a significant reduction in Ki67-positive cells in the knockdown tumors compared to the controls (Fig. 4F). The data confirm that CDK7 knockdown substantially reduces in vivo tumor growth.

Fig. 4.

CDK7 knockdown significantly decreased tumor growth in vivo. A-B Tumor images and growth curves for nude mice with HGC-27 xenografts after CDK7 knockdown (n = 5/group) are shown. C Representative tumor weights in nude mice bearing HGC-27 xenografts after CDK7 knockdown (n = 5/group). D Body weight curves of nude mice after CDK7 knockdown (n = 5/group) are shown. E CDK7 expression in subcutaneous tumors was visualized using IHC (scale bar: 100 μm). F Ki67 expression in subcutaneous tumors was analyzed using IHC (scale bar: 100 μm), showing significant results (**P < 0.01)

THZ1 causes global transcriptional downregulation and preferentially targets the Hippo pathway-related transcripts in GC cells

CDK7 is a transcriptional kinase that regulates RNA Pol II activity by phosphorylating its C-terminal domain (CTD) at Serine 5 during the initiation phase and Serine 2 during the elongation phase [1, 28]. We first investigated the impact of THZ1 on CTD phosphorylation. THZ1 treatment reduced CTD phosphorylation at both Serine 2 and Serine 5 in a dose-dependent manner in GC cells, with greater sensitivity observed compared to GES-1 cells (Fig. 5A and Fig. S6D, S6E).

Fig. 5.

THZ1 causes global transcriptional downregulation and preferentially targets the Hippo pathway-related transcripts in GC cells. A Western blot analysis of RNA polymerase II (RNA Pol II)-CTD phosphorylation in GES-1, AGS and HGC-27 cells treated with DMSO or THZ1 for 6 h. B Heatmap of gene expression changes in GES-1 (DMSO) and AGS (DMSO or THZ1) cells. C Volcano plots showing differentially expressed genes (DEGs) from two comparisons: GES-1 vs. AGS (left) to identify genes dysregulated in GC, and DMSO vs. THZ1 (AGS)(right) to identify genes responsive to THZ1 treatment in AGS cells. D Venn diagram showing the 979 THZ1-sensitive genes commonly upregulated in AGS cells and downregulated following THZ1 treatment. E Gene ontology (GO) enrichment analysis of the 979 THZ1-sensitive genes presented in (D). F KEGG pathway enrichment analysis of the 979 THZ1-sensitive genes presented in (D). G Representative heatmaps of the Hippo pathway-related genes upregulated in AGS and downregulated by THZ1 treatment

THZ1’s preferential inhibition of CTD phosphorylation in GC cells prompted us to hypothesize that it selectively disrupts RNA Pol II-mediated transcription. To test this, we performed RNA-seq on the following three groups: GES-1, AGS, and AGS treated with THZ1. We identified 3741 genes upregulated in AGS relative to GES-1 (GES-1 vs. AGS), and 3039 genes downregulated in AGS cells upon THZ1 treatment compared to the DMSO control (DMSO vs. THZ1) (Fig. 5B, C). A Venn diagram identified 979 THZ1-sensitive genes, defined as those upregulated in AGS cells and downregulated by THZ1 treatment (Fig. 5D). To investigate the functional implications, we performed enrichment analysis on these 979 genes. GO terms were significantly associated with nuclear processes of RNA Pol II-mediated transcription (Fig. 5E), while KEGG analysis identified enrichment in oncogenic pathways, particularly the Hippo pathway (Fig. 5F, G). These data demonstrate that THZ1 inhibits GC progression by attenuating RNA Pol II-mediated transcription, primarily through targeting oncogenic pathways essential for tumorigenesis.

Identification and characterization of SE-associated oncogenes in GC

Super-enhancer (SE)-linked oncogenes are key cancer drivers and have been shown to be highly sensitive to THZ1 [9]. Since SEs are characterized by strong H3K27ac enrichment, we analyzed H3K27ac ChIP-seq data from GEO to define the SE landscape in AGS cells. Our analysis uncovered 1,092 SE-associated genes (Fig. 6A) and found that GO terms for cell migration, transcriptional regulation by RNA Pol II, and DNA-templated transcription are significantly enriched (Fig. 6B). Integration of transcriptomic datasets with SE profiling in GC cells revealed pivotal oncogenes (Fig. 6C). Among the 109 genes identified, KEGG pathway analysis demonstrated significant enrichment in the Hippo pathway (Fig. 6D). After further analysis, four candidate oncogenes were identified and confirmed by RT-qPCR analysis (Fig. 6E). The corresponding H3K27ac ChIP-seq and RNA-seq profiles for these candidates were presented in Fig. 6F and G.

Fig. 6.

Characterization of SE landscapes in AGS, and identification of critical oncogenes in GC. A The H3K27ac ChIP-seq signal defines AGS enhancer characteristics and rank. B SE-associated genes in AGS cells underwent GO analysis. C A Venn diagram highlights the shared elements between transcriptomic and SE profiling data. D The overlaps in (C) underwent KEGG pathway enrichment analysis. E Relative expression of candidate genes after 6 h of THZ1 treatment in AGS detected by RT-qPCR. F Gene tracks depict H3K27ac ChIP–seq profiles of candidate oncogenes. G RNA-seq profiles of candidate oncogenes are shown, with data as mean ± SD and ***P < 0.001

SKIL is indispensable for GC malignant progression and its elevation is associated with poor prognosis in patients with GC

The above results suggest that THZ1 curbs GC progression by preferentially targeting transcripts related to the Hippo pathway, a key driver of tumorigenesis [17] known to be both frequently dysregulated and essential in GC [24, 39]. Notably, among the candidate oncogenes, SKIL is a known regulator of the Hippo pathway that promotes oncogenic transformation and EMT [31, 50]. Supporting this, SKIL downregulation in AGS cells reduced the protein levels of TAZ, a key Hippo pathway transcriptional coactivator, without affecting its mRNA (Fig. S4). We therefore focused on SKIL’s role in GC. Consistent with an oncogenic function, SKIL was significantly upregulated in GC (Fig. 7A and Fig. S6F), and high SKIL expression correlated with poor patient prognosis (Fig. 7B). Functional experiments demonstrated that SKIL knockdown (Fig. 7C, Fig. S5A, B and Fig. S6G) inhibits GC cell proliferation (Fig. 7D, E and Fig. S5C, D) and migration (Fig. 7F and Fig. S5E). These results establish SKIL as an overexpressed oncogene in GC, essential for proliferation and migration.

Fig. 7.

SKIL is indispensable for GC malignant progression and its elevation is associated with poor prognosis in patients with GC. A SKIL expression in GC cell lines (GES-1, AGS, HGC-27, and MKN45) was assessed using RT-qPCR and Western blotting. B Kaplan-Meier survival curves were used to examine the link between SKIL expression and GC patient prognosis. C SKIL expression was assessed via RT-qPCR and Western blotting after knockdown. D Viability of GC cells after SKIL knockdown was evaluated by CCK8 assays. E GC cell proliferation was quantitatively evaluated through a colony formation assay subsequent to SKIL silencing. F GC cell migration after SKIL knockdown was assessed via wound healing assay (scale bar = 500 μm). Data are shown as mean ± SD, with *P < 0.05, **P < 0.01, ***P < 0.001

SKIL overexpression partially reverses the inhibitory effect of THZ1 on malignant progression of GC cells

Since SKIL knockdown effectively inhibited GC malignant progression, we next investigated whether THZ1 exerts its anti-tumor effects by targeting SE-driven SKIL expression. IHC analysis confirmed reduced SKIL expression in subcutaneous tumor tissues from nude mice following THZ1 administration or CDK7 silencing (Fig. 8A, B). To further verify our assumption, we overexpressed SKIL in GC cells via lentiviral transfection, generating four experimental groups: vector DMSO, vector THZ1, OE DMSO and OE THZ1. Successful SKIL overexpression was confirmed by RT-qPCR and Western blot (Fig. 8C, D and Fig. S6H). Functional assays demonstrated that SKIL overexpression reverses THZ1’s inhibitory effect on GC cell proliferation (Fig. 8E, F) and migration (Fig. 8G). These results indicate that SKIL overexpression attenuates the anti-tumor efficacy of THZ1 in GC cells.

Fig. 8.

SKIL overexpression partially reverses the inhibitory effect of THZ1 on the proliferation and migration of GC cells. A-B IHC analysis revealed SKIL expression in subcutaneous tumors of nude mice after THZ1 treatment or CDK7-knockdown (scale bar: 100 μm). C The mRNA expression of SKIL after SKIL overexpression in DMSO and THZ1-treated group. D Western blot for SKIL protein levels after SKIL overexpression in DMSO and THZ1-treated group. E Viability of GC cells after SKIL overexpression in DMSO and THZ1-treated group evaluated by CCK8 assays. F Proliferation ability of GC cell lines after SKIL overexpression in DMSO and THZ1-treated group detected by colony formation assay. G The wound healing assay assessed GC cell migration after SKIL overexpression in DMSO and THZ1 groups (scale bar: 500 μm)

Discussion

The poor prognosis and limited treatment options for GC highlight our incomplete understanding of its complex etiology. Despite its heterogeneity, GC progression is driven by a common oncogene-dependent transcriptional program, the aberrant activation of which leads to cancer hallmarks such as sustained proliferation, invasion, and metastasis [14, 15]. Thus, deciphering the aberrant transcriptional landscape in GC provides a rationale for novel therapeutic discovery. Our studies demonstrated that targeting the SE-driven oncogenic transcriptional addiction represents an effective approach against GC.

Transcriptional CDKs, particularly CDK7 and CDK9, regulate gene expression by phosphorylating the RNA Pol II CTD to facilitate transcription initiation and elongation [28, 49]. CDK7 inhibition has shown promise as a therapeutic strategy in several cancers. For instance, dual targeting of CDK7 and CDK9 was effective in glioblastoma regardless of temozolomide sensitivity [3], while CDK7 inhibition suppressed gallbladder cancer by disrupting transcriptional addiction [19]. THZ1, a covalent CDK7 inhibitor, has demonstrated antitumor activity in other cancers, but its effects and molecular mechanisms on GC remain unclear. Our study demonstrated that GC cells are highly vulnerable to THZ1 due to a CDK7-mediated transcriptional dependency driven by GC-specific SEs, thereby establishing CDK7 inhibition as a promising therapeutic strategy that directly targets GC’s core oncogenic transcriptional addiction.

SEs are cell type-specific enhancer clusters that activate oncogene expression, thereby driving tumorigenesis [7, 22, 40, 47]. In GC, SEs play a critical role in facilitating malignant phenotypes. For instance, genomic analysis of 168 GC cases identified ZFP36L2, whose expression is driven by an SE-associated tandem duplication that promotes GC progression [41]. Additionally, micro-scale chromatin profiling of primary GC tissues revealed that CDX2 and HNF4α co-occupy SE regions, activating oncogenes and driving tumor progression [33]. Importantly, SE landscapes exhibit subtype-specific patterns in GC, as demonstrated by distinct regulatory profiles in mesenchymal-type GC and differential SE distributions between EMT and non-EMT subtypes [18, 37]. Therefore, characterizing these subtype-specific SEs is essential for identifying key drivers and deciphering GC tumorigenesis mechanisms. Here, we characterized the subtype-specific SE landscape and identified 1092 SE-associated genes in AGS cells. Future studies will characterize the SE landscape in other GC subtypes to fully elucidate SE-driven tumorigenic mechanisms.

Dysregulation of multiple pathways plays an essential role in GC tumorigenesis and progression [29]. We found that CDK7 inhibition preferentially targets the Hippo pathway-related transcripts, and key SE-driven oncogenes in AGS cells are enriched in this pathway. Notably, among these newly identified oncogenes in GC, SKIL is closely associated with the Hippo pathway and functions as its regulator [31, 50]. It exhibits context-dependent roles in tumorigenesis [11], with evidence supporting its elevated expression in multiple cancers [6, 31, 46, 51], as well as tumor-suppressive functions in certain contexts [21, 35]. Consistent with its pro-oncogenic role, we found that SKIL is elevated and sensitive to THZ1 treatment in AGS cells. Functionally, its knockdown effectively inhibited GC malignant progression, whereas its overexpression attenuated THZ1-induced suppression.

MST1/2, LATS1/2, YAP1, and TAZ constitute the core components of the Hippo pathway [12]. The kinases MST1/2 and LATS1/2 phosphorylate YAP1/TAZ to retain them in the cytoplasm, whereas unphosphorylated YAP1/TAZ enter the nucleus to trigger oncogene expression and facilitate tumorigenesis [25]– [26]. Hippo pathway alterations are common in GC, characterized by reduced MST1/2 and LATS1/2 coupled with increased YAP1/TAZ [32]. SKIL contributes to this dysregulation by interacting with the Hippo kinase complex to inhibit TAZ phosphorylation, thereby enhancing its oncogenic function [31, 50]. We observed that SKIL downregulation in AGS cells reduces TAZ protein levels without affecting its mRNA, indicating post-transcriptional regulation. Therefore, further investigation will be directed toward determining whether SKIL regulates TAZ via post-translational modifications and elucidating the underlying mechanisms involved.

In summary, our findings establish CDK7 inhibition as a promising therapeutic strategy for GC by targeting transcriptional addiction. Integrated analysis of the transcriptional and SE landscape demonstrates that THZ1 suppresses tumorigenesis by disrupting oncogenic transcriptional circuits, revealing its potent therapeutic potential.

Conclusions

Our findings demonstrate that THZ1 suppresses GC progression by targeting SE-driven SKIL transcription, thus establishing CDK7 inhibition as a promising therapeutic strategy. Future studies will focus on delineating the crosstalk between SKIL and the Hippo pathway in GC tumorigenesis and identifying novel subtype-specific SEs in other GC subtypes.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Bingxue Lan: Research design, study execution, manuscript drafting, data analysis, and funding acquisition. Tianli Zhou: Research, verification, manuscript drafting, and data organization. Li Pan: Research, verification, and data organization. Miaomiao Cui: Research and data organization. Qiqi Tan: Research and dataset management. Tingwei Pu: Dataset organization. Lianhui Ran: Research and data management. Sixi Wei: Reviewing and editing text. Xu Zhu: Reviewing and editing text, conducting analyses, and software. Hai Huang: Idea development, editorial review, oversight, resource management, project coordination, and funding acquisition.

Funding

The project was funded by the National Natural Science Foundation of China (82060442, 82460151), Guizhou Provincial Science and Technology Projects (qkhjc-ZK[2024]YB-238), Science and Technology Foundation of Guizhou Provincial Health Commission (gzwkj2024-313).

Data availability

The corresponding author will provide all study data and materials upon reasonable request.

Declarations

Ethics approval and consent to participate

All experimental procedures involving animals adhered to the National Institutes of Health guidelines and were authorized by the Animal Research Ethics Committee of Guizhou Medical University (approval number 2403428). Informed consent was obtained from all subjects. The study was approved by the Ethics Committee of the Affiliated Hospital of Guizhou Medical University [approval number 2024 (343)].

Competing interests

This work is original and unpublished in any other journal. All co-authors approved this publication, contributed to the research, and report no conflicts of interest.