Mevalonate pathway rewiring driven by enhancer remodelling confers resistance to KRAS inhibitors in colorectal cancer Authors: Yaoyu Guo, Yi Zhong, Peishan Hu, Yufeng Chen, Wenjun Guo, Jianfeng Chen, Peiyong Guan, Zhengran Zhou, Fang Zhu, Xian Zeng, Jiuping Gao, Qiqing Xiong, Zhaoliang Yu, Chuling Hu, Zerong Cai, Xiaoyu Xie, Jing Han Hong, Jason Yongsheng Chan, Wenyu Wang, Dechang Diao, Xiaojun Xia, Qiang Yu, Choon Kiat Ong, Yijun Gao, Xiaojian Wu, Jing Tan Published: 2026 ABSTRACT AbstractKRAS inhibitors (KRASi) have emerged as promising new cancer therapeutics for KRAS-mutant cancers; however, resistance remains a potential clinical challenge. Here, we show that reactivation of ERK is a hallmark of KRASi-resistant colorectal cancers (CRCs) and further demonstrate that enhancer remodeling rewires cholesterol biosynthesis through the mevalonate (MVA) pathway to confer this resistance. Mechanistically, enhancer remodeling activates MVA pathway, which facilitates the trafficking of KRAS to the membrane and sustains the MAPK signaling despite KRAS inhibition. Pharmacological inhibition of the MVA pathway with statins effectively blocks KRAS localization to the cell membrane, overcoming KRASi resistance in CRC. Together, these findings identify epigenetic-metabolic coupling of cholesterol biosynthesis as a mechanism of KRASi resistance and highlight targetable metabolic vulnerability in KRAS-mutant CRC.Subject terms: Colorectal cancer, Cancer metabolism, Cancer therapeutic resistance Introduction Colorectal cancer (CRC) is the third most prevalent malignancy globally and the second leading cause of cancer-related mortality1. Oncogenic KRAS mutations, the most frequently altered proto-oncogenes in CRC, occur in ~50% of cases and drive tumor progression through constitutive activation of the MAPK signaling cascade2–4. Clinically, KRAS-mutant tumors exhibit intrinsic resistance to multiple standard therapies, including chemoradiotherapy, anti-EGFR therapy, and immune checkpoint blockade, resulting in significantly reduced progression-free and overall survival compared with their wild-type counterparts5–9. This therapeutic recalcitrance highlights the urgent need for precision strategies that directly target KRAS-driven oncogenic signaling while overcoming adaptive resistance mechanisms. Recently developed KRAS-G12C inhibitors (G12Ci) represent a novel therapeutic approach for KRAS-mutant cancers. While these agents are now incorporated into the National Comprehensive Cancer Network (NCCN) guidelines for KRAS-mutant CRC10, their clinical utility remains limited because of the low prevalence of KRAS-G12C mutations ( ~ 3%) in CRC3. To expand the therapeutic options, pan-KRAS (pan-KRASi), such as BI-2865 and BI-2493, have emerged as promising therapeutic alternatives, targeting a broader spectrum of KRAS variants, including somatic mutations in G12D/V, G13D, and Q61H11 and thereby potentially benefiting >90% of KRAS-mutant CRC cases. However, despite its preclinical promise, pan-KRASi in patient-derived models and clinical cohorts remains largely elusive12,13. In addition, the molecular determinants of the therapeutic response, including predictive biomarkers and resistance mechanisms, remain poorly characterized, posing a formidable barrier to the rational clinical translation of pan-KRASi in CRC. Although clinical trials of G12Ci have demonstrated objective response rates (ORRs) of 37–45% in non-small cell lung cancer (NSCLC), response rates in CRC are significantly lower (7–18%), suggesting the presence of intrinsic and acquired resistance14–16. Similarly, emerging evidence indicates that pan-KRASi also encounters substantial resistance due to secondary mutations in components of the MAPK pathway and genetic alterations in other oncogenes12, implying that pan-KRASi monotherapy may be insufficient for effectively treating KRAS-mutant CRC. To date, multiple mechanisms of resistance to KRAS-targeted therapies have been identified, including secondary mutations, gene amplification, and compensatory activation of KRAS upstream and downstream signaling pathways16. For instance, the KRAS-Y96D mutation in NSCLC has been shown to prevent KRAS inhibitor binding, thereby sustaining ERK activity and leading to the development of resistance17,18. In addition, amplification of upstream oncogenes, such as MET, has been implicated in mediating resistance to KRAS inhibitors19. Furthermore, compensatory activation of receptor tyrosine kinases, including EGFR, ERBB2/3, FGFR1, or AXL, has been recognized as a major driver of resistance20–23. The combination of KRAS G12C inhibitors and anti-EGFR antibodies overcomes adaptive resistance and demonstrates synergistic anti-tumor activity in preclinical models20,24. These findings translate into improved efficacy and durable responses in clinical trials25,26, culminating in FDA approval of combination therapy for colorectal cancer. Moreover, tumor differentiation status has been implicated in modulating the sensitivity to KRAS-targeted therapy in both NSCLC and CRC27,28. These findings underscore the complexity of resistance mechanisms and highlight the urgent need for systematic investigations to develop more effective combination therapeutic strategies. In this study, we establish a panel of CRC patient-derived organoids (PDOs) with differential sensitivity to pan-KRASi, providing a pathologically relevant ex vivo model to dissect both intrinsic and acquired resistance mechanisms. Integrated epigenomic and transcriptomic profiling reveals that enhancer remodeling reprograms cholesterol biosynthesis through activation of the MVA (mevalonate) pathway, thereby mediating KRASi resistance. Pharmacological inhibition of the MVA pathway using statins restores therapeutic sensitivity in resistant PDOs, suggesting a critical role for epigenetically driven metabolic reprogramming in KRAS inhibitor resistance. These findings provide a strong rationale for combining KRAS inhibition with mevalonate pathway blockade to enhance KRASi efficacy and overcome drug resistance in CRC. Results In vitro effects of pan-KRAS inhibitors in preclinical CRC models To systematically assess the therapeutic vulnerability of CRC to pan-KRASi, we established a cohort of 22 patient-derived organoids (PDOs) from primary CRC specimens obtained from the Sixth Affiliated Hospital of Sun Yat-sen University and the Sun Yat-sen University Cancer Center (Fig. 1a; and Supplementary Fig. 1a). Genomic analysis identified a diverse spectrum of KRAS mutations, predominantly G12X substitutions (G12D: 22.7%; G12C: 9.1% and G12S: 4.5%) and non-G12 variants (G13D: 18.2% and Q61H: 13.6%) (Fig. 1b, c; and Supplementary Table 3), consistent with the high prevalence of codon 12/13 mutations in CRC reported in recent genomic studies3. Half-maximal inhibitory concentration (IC50) analysis demonstrated that approximately half of the KRAS-mutant PDOs were sensitive to the pan-KRAS inhibitor BI-2865 (IC50 < 5 μM), whereas the remaining PDOs exhibited intrinsic resistance (IC50 > 5 μM). Notably, no significant correlation was observed between KRAS mutation type and drug sensitivity (Fig. 1d). These findings were further confirmed in a panel of commercially available CRC cell lines, where resistance to BI-2865 was similarly observed in a subset of cell lines with KRAS mutants (Fig. 1e; and Supplementary Fig. 1b, c). Western blot analysis showed that resistant CRC models required higher BI-2865 concentrations to effectively suppress ERK phosphorylation (p-ERK) compared with their sensitive counterparts (Fig. 1f, g). In line with previous reports20,29,30, prolonged exposure to BI-2865 resulted in marked rebound activation of p-ERK, a compensatory effect that was particularly prominent in resistant CRCs (Fig. 1h, i). Next, we constructed an ERK kinase translocation reporter (ERK-KTR) and transfected it into the DLD1 cells (Supplementary Fig. 1d). By monitoring changes in GFP (green fluorescent protein) fluorescence localization, consistently, we observed that BI-2865 transiently inhibited ERK phosphorylation in the short term, but was reactivated upon prolonged treatment (Supplementary Fig. 1e). These findings suggest that aberrant p-ERK reactivation may serve as key mechanism underlying pan-KRASi resistance. Pan-KRAS Inhibitors Exhibit Selective Anti-Tumor Efficacy in CRC Models. a Representative hematoxylin and eosin (H&E)-stained section of primary CRC specimens illustrating histopathological features of CRC1, CRC2, CRC6, CRC11, CRC12, and CRC14 (left panel, scale bars, 300 μm). Bright-field microscopy of PDOs (middle panel, scale bars, 25 μm). Confocal immunofluorescence images of PDOs labeled with intestinal lineage markers CDX2 (green), CK20 (red), and DAPI (blue) (right panel, scale bars, 300 μm). b Clinicopathological and genomic characteristics of PDO cohorts, including mutational profiles of MAPK pathway genes, microsatellite instability (MSI) status, and clinical tumor stage. c Schematic of KRAS genomic architecture, annotated with mutation frequency observed in PDOs (values in parentheses indicate case counts for each variant). d, e Dose-dependent growth inhibitory effects of pan-KRAS inhibitor BI-2865 on 22 PDOs and 12 commercial CRC cell lines. IC50 values (y-axis) were derived from ATP-based viability assays (n = 3 independent experiments) after 96-h drug exposure. Dashed line denotes sensitivity threshold: IC50 ≤ 5 μM. f, g Immunoblot analysis of phospho-ERK (p-ERK) in BI-2865 treated sensitive versus intrinsically resistant CRC models after 2-h treatment (n = 3 independent experiments). h, i Time-course immunoblot analysis of p-ERK dynamics in sensitive versus resistant models treated with 5 μM BI-2865 (n = 3 independent experiments). Given previous reports linking p-ERK hyperactivation to somatic mutations in components of the MAPK pathway31, we systematically analyzed genomic datasets from the Catalogue of Somatic Mutations in Cancer (COSMIC)(https://cancer.sanger.ac.uk/cosmic/)32 and the Broad Institute Cancer Cell Line Encyclopedia (CCLE)(https://depmap.org/portal/ccle/)33 to assess mutations in H/N-RAS, BRAF, MEK1/2, ERK1/2, and NF1 in commercial CRC cell lines. This analysis revealed no significant correlation between MAPK pathway mutations and CRC resistant patterns (Supplementary Fig. 1b; Supplementary Fig. 1f). Similarly, mutations in the MAPK pathway genes in our PDO cohort did not correlate with drug resistance (Fig. 1b; and Supplementary Fig. 1g). In addition, the baseline activity of p-ERK did not correlate with sensitivity to BI-2865 in either PDOs or commercial cell lines (Supplementary Fig. 1h, i). Collectively, these findings suggest that resistance to pan-KRASi in CRC is unlikely to be driven by somatic mutations in the MAPK pathway, and mediated by alternative mechanisms, potentially involving epigenetic alterations. Aberrant ERK reactivation dynamics underlie acquired resistance to KRAS inhibitors in CRC Clinical studies have established acquired resistance as a primary cause of therapeutic failure of KRAS-G12C inhibitor (G12Ci) in CRC34; however, the resistance mechanisms underlying pan-KRASi remain poorly characterized. To investigate the molecular basis of acquired resistance to pan-KRASi in CRC, we subjected one PDO (PDO-CRC5) harboring a KRAS-G13D mutation to chronic exposure to BI-2865 to generate acquired resistance, followed by phenotypic validation (Fig. 2a). Short-term treatment with BI-2865 for 2 h effectively suppressed p-ERK levels in both sensitive and acquired-resistant PDOs, whereas the resistant models required higher drug concentrations to achieve comparable inhibition (Fig. 2b). However, consistent with the observations in Fig. 1h, prolonged BI-2865 treatment induced a pronounced p-ERK rebound in resistant PDOs, with p-ERK levels surpassing those of the untreated controls (Fig. 2c). To further validate this phenomenon, we established additional BI-2865 resistant cell line models and confirmed their resistance profiles using colony formation, cell viability, and apoptosis assays (Fig. 2d–j). Similarly, while high-dose BI-2865 initially suppressed p-ERK (Fig. 2k), a rapid rebound in p-ERK levels was observed within 24 h (Fig. 2l). To investigate the molecular basis of p-ERK rebound in acquired resistance models, we performed whole-exome sequencing (WES) of LS174T parental (LS174T-P) and resistant (LS174T-R) cell lines. Mutational profiling analysis revealed no secondary mutations in KRAS or other MAPK pathway components in LS174T-R (Supplementary Table 4), which is consistent with previous clinical observations that non-genetic adaptive mechanisms frequently drive resistance to KRAS-targeted therapies35. Taken together, these findings suggest that acquired resistance to KRAS inhibitors in CRC is predominantly driven by epigenetic remodeling rather than genetic alterations, highlighting the need for therapeutic strategies targeting epigenetic plasticity to overcome KRASi resistance. Sustained ERK Signaling Drives Acquired Resistance to Pan-KRASi in CRC Models. a Morphological comparison of KRASi-sensitive (CRC5-P) versus resistant (CRC5-R) organoids under BI-2865 treatment (0, 1, 5 μM; 10 days). Scale bar: 150 μm. b, c Dose- and time-dependent ERK1/2 phosphorylation dynamics in CRC5-P versus CRC5-R organoids (n = 3 independent experiments). d, e LS174T-P, LS174T-R, GP2D-P, and GP2D-R cells were treated with the indicated concentrations of BI-2865 (n = 3 independent experiments). The number of viable cells was measured at 96 h. f, g Growth curves of LS174T-P, LS174T-R, GP2D-P, and GP2D-R treated with vehicle or 5 μM BI-2865 (n = 3 independent experiments). h, i Representative images (left) and quantification (right) of colony formation assay of parental/resistant LS174T and GP2D cells treated with vehicle or BI-2865 (1 or 5 μM) (n = 3 independent experiments). j Quantification of the apoptotic cells among LS174T-P/R and GP2D-P/R cells treated with vehicle or 5 μM BI-2865 for 72 h, analyzed by flow cytometry. Results of Fig. d–j are represented as mean ± SD of 3 independent experiments. k Immunoblot of ERK phosphorylation in LS174T-R and GP2D-R cells treated with increasing concentrations of BI-2865 for 2 h (n = 3 independent experiments). l Immunoblot of ERK phosphorylation in LS174T-R and GP2D-R cells treated with 5 μM BI-2865 at indicated time points (n = 3 independent experiments). Data were analyzed by two-sided two-way ANOVA in (f, g), by two-sided one-way ANOVA in (h, i) and by two-sided unpaired Student’s t-test in (j). All p values are shown in the figure. Enhancer remodeling epigenetically activates cholesterol biosynthesis to drive KRASi resistance To investigate whether epigenetic plasticity contributes to resistance to KRAS inhibitors, we performed chromatin immunoprecipitation–sequencing (ChIP-Seq) for the active enhancer mark H3K27ac and the promoter-associated mark H3K4me3 to compare chromatin accessibility and transcriptional regulatory landscape between resistant and sensitive cells. The overall distributions of H3K27ac and H3K4me3 across the genome were similar between the parental and resistant LS174T cells (Fig. 3a). Both H3K27ac and H3K4me3 peaks were enriched at transcription start sites (TSSs) (Fig. 3b). There were 3178 gained and 3115 lost enhancers in the resistant LS174T cell line. Moreover, 1019 lost and 1131 gained promoters were observed in the resistant LS174T cell line (Fig. 3c). Gained enhancer regions in LS174T resistant cells showed elevated H3K27ac levels, whereas in lost regions, H3K27ac was reduced. Concordant changes in H3K4me3 and H3K27ac levels were also observed in the gained or lost promoter regions (Fig. 3d). These data indicate that widespread enhancer reprogramming is associated with acquired resistance to KRAS inhibitors in CRC. Enhancer remodeling drives transcriptional upregulation of the mevalonate/cholesterol biosynthesis axis in pan-KRASi resistant CRC models. a Genomic distribution of H3K27ac and H3K4me3 peaks in LS174T-P and LS174T-R cells. b Plot of the average level of H3K27ac and H3K4me3 peaks centered at the TSS in LS174T-P and LS174T-R cells. c Venn diagram showing the number and overlaps of enhancers and promoters between LS174T-P and LS174T-R cells. d Distribution of histone markers surrounding the summit of H3K27ac peaks, in gained or lost enhancer regions and promoter regions. Black borders, LS174T-P; red borders, LS174T-R. Each row represents 1 peak centered at the midpoint between two 2-kb flanking regions (n = 1 for ChIP-seq). e Scatterplot of the correlation between H3K27ac mark intensity and fold change of corresponding gene expressions (n = 2 for RNA-seq and n = 1 for ChIP-seq). f WiKi pathway analysis enriched by downregulated genes in LS174T-R (relative to LS174T-P) showed that the cholesterol biosynthesis pathway was significantly enriched (Permutation test). NES, normalized enrichment score. g–j The results showed relative mRNA level of HMGCS1, FDPS, SQLE and LSS in 10 commercial cell lines. Results are represented as mean ± SD of 3 independent experiments. k Representative IHC images for HMGCS1 in primary CRC tissues of CRC4 (sensitive) and CRC14 (resistant). Scale bar, 50μm. l IHC scores of HMGCS1 in primary tissues of sensitive (n = 6) and resistant (n = 6) CRC. m Reactome pathway analysis enriched by up-regulated genes in resistant CRC showed that the cholesterol biosynthesis pathway is significantly enriched (Permutation test). Comparison of serum total cholesterol (CH) n low-density lipoprotein cholesterol (LDL-C) o levels and body mass index (BMI) p, between resistant (n = 6) and sensitive (n = 7) CRC. Results of Fig. i, n–p are represented as mean ± SD. Data were analyzed by two-sided unpaired Student’s t-test in (i, n–p). All p values are shown in the figure. To determine whether the transcriptional changes are causally linked to alterations in chromatin modifications, we performed RNA sequencing (RNA-seq) in LS174T-P and LS174T-R cells. Integrative analysis of transcriptomic and chromatin profiling data revealed that changes in H3K27ac enrichment at regulatory elements were significantly correlated with corresponding fold changes in the expression of their putative target genes (Fig. 3e). By integrating histone modification (H3K27ac) and transcriptome data, we identified 657 upregulated and 644 downregulated genes in resistant cells compared to sensitive cells, based on stringent thresholds for enhancer activation/suppression and transcriptional changes ( | M_value | ≥1 for H3K27ac signal difference; |log2FC | ≥1 and adjusted P ≤ 0.05 for gene expression; Supplementary Fig. 2a). Notably, the gain of H3K27ac signal at enhancers was strongly associated with transcriptional upregulation, whereas loss of H3K27ac correlated with gene downregulation, suggesting a direct epigenetic regulatory mechanism underlying drug resistance. Intriguingly, gene set enrichment analysis (GSEA) of the Wiki pathway identified significant upregulation of the cholesterol biosynthesis pathway in the resistant LS174T cell line (Fig. 3f; and Supplementary Fig. 2b), which was validated by ChIP-qPCR and RT-qPCR (Supplementary Fig. 2c, d). Consistently, higher cholesterol level and HMGCR enzyme activity were observed in LS174T-R cells (Supplementary Fig. 2e, f). Furthermore, cholesterol biosynthesis-related genes exhibited elevated expression levels in primary BI-2865 resistant cells compared with their sensitive counterparts (Fig. 3g-j; Supplementary Fig. 2g). We further investigated the expression of HMGCS1, an enzyme involved in the initial steps of cholesterol biosynthesis in clinically acquired CRC samples. Notably, primary tumor samples corresponding to pan-KRASi resistant PDOs demonstrated significantly elevated HMGCS1 expression levels compared to their sensitive counterparts (Fig. 3k, l). Consistently, transcriptomic profiling of CRC specimens corresponding to resistant PDOs demonstrated significant enrichment of cholesterol biosynthesis compared to that in the sensitive group (Fig. 3m). Consistent with the epigenetic remodeling observed in cell lines with acquired resistance, intrinsically resistant PDO-CRC15 exhibited higher basal H3K27ac enrichment at cholesterol biosynthesis related genes compared with the sensitive PDO-CRC5, as determined by ChIP-qPCR (Supplementary Fig. 2h). Given the established correlation between cholesterol biosynthesis and clinical indicators such as serum total cholesterol (TC) and body mass index (BMI)36,37, we assessed these parameters in patients for whom PDOs were derived. Patients with BI-2865 resistant PDOs exhibited significantly elevated serum TC and low-density lipoprotein cholesterol (LDL-C) levels compared to their drug-sensitive counterparts (Fig. 3n, o; and Supplementary Table 3), along with a higher prevalence of obesity (Fig. 3p; and Supplementary Table 3). To substantiate the enhancer’s regulatory role in cholesterol biosynthesis gene expression, DLD1 cells were treated with THZ1, a CDK7 inhibitor widely recognized as a canonical enhancer suppressor38. Intriguingly, RNA-seq analysis revealed that BI-2865 induced significant upregulation of cholesterol biosynthesis pathways in therapy-resistant DLD1 cells (Supplementary Fig. 2i and Supplementary Fig. 2l). THZ1 treatment markedly diminished both the transcriptional induction of cholesterol biosynthesis pathways and the concomitant enrichment of H3K27ac (Supplementary Fig. 2j, k, m, n). These findings were corroborated by ChIP-qPCR, RT-qPCR, quantitative assessment of cellular cholesterol levels, and measurement of HMGCR enzymatic activity (Supplementary Fig. 2o–r). Furthermore, to evaluate KRASi-induced transcription dynamics, we treated cells with BI-2865 over a 24-h time course. Rapid upregulation of HMGCS1, HMGCR and FDPS were evident in resistant cells, contrasting with the time-dependent decline in sensitive lines (Supplementary Fig. 2s–v). This distinct response identifies cholesterol biosynthesis activation as a hallmark of resistance. Collectively, these findings demonstrate that enhancer remodeling orchestrates this metabolic rewiring to facilitate pan-KRASi resistance in CRC. Enhancer Inhibitors Exhibit Synergistic Effects with Pan-KRASi in CRC To determine whether enhancer remodeling contributed to pan-KRASi resistance in CRC, we performed combination index assays using the Chou-Talalay model in DLD1 and LS174T-R cells. Three CDK7 inhibitors (CDK7i) THZ1, YKL-5-124, and samuraciclib exhibited significant synergy with BI-2865 (Supplementary Fig. 3a, b). This synergy was further validated through cell proliferation, colony formation, and apoptosis assays (Supplementary Fig. 3c–h), indicating synergistic antitumor effects. In addition, we investigated the combination effects of THZ1 and RMC-6236—a RAS(ON) inhibitor currently in Phase III clinical trials39—in DLD1 and LS174T-R cells. The colony formation data indicated that THZ1 significantly enhanced anti-tumor effects of RMC-6236 (Supplementary Fig. 3i–j). Notably, similar phenotype was observed in a pan-KRASi resistant PDO (Supplementary Fig. 3k). In this context, BI-2865 treatment triggered the upregulation of cholesterol biosynthesis genes (HMGCS1, HMGCR and FDPS), whereas co-treatment with THZ1 successfully blunted this transcriptional induction (Supplementary Fig. 3l-m). Consistently, JQ1, a BET inhibitor that targets enhancer activity, demonstrated combinatorial effects comparable to those of CDK7i (Supplementary Fig. 3n–s). Western blot analysis confirmed that both THZ1 and JQ1 effectively suppressed BI-2865 induced p-ERK reactivation (Supplementary Fig. 3t–u). Taken together, these results indicate that enhancer remodeling plays a pivotal role in KRASi resistance in CRC, and targeting enhancer activity represents a promising therapeutic strategy to overcome resistance to KRAS-targeted therapy. The mevalonate pathway-driven protein farnesylation promotes pan-KRASi resistance in CRC Cholesterol is a well-established metabolic factor that contributes to tumor drug resistance40. To determine whether cholesterol directly induces resistance to KRASi, we depleted cholesterol in DLD1 cells using methyl-β-cyclodextrin (MβCD) and observed that this intervention did not alter their drug-resistant phenotype (Supplementary Fig. 4a, b). Similarly, supplementation of pan-KRASi sensitive LS174T cells with exogenous cholesterol failed to induce resistance (Supplementary Fig. 4c, d). Notably, neither cholesterol depletion nor supplementation affected the p-ERK levels (Supplementary Fig. 4e, f), suggesting that cholesterol itself does not directly mediate pan-KRASi resistance. Cholesterol biosynthesis is a multi-step biochemical process comprising the mevalonate (MVA) pathway and the downstream sterol synthesis pathway, with the MVA pathway providing isoprenoids required for protein prenylation (Fig. 4a)41. To delineate the contributions of these pathways to pan-KRASi resistance, we selectively silenced the key genes HMGCR and FDPS in the MVA pathway and SQLE and DHCR7 in the downstream sterol synthesis pathway. All four genes were significantly upregulated in the resistant cells (Fig. 4a). Depletion of HMGCR and FDPS effectively suppressed ERK phosphorylation (Fig. 4b, c), mimicking the effects observed with THZ1 and JQ1 treatment, whereas SQLE or DHCR7 depletion had no comparable effects (Supplementary Fig. 4g). Furthermore, HMGCR and FDPS silencing reversed pan-KRASi resistance in DLD1 and LS174T-R (Fig. 4d, e), whereas depletion of SQLE or DHCR7 had no such effect (Supplementary Fig. 4h). These findings indicate that pan-KRASi resistance in CRC is driven by the MVA pathway, rather than the downstream sterol synthesis pathway. MVA Pathway-mediated Protein Farnesylation Promotes Pan-KRASi Resistance in CRC. a Differential expression of cholesterol biosynthesis genes as determined by LS174T parental vs. resistant RNA-seq. Pathway map is colored by log2FC. b, c Effect of HMGCR and FDPS silencing on ERK phosphorylation in DLD1 and LS174T-R cells (n = 3 independent experiments). d, e Representative images (left) and quantification (right) of colony formation assay in DLD1 and LS174T-R cells treated with BI-2865 (5 μM for DLD1 and 10 μM for LS174T-R) with/without HMGCR and FDPS knockdown. Results are represented as mean ± SD of three independent experiments. f Immunoblot of ERK phosphorylation in DLD1 cells treated with inhibitors of MVA pathway enzymes: HMGCR inhibitor (statin: rosuvastatin calcium, 20 μM), farnesyltransferase inhibitor (FTI: FTI-277, 5 μM), geranylgeranyltransferase inhibitor (GGTI: GGTI-2418, 5 μM), squalene epoxidase inhibitor (ter: terbinafine, 20 μM) or lanosterol synthase inhibitor (BIBB: BIBB-515, 2 μM) (n = 3 independent experiments). g Representative images (left) and quantification (right) of colony formation assay in DLD1 cells treated with 5 μM BI-2865 with or without MVA pathway inhibitors. Results are represented as mean ± SD of three independent experiments. h Immunoblot of ERK phosphorylation in GP2D cells treated with 5 μM BI-2865 with or without MVA intermediates. MVA, mevalonic acid (100 μM); FPP, farnesyl pyrophosphate (2 μM); GGPP, geranylgeranyl pyrophosphate (2 μM); EOS, 2,3-epoxysqualene (2 μM); CH, cholesterol (10 μM) (n = 3 independent experiments). i Representative images (left) and quantification (right) of colony formation assay in GP2D cells treated with BI-2865 with or without MVA intermediates. Results are represented as mean ± SD of three independent experiments. j, k The growth curves of DLD1 and LS174T-R treated with BI-2865 (5 μM for DLD1 and 10 μM for LS174T-R, optimal concentrations were selected from the moderate CI range) in combination with rosuvastatin calcium (20 μM) for 5 d. Data are presented as the mean ± SD (n = 3). l Representative images of CRC15 organoid treated with DMSO, BI-2865 (5 μM), rosuvastatin calcium (5 μM), or combination (n = 3 independent experiments). (Scale bar, 150 μm.). Data were analyzed by two-sided one-way ANOVA in (d, e, g and i) and by two-sided two-way ANOVA in (j, k). All p values are shown in the figure. To further validate these phenotypes, we employed pharmacological inhibitors targeting distinct segments of the cholesterol biosynthesis pathway. Targeting HMGCR (the rate-limiting enzyme of the MVA pathway) by using rosuvastatin calcium effectively attenuated p-ERK reactivation and sensitized cells to BI-2865 and RMC-6236 treatment (Supplementary Fig. 4i, j). However, inhibition of the downstream enzymes SQLE (terbinafine) and LSS (BIBB-515) did not exhibit such an effect. Notably, pharmacological blockade of protein farnesylation using FTI-277 effectively recapitulated the therapeutic synergy observed with statins; however, geranylgeranylation inhibition with GGTI-2418 had no similar effects (Fig. 4f, g), which is consistent with previous reports42. Conversely, supplementation of pan-KRASi sensitive GP2D cells with mevalonic acid (MVA) and farnesyl pyrophosphate (FPP), but not geranylgeranyl pyrophosphate (GGPP), 2,3-epoxysqualene, or lanosterol, induced p-ERK reactivation (Fig. 4h) and conferred partial resistance to BI-2865 (Fig. 4i). To further assess the impact of KRAS farnesylation on the response to KRAS inhibitors, we treated DLD1 and LS174T-R cells with FTI-277. Subsequent co-IP analysis showed that FTI-277 blocks KRAS farnesylation and concomitantly reduces the IC50 of BI-2865 (Supplementary Fig. 4k–m), suggesting that MVA pathway mediated KRAS farnesylation may contribute to resistance to pan-KRAS inhibitors. Meanwhile, organoid assays likewise demonstrated that FTI-277 can potentiate the efficacy of BI-2865 and suppress the reactivation of p-ERK (Supplementary Fig. 4n–p). To assess the therapeutic potential of targeting the MVA pathway, we evaluated the antitumor efficacy of rosuvastatin calcium and atrovastatin calcium in combination with BI-2865. Statins co-treated with BI-2865 significantly suppressed the proliferation of pan-KRASi resistant CRC cells and PDOs (Fig. 4j–l; and Supplementary Fig. 4q–r). Collectively, these findings demonstrate that upregulated cholesterol biosynthesis promotes pan-KRASi resistance, primarily through MVA pathway-driven protein farnesylation. Targeting protein farnesylation via MVA pathway inhibition represents a promising therapeutic strategy for overcoming pan-KRASi resistance in CRC. Mevalonate pathway mediated KRAS membrane trafficking drives pan-KRASi resistance The KRAS protein requires farnesyl pyrophosphate, synthesized via the mevalonate pathway, to undergo farnesylation, a post-translational modification essential for membrane anchoring and the subsequent activation of downstream signalling (Fig. 5a)43. To elucidate the correlation between KRAS membrane trafficking and KRASi resistance, we employed pan-KRASi resistant (DLD1 and HCT116) and sensitive (GP2D) cells. Our findings revealed that BI-2865 treatment induced a time-dependent, progressive accumulation of KRAS at the plasma membrane in DLD1 and HCT116 cells, an effect that was absent in GP2D cells (Fig. 5b–e; and Supplementary Fig. 5a, b). Crucially, this accumulation was specific to KRAS, with no similar changes observed for HRAS or NRAS, both of which are common oncogenic RAS family members that require prenylation for proper function44. Furthermore, this membrane enrichment of KRAS was concomitant with its activation, a phenomenon not observed for HRAS or NRAS (Supplementary Fig. 5c–e). These specific KRAS changes were further corroborated in acquired resistant LS174T-R cells, where BI-2865 similarly induced KRAS activation and ERK phosphrylation, while maintaining HRAS and NRAS suppression (Supplementary Fig. 5f). This evidence strongly suggests that altered KRAS membrane trafficking is a critical mechanism driving drug resistance. Pharmacological Inhibition of MVA Pathway Abrogates KRAS Plasma Membrane Trafficking. a Schematic of MVA pathway-dependent KRAS membrane trafficking via farnesylation-mediated lipid anchoring. (Created in BioRender. Guo, E. (https://biorender.com/gj7ehxc) b Temporal dynamics of KRAS subcellular localization in HCT116, DLD1 and GP2D cells treated with 5 μM BI-2865 at the indicated time points. Immunofluorescence co-staining of KRAS (red), ATP1A1 (plasma membrane marker, green), and Nuclei (DAPI, blue) (n = 3 independent experiments). Scale bar, 10 μm. Subcellular fractionation immunoblot analysis of cytosol and membrane KRAS, HRAS and NRAS protein levels in c, HCT116, d DLD1 and e GP2D cells treated with BI-2865 (5 μM) for 0, 2, 8 and 24 h (n = 3 independent experiments). f Representative confocal images of DLD1 cells treated with 5 μM BI-2865, 20 μM rosuvastatin calcium or combination for 24 h. Cells were stained with anti-KRAS antibody (red), anti-ATP1A1 (green) and DAPI (blue) (n = 3 independent experiments). Scale bar, 10 μm. g Subcellular fractionation immunoblot analysis of cytosol and membrane KRAS, HRAS and NRAS protein levels in DLD1 cells treated with BI-2865 (5 μM), rosuvastatin calcium (20 μM) or combination for 24 h (n = 3 independent experiments). h Immunoprecipitation with anti-pan-farnesylation antibody followed by immunoblot with KRAS antibody using cell extracts form DLD1 cells treated with BI-2865 (5 μM), rosuvastatin calcium (20 μM) or combination for 24 h (n = 3 independent experiments). i Representative confocal images of LS174T and GP2D cells treated with 5 μM BI-2865, 100 μM mevalonic acid or combination for 24 h. Cells were stained with anti-KRAS antibody (red), anti-ATP1A1 (green) and DAPI (blue) (n = 3 independent experiments). Scale bar, 10 μm. Immunoblot of cytosol and membrane KRAS, HRAS and NRAS protein levels in j, LS174T and k GP2D cells treated with BI-2865 (5 μM) with or without MVA (mevalonic acid, 100 μM) addition for 24 h (n = 3 independent experiments). l Immunoblot of KRAS/HRAS/NRAS activation and ERK phosphorylation in LS174T and GP2D cells treated with BI-2865 (5 μM) with or without MVA (mevalonic acid, 100 μM) addition for 24 h (n = 3 independent experiments). Notably, co-treatment with statin effectively prevented KRAS accumulation in plasma membrane in DLD1 and LS174T-R cells. This prevention was specific to KRAS, as HRAS and NRAS localization remained unaffected (Fig. 5f; and Supplementary Fig. 5g-i). These results were further corroborated by subcellular fractionation (Fig. 5g; Supplementary Fig. 5j). We used immunoprecipitation assay for further validation, and found that BI-2865 promotes KRAS farnesylation, while statin co-administration suppresses this process (Fig. 5h). In addition, KRAS membrane trafficking correlated with its activation state and p-ERK reactivation (Supplementary Fig. 5k), reinforcing the notion that membrane-localized KRAS sustains oncogenic ERK activation. Similarly, co-treatment with FTI-277 prevented KRAS membrane accumulation in resistant DLD1 cells (Supplementary Fig. 5l). Consistently, MVA supplementation in pan-KRASi sensitive LS174T and GP2D cells selectively enhanced KRAS membrane trafficking, while HRAS and NRAS remained unaffected (Fig. 5i, k; Supplementary Fig. 5m–p). This effect also led to p-ERK reactivation, consequently alleviating pan-KRASi induced ERK pathway inhibition (Fig. 5l). Collectively, these findings establish that MVA pathway-driven KRAS membrane trafficking plays a critical role in pan-KRASi resistance, while statins inhibit the MVA pathway-mediated KRAS membrane anchoring, thereby overcome the resistance to pan-KRASi in CRC. Statins restore pan-KRASi sensitivity and enhance therapeutic efficacy in preclinical CRC models To assess the therapeutic efficacy and safety profile of pan-KRAS inhibitor BI-2493, an optimized in vivo-ready structural analog of BI-2865, we conducted comprehensive preclinical evaluations using xenograft models, both as monotherapy and in combination with statins. In the LS174T-R xenograft model, combination therapy with BI-2493 and statins significantly inhibited tumor growth compared with monotherapy, with modest body weight change (Fig. 6a, b, and Supplementary Fig. 6a). Serum biochemical analysis and histopathological examination of organ tissues revealed no evidence of hepatorenal toxicity or pathological alterations (Supplementary Fig. 6b–f), indicating favorable biosafety profiles for the combination regimen. In vivo therapeutic synergy of statin and pan-KRASi in CRC resistance models. a, b The tumor growth curve (a) and weight of tumor (b) with treatment of vehicle, BI-2493, rosuvastatin calcium, or combination in LS174T-R xenografts (n = 6 per group). Error bars represent ± SD. c, d The tumor growth curve (c) and weight of tumor (d) with treatment of vehicle, BI-2493, rosuvastatin calcium, or combination in PDX-CRC15 xenografts (n = 8 per group). Error bars represent±SD. e Representative IHC staining of Ki-67 and p-ERK in LS174T-R xenografts (Scale bar, 100 μm). f Relative Ki-67 and p-ERK expression in four groups of LS174T-R xenograft tumors (n = 6 per group). Data are presented as the mean ± SD. g Representative IHC staining of Ki-67 and p-ERK in PDX-CRC15 (Scale bar, 100 μm). h Relative Ki-67 and p-ERK expression in four groups of PDX-CRC15 (n = 8 per group). Data were analyzed by two-sided two-way ANOVA in (a, c) and by two-sided one-way ANOVA in (b, d, f and h). This therapeutic synergy was further validated in a CRC patient-derived xenograft (PDX) model with a KRAS-mutant (PDX-CRC15), where co-administration of pan-KRASi and statins resulted in robust tumor suppression with minimal side effects (Fig. 6c, d; Supplementary Fig. 6g). Immunohistochemistry (IHC) analysis showed that combination therapy markedly reduced Ki-67, a proliferation index, and concurrently suppressed aberrant ERK phosphorylation within the tumor tissues (Fig. 6e–h). Collectively, these findings demonstrate that statins could enhance the anti-tumor efficacy of pan-KRASi, highlighting a clinically translatable strategy to improve therapeutic outcomes in KRAS-mutant CRC. Combining statins with clinically relevant G12Ci enhances therapeutic efficacy in resistant CRC preclinical models Given the ongoing clinical development of pan-KRAS inhibitors, we further investigated the therapeutic potential of combining statins with the FDA-approved G12C inhibitor sotorasib. We first established the acquired sotorasib-resistant cell line SW837-R and validated its drug resistance through ATP viability, colony formation and apoptosis assays (Fig. 7a–d). Similar to the pan-KRASi resistant models, resistant SW837-R cells required higher sotorasib concentrations to suppress p-ERK levels (Fig. 7e). Notably, compared to their parental counterparts, SW837-R cells exhibited a rapid rebound in p-ERK signaling within 24 h of treatment (Fig. 7f). In vitro assays demonstrated that the combination of sotorasib with statin or CDK7i significantly inhibited tumor growth in G12Ci-resistant cell lines harboring the G12C mutation (Fig. 7g–i; and Supplementary Fig. 7a–d). Similarly, we observed robust p-ERK reactivation following sotorasib treatment in SW837-R cells, whereas statin co-treatment effectively suppressed this activation (Fig. 7j). The combinatorial effect of sotorasib plus statin was also verified in SW837-R xenografts (Fig. 7k; and Supplementary Fig. 7e), with no evidence of additional toxicity (Supplementary Fig. 7f–j). IHC analyses mirrored the effects observed with pan-KRASi statin combinations, demonstrating synergistic suppression of tumor proliferation and ERK activity (Fig. 7l–m). Pharmacological Synergy between G12Ci and Statins in Preclinical CRC Models. a SW837-P and SW837-R cells were treated with the indicated concentrations of sotorasib for 96 h. b Growth curves of SW837-P and SW837-R treated with vehicle or 1 μM sotorasib. c Representative images (left) and quantification (right) of colony formation assay of SW837-P/R cells treated with vehicle or sotorasib (0.25 or 1 μM). d Quantification of the apoptotic cells in SW837-P/R cells treated with vehicle or 1 μM sotorasib for 72 h, analyzed by flow cytometry. e, f Immunoblot of ERK phosphorylation in SW837-P/R cells after 2-h dose escalation or time-course of 1 μM sotorasib treatment for 2 h (n = 3 independent experiments). g, h The growth curves of SW837-R and SW1463 cells treated with G12C inhibitor (sotorasib 5 μM for SW837-R and 200 nM for SW1463) with or without 20 μM rosuvastatin calcium for 5 d. i Representative images (left) and quantification (right) of colony formation assay in SW837-R and SW1463 cells treated with sotorasib (5 μM for SW837-R and 200 nM for SW1463), rosuvastatin calcium (20 μM) or combination. Results of Fig. a–d and g–i are represented as mean ± SD of three independent experiments. j Immunoblot of ERK phosphorylation in SW837-R cells treated with sotorasib (5 μM) and rosuvastatin calcium (20 μM) (n = 3 independent experiments). k In vivo efficacy of SW837-R xenografts under sotorasib (100 mg/kg), rosuvastatin calcium (50 mg/kg) or combination treatment: (left) tumor growth curve and (right) tumor weight (n = 5 per group). l Representative IHC staining of Ki-67 and p-ERK in SW837-R xenografts (Scale bar, 100 μm). m Relative Ki-67 and p-ERK expression in four groups of SW837-R xenograft tumors (n = 5 per group). n The tumor growth curve (left) and weight of tumor (right) with treatment of vehicle, sotorasib (100 mg/kg), rosuvastatin calcium (50 mg/kg), or combination in PDX-CRC16 xenografts (n = 8 per group). oRepresentative IHC staining of Ki-67 and p-ERK in PDX-CRC16 (Scale bar, 100 μm). p Relative Ki-67 and p-ERK expression in four groups of PDX-CRC16 (n = 8 per group). q Tumour volumes over the treatment course of the SW837 xenograft model treated with sotorasib (50 mg/kg), rosuvastatin calcium (50 mg/kg) or their combination (left) (n = 6 per group). Kaplan-Meier curves of SW837 xenografts after indicated treatment. r Representative IHC staining of Ki-67 and p-ERK in SW837 xenograft (Scale bar, 100 μm). s Relative Ki-67 and p-ERK expression in four groups of SW837 xenograft (n = 6 per group). Error bars in Fig. k, m, n, p, q and s represent mean ± SD. Data were analyzed by two-sided two-way ANOVA in (b, g, h, left k, left n and left q), by two-sided one-way ANOVA in (c, i, right k, m, right n, p and s), by two-sided unpaired Student’s t-test in (d) and by two-sided log-rank tests (right q). All p values are shown in the figure. To reinforce the clinical translatability of the KRASi-statin combination therapy, we further validated this effect in PDOs with KRAS-G12C mutation (Supplementary Fig. 7k-l). Furthermore, drug combination effects were observed to significantly inhibit tumor growth in the KRAS-G12C PDX model (Fig. 7n), with IHC analysis demonstrating trends similar to those observed in the aforementioned experiments (Fig. 7o-p). Despite a subset of KRAS-mutant CRC patients shows the initial favorable response to sotorasib, responses are often short-lived due to rapid development of resistance25,45. To determine whether statin could delay resistance to KRAS inhibition, we tested sotorasib and statin as single agents and in combination in the SW837 xenograft model. While the tumors were able to adapt and outgrow sotorasib treatment alone, the combination treatment maintained durable control and prolonged survival compared to single-agent group (Fig. 7q), suggesting that statin could delay resistance to sotorasib in CRC. Similarly, the IHC analysis of SW837 xenograft reflect a trend consistent with previous results (Fig. 7r, s). Taken together, these findings highlight the broad therapeutic potential of statins in overcoming resistance to KRAS-targeted therapies, providing a clinically translatable strategy to mitigate G12Ci resistance in KRAS-mutant CRC. Discussion Despite the therapeutic potential of KRASi in targeting KRAS-mutant cancers, the emergence of drug resistance remains a formidable obstacle to the effective treatment of KRAS-mutant CRC. In this study, we identified ERK reactivation as a defining feature of KRASi resistance, which occurs independently of secondary mutations in KRAS or other MAPK pathway components. Mechanistically, we demonstrated that enhancer remodeling acts as a key driver of resistance by upregulating the MVA pathway. This upregulation facilitates KRAS membrane trafficking via protein farnesylation, thereby sustaining p-ERK activation and downstream oncogenic signaling. This epigenetic-metabolic coupling represents a non-genetic mechanism of drug resistance. Importantly, pharmacological blockade of the MVA pathway using statins restored sensitivity to KRASi and enhanced therapeutic efficacy in cellular, PDO, and PDX models (Fig. 8). In addition, the inhibition of enhancer activity through CDK7 or BET inhibitors effectively suppresses ERK reactivation and acted synergistically with KRASi. Notably, co-treatment with the FDA-approved KRAS-G12C inhibitor sotorasib and statins exhibits robust tumor suppression, supporting the clinical translation of this combination strategy in KRAS-mutant CRC. From a therapeutic perspective, our study provides compelling preclinical evidence for combining statins with KRASi as a viable approach to augment therapeutic responses and overcome resistance. Given the established safety profile and widespread clinical use of statins, this combination strategy holds strong translational potential for patients with KRAS-mutant CRC. Schematic Model Illustrating the Mechanistic Role of Statins in Overcoming KRASi Resistance in CRC. In KRASi-sensitive cells, the mevalonate (MVA) pathway exhibits low basal activity, resulting in diminished farnesylation-dependent KRAS membrane trafficking. This metabolic state permits sustained suppression of ERK signaling by KRASi-mediated inhibition. Conversely, in KRASi-resistant cells, enhancer remodeling induces transcriptional upregulation of MVA pathway components, driving excessive farnesylation and compensatory KRAS membrane trafficking. This aberrant location facilitates compensatory reactivation of ERK, thereby conferring resistance to KRASi. Pharmacological inhibition of the MVA pathway by statins blocks KRAS membrane trafficking, effectively abrogating KRAS-dependent ERK hyperphosphorylation and overcoming the resistance to KRAS in CRC. (Created in BioRender. Guo, E. (https://biorender.com/0dsu4ii). While previous studies have implicated secondary mutations in KRAS or MAPK pathway components as the primary mechanism of KRASi resistance35, our WES data indicate that resistance in CRC is predominantly driven by epigenetic reprogramming rather than genetic alterations, recapitulating the non-genetic resistance observed in 30% of CRC patients following dual inhibition of KRAS and EGFR24. Epigenetic mechanisms, including histone modification and DNA methylation, can modulate chromatin accessibility and transcriptional programs that support cell survival under therapeutic conditions. Enhancer remodeling, largely mediated by H3K27ac modifications, dynamically rewires oncogenic or pro-survival gene networks by activating alternative regulatory elements to sustain drug resistance. Despite its central role in resistance to MEK inhibitors46,47, enhancer-driven modulation of KRASi response has not been previously described. A recent study revealed that targeting EZH2, a histone methyltransferase, synergizes potently with RAS/MAPK pathway inhibitors to enhance antitumor efficacy in CRC, highlighting the therapeutic promise of combined epigenetic and KRAS inhibition. In our study, the observation of synergistic antitumor effects between KRASi and CDK7 or BET inhibitors underscores the therapeutic value of targeting enhancer-mediated transcriptional rewiring. These findings warrant further investigation of additional epigenetic modulators that may reverse resistance phenotypes. Emerging evidence has highlighted the interplay between epigenetic remodeling, metabolic reprogramming, and therapeutic resistance. For instance, loss of EZH2 has been shown to reprogram branched-chain amino acid (BCAA) metabolism, creating metabolic vulnerabilities in cancer-initiating cells48. Similarly, epigenetically induced lactate production has been implicated in EGFR inhibitor resistance through the activation of the EGR1/NNMT/ALDH3A1 axis49. Our findings demonstrate that epigenetic alterations in KRAS-resistant CRC led to MVA pathway upregulation and that pharmacological inhibition of this pathway using statins achieved comparable efficacy to enhancer-targeting strategies when combined with KRAS inhibitors. The MVA pathway plays a critical role in cholesterol biosynthesis and provides substrates necessary for prenylation, which is essential for membrane localization of signaling proteins. Gao et al. demonstrated that chemoresistant small cell lung cancer (SCLC) depends on the MVA–GGPP pathway to sustain Rab7a localization to plasma membrane, thereby evading ROS-induced apoptosis50. Similarly, Saito et al. reported that MVA pathway activation in hepatocellular carcinoma (HCC) promotes resistance to sorafenib and anti-PD-1 therapy via TAZ-TEAD2-mediated signaling51. These findings emphasize the therapeutic relevance of targeting the MVA cascade. Statins, inhibitors of HMG-CoA reductase (HMGCR), have demonstrated synergistic anti-tumor activity when combined with EGFR inhibitors in lung adenocarcinoma52. Moreover, clinical studies suggest they may enhance gefitinib efficacy in resistant non-small cell lung cancer (NSCLC) patients53. Although previous study reported that sotorasib treatment in CRC induces p-ERK reactivation and subsequent therapeutic resistance30, the transcriptional and metabolic mechanisms driving this feedback response have remained incompletely defined. Previous reports indicate that wild-type RAS isoforms, including HRAS and NRAS, can mediate MAPK pathway reactivation, thereby contributing to G12Ci resistance54. Here, we identified a critical feedback loop linking epigenetic reprogramming to metabolic adaptation, providing a compelling rationale for the dual inhibition of epigenetic regulators and the mevalonate pathway to overcome KRAS inhibitor resistance. Future research should address several important issues. First, prospective clinical trials are warranted to evaluate the efficacy of statin-based combination therapies in patients with KRAS-mutant CRC, particularly those who have developed resistance to KRAS-G12C inhibitors. Second, biomarker-driven studies are needed to identify patients most likely to benefit from such strategies, potentially through metabolic profiling of the MVA pathway in tumor or plasma samples. The observed correlation between plasma cholesterol levels and KRASi response supports the clinical utility of circulating lipid biomarkers in guiding patient stratification. Furthermore, alternative strategies to disrupt KRAS membrane trafficking, such as farnesyl transferase inhibitors (FTIs), merit exploration for their synergistic effects with KRASi. Finally, while enhancer remodeling appears central to resistance, the specific transcriptional regulators involved remain poorly defined. Integrative chromatin accessibility and transcriptomic analyses are necessary to elucidate the epigenetic landscape that underlies resistance. In summary, this study identified enhancer remodeling-driven activation of the MVA pathway as a potential mechanism of KRASi resistance in CRC, with KRAS membrane trafficking via farnesylation emerging as a critical mediator of sustained oncogenic signaling. We established statins as promising agents to counteract this resistance and enhance the efficacy of KRASi. These findings provide a strong rationale for clinical translation and support the development of metabolism- and epigenetic-based combination therapies for KRAS-mutant CRC. Methods Ethical statement Archived patient samples were obtained from Sixth Affiliated Hospital, Sun Yat-sen University with an approval from the medical ethics committee (Approval No. 2018ZSLYEC-082). All participated patients were fully informed, followed by documentation in a written, signed, and dated informed consent form. All animal studies were conducted in compliance with animal protocols approved by the Animal Care and Ethics Committee of Sun Yat-sen University and followed the National of Health Guidelines on the Care and Use of Animals (SYSU-IACUC-2025-001159). The ethical authorization limits tumor size to a maximum size of 2000 mm3, and this limit was not exceeded in any experiment conducted for this study. Cell lines and reagents All commercial cell lines were obtained from the ATCC. All cell lines were authenticated via short tandem repeat fingerprinting prior to experimentation. Human tumor cell lines were cultured in DMEM (LS174T, GP2D, HCT116, SW837, and LoVo), RPMI 1640 (DLD1, HCT8, HCT15, LS513, SW480, and SW620), or DMEM/F12 (SW1463) supplemented with 10% fetal bovine serum (FBS, HyClone, GE Healthcare Life Sciences) and 1% penicillin/streptomycin (Gibco). The cell lines were cultured at 37 °C in a humidified incubator with 5% CO2 and routinely tested for mycoplasma contamination every month. Cells used in the experiments were between 3 and 30 passages post-thaw. The following compounds were purchased from Targetmol (Boston, MA): BI-2493 (T72061), sotorasib (T8684), THZ1 (T3664), YKL-5-124 (T22461), samuraciclib (T61835), JQ-1 (T2110), rosuvastatin calcium (T1510), FTI-277 (T62675), GGTI-2418 (T11396), terbinafine (T6702), and BIBB-515 (T4039). BI-2865 (HF04235) was purchased from Hi-Future (Nantong, China). For the colony formation assay, single-cell suspensions (20,000 cells per well) were seeded in 6-well plates, and the medium containing various drugs was replaced every three days over a 10-day culture period. The colonies were then fixed with methanol, stained with 0.5% crystal violet in 20% methanol, and imaged. Colony quantification was performed using the ImageJ/Fiji software (https://fiji.sc/). For the cell viability assay, the cells were plated in triplicate in 96-well plates at the optimal seeding density. After 24 h, the cells were treated with varying concentrations of the drugs and incubated at 37 °C for 96 h. The number of viable cells was determined using the CellTiter-Glo Luminescent Cell Viability Assay (#G7573, Promega, Madison, WI, USA), following the manufacturer’s instructions. Patient Samples Primary colorectal cancer specimens were obtained from the Sun Yat-sen University Sixth Affiliated Hospital and Sun Yat-sen University Cancer Center. Nineteen CRC samples confirmed to harbor KRAS mutations and three KRAS wild type CRC samples were used to establish PDOs. The study was approved by the Clinical Research Ethics Committee of the Sixth Affiliated Hospital of Sun Yat-sen University (Approval No. 2018ZSLYEC-082) and conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants prior to sample collection. Tumor isolation and patient-derived organoid culture The methods for developing and cultivating PDOs were adapted from a previously published protocol55. In brief, CRC specimens were minced and incubated in digestion buffer (HBSS containing 1 mg/mL collagenase from Clostridium histolyticum, 0.1 mg/mL DNase I, 3 mM calcium chloride, 100 µg/mL primocin, and 10 µM Y-27632) at 37 °C for 60 min with orbital shaking. After incubation, large fragments were removed by passing the mixture through a 100-µm cell strainer. The cells were centrifuged and washed twice with PBS. After counting, cells were embedded in ice-cold Matrigel and seeded into 24-well plates. Matrigel polymerization was allowed to occur at 37 °C for at least 15 min. CRC culture medium (Advanced DMEM/F12 supplemented with 2 mM GlutaMAX, 10 mM HEPES, 100 U/mL penicillin and streptomycin, 1X B27, 1.25 mM n-Acetyl Cysteine, 10 nM Gastrin I, 500 nM A-83-01, 3 µM SB202190, 100 ng/mL recombinant human RSpondin, 100 ng/mL recombinant human Noggin, 50 ng/mL recombinant human EGF, 10 nM prostaglandin E2, 100 µg/mL primocin, and 10 mM nicotinamide) was added and refreshed every two to three days. For PDOs that grew successfully, the organoids were passaged every 1-2 weeks depending on their growth rate. The CRC culture medium was removed, and the Matrigel dome was degraded with PBS. The PBS-Matrigel-organoid mixture was collected, centrifuged at 300 g for 5 min, and PBS was discarded. TrypLE Express enzyme (#12604013, Gibco) was added and the organoids were incubated at 37 °C for 10 min. After adding PBS, the cells were centrifuged again, re-washed with PBS, and reseeded in Matrigel in a 24-well plate. The culture medium was then added following matrigel polymerization. Bright-field images were captured using a Leica DMIL LED fluorescence microscope. Some of the established PDOs were dissociated and cryopreserved for sequencing in a frozen medium (70% FBS + 20% culture medium + 10% DMSO) in liquid nitrogen. Establishment of cell lines and PDOs with acquired resistance to KRAS inhibitors LS174T and GP2D cells were grown in the presence of increasing concentrations of BI2865. SW837 cells were grown in the presence of increasing concentrations of sotorasib. Subconfluent cells were treated with BI-2865 or sotorasib at the IC50 dose for two 3-day periods, and surviving cells were maintained for another 4 days. This process was repeated once, and the remaining cells were cultured in the presence of the drug for three additional passages. Resistant cells were maintained in the presence of BI-2865 or sotorasib, and passaged parental cells were used for various cellular assays. PDO-CRC5 organoids were maintained in increasing concentrations of BI-2865 for two 5-day periods, followed by passaging at 5-day intervals without drug exposure. This process was repeated three times, and the remaining PDOs were maintained in the presence of BI-2865. Apoptosis assay Cells (1 × 105 cells were seeded in a 6-well plate and treated with the specified agents for 72 h. Apoptotic cells were quantified using the Annexin V–FITC Apoptosis Detection Kit (Vazyme), following the manufacturer’s instructions. All experiments were conducted in triplicates. Data acquisition and analysis were performed using a Spectral Cell Analyzer SP6800Z (Sony). Representative flow cytometry gating of apoptosis assay is shown in Supplementary Fig. 8. Combination index analysis BI2865 resistant cells were simultaneously exposed to the indicated concentrations of BI2865 and CDK inhibitors or JQ1 for 96 h, and cell viability was subsequently assessed using CellTiter-Glo reagent (Promega). Combination index values were determined by the inhibition rate of the cells and calculated using the CalcuSyn software (http://www.biosoft.com/w/calcusyn.htm). Combination index values less than 1 indicated a synergistic interaction between compounds, whereas those equal to or more than 1.0 indicated additivity or antagonism. Immunoblotting and immunoprecipitation The detailed procedure of immunoblotting was described in a previous publication56. In brief, cell pellets were lysed with RIPA buffer (Roche) to extract proteins. Protein concentrations were measured using the Bradford assay (Bio-Rad). Equal amounts of protein were loaded onto SDS-PAGE gels and subsequently transferred to PVDF membranes (Bio-Rad). After blocking the membranes with 5% BSA (Sigma-Aldrich), they were incubated with the appropriate primary and secondary antibodies. Immunoblotting was developed using ECL Western Blotting Detection Reagents (GE Healthcare Life Sciences) and visualized with a Bio-Rad ChemiDoc MP imaging system. Details of antibodies and dilutions in this study were provided in Supplementary table 1. Membrane and cytoplasmic proteins were isolated using the Cell Membrane Protein and Cytoplasmic Protein Extraction Kit (P0033, Beyotime) according to the manufacturer’s instructions. Extracted samples were mixed with membrane protein solubilization buffer (62.5 mM Tris-HCl, pH 6.8; 15% SDS; 8 M urea; 10% glycerol; 100 mM DTT) and 4× loading buffer (150 mM Tris-HCl, pH 6.8; 12% SDS; 30% glycerol; 6% 2-mercaptoethanol; 0.02% bromophenol blue), incubated for 30 min at 37 °C, and then subjected to immunoblot analysis. For immunoprecipitation (IP) analysis, cells were lysed in co-IP lysis buffer (50 mM Tris-HCl, 0.5% NP-40, 120 mM NaCl, 1 mM EDTA, pH 8.0), containing proteasome inhibitors. Cell lysates were incubated with indicated antibodies at 4 °C overnight and then incubated with protein G beads (GE Lifesciences) for additional 4 h. The immunoprecipitates were washed four times with lysis buffer, followed by IB analysis. RAS-GTP Pulldown Following the specified treatment, RAS activity was evaluated via RAF-RBD pulldown (Beyotime) followed by immunoblotting. Both the pulldown samples and whole-cell lysates were separated on a 12% SDS-PAGE gel, and western blot analysis was conducted using KRAS (Santa Cruz, sc-517599), HRAS (Proteintech, 18295-1-AP) and NRAS (Proteintech, 10724-1-AP) specific antibodies. RNA-Seq and data analysis RNA was extracted from CRC organoids or cell lines (LS174T-P, LS174T-R, or DLD1 after 24 h of treatment) using the RNeasy Mini Kit (Qiagen). RNA sequencing libraries were prepared using the TruSeq Stranded RNA HT kit (96 samples) and Ribo-Zero Gold (Illumina) according to the manufacturer’s protocol. Sequencing was performed using the Novaseq S6000 system (Illumina). Library integrity was assessed using an Agilent 2100 Bioanalyzer. Reads that passed quality control were obtained using the fastp software (version 0.12.5; ref. 57) and aligned to the human reference genome (GRCh37, hg19) with STAR (github.com/alexdobin/STAR) using GENCODE annotation (gencode.v19.annotation.gtf) and the settings recommended in the STAR manual. The gene expression levels were quantified using RSEM (version 1.3.3; github.com/deweylab/RSEM). Differential gene expression was analyzed using DESeq2 (version 1.28.1; |log2 fold change | ≥ 1 and adjusted P-value ≤ 0.05; github.com/thelovelab/DESeq2). Gene set enrichment analysis (GSEA) was performed using GSEA (version 4.0.4; www.gsea-msigdb.org/gsea). Custom R scripts were used to generate visualizations. qRT-PCR Total RNA was extracted using the RNeasy Mini Kit (Qiagen), and cDNA was synthesized using the TranScript All-in-One First-Strand cDNA Synthesis SuperMix for RT-PCR (one-step gDNA removal). Quantitative real-time PCR (qRT-PCR) was performed according to the instructions provided with PerfectStart Green qPCR SuperMix (TransGen Biotech). 18S rRNA served as an endogenous housekeeping gene for normalization. The primer pairs used for qRT-PCR are listed in Supplementary table 2. The mRNA expression levels of target genes were calculated as the average of the Ct values obtained from the respective primer pairs. Data are presented as relative gene expression. ChIP and ChIP-qPCR Cells were treated with BI-2865 and CDK7 inhibitors, as specified, followed by cross-linking for 10 min at room temperature using 1% formaldehyde. The reaction was then quenched with 0.125 M glycine for 5 min. The cross-linked cells were then washed with TBSE buffer (20 mM Tris-HCl pH 7.5, 1 mM EDTA, 150 mM NaCl) and stored at −80 °C for later use. To prepare for chromatin immunoprecipitation (ChIP), the frozen cross-linked cells were lysed in a buffer containing 50 mM Tris-HCl (pH 8.0), 10 mM EDTA, and 1% SDS and sonicated on ice using a Bioruptor (Diagenode) for 16 cycles (30 s on, 30 s off). The sonicated lysates were pre-cleared by incubation with protein G Dynabeads (Invitrogen) for 1 h at 4 °C. The lysates were incubated overnight at 4 °C with magnetic beads conjugated to antibodies targeting H3K27ac (Abcam ab4729) and H3K4me3 (EMD Millipore 07-473). Following incubation, immunoprecipitates were washed four times to remove non-specific binding. The bound complexes were eluted in elution buffer (50 mM Tris-HCl, pH 8.0; 10 mM EDTA; and 1% SDS) and subjected to reverse cross-linking. This was achieved by incubation with pronase at 42 °C for 2 h followed by incubation at 68 °C for 6 h. After reverse cross-linking, the DNA was purified using phenol-chloroform-isoamyl alcohol (Thermo Fisher Scientific) and precipitated with ethanol. For ChIP-qPCR analysis, purified DNA was used in PCR reactions with SYBR Green PCR Master Mix (Applied Biosystems). Primer sequences for the specific genomic regions analyzed are provided in Supplementary table 2. The enrichment of the target genomic regions was calculated by comparing the relative amounts of DNA in the immunoprecipitates with the input DNA, and the data were normalized to the respective control input values. ChIP-Seq ChIP and input DNA were amplified using the GenomePlex Single Cell Whole Genome Amplification Kit (Sigma-Aldrich). Following digestion with BpmI (New England Biolabs), 30 ng of ChIP DNA was used for library preparation using the NEBNext ChIP-Seq Library Prep Master Mix Set for Illumina (New England Biolabs) according to the manufacturer’s instructions. The resulting libraries were sequenced using an Illumina HiSeq 2500 system. ChIP-Seq data analysis For preprocessing, the quality of next-generation raw data was verified using FastQC (version 0.11.8; ref. 58). ChIP-Seq data were trimmed to 10 bp from both ends to remove adapter sequences using the Trim Galore (version 0.6.3, https://www.bioinformatics.babraham.ac.uk/projects/trim_galore). Trimmed reads were mapped to the human reference genome hg19 (hs37d5) using BWA-MEM version 0.7.17-r118859. Mappings with mapping qualities greater than 10 were retained and sorted using SAMtools (version 1.9; samtools.sourceforge.net), and duplicates were removed using the Picard toolkit (MarkDuplicates version 2.20.3, http://broadinstitute.github.io/picard/). The peaks of H3K27ac (narrow) and H3K4me3 (narrow) were called using MACS2 (version 2.1.2; settings: -f BAMPE -q0.01 -g hs -B --trackline --SPMR; macs3-project.github.io/MACS/). Regions of peaks within the ENCODE Blacklist were excluded60. Peaks were annotated using ChIPseeker (version 1.20.061;) and visualized using ngs.plot.r (version 2.61; github.com/shenlab-sinai/ngsplot). Promoters were regions with H3K27ac + /H3K4me3+ peaks and within ±2 kb of any transcription start site (TSS) in refTSS (version 3.162). Enhancers were regions outside ±2 kb of all TSSs and H3K27ac + /H3K4me3+ peaks. BEDOPS (github.com/bedops/bedops) and BEDTools (github.com/arq5x/bedtools2) were used to process BED files. GREAT (github.com/jokergoo/rGREAT) was used to identify the target genes of promoters and enhancers (the single nearest gene within 1000 kb). To quantify changes in histone modifications (e.g., H3K27ac and H3K4me3), we identified significant peak regions in the parental (P) and resistant (R) cells using MACS2 (*q* <0.01). BEDTools intersect was then applied to extract peaks shared between P and R ( ≥ 100 bp overlap), which served as candidate promoter and enhancer regions, respectively. Read counts within these regions were quantified using HTSeq-count (version 0.11.263;) and normalized to counts per million (CPM) for cross-sample comparisons. Customized R scripts and Adobe Illustrator were used to generate better visualization. Detection of cholesterol levels in cells A total of 100,000 cells were counted, lysed, and quantified using the Amplex Red Cholesterol Kit (Beyotime Biotechnology, cat# S0211S) following the manufacturer’s instructions. Animal experiments Animal studies were conducted in accordance with protocols approved by the Institutional Animal Care and Use Committee of Sun Yat-sen University. Female BALB/c nude mice and NOD/SCID mice (5–6 weeks old) were obtained from the Beijing Vital River Laboratory Animal Technology Co. All experimental mice were maintained in a specific-pathogen-free (SPF) animal centre with regulated environmental conditions. The ambient temperature was kept at 22 ± 2°C with a relative humidity of 55 ± 10%. An automated 12-h light/12-h dark cycle was implemented, and the mice were provided with free access to food and water throughout the study. Only female mice were used for xenograft in this study based on our experiences and previous publications. Sex was not considered in study design, and no sex-based analyses was performed. The tumor volume was measured using a Vernier caliper and calculated using the following formula: tumor volume = width² × length × 0.537. Mice with similar tumor burdens were randomly assigned to different treatment groups. For the LS174T-R tumor xenograft model, 1.5 × 106 cells were subcutaneously injected into the right flank of the BALB/c nude mice. For the patient-derived xenograft (PDX) models, PDX-CRC15 tumor masses were implanted into 40 NOD/SCID mice. Once the tumors reached ~100 mm³, the mice were randomly assigned to one of four treatment groups: (a) vehicle, (b) BI2493, (c) rosuvastatin calcium, or (d) a combination of BI2493 and rosuvastatin calcium. For the SW837 and SW837-R tumor xenograft model, 3 × 106 cells were subcutaneously injected into the right flank of the BALB/c nude mice. When the tumors reached ~100 mm³, the mice were randomly assigned to one of four treatment groups: (a) vehicle, (b) sotorasib, (c) rosuvastatin calcium, or (d) a combination of sotorasib and rosuvastatin calcium. For the patient-derived xenograft (PDX) models, PDX-CRC16 tumor masses were implanted into 40 NOD/SCID mice. Once the tumors reached ~100 mm³, the mice were randomly assigned to one of four treatment groups: (a) vehicle, (b) sotorasib, (c) rosuvastatin calcium, or (d) a combination of sotorasib and rosuvastatin calcium. All the drugs were prepared in distilled H₂O containing 2% DMSO, 30% PEG 300, and 5% Tween-80. The drug dosages were administered as follows: BI2865 at 90 mg/kg daily by oral gavage (i.g.), sotorasib at 50–100 mg/kg daily by oral gavage (i.g.), and rosuvastatin calcium at 50 mg/kg daily by oral gavage (i.g.). PBS (1×) was used as vehicle control. Tumor volumes and body weights were monitored 2–3 times per week until the tumors reached 1000–1500 mm³. For survival curve analysis, treatment was prolonged until the tumor reached a total volume of 1000 mm3. The mice were euthanized by CO₂ inhalation and the tumors were harvested for further analysis. Immunohistochemistry (IHC) The slides were deparaffinized and rehydrated through a series of graded alcohols before being subjected to antigen retrieval was performed using a solution of 10 mM sodium citrate and 0.05% Tween-20 (pH 6.0) for 25 min. Endogenous peroxidase activity was inhibited by incubating slides with 0.3% hydrogen peroxide for 10 min at room temperature. The slides were then incubated with the primary antibody overnight at 4 °C. After incubation, the sections were stained with diaminobenzidine (DAB) for visualization. The antibodies used in this study are listed in the Supplementary table 1. Immunoreactivity was evaluated using a semi-quantitative scoring system64. Each slide was independently assessed by two trained researchers. Staining intensity was graded on a scale of 1 to 4 (1 = weak, 2 = moderate, 3 = strong, 4 = very strong). The percentage of positive tumor cells was estimated in 25% increments (1 = 25%, 2 = 50%, 3 = 75%, 4 = 100%). The final score for each field was calculated by multiplying the intensity score by the proportion score. The overall score for each case was defined as the average of multiple high-power fields from representative tumor regions. Immunofluorescence (IF) For immunofluorescence (IF) staining of cell lines, cells were fixed with paraformaldehyde at 4 °C for 1 h. After fixation, the cells were blocked with 1% bovine serum albumin (BSA) for 1 h at room temperature to prevent non-specific binding. The cells were incubated overnight with the primary antibody at 4 °C. Following this, the secondary antibodies were applied for 1 h at room temperature. Finally, the cells were stained with DAPI (MIKX, CE378) for 10 min to counterstain nuclei. For immunofluorescence of patient-derived organoids (PDOs), a previously published protocol was followed65. Briefly, organoids were recovered from Matrigel using an ice-cold cell recovery solution (Corning) and fixed with 4% paraformaldehyde (PFA) at 4 °C for 45 min. After fixation, the organoids were blocked and incubated sequentially with primary and secondary antibodies for immunolabeling. To facilitate imaging, a fructose–glycerol clearing solution was applied to the organoids. Imaging was performed using a Nikon AX/AX R with an NSPARC confocal microscope. The antibodies used in this study are listed in the Supplementary table 1. RNA interference Cells were plated in a 6-well plate prior to transfection with the specified siRNAs. Transfection was carried out using Lipofectamine RNAi-MAX (Thermo Fisher Scientific), according to the manufacturer’s protocol. The siRNAs, including the non-targeting control (siNC), were purchased from RiboBio (Guangzhou RiboBio, CO). The sequences of all siRNAs are listed in Supplemental Table 2. ERK-KTR stable cell line generation The pLentiCMV Puro DEST ERKKTRClover (plasmid #59150) plasmid was purchased from Addgene. com66. Viruses were generated by cotransfection of the packaging vector psPAX2, envelope vector pMD2.G, and target plasmids into 293 T cells using Lipofectamine 2000 (Thermo Fisher Scientific) according to the manufacturer’s instructions. After 48 h of transfection, virus-containing medium was harvested through a 0.45μm nylon filter and added to DLD1 cells, which were sorted with puromycin for further analysis. DLD1 cells stably expressing ERK-KTR-GFP were imaged using a Zeiss LSM880 inverted confocal microscope after treatment with DMSO or BI-2865. Statistics and reproducibility All experiments were conducted with a minimum of three biological replicates, and the data are expressed as the mean ± standard deviation (SD). Statistical analyses were performed using GraphPad Prism version 8.0, with the exception of the Fisher’s exact test, which was performed using SPSS version 24.0. Detailed data processing, sample size and statistical methods for each result were shown in the corresponding figure legends. All p values were two-sided and p value ≤ 0.05 was considered as statistically significant. No statistical methods were used to predetermine sample size. For cell culture experiments, blinding was not possible. For animal experiments, mice were randomized into different groups prior to treatment initiation and investigators who performed drug treatment were blinded to the mice group allocation. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information Reporting Summary Transparent Peer Review file Source data Source Data Acknowledgements The authors thank all the patients who donated samples for this study. Author contributions Y.G., Y.Z., P.H., and Y.C. contributed equally to this study. Y.Guo and J.T. conceived, designed, and supervised this study. P.H. and Z.Z. contributed to the establishment of CRC organoids. Y.G., Y.C., and X.X. contributed to the acquisition and analysis of clinical data. Most in vitro and in vivo experiments have been performed by Y.G. and W.G. Y.G., Y.Z., P.G., J.C., and C.H. analyzed and interpreted the data (e.g., statistical analysis, biostatistics, and computational analysis). F.Z., X.Z., R.X., J.G., H.L., C.H., L.L., and Q.X. provided material and technical support and helped with parts of the involved experiments. Z.Y., P.D., Z.C., J.H.H, J.Y.C., W.W., X.Xia, D.D., C.K.O. and Q.Y. supervised the study and contributed to manuscript writing. J.C, Y.Guo, Y.Gao, X.W. and J.T. wrote and revised the manuscript. All authors reviewed the manuscript and provided consent to publish this study. Peer review Peer review information Nature Communications thanks Youjun Li, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available. Funding This study was funded by the National Key Research and Development Program of China (No.2022YFA1304000), National Natural Science Foundation of China (82320108015, 82170188, 82470143, 8257115431 and 8250114178), Guangzhou Science and Technology Program (2023B01J1004), Guangdong Provincial Key Areas R&D Program (2023B1111040003) and National Key Clinical Discipline. Data availability The raw RNA-seq and ChIP-seq data of the cell lines reported in this paper have been deposited in the Gene Expression Omnibus (GEO) under accession number GSE306909. The whole-exome sequencing (WES) data have been deposited in the NCBI Sequence Read Archive (SRA) under BioProject accession PRJNA1455728. The CRC genomic data are part of the Cancer Omics Atlas of CHinese cohort-Colorectal Cancer (COACH-CRC) project (https://www.icgc-argo.org/page/114/cgcc), a study led by co-corresponding author Prof. Xiaojian Wu under the framework of the International Cancer Genome Consortium Accelerating Research in Genomic Oncology (ICGC-ARGO). These data have been deposited in the Genome Variation Map (GVM) at the National Genomic Data Center (NGDC), China National Center for Bioinformation (accession no. GVM000630). Due to patient privacy regulations and legal restrictions on the sharing of human genomic data, the COACH-CRC cohort data are available under restricted access. Access will be granted to qualified researchers affiliated with recognized academic or non-profit institutions, for non-commercial research purposes and subject to approval of a data access agreement and, where applicable, ethics committee or institutional review board (IRB) approval. Requests for access should be directed to the corresponding author, Dr. Xiaojian Wu (wuxjian@mail.sysu.edu.cn), together with an appropriate data access agreement (Supplementary Note 1). We will respond to all data access requests within 15 working days. Once approved, access to the data will be provided via GVM for at least 12 months from the date of publication, subject to continued operation of the repository and compliance with the agreed terms of use. Additionally, the CRC cell line mutation publicly available data used in this study are available in the Broad Institute Cancer Cell Line Encyclopedia (CCLE) (https://depmap.org/portal/ccle/)33 and Catalogue of Somatic Mutations in Cancer (COSMIC) (https://cancer.sanger.ac.uk/cosmic/)32. The remaining data are available within the Article, Supplementary Information or Source Data file. Source data are provided with this paper. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. These authors contributed equally: Yaoyu Guo, Yi Zhong, Peishan Hu, Yufeng Chen. Contributor Information Yijun Gao, Email: gaoyj@sysucc.org.cn. Xiaojian Wu, Email: wuxjian@mail.sysu.edu.cn. Jing Tan, Email: tanjing@sysucc.org.cn. Supplementary information The online version contains supplementary material available at 10.1038/s41467-026-73805-7. References Associated Data Supplementary Materials Supplementary Information Reporting Summary Transparent Peer Review file Source Data Data Availability Statement The raw RNA-seq and ChIP-seq data of the cell lines reported in this paper have been deposited in the Gene Expression Omnibus (GEO) under accession number GSE306909. The whole-exome sequencing (WES) data have been deposited in the NCBI Sequence Read Archive (SRA) under BioProject accession PRJNA1455728. The CRC genomic data are part of the Cancer Omics Atlas of CHinese cohort-Colorectal Cancer (COACH-CRC) project (https://www.icgc-argo.org/page/114/cgcc), a study led by co-corresponding author Prof. Xiaojian Wu under the framework of the International Cancer Genome Consortium Accelerating Research in Genomic Oncology (ICGC-ARGO). These data have been deposited in the Genome Variation Map (GVM) at the National Genomic Data Center (NGDC), China National Center for Bioinformation (accession no. GVM000630). Due to patient privacy regulations and legal restrictions on the sharing of human genomic data, the COACH-CRC cohort data are available under restricted access. Access will be granted to qualified researchers affiliated with recognized academic or non-profit institutions, for non-commercial research purposes and subject to approval of a data access agreement and, where applicable, ethics committee or institutional review board (IRB) approval. Requests for access should be directed to the corresponding author, Dr. Xiaojian Wu (wuxjian@mail.sysu.edu.cn), together with an appropriate data access agreement (Supplementary Note 1). We will respond to all data access requests within 15 working days. Once approved, access to the data will be provided via GVM for at least 12 months from the date of publication, subject to continued operation of the repository and compliance with the agreed terms of use. Additionally, the CRC cell line mutation publicly available data used in this study are available in the Broad Institute Cancer Cell Line Encyclopedia (CCLE) (https://depmap.org/portal/ccle/)33 and Catalogue of Somatic Mutations in Cancer (COSMIC) (https://cancer.sanger.ac.uk/cosmic/)32. The remaining data are available within the Article, Supplementary Information or Source Data file. Source data are provided with this paper.