Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 May 7.
Published in final edited form as: Gastroenterology. 2025 Jul 7;170(1):70–88. doi: 10.1053/j.gastro.2025.07.003

Chemical perturbations impacting histone acetylation govern colorectal cancer differentiation

Pornlada Likasitwatanakul 1,2,3,4,#, Zhixin Li 1,2,3,#, Paul Doan 1,3,#, Sandor Spisak 1, Akhouri Kishore Raghawan 1,2,3, Qi Liu 2,6, Priscilla Liow 2,6, Sunwoo Lee 2,6, David Chen 1,5, Pratyusha Bala 1,2,3, Pranshu Sahgal 1,2,3, Daulet Aitymbayev 1,3, Jennifer S Thalappillil 1,7, Malvina Papanastasiou 3, William Hawkins 3, Steven A Carr 3, Haeseong Park 1,7, James M Cleary 1,7, Jun Qi 2,6,*, Nilay S Sethi 1,2,3,7,*
PMCID: PMC13148549  NIHMSID: NIHMS2158553  PMID: 40633623

Abstract

BACKGROUND and AIMS:

Aberrant epigenetic programs that suppress differentiation and enhance plasticity drive colorectal cancer (CRC), yet the molecular determinants underlying these processes remain elusive. We aimed to identify and characterize epigenetic regulators of CRC differentiation, uncovering mechanisms that reprogram cancer cell states.

METHODS:

A small molecule library targeting epigenetic regulators was screened using an endogenous dual reporter system. We evaluated lead compounds in mouse and human CRC models via histopathology, cellular assays, epigenetic studies, mass-spectrometry–based histone modification profiling, and single cell RNA-sequencing. Integrative analyses of drug-induced chromatin dynamics, gene expression, target engagement, and histone marks elucidated molecular mechanisms. Focused genetic screens were conducted to identify regulators of HDAC1/2-mediated differentiation.

RESULTS:

We found that inhibition of histone deacetylase (HDAC) 1/2 catalytic domain promotes CRC differentiation and suppresses tumor growth. Unbiased profiling of histone modifications identified H3K27ac and H3K9ac as critical regulatory marks, with genome-wide analyses demonstrating their enrichment at HDAC1/2-bound regions associated with open chromatin and upregulated differentiation genes. Disrupting H3K27ac by targeted degradation of acetyltransferase EP300 reversed the differentiation phenotype induced by HDAC1/2 inhibition in a patient-derived CRC organoid. Genetic screens revealed that DAPK3 contributes to H3K27ac-mediated CRC differentiation induced by HDAC1/2 inhibition.

CONCLUSIONS:

Our findings establish histone acetylation as a chemically targetable mechanism governing CRC cell fate and demonstrate that epigenetic reprogramming can be leveraged as a therapeutic strategy. By identifying HDAC1/2 inhibition as a driver of differentiation and revealing H3K27ac as a key regulatory mark, this study provides a framework for targeting chromatin-modifying enzymes to counteract CRC plasticity and improve treatment outcomes.

Keywords: stem cell, intestinal differentiation, colorectal cancer, epigenetic regulation

INTRODUCTION

Colorectal cancer (CRC) ranks third in global incidence and is increasingly diagnosed in younger adults.1. Metastatic disease relies on cytotoxic chemotherapy, yet responses are short-lived, and targeted agents underperform relative to other tumors with the same drivers (e.g., KRASG12C, BRAFV600E)2–6. Evidence shows that CRC progression and poor responses to targeted therapies stem from aberrant cell state plasticity7–9, where cancer cells reversibly shift among different phenotypic states10. Impaired differentiation is a key mechanism by which cancer cells unlocks phenotypic plasticity7, 10, 11. The transcriptional and epigenetic programs that underlie impaired differentiation in CRC remain incompletely understood.

Although cancer is traditionally viewed through the lens of genetic mutations, aberrant epigenetic activity - encompassing DNA methylation, histone modifications, nucleosome positioning, and noncoding RNA changes - also drives tumor initiation and progression10, 12. Targeting these epigenetic regulators has proven clinically effective: HDAC inhibitors, especially those against HDAC1/2 or pan-HDACs, induce differentiation and exhibit powerful anti-leukemic effects13, 14. Five such agents are FDA-approved for hematologic malignancies, either alone or in combination. However, despite strong preclinical efficacy, HDAC inhibitors have thus far underperformed in patients with late-stage CRC15.

Inspired by the success of therapeutics that overcome arrested differentiation in leukemias driven by BCR-ABL and IDH1/216, 17, we seek to uncover how differentiation can be restored in CRC. Our lab has shown that SOX9 promotes CRC by engaging enhancers and reprogramming the epigenetic landscape, enforcing a stem-like transcriptional state that blocks differentiation7, 11. These findings form the basis of a newly engineered reporter system that emits fluorescent signals from the endogenous SOX9 and KRT20 genomic loci of CRC cell lines, broadcasting aberrant stem cell-like and differentiation activity, respectively18. Screening a library of well-annotated small molecules against this dual-reporter system, we set out to find and investigate agents that promote CRC differentiation by modulating epigenetic regulation.

MATERIALS AND METHODS

Inactive MRK60 synthesis

(details in Supplementary Methods)

graphic file with name nihms-2158553-f0007.jpg

Synthesis of Biotin-MRK60

graphic file with name nihms-2158553-f0008.jpg

Synthesis of Biotin-iMRK60

graphic file with name nihms-2158553-f0009.jpg

Endogenous reporter cell lines

Endogenous reporter cell lines were adopted from our prior study18. HT115KRT20-mKate2, and LS180SOX9-GFP, HT29KRT20-GFP, HT29KRT20-GFP single reporters, and HT29SOX9-mKate2/KRT20-GFP dual reporter were used in this study.

Chem-sequencing (Chem-seq)

Chem-seq is a ChIP-based method to identify the sites bound by small chemical molecules throughout the human genome. The protocol is adapted from Anders, et al. 201419. Biotinylated MRK60 and iMRK60 were synthesized for this experiment. Anti-biotin antibody was used to pull down the biotinylated compounds and their complex.

For the in-vivo Chem-seq, either biotinylated MRK60 or biotinylated iMRK60 10 uM were added for 1 or 2 hours in an incubator before fixing the cells. For the in-vitro Chem-seq, the chromatin lysate after the sonication step was added with either biotinylated MRK60 or biotinylated iMRK60 10 uM and was incubated on a 4°C rocker overnight.

Cells were fixed with formaldehyde at final concentration of 1% for 10 minutes on a rocker at room temperature before quenching with 1.25 M Glycine at 10% of the media volume for 5 minutes. Cells were collected with cell scraper and were centrifuged at 600 g for 4 minutes at 4°C. Pellets were washed with cold PBS for 3 times. 1.1 mL of the sonication buffer was added and transfer to shearing tubes. Chromatin was sheared with a Covaris E220 machine with the following settings: Peak Incident Power 150, Duty Cycles 5%, Cycles per Burst 200, Time 300 seconds. Lysates were centrifuged at 13,000 g at 4°C for 5 minutes and the supernatants were collected. 75 μL of Pierce™ High Capacity Streptavidin agarose per reaction was washed with PBS at room temperature 3 times before use. The agarose was mixed with the chromatin lysate 450 μL and incubated at room temperature on a rocker for 15 minutes. Washing, eluting DNA-protein complex, reverse crosslinking, and DNA isolation were done as previously described.

Histone profiling

HT29 cells were treated with DMSO, MRK60 and iMRK60 for 24 or 48 hours. Histones were extracted using a standard protocol20 and assessed for purity by SDS-PAGE. Each 10 μg sample was propionylated, desalted, and digested overnight with trypsin, followed by a second propionylation at the peptide level. Peptides were desalted using C18 Sep-Pak cartridges (Waters), and isotopically labeled H3 and H4 reference peptides were spiked in prior to mass spectrometry. Peptides were separated by C18 chromatography (EASY-nLC 1000, Thermo Scientific) and analyzed by PRM on a Q Exactive™ Plus (Thermo Scientific). Peak areas were extracted and integrated using Skyline. Ratios of endogenous to heavy peptides were log2-transformed and normalized to unmodified regions of H3 or H4. Data were then normalized to DMSO controls. Spearman rank correlation was used to generate similarity matrices across samples. Detailed sample preparation protocols are available at:

https://panoramaweb.org/labkey/wiki/LINCS/Overview%20Information/page.view?name=sops.

MRK60 resistance CRISPR screen – Experiment and Analysis

HT29KRT20-GFP cells were infected with lentivirus at a concentration that would yield a multiplicity of infection at ~0.3. After 3 days of complete selection with puromycin (3 μg/mL), cells were split into two arms: MRK60 (5 μM) and DMSO. After 4 days of drug treatment, cells were washed with PBS and selected by FACS for the top and bottom 10% GFP+ subpopulations to identify gene perturbations that would inhibit the induction of differentiation and KRT20 by MRK60. To identify gene perturbations that would confer survival to MRK60-treated cells, the unsorted populations were collected. The genomic DNA was purified using NucleoSpin Blood kit (Takara, 740951.250) following manufacturer’s protocol.

A two-step PCR protocol was performed to construct the indexed libraries of amplicons encoding sgRNA sequences present in each cell population. PCR #1 consisted of 1 μg of purified genomic DNA, 25 μL of 2X NEBNext UltraII Q5 PCR Master Mix (New England Biolabs, M0544X), 5 μL of DMSO, 1 μM of CRS1F primer and 1 μM of CRS1Rc primer in a total reaction volume of 50 μL. The PCR #1 thermocycler program was 30 seconds at 98°C; 28 cycles of denaturation (30 seconds at 98°C), annealing (30 seconds at 53°C), and extension (30 seconds at 72°C); final extension for 10 minutes at 72 C. The PCR #1 product was purified with AMPure XP beads (1.0X) (Beckman Coulter) and eluted with 10μL of DNase-free water. All PCR #1 products from the same cell population were pooled. To index the samples, PCR #2 consisted of 5 μL of the pooled, purified PCR #1 product, 25 μL of 2X NEBNext UltraII Q5 PCR Master Mix (New England Biolabs, M0544X), 0.6 μM of P5A1–6 primer and 0.6 μM of P7GR4–6 primer in a total reaction volume of 50 μL. The PCR #2 thermocycler program was 30 seconds at 98°C; 10 cycles of denaturation (10 seconds at 98°C), annealing (10 seconds at 62°C), and extension (30 seconds at 72°C); final extension for 2 minutes. The PCR #2 product was purified with AMPure XP beads (1.0X) (Beckman Coulter), and eluted in 50 μL of DNase-free water. The concentration was determined using a Qubit fluorometer (Thermo Fisher), and the library size was confirmed by gel electrophoresis and TapeStation (Agilent). All indexed libraries were pooled in equimolar amounts for next-generation sequencing. The sequences of primers can be found in Supplementary Table 1.

The counts of sgRNA in each subpopulation were detected and quantified by paired-end 150 base-pair sequencing using Illumina NovaSeq X Plus platform. The demultiplexing and quantification of raw read counts were done using a custom Python 3 script. The Aho-Corasick string-search algorithm was used to (1) demultiplex FASTQ files and assign each read to a known cell population by searching for the associated barcode in each read and to (2) count the number of sgRNA in each cell population by searching for the sgRNA sequence in each read without allowing for any mismatches. The raw read counts of sgRNA in each cell population can be found in Supplementary Table 4.

The raw read count data were transformed into log2 counts then normalized by the library read depth. sgRNAs whose counts were 3 standard deviations below the mean in the plasmid library were excluded. The replicate-averaged log2 fold-change (LFC) values were z-score normalized by the distribution of the negative control sgRNAs’ LFC values. sgRNAs whose scores were opposite in sign to the median score of all sgRNAs targeting the same gene were excluded, likely due to technical noise. The gene-level scores were derived by dividing the sum of sgRNA-level scores by the square root of the number of sgRNAs targeting the same gene.

RESULTS

A chemical library screen identifies the HDAC1/2 inhibitor MRK60 as an inducer of CRC differentiation

SOX9 drives developmental reprogramming and blocks differentiation, promoting colorectal cancer (CRC) initiation7, 11. Disrupting SOX9 in human CRC and genetic mouse models leads to tumor regression by inducing differentiation, which is faithfully captured by KRT20 expression21. To track these processes in real-time, we developed a dual-reporter system by genome-editing CRC cell lines to insert fluorescent markers: mKate2 at the SOX9 locus (for aberrant stem-like activity) and GFP at the KRT20 locus (for differentiation)18 (Figure 1A and Supplementary Figure 1A). Recognizing the pivotal role of epigenetic regulation in intestinal stem cell function - and its disruption in CRC - we conducted a focused drug screen of 31 well-characterized small molecule inhibitors and degraders targeting key epigenetic regulators, using two CRC reporter cell lines. The library includes synthetic compounds that inhibit or degrade key epigenetic regulators, including ATP-dependent chromatin remodelers (e.g., mSWI/SNF subunits) and covalent modifiers (e.g., HDACs) (Supplementary Table 1). In the initial screen, compounds were tested at 10 μM, identifying 11 hits that reduced cell viability, suppressed stem cell activity, and promoted differentiation, based on performance relative to DMSO controls (Figure 1B and Supplementary Figure 1B). We then performed a secondary screen to generate dose-response curves for these 11 compounds. Six emerged as promising, including two SMARCA2/4 inhibitors and three HDAC inhibitors (Figure 1C).

Figure 1. Identification of HDAC1/2 inhibitor MRK60 as a CRC differentiating agent.

Figure 1.

Open in a new tab

A. Schematic of endogenous reporter knock-in at SOX9 and KRT20 genomic loci. E = exon.

B. Schematic of primary drug screen; 3 phenotypic readouts of differentiation (HT115KRT20-mKate2), stem cell (LS180SOX9-GFP), and viability across 3 cell lines. Venn diagram shows hits (> DMSO)

C. CellTiter-Glo® viability assay showing dose-response curves of LS180 and HT29 CRC cells treated with 8 doses of 11 drugs at 48 hours

D. Representive GFP images of HT29KRT20-GFP cells treated with DMSO or indicated concentrations of MRK60 (top); Immunoblot of KRT20, SOX9, and GAPDH in LS180 cells treated with DMSO or indicated concentrations of MRK60 (bottom)

E. Normalized mRNA expression of KRT20 and SOX9 in MRK60-treated LS180 cells by RT-qPCR.

F. AUC scores across different drugs in 50 human CRC cell lines in the CTRPv2 database.

G. HDAC1 and HDAC2 mRNA expression and ISC gene signature37 in normal and CRC tissues from a large cohort of CRC patients (TCGA).

H. HDAC1 and HDAC2 mRNA expression in a patient sample of CRC, comparing tumor and adjacent normal regions, using spatial transcriptomics (Visium HD platform)66.

We next assessed whether the top six compounds could induce CRC differentiation in a dose-dependent manner by measuring KRT20 mRNA and protein expression (Supplementary Figure 1C). HDAC1/2/3 inhibitors—particularly the HDAC1/2-selective inhibitor MRK6022—consistently induced KRT20 expression in a dose-dependent fashion across CRC cell lines (Figures 1D-E and Supplementary Figure 1C-F). Analysis of drug sensitivity data from the CTRPv2 database23 showed that HDAC1/2 inhibitors had stronger anti-tumor effects than EZH2 or DOT1L inhibitors across 62 CRC cell lines (Figure 1F; SMARCA2/4 or dBET6 inhibitors not tested in CTRPv2). Furthermore, a pan-cancer analysis of a SMARCA2/4 degrader showed limited efficacy in CRC24.

To assess the clinical relevance of our findings, we analyzed HDAC1 and HDAC2 expression in normal and cancer tissue from TCGA and Orouji et al.25 bulk RNA-seq datasets. Both HDAC1 and HDAC2 were significantly overexpressed in CRC tumors compared to normal tissue and positively associated with a stem cell signature (Figure 1G and Supplementary Figure 1G). Spatial transcriptomics of a CRC patient sample further confirmed strong, uniform HDAC2 and moderate HDAC1 expression in tumor tissue relative to adjacent normal regions (Figure 1G). Tumors with high HDAC1 or HDAC2 expression exhibited stronger stem cell-like programs (Figure 1H and Supplementary Figure 1H). Interestingly, higher HDAC2 expression was associated with increased aneuploidy in TCGA samples (Supplementary Figure 1I; P = 8.3 × 10−4). These results support a model whereby elevated HDAC1/2 expression in CRC tumors is associated with stem-cell like features, consistent with a therapeutic potential of targeting these enzymes to promote differentiation.

HDAC1/2 inhibition induces differentiation and impairs tumor growth across CRC models

To test whether MRK60-induced differentiation relies on HDAC1/2 enzymatic inhibition, we synthesized an inactive analog, iMRK60, by modifying its zinc-binding amine to a non-functional amide group22 (Figure 2A). In CRC cell lines, MRK60 increased KRT20 expression, indicating differentiation, while iMRK60 had no effect (Figures 2B and Supplementary Figure 2A-B). Consistent with this, iMRK60 did not affect CRC cell proliferation, as shown by bioluminescent assays (Figures 2C and Supplementary Figure 2C-D).

Figure 2. MRK60 induces differentiation and impairs tumor growth across CRC models.

Figure 2.

Open in a new tab

A. Structures of MRK60 and its inactive form iMRK60

B. Immunoblots of KRT20, SOX9, and GAPDH (loading control) in LS180 treated with DMSO, iMRK60 (2 μM), or MRK60 (2 μM) at indicated days

C. CellTiter-Glo® viability assay of LS180 treated with iMRK60 (left) or MRK60 (right) relative to DMSO at indicated days.

D. CellTiter-Glo® viability assay (left) and representative phase contrast images (right) of MRK60 (2 μM), iMRK60 (2 μM), or DMSO treated Cdx2CreERT2; Apcf/f colon adenoma organoids for 4 days.

E. Immunoblot of KRT20, SOX9, and GAPDH in Cdx2CreERT2; Apcf/f colon adenoma organoids treated with indicated doses of MRK60.

F. CellTiter-Glo® viability assay of patient-derived adenoma organoid from a patient with FAP treated with iMRK60 or MRK60 for 72 hours.

G. H&E and KRT20 immunohistochemistry of FAP adenoma treated with iMRK60 or MRK60 for 5 days. Percentage of differentiated organoids (light microscopy) are displayed below images.

H. RT-qPCR of differentiation markers (KRT20, DPP4) and stem cell markers (LGR5, ASCL2) in FAP adenoma treated with iMRK60 or MRK60 for 72 hours. ACTB was used for normalization.

I. Schematic of Apcf/f; KrasG12D mouse colon organoid in vivo xenograft experiment. Tumor xenograft volume growth curve and waterfall plot of individual tumor responses.

J. ZIP synergy score heatmap of HT29 treated with MRK60 and 5-FU at indicated doses for 5 days.

We tested MRK60 in three neoplastic colon organoid models, which capture the spectrum of cell states and are valuable for drug testing26. In mouse adenoma organoids (derived from tamoxifen-induced Cdx2CreERT2; Apcf/f; R26tdT genetically engineered mouse colons), MRK60—but not its inactive analog iMRK60—reduced proliferation, increased Krt20, and decreased the stem cell marker Lgr5 (Figure 2D-E and Supplementary Figure 2F). Similarly, in human adenoma organoids from a patient with familial adenomatous polyposis (FAP), MRK60 reduced viability, elevated KRT20, and suppressed LGR5, consistent with differentiation; these effects were absent with iMRK60 (Figure 2F-H). In patient-derived CRC organoids, MRK60 also reduced viability by 50%, though the effect was less pronounced than the 75% reduction observed in FAP adenomas (Supplementary Figure 2G).

To assess MRK60’s anti-tumor activity in vivo, we treated nude mice bearing ApcKO; KrasG12D colon organoids in their flanks. Mice were randomized to receive vehicle or daily intraperitoneal MRK60 beginning 21 days after implantation. After 10 days, MRK60-treated tumors showed shrinkage or stabilization, while control tumors doubled in size (Figure 2I). In Lgr5Cre;ApcF/F;R26tdT mouse model of intestinal polyposis, MRK60 treatment induced Krt20 expression, with a modest increase in protein levels by immunohistochemistry (IHC) (Supplementary Figure 2H).

We next tested whether MRK60 could enhance the effects of existing CRC therapies. Two CRC cell lines were treated with MRK60 in combination with three chemotherapies (5-FU, oxaliplatin, SN38) and two targeted agents (Regorafenib and TAS-102). Of these, only 5-FU synergized with MRK60 (Figure 2J and Supplementary Figure 2I). Synergy, quantified by ZIP scoring, was observed at 1 μM 5-FU—matching the dose at which the MRK60 + 5-FU combination maximally increased KRT20 expression in our reporter system (Figure 2J). In patient-derived CRC organoids, the combination showed an additive anti-tumor effect, with increased KRT20 and decreased LGR5 expression (Supplementary Figure 2K-M).

MRK60 induces differentiation via on-target HDAC1/2 inhibition

To confirm that MRK60 promotes differentiation specifically by on-target HDAC1/2 inhibition, we performed biochemical, genetic, and chemical assays (Figure 3A). Using unbiased proteomics, we evaluated MRK60’s direct protein interactions. Biotinylated MRK60 and iMRK60 (see Methods) were separately incubated with CRC lysates, followed by streptavidin pulldowns and mass spectrometry (Figure 3B). MRK60 preferentially bound HDAC1 and HDAC2 compared to iMRK60 in two CRC cell lines (Figure 3B-C and Supplementary Figure 3A-B). Streptavidin pulldowns followed by immunoblot confirmed a specific interaction between active MRK60 and HDAC1 (Figure 3D). MRK60 also bound the kinase activator GCN1, but CRISPR knockout indicated that GCN1 was not required for MRK60-induced differentiation (Figures 3B and Supplementary Figure 3B-C). Individual knockout of HDAC1 or HDAC2 modestly increased differentiation (measured by KRT20), but this effect was less robust than MRK60’s dual inhibition (Supplementary Figure 3D). Attempts at simultaneous HDAC1/2 genetic disruption (tandem sgRNAs or shRNAs in same vector, or sequential approaches with individual vectors) failed, likely due to synthetic lethality27. Additionally, genetic knockouts typically do not replicate the complete effects of chemical inhibition, as HDAC1/2 have known non-enzymatic roles beyond their deacetylase activity28.

Figure 3. MRK60-induced differentiation is mediated by on-target HDAC1/2 inhibition.

Figure 3.

Open in a new tab

A. Schematic of strategies to validate on-target effect of MRK60 on HDAC1/2 inhibition.

B. Schematic of MRK60 IP followed by mass spectometry workflow (left); density plots of unique peptides bound to biotinylated MRK60 (y-axis) or iMRK60 (x-axis) in HT29 cells (right).

C. Venn diagram showing proteins detected by co-IP mass spectometry in HT115 and HT29.

D. Immunoblots of HDAC1 and GAPDH in HT115 and HT29 following co-IP with biotin control, iMRK60, biotinylated iMRK60, MRK60, or biotinylated MRK60.

E. Heatmaps of iMRK60 and MRK60 binding at genomic loci co-bound by HDAC1/2 using Chem-seq protocol in live CRC cells.

F. A table displaying ability of 13 inhibitors with indicated HDAC inhibitory profiles and specificity to induce differentiation in CRC cell lines upon treatment.

G. Immunoblots of KRT20, SOX9, and GAPDH in LS180 treated with indicated HDAC inhibitors.

To determine where the MRK60–HDAC1/2 drug-protein complex interacts with DNA, we employed Chem-seq19, a method designed to identify genomic regions bound by chemical probes interacting with DNA-binding proteins. CRC cells were treated separately with biotin-labeled MRK60 or iMRK60, followed by streptavidin pulldown and DNA sequencing. MRK60, but not iMRK60, showed specific association with HDAC1/2-bound genomic regions, and this binding intensified from 6 to 20 hours (Figure 3E and Supplementary Figure 3E).

Finally, we chemically evaluated whether inhibition of specific HDAC family members induces CRC differentiation. We treated CRC cells with a panel of HDAC inhibitors, including six class I (HDAC1,2,3,8), two class II (HDAC4–7,9,10), and four pan-HDAC inhibitors (Figure 3F). Differentiation, assessed by KRT20 induction, strongly correlated with class I HDAC inhibition, particularly targeting HDAC1, HDAC2, or HDAC3 (Figures 3F-G and Supplementary Figure 3F). Among these, MRK60, Romidepsin29, MS-27530, Cl-99431, and SR-437032 most effectively induced differentiation, whereas HDAC3 inhibitors LW333 and RGFP-96634 exhibited varying abilities to induce KRT20 (Figures 3F-G and Supplementary Figure 3F). Class II and pan-HDAC inhibitors similarly did not reliably induce KRT20 (Figures 3F-G and Supplementary Figure 3F). Additionally, the CoREST complex degrader UM-17135 did not trigger differentiation, suggesting that HDAC co-repressor complexes are less critical for this phenotype (Supplementary Figure 3G).

Integrative analyses of HDAC1/2 inhibition on chromatin accessibility and gene regulation

To uncover how HDAC1/2 inhibition promotes CRC differentiation, we analyzed gene expression changes in CRC cells treated with MRK60 using RNA-seq. We identified 6,606 differentially expressed genes, evenly divided between up- and downregulated genes (Supplementary Figure 4A), consistent with prior studies36. Gene-set enrichment analysis (GSEA) revealed robust induction of colon differentiation gene signatures37 following MRK60 treatment (Figure 4A), accompanied by suppression of intestinal stem cell signatures (Supplementary Figure 4B-D). Notably, known HDAC-related programs such as apoptosis or autophagy were not significantly activated (Supplementary Figure 4E).

Figure 4. Gains in H3K27ac mediated by HDAC1/2 inhibition are associated with CRC differentiation.

Figure 4.

Open in a new tab

A. Gene set enrichment analysis (GSEA) of differentiated enterocyte signature41 in RNA-seq profiles of MRK60 (2 μM) and DMSO treated HT115 cells; Normalized enrichment score (NES) and false discovery rate (FDR) (left). Heatmap of enriched differentiation genes (right).

B. GSEA of differentiated enterocyte signature41 in ATAC-seq profiles of MRK60 (2 μM) and DMSO treated HT115 cells; peaks around TSS (±2kb) were annotated to nearest gene (left). Heatmap of accessibility peaks near enriched differentiation genes across drug treatments (right).

C. GSEA dot plot showing enrichment of gene sets associated with differential chromatin accessibility peaks and HDAC1/2 binding in RNA-seq of MRK60 (2 μM) vs. DMSO in HT115 cells. Dot size indicates enriched gene proportion in gene set. Adjusted P = .01 (dashed red line).

D. Bar plots of the number of differential accessbility peaks induced by MRK60 (2 μM); HDAC1/2 bound subset of differential accessibility peaks are highlighted in red.

E. Heatmap of gained (red) and lost (blue) histone PTMs induced by MRK60 (2 μM) in HT29 cells by mass spectometry; H3K9ac (green),H3K27ac (red), and H4K8ac (blue).

F. Histone-enriched immunoblots of H3K9ac, H3K27ac, H4K8ac, and histone H3 in MRK60 (2μM)-treated HT115 cells.

G. Bar plots of differential peaks of H3K27ac, H3K9ac and H4K8ac induced by MRK60 (2μM) in HT115 cells; HDAC1/2-bound subset (red).

H. (Left) GSEA dot plot showing enrichment of gene sets associated with differential peaks of H3K9Ac, H3K27Ac, or H4K8Ac and HDAC1/2 binding in RNA-seq of MRK60 (2μM) vs. DMSO in HT115 cells. Dot size indicates enriched gene proportion in gene set. Adjusted P = .01 (dashed red line). (Right) GSEA plot showing enrichment of gene sets of differential peaks of H3K27ac (top) or H3K9ac (bottom), and HDAC1/2 binding in RNA-seq of MRK60 (2μM) vs. DMSO in HT115

I. Schematic summarizing the association between HDAC1/2 inhibition and increase in chromatin accessibility, deposition of H3K27ac and H3K9ac, and differentiation gene expression.

J. Multi-omics scatter plot displaying global changes in chromatin accessibility (ATAC-seq, y-axis) and gene expression (RNA-seq, x-axis) in HT115 cells treated with MRK60 (2 μM) compared to DMSO. Enterocyte differentiation genes (blue) with contour density map to indicate distribution.

K. Density plot of distribution of histone mark signals (H3K27ac, H3K9ac, and H4K8ac) at enterocyte differentiation genes. Log2 fold changes between MRK60 (2 μM) and DMSO in HT115 cells. Areas under curves are color-coded by histone mark.

L. Integrative Genomic Viewer (IGV) tracks for CUT&RUN of HDAC1, HDAC2, H3K27ac, H3K9ac, H4K8ac, and RNA-seq at KRT20 and DPP4 genomic loci in HT115 cells +/− MRK60.

M. Boxplot of H3K27ac ChIP-seq signals at genomic loci co-bound by HDAC1/2 and MRK60 in normal and CRC tumors from patients25

Histone acetylation influences chromatin accessibility, prompting us to examine how HDAC1/2 inhibition alters the chromatin landscape using ATAC-seq. MRK60 treatment resulted in 15,677 regions of gained accessibility and 16,822 regions of decreased accessibility (Supplementary Figure 4F), mirroring the balanced transcriptional changes observed by RNA-seq. Regions that gained accessibility were enriched for genes associated with differentiation (Figure 4B). Although globally accessible regions correlated more with gene upregulation than closed regions did with downregulation, neither trend reached statistical significance (Figure 4C), possibly due to the broader and more complex nature of chromatin remodeling relative to gene expression. To refine these associations, we performed HDAC1/2 CUT&RUN. Surprisingly, 72.5% of regions with reduced accessibility overlapped HDAC1/2 binding sites, compared to only 21.1% of regions that gained accessibility (Figure 4D) - a pattern contrary to the expectation that HDAC inhibition generally opens chromatin. However, the subset of 3,308 HDAC1/2-bound regions with increased accessibility was significantly associated with gene upregulation by GSEA, while the 12,204 HDAC1/2-bound regions that closed did not correlate with transcriptional repression (Figure 4C and Supplementary Figure 5A-C). Furthermore, MRK60-bound regions identified by Chem-seq overlapped extensively with HDAC1/2-bound loci from CUT&RUN, confirming that MRK60 engages HDAC1/2 at chromatin without displacing them (Supplementary Figure 3E). Together, these results suggest that MRK60 induces differentiation by directly targeting HDAC1/2 to open chromatin at specific regulatory sites, leading to transcriptional activation.

Chemical inhibition of HDAC1/2 leads to gains in acetylation of specific histones.

HDACs regulate gene expression by removing acetyl groups from lysine residues on histone and non-histone proteins. To link MRK60-induced transcriptional and chromatin changes to HDAC1/2’s direct enzymatic function, we profiled histone posttranslational modifications (PTMs) by mass spectrometry38. Following histone-specific immunoprecipitation, we analyzed 79 PTMs - including acetylation, methylation, and ubiquitination - across two CRC cell lines treated with MRK60 or iMRK60 (Figure 4E). As expected, MRK60 increased acetylation at multiple H3 and H4 lysine residues, with the most prominent gains at H3K27ac, H3K9ac, and H4K8ac - highlighting them as candidate marks linked to differentiation. (Figure 4E). We validated these findings via histone-enriched immunoblots, which confirmed a dose-dependent increase in these acetylation marks upon MRK60 treatment (Figures 4F and Supplementary Figure 5D).

Acetylation of H3K27 mediated by HDAC1/2 inhibition is associated with CRC differentiation.

To assess locus-specific changes in histone acetylation and their relationship to chromatin accessibility and gene expression, we profiled H3K27ac, H3K9ac, and H4K8ac in CRC cells treated with MRK60 using CUT&RUN39. Quality control confirmed successful enrichment (fingerprint analysis40), strong signal over background, and high reproducibility across replicates (Supplementary Figure 5E), and replicates were consistent (Supplementary Figure 5F). Regions with MRK60-induced changes in histone acetylation largely overlapped with HDAC1/2 binding sites and were distributed across active gene bodies (48.3%), distal intergenic regions (30.8%), and promoters (28.0%) (Supplementary Figure 5G). Most changes in H3K27ac, H3K9ac, and H4K8ac occurred at HDAC1/2-bound loci (Figure 4G), suggesting a direct link between HDAC1/2 activity and these chromatin marks. This pattern was recapitulated in cells treated with Romidepsin, a second HDAC1/2 inhibitor (Supplementary Figure 5H). While gains and losses in H3K27ac and H3K9ac were balanced, H4K8ac was unexpectedly reduced across most regions following HDAC1/2 inhibition with either compound (Figures 4G and Supplementary Figure 5H-J), despite its increased global abundance (Figure 4E-F). Immunofluorescence confirmed that these shifts were not due to changes in histone localization (Supplementary Figure 5I).

We next examined how histone acetylation changes relate to chromatin accessibility. HDAC1/2-bound regions that became more accessible following inhibition showed increased H3K27ac and H3K9ac, whereas regions that lost accessibility did not display corresponding acetylation changes. In contrast, H4K8ac levels were significantly reduced at both gained and lost accessible sites, suggesting it does not correlate with chromatin accessibility (Supplementary Figure 5K).

We then integrated histone acetylation, chromatin accessibility, and gene expression data to construct a comprehensive molecular portrait of HDAC1/2 inhibition. HDAC1/2-bound regions with increased H3K27ac and H3K9ac were strongly associated with transcriptional upregulation (Figure 4H). Of these, regions gaining H3K27ac—more than H3K9ac or H4K8ac—showed the strongest correlation with increased chromatin accessibility and activation of differentiation-associated genes (Figure 4J-L). These findings suggest that HDAC1/2 inhibition drives H3K27ac accumulation at key regulatory elements, promoting chromatin opening and transcription of differentiation programs. In contrast, regions that lost H3K27ac - often with modest decreases in H3K9ac and H4K8ac - were linked to chromatin closing and repression of stemness-related genes (Supplementary Figure 5L-M).

To further connect H3K27ac with CRC differentiation, we treated HT29 cells with selective inhibitors of HDAC1/2 and HDAC3. Only HDAC1/2 inhibitors—MRK60 and Romidepsin—induced both KRT20 and H3K27ac, while iMRK60 and two HDAC3 inhibitors had no effect (Supplementary Figure 5N). These results confirm that selective HDAC1/2 inhibition drives CRC differentiation through H3K27ac accumulation.

Of clinical relevance, HDAC1/2- and MRK60-bound genomic regions in normal colon tissue (n = 19) exhibit H3K27ac enrichment, encompassing nearly all annotated colon differentiation genes; this H3K27ac signature is modestly but significantly reduced in CRC tumors (n = 32) (Figure 4M)25. Global H3K27ac levels in CRC were heterogeneous, with roughly equal number of regions gained (41,830) and lost (42,441) compared to normal tissue (Supplementary Figure 5O). These findings support a model in which HDAC1/2 inhibition restores H3K27ac at key regulatory sites, increases chromatin accessibility, and reactivates differentiation gene expression in CRC.

Histone acetyltransferase EP300 is required for CRC differentiation mediated by HDAC1/2 inhibition.

To test whether H3K27ac accumulation is required for CRC differentiation following HDAC1/2 inhibition, we disrupted EP300 - the primary acetyltransferase responsible for H3K27 acetylation - and asked whether this would reverse the differentiation phenotype. CRC cells treated with MRK60 were exposed to JQAD1, a selective EP300 degrader recently developed by our group28. JQAD1 reduced MRK60-induced KRT20 expression in a dose-dependent manner (Figure 5A and lowered H3K27ac levels, as confirmed by H3-enriched immunoblots (Figure 5B). In our HT29KRT20-GFP differentiation reporter line, MRK60 induced a ~3-fold increase in GFP+ cells, which was reduced by ~2-fold with JQAD1 co-treatment—bringing levels nearly back to baseline (Figure 5C). Consistently, in a previously published epigenetic regulator screen18, EP300 knockout emerged as one of the strongest hits sustaining a stem-like state, as measured by our dual reporter system (Figure 5D).

Figure 5: H3K27ac is required for differentiation reprogramming by HDAC1/2 inhibition.

Figure 5:

Open in a new tab

A. Immunoblot of KRT20 and GAPDH proteins in LS180 treated with MRK60 (2 μM) and indicated concentrations of JQAD1 (0.4 – 10 μM).

B. Histone-enriched immunoblot of H3K27ac and histone H3 in HT29 treated with MRK60 (5 μM) +/− JQAD1 (10 μM).

C. Quantification of GFP+ percentage in HT29KRT20-GFP reporter cell line treated with MRK60 (5 μM) +/− JQAD1 (10 μM).

D. Ranked log2 fold change of sgRNA distribution in GFPhigh (perturbation promotes differentiation) and mKate2high (perturbation promotes stem cell) sorted cell fractions of published CRISPR-Cas9 screen targeting epigenetic regulators (78 genes 542 sgRNAs) using HT29SOX9-mKate2/KRT20-GFP reporter cell line16; select perturbations indicated in colored circles.

E. Schematic of scRNA-seq experiment in patient-derived CRC organoids treated with DMSO, MRK60, or MRK60+JQAD1 combination (left). UMAPs of scRNA-seq colored by treatment group (middle) or cell type (right).

F. Proportion of cell types in CRC organoids treated with DMSO, MRK60, or MRK60+JQAD1.

G. Trajectory analysis of colonocytes in scRNA-seq experiment colored by treatment group.

H. Expression of colonocyte differentiation signatures along drug response trajectory. Density plot shows histogram of cells in different treatment along pseudotime. ** P = .0042 ; *** P = .0002

I. Violin plot of colonocyte differentiation gene signature41 in scRNA-seq experiment by treatment groups. **** P = 7.0 × 10−6 (left), **** P = 8.4 × 10−8 (right)

J. Violin plot of gene signatures of MRK60-induced gains in H3K27ac bound by HDAC1/2 in scRNA-seq experiment by treatment groups. **** P = 5.8 × 10−10 (left), **** P = 2.1 × 10−5(right)

K. Integrative heatmap of expression changes in stem cell and differentiation genes associated with H3K27ac, H3K9ac, and H4K8ac in bulk and single-cell RNA data sets with MRK60 +/− JQAD1

Bulk RNA-seq cannot resolve cellular heterogeneity, so to dissect how histone acetylation regulates stem and differentiated states in CRC, we performed single-cell RNA-seq (scRNA-seq) on patient-derived organoids treated with DMSO, MRK60, or MRK60+JQAD1. We profiled 33,331 high-quality cells and identified three main epithelial populations: intestinal stem cells (ISCs, 24.9%), transit amplifying cells (TA, 19.1%), and absorptive colonocytes (ACC, 55.9%) (Figure 5E). Cell type identities were confirmed by comparing to a reference human scRNA-seq dataset41, and all clusters contained cells from each treatment group, ruling out dissociation bias. We detected an average of 3,256 unique genes and 10,142 transcripts per cell (Supplementary Figure 6A-B).

MRK60 treatment reduced the proportion of ISC and TA cells while increasing the percentage of differentiated colonocytes (Figure 5F-H). This shift was accompanied by an increase in cells arrested in G1 (Supplementary Figure 6C), consistent with the post-mitotic state of differentiated cells. MRK60 also upregulated differentiation genes, with the strongest associations linked to H3K27ac gains at HDAC1/2-bound regions—more so than H3K9ac or H4K8ac—supporting prior findings (Figures 5I-K and Supplementary Figure 6F-G).

To test whether H3K27ac is required for this differentiation effect, we degraded EP300 using JQAD1. scRNA-seq showed that combination treatment partially reversed MRK60’s impact—restoring ISC and TA populations and reducing differentiated colonocytes (Figure 5F). Compared to DMSO, MRK60 and combination treatment produced 3,006 and 3,831 unique DEGs, respectively (FDR<0.01, LFC>1.5). Many MRK60-induced gene changes were reversed by JQAD1 cotreatment (Supplementary Figure 6D). Importantly, DEG profiles from scRNA-seq correlated strongly with those from bulk RNA-seq (Supplementary Figure 6E-G), confirming the robustness of these findings across platforms and CRC models.

We next mapped a drug response trajectory for colonocytes and found that MRK60-treated cells progressed along a differentiation continuum, while co-treatment with JQAD1 shifted cells back toward stem and transit amplifying states (Figure 5G-H). JQAD1 suppressed the colonocyte differentiation gene signatures induced by HDAC1/2 inhibition (Figure 5H-I), and these downregulated genes were most strongly enriched at HDAC1/2-bound H3K27ac sites, with a weaker association at H3K9ac regions (Figures 5J-K and Supplementary Figure 6G). Consistent with this, GSEA revealed opposing effects of MRK60 and MRK60+JQAD1 on stem42 and differentiation41 gene signatures (Figures 5I and Supplementary Figure 6H). These data indicate that chemically induced gains in H3K27ac—and the resulting CRC differentiation—can be reversed by targeted degradation of its writer, EP300.

DAPK3 is induced by HDAC1/2 inhibition and a functional regulator of differentiation.

Beyond direct activation of differentiation genes via H3K27ac, we asked whether HDAC1/2 inhibition also induces upstream regulators that facilitate differentiation. To identify candidates, we focused on genes that met three criteria following MRK60 treatment: (1) bound by HDAC1/2 (via both CUT&RUN and Chem-seq), (2) gained H3K27ac, and (3) showed increased expression (Figure 6A). We then performed a pooled CRISPR-Cas9 screen in HT29KRT20-GFP cells to determine which of these genes, when disrupted, could reverse MRK60-induced loss of viability and differentiation (Figures 6B-C). Among 27 shortlisted candidates, DAPK3 (Death-Associated Protein Kinase 3) - a chromatin- and apoptosis-associated kinase involved in cell cycle regulation and survival43 - emerged as a potential regulator. DAPK3-targeting sgRNAs ranked 3rd and 4th in reversing MRK60-induced loss of viability and KRT20 expression, respectively (Figure 6C and Supplementary Figure 7A).

Figure 6: DAPK3 is upregulated and required for differentiation mediated by HDAC1/2 inhibition.

Figure 6:

Open in a new tab

A. Venn diagram showing genes bound by HDAC1/2 and MRK60, upregulated by MRK60, and gained H3K27ac upon MRK60 treatment.

B. Schematic diagram showing CRISPR-Cas9 genetic screen in HT29KRT20-GFP reporter cells.

C. Ranked normalized Z-scored log2 fold change (LFC) plots of sgRNA targeting 27 genes in HT29KRT20-GFP reporter screen with respect to viability (top) and GFP/KRT20 (bottom) readouts.

D. mRNA expression of KRT20 and DAPK3 in HT115 cells engineered to suppress DAPK3 by CRISPRi knockdown treated with 2μM MRK60 for 72 hours.

E. Protein expression of KRT20 and DAPK3 in HT115 cells engineered to suppress DAPK3 by CRISPRi knockdown treated with 2μM MRK60 for 72 hours.

F. IGV snapshot depicting HDAC1/2 binding, H3K27ac, H3K9ac, and H4K8ac CUT&RUN peaks and ATAC-seq profiles at DAPK3 genomic locus.

G. DAPK3 mRNA expression in normal and CRC tissue from patients in TCGA.

H. IGV plot of H3K27ac peaks at DAPK3 genomic locus in normal and CRC; box plot quantification.

I. Scatter plot of GSEA results: NES of pathways significantly altered in HT115-NTC cells treated with MRK60 (2 μM) / DMSO over 5 days (x-axis); NES of pathways significantly altered in MRK60 (2 μM)-treated HT115 control or DAPK3 KD (y-axis).

J. Immunoblots of Cdx2CreERT2; Apcf/f; R26tdT adenoma mouse organoids treated with 5 μM MRK60 and indicated doses of Ruxolitinib at 72 hours.

K. Schematic of a proposed epigenetic & differentiation mechanism mediated by HDAC1/2 inhibition

HDAC1/2 inhibition consistently upregulated DAPK3 expression across all CRC cell lines and organoid models tested (Figures 6D-E and Supplementary Figure 7B-E). At the DAPK3 genomic locus, MRK60 treatment increased H3K27ac and mRNA levels, reduced HDAC1/2 binding and H4K8ac, but had no effect on H3K9ac or chromatin accessibility (Figure 6F). To assess the functional role of H3K27ac in regulating DAPK3, we reanalyzed our scRNA-seq data from MRK60 and JQAD1-treated organoids JQAD1 (Figure 5E). DAPK3 was induced by MRK60 in both stem-like and differentiated colonocytes, but EP300 degradation significantly reduced DAPK3 expression only in differentiated cells, suggesting H3K27ac is required to sustain DAPK3 in that compartment (Supplementary Figure 7F). Of clinical importance, DAPK3 expression is markedly reduced in CRC tumors compared to normal colon tissue (Figure 6G), paralleling a loss of H3K27ac at its enhancer (Figure 6H). This concordant decline in expression and enhancer activity underscores DAPK3 as a differentiation-linked tumor suppressor epigenetically silenced in CRC.

CRISPRi-mediated suppression of DAPK3 partially reversed MRK60-induced differentiation in two CRC cell lines, validating results from our genetic screen (Figure 6D-E and Supplementary Figure 7B-E). Consistently, analysis of 35 CRC cell lines from the BROAD CTD2 drug screen showed that lower DAPK3 expression correlated with resistance to MRK60 (Supplementary Figure 7G, P = .045). To further define DAPK3’s role, we performed bulk RNA-seq on HT-115 cells with either non-targeting control (NTC) or DAPK3 knockdown, treated for 5 days with DMSO or MRK60. GSEA revealed that differentiation-associated transcriptional programs strongly induced by MRK60 in control cells were largely lost in DAPK3-deficient cells (Figure 6I), underscoring DAPK3 as a key mediator of HDAC1/2 inhibition–driven differentiation. In parallel, gene sets linked to proliferation were enriched in DAPK3-deficient cells, suggesting that loss of DAPK3 enables escape from MRK60’s anti-tumor effects.

Ruxolitinib, a JAK1/2 kinase inhibitor, also targets DAPK3 with a reported Kd of 89 nM (compared to 3.4nM for JAK1, 0.036nM for JAK2, and 2nM for JAK3). Notably, DAPK3 and JAK3 were the only Ruxolitinib targets upregulated by MRK60 (Supplementary Figure 7H). We therefore tested whether Ruxolitinib could reverse the CRC differentiation phenotype induced by HDAC1/2 inhibition. Dose-dependent Ruxolitinib treatment reduced MRK60-induced KRT20 expression in both Cdx2-ApcKO adenoma organoids and CRC cell lines (Figures 6J and Supplementary Figure 7D-E), while stem cell markers increased with combination treatment. Importantly, JAK/STAT pathway genes were not associated with the differentiation phenotype; they were downregulated by MRK60, and their expression remained unchanged or suppressed by Ruxolitinib in both organoids and HT-115 cells (Supplementary Figure 7I-J). Furthermore, genetic knockout of JAK1, JAK2, JAK3, or STAT3 did not affect MRK60-induced KRT20 expression (Supplementary Figure 7K-L), supporting a JAK-independent mechanism for Ruxolitinib’s effects. Consistent with its impact on differentiation, Ruxolitinib antagonized MRK60’s anti-tumor effects, showing dose-dependent anti-synergy in viability assays based on ZIP score analysis (Supplementary Figure 7M). Collectively, these data indicate that DAPK3 is a functional component of HDAC1/2 inhibition mediated CRC differentiation (Figure 6K).

DISCUSSION

Genetic and epigenetic alterations cooperate to drive CRC initiation, progression, and therapy resistance by enforcing a stem-like transcriptional program that underlies tumor plasticity7, 9. While the cascade of APC, TP53, SMAD4, and KRAS mutations is well-defined, our understanding of how epigenetic dysregulation fuels CRC plasticity remains incomplete and increasingly important44. To functionally interrogate these non-genetic mechanisms, we engineered CRC cell lines with dual endogenous reporters at the SOX9 and KRT20 loci, yielding simultaneous fluorescent readouts of aberrant stem-like and differentiated states18. Leveraging this platform, we will screen inhibitors and degraders of epigenetic regulators - borrowing from successful leukemia paradigms17 - to pinpoint agents that suppress aberrant stem-cell activity and reactivate CRC differentiation.

Using our dual-reporter and viability screens, we identified both HDAC1/2 inhibitors and SMARCA2/4 degraders as differentiation inducers, and we chose to focus on HDAC1/2 inhibitors for their consistent activity across models and proven clinical use in hematologic malignancies. HDAC inhibitors Vorinostat and Romidepsin have advanced to phase II trials in late-stage CRC patients45. Unfortunately, their clinical success is limited by cardiac and hematologic toxicities. By isolating the specific acetylation changes that drive CRC differentiation and antitumor effects, we can guide the design of therapies that retain efficacy while reducing untoward side effects. We concentrated on histone acetylation since (1) transcriptional reprogramming underlies differentiation and (2) loss of histone acetylation is a hallmark of human cancers46.

HDAC inhibitors drive antitumor activity through cell-cycle arrest, apoptosis, DNA damage responses47–50 and, importantly, by reactivating silenced differentiation programs. Oncogenic proteins complex with, co-opt, and aberrantly recruit HDACs to repress genes that obstruct tumorigenesis16, 51- a repression that HDAC inhibitors can reverse52. In preclinical models, blocking HDAC6 in glioblastoma or HDAC3 in glioma stem cells induces differentiation and curbs self-renewal53; inhibiting HDAC854, 55 - or HDAC1/256, with or without retinoic acid - in neuroblastoma triggers differentiation and growth arrest. Similarly, in AML, HDAC inhibition promotes maturation via distinct molecular pathways. Despite these varied contexts, a common mechanism for how HDAC inhibitors unleash differentiation across cancers has yet to be defined.

In our study, HDAC1/2 inhibition drove broad differentiation signals that were partly dependent on H3K27ac. Although these inhibitors increased acetylation across multiple histone sites, only H3K27ac at HDAC1/2-bound enhancers tracked with a differentiation gene signature. Distinct histone acetylation marks are associated with CRC prognosis and subtypes57, and CRC exhibits a unique enhancer-associated H3K27ac landscape compared to normal colon25, 58. Crucially, EP300 degradation - removing the principal H3K27 acetyltransferase - partially reversed HDAC1/2-inhibitor–induced differentiation, confirming H3K27ac’s functional role. Together, these data suggest that a dynamic balance between EP300 and HDAC1/2 activity governs stem-cell versus differentiation programs in CRC and, perhaps, normal intestinal homeostasis. Selectively enhancing H3K27ac, rather than broadly blocking HDAC1/2, may therefore offer targeted antitumor activity with reduced toxicity.

HDAC1/2 inhibition not only increases H3K27ac at differentiation genes but also upregulates DAPK3- a p53-linked mediator of therapy resistance59, 60. Inhibiting DAPK3 genetically or pharmacologically partially blocks HDAC1/2-driven differentiation, implicating it as a critical effector. Defining the DAPK3-regulated phosphor-proteome and its precise role in CRC differentiation remains an essential next step.

Drug toxicity may prevent achieving the HDAC1/2 inhibitor exposures needed for differentiation, contributing to disappointing trial results. Combining selective HDAC1/2 inhibitors with standard CRC chemotherapies (e.g., 5-FU, oxaliplatin, irinotecan) could lower systemic doses, reduce toxicity, and amplify antitumor efficacy, as shown by preclinical synergy with these agents61, 62. Because CRC resistance to targeted therapies - such as adaptive EGFR activation in KRAS-mutant tumors2, 3 - stems from epigenetic plasticity, enforcing differentiation via HDAC1/2 blockade may prevent or delay resistance. Supporting this, EZH2 inhibitors cooperate with RAS-pathway blockers to induce differentiation and tumor regression63, and the HDAC inhibitor chidamide plus bevacizumab sensitizes refractory CRC to PD-1 blockade in a phase II trial64. To minimize systemic toxicity, tumor-targeted delivery (e.g., antibody–drug conjugates) and biomarker-guided patient selection could further enhance clinical translation.

STUDY LIMITATIONS

While our analysis focused on histone PTMs, HDACs are also known to modify non-histone proteins65, and we cannot exclude their contribution to the observed differentiation phenotype. Additionally, disentangling cell cycle regulation from differentiation remains challenging. Although differentiation markers are upregulated prior to reductions in proliferation, the precise relationship between these processes during HDAC1/2 inhibition in CRC is not fully resolved. Finally, broader conclusions regarding H3K27ac changes in patient tissue will require analysis of larger sample cohorts.

Supplementary Material

Supplementary Material
Supplemental Table 1
Supplemental Table 2
Supplemental Table 3
Supplemental Table 4
Supplemental Table 5

Supplementary Information is available for this paper

Acknowledgements

We regret that space constraints prevent us from citing every foundational contribution (118 references in the original manuscript), and we extend our sincere apologies to those whose work is unlisted. We are grateful to all members of the Sethi laboratory for their unwavering support and insightful feedback; to Rameen Beroukhim for invaluable discussions and critical manuscript review; and to our colleagues at the Center for Cancer Epigenetics, especially Kornelia Polyak, Myles Brown, and Ramesh Shivdasani and Matthew Freedman, for their expertise, encouragement, and use of Celigo™ High Throughput Micro-Well Image Cytometer; Christine Perret for kindly sharing the Apcflox/flox mice developed in her laboratory; Aniket Gad and Lay-Hong Ang for assistance with immunohistochemistry assays; Shuqiang Li and Kenneth Livak for assistance with scRNA-seq; Dana-Farber/Harvard Cancer Center for the use of the Specialized Histopathology Core, which provided histology and immunohistochemistry service; Harvard Digestive Disease Center and NIH grant P30DK034854 for core services, resources, technology, and expertise. Dana-Farber/Harvard Cancer Center is supported in part by an NCI Cancer Center Support Grant # NIH 5 P30 CA06516. S.S. was supported by the National Research, Development and Innovation Office (grants FK142835, EXCELLENCE_24 151330, and MILAB-AINL RRF-2.3.1–21-2022–00004), the Bolyai János Research Fellowship of the Hungarian Academy of Sciences, and the Hungarian Research Network (HUN-REN KSZF-160/2024). This work was funded by the Colorectal Cancer Alliance, Virtual Scholar Award from the Department of Defense (CA201084), Bridge Project (DF/HCC-MIT), and 1R01CA292507 to N.S.S and philanthropic support from the Jimmy Fund Walk (Opiela Family), Craig Baskin, Dave & Carol Fischer, and Howard & Wendy Cox to the Sethi Lab.

Footnotes

Declarations of Interest

J.M.C receives research funding to his institution from Amgen, Merus, Servier, and Bristol Myers Squibb. He receives research support from Merck, AstraZeneca, Esperas Pharma, Bayer, Tesaro, Arcus Biosciences, and Pyxis; he has also received honoraria for being on the advisory boards of Incyte and Blueprint Medicines and for serving on the data safety monitoring committee for Astrazeneca. He has given educational talks sponsored by Bayer, Bristol Myers Squibb, Lilly, Merck, AstraZeneca, and Genentech. S.S. and N.S.S. are co-inventers on patent US20240280561A1 published on August 22nd, 2024 involving part of this work. N.S.S. receives research funding from Novartis, is a consultant for Dewpoint Therapeutics, and is on the scientific advisory board for Astrin Biosciences and

Data Transparency:

All data that support the findings of this study are available within the paper and its supplementary files. Raw and processed sequencing data are available in the Sequence Read Archive (SRA) at the NCBI Center with the accession number PRJNA1142083 and Gene Expression Omnibus (GEO) with the accession number GSE273626. All processed data are also available at https://doi.org/10.5281/zenodo.13145655. Source data are provided with this paper. All other data and bioinformatics code supporting the findings of this study are available from the corresponding author on reasonable request.

REFERENCES

  • 1.Siegel RL, Miller KD, Jemal A. Cancer statistics, 2018. CA Cancer J Clin 2018;68:7–30. [DOI] [PubMed] [Google Scholar]
  • 2.Amodio V, Yaeger R, Arcella P, et al. EGFR Blockade Reverts Resistance to KRAS(G12C) Inhibition in Colorectal Cancer. Cancer Discov 2020;10:1129–1139. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Fakih MG, Salvatore L, Esaki T, et al. Sotorasib plus Panitumumab in Refractory Colorectal Cancer with Mutated KRAS G12C. N Engl J Med 2023;389:2125–2139. [DOI] [PubMed] [Google Scholar]
  • 4.Gupta PB, Pastushenko I, Skibinski A, et al. Phenotypic Plasticity: Driver of Cancer Initiation, Progression, and Therapy Resistance. Cell Stem Cell 2019;24:65–78. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Arozarena I, Wellbrock C. Phenotype plasticity as enabler of melanoma progression and therapy resistance. Nat Rev Cancer 2019;19:377–391. [DOI] [PubMed] [Google Scholar]
  • 6.Yaeger R, Weiss J, Pelster MS, et al. Adagrasib with or without Cetuximab in Colorectal Cancer with Mutated KRAS G12C. N Engl J Med 2023;388:44–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Bala P, Rennhack JP, Aitymbayev D, et al. Aberrant cell state plasticity mediated by developmental reprogramming precedes colorectal cancer initiation. Sci Adv 2023;9:eadf0927. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Moorman A, Benitez EK, Cambulli F, et al. Progressive plasticity during colorectal cancer metastasis. Nature 2025;637:947–954. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Vasquez EG, Nasreddin N, Valbuena GN, et al. Dynamic and adaptive cancer stem cell population admixture in colorectal neoplasia. Cell Stem Cell 2022;29:1213–1228 e8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Hanahan D Hallmarks of Cancer: New Dimensions. Cancer Discov 2022;12:31–46. [DOI] [PubMed] [Google Scholar]
  • 11.Liang X, Duronio GN, Yang Y, et al. An Enhancer-Driven Stem Cell-Like Program Mediated by SOX9 Blocks Intestinal Differentiation in Colorectal Cancer. Gastroenterology 2022;162:209–222. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Parreno V, Loubiere V, Schuettengruber B, et al. Transient loss of Polycomb components induces an epigenetic cancer fate. Nature 2024;629:688–696. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Baylin SB, Jones PA. Epigenetic Determinants of Cancer. Cold Spring Harb Perspect Biol 2016;8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.San José-Enériz E, Gimenez-Camino N, Agirre X, et al. HDAC Inhibitors in Acute Myeloid Leukemia. Cancers (Basel) 2019;11. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Singh A, Patel P, Jageshwar, et al. The Safety, Efficacy and Therapeutic Potential of Histone Deacetylase Inhibitors with Special Reference to Panobinostat in Gastrointestinal Tumors: A Review of Preclinical and Clinical Studies. Curr Cancer Drug Targets 2018;18:720–736. [DOI] [PubMed] [Google Scholar]
  • 16.Ablain J, de The H. Revisiting the differentiation paradigm in acute promyelocytic leukemia. Blood 2011;117:5795–802. [DOI] [PubMed] [Google Scholar]
  • 17.de The H Differentiation therapy revisited. Nat Rev Cancer 2018;18:117–127. [DOI] [PubMed] [Google Scholar]
  • 18.Spisak S, Chen D, Likasitwatanakul P, et al. Identifying regulators of aberrant stem cell and differentiation activity in colorectal cancer using a dual endogenous reporter system. Nat Commun 2024;15:2230. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Anders L, Guenther MG, Qi J, et al. Genome-wide localization of small molecules. Nat Biotechnol 2014;32:92–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Rossmann MP, Hoi K, Chan V, et al. Cell-specific transcriptional control of mitochondrial metabolism by TIF1gamma drives erythropoiesis. Science 2021;372:716–721. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Dow LE, O’Rourke KP, Simon J, et al. Apc Restoration Promotes Cellular Differentiation and Reestablishes Crypt Homeostasis in Colorectal Cancer. Cell 2015;161:1539–1552. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Methot JL, Chakravarty PK, Chenard M, et al. Exploration of the internal cavity of histone deacetylase (HDAC) with selective HDAC1/HDAC2 inhibitors (SHI-1:2). Bioorg Med Chem Lett 2008;18:973–8. [DOI] [PubMed] [Google Scholar]
  • 23.Rees MG, Seashore-Ludlow B, Cheah JH, et al. Correlating chemical sensitivity and basal gene expression reveals mechanism of action. Nat Chem Biol 2016;12:109–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Xiao L, Parolia A, Qiao Y, et al. Targeting SWI/SNF ATPases in enhancer-addicted prostate cancer. Nature 2022;601:434–439. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Orouji E, Raman AT, Singh AK, et al. Chromatin state dynamics confers specific therapeutic strategies in enhancer subtypes of colorectal cancer. Gut 2022;71:938–949. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Crespo M, Vilar E, Tsai SY, et al. Colonic organoids derived from human induced pluripotent stem cells for modeling colorectal cancer and drug testing. Nat Med 2017;23:878–884. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Zhang Y, Remillard D, Onubogu U, et al. Collateral lethality between HDAC1 and HDAC2 exploits cancer-specific NuRD complex vulnerabilities. Nat Struct Mol Biol 2023;30:1160–1171. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Durbin AD, Wang T, Wimalasena VK, et al. EP300 Selectively Controls the Enhancer Landscape of MYCN-Amplified Neuroblastoma. Cancer Discov 2022;12:730–751. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Ueda H, Manda T, Matsumoto S, et al. FR901228, a novel antitumor bicyclic depsipeptide produced by Chromobacterium violaceum No. 968. III. Antitumor activities on experimental tumors in mice. J Antibiot (Tokyo) 1994;47:315–23. [DOI] [PubMed] [Google Scholar]
  • 30.Saito A, Yamashita T, Mariko Y, et al. A synthetic inhibitor of histone deacetylase, MS-27–275, with marked in vivo antitumor activity against human tumors. Proc Natl Acad Sci U S A 1999;96:4592–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.LoRusso PM, Demchik L, Foster B, et al. Preclinical antitumor activity of CI-994. Invest New Drugs 1996;14:349–56. [DOI] [PubMed] [Google Scholar]
  • 32.Zutz A, Chen L, Sippl F, et al. Epigenetic Compound Screening Uncovers Small Molecules for Reactivation of Latent HIV-1. Antimicrob Agents Chemother 2020;65. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Gryder BE, Wu L, Woldemichael GM, et al. Chemical genomics reveals histone deacetylases are required for core regulatory transcription. Nat Commun 2019;10:3004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Malvaez M, McQuown SC, Rogge GA, et al. HDAC3-selective inhibitor enhances extinction of cocaine-seeking behavior in a persistent manner. Proc Natl Acad Sci U S A 2013;110:2647–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Fares I, Chagraoui J, Gareau Y, et al. Cord blood expansion. Pyrimidoindole derivatives are agonists of human hematopoietic stem cell self-renewal. Science 2014;345:1509–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Li Y, Seto E. HDACs and HDAC Inhibitors in Cancer Development and Therapy. Cold Spring Harb Perspect Med 2016;6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Haber AL, Biton M, Rogel N, et al. A single-cell survey of the small intestinal epithelium. Nature 2017;551:333–339. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Geffen Y, Anand S, Akiyama Y, et al. Pan-cancer analysis of post-translational modifications reveals shared patterns of protein regulation. Cell 2023;186:3945–3967 e26. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Skene PJ, Henikoff S. An efficient targeted nuclease strategy for high-resolution mapping of DNA binding sites. Elife 2017;6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Ramírez F, Ryan DP, Grüning B, et al. deepTools2: a next generation web server for deep-sequencing data analysis. Nucleic Acids Research 2016;44:W160–W165. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Burclaff J, Bliton RJ, Breau KA, et al. A Proximal-to-Distal Survey of Healthy Adult Human Small Intestine and Colon Epithelium by Single-Cell Transcriptomics. Cell Mol Gastroenterol Hepatol 2022;13:1554–1589. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Muñoz J, Stange DE, Schepers AG, et al. The Lgr5 intestinal stem cell signature: robust expression of proposed quiescent ‘+4’ cell markers. Embo j 2012;31:3079–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Weitzel DH, Chambers J, Haystead TA. Phosphorylation-dependent control of ZIPK nuclear import is species specific. Cell Signal 2011;23:297–303. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Wainwright EN, Scaffidi P. Epigenetics and Cancer Stem Cells: Unleashing, Hijacking, and Restricting Cellular Plasticity. Trends Cancer 2017;3:372–386. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Whitehead RP, Rankin C, Hoff PM, et al. Phase II trial of romidepsin (NSC-630176) in previously treated colorectal cancer patients with advanced disease: a Southwest Oncology Group study (S0336). Invest New Drugs 2009;27:469–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Fraga MF, Ballestar E, Villar-Garea A, et al. Loss of acetylation at Lys16 and trimethylation at Lys20 of histone H4 is a common hallmark of human cancer. Nat Genet 2005;37:391–400. [DOI] [PubMed] [Google Scholar]
  • 47.Yamaguchi J, Sasaki M, Sato Y, et al. Histone deacetylase inhibitor (SAHA) and repression of EZH2 synergistically inhibit proliferation of gallbladder carcinoma. Cancer Sci 2010;101:355–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Zupkovitz G, Grausenburger R, Brunmeir R, et al. The cyclin-dependent kinase inhibitor p21 is a crucial target for histone deacetylase 1 as a regulator of cellular proliferation. Mol Cell Biol 2010;30:1171–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Zhang J, Zhong Q. Histone deacetylase inhibitors and cell death. Cell Mol Life Sci 2014;71:3885–901. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Miller KM, Tjeertes JV, Coates J, et al. Human HDAC1 and HDAC2 function in the DNA-damage response to promote DNA nonhomologous end-joining. Nat Struct Mol Biol 2010;17:1144–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Liu Y, Cheney MD, Gaudet JJ, et al. The tetramer structure of the Nervy homology two domain, NHR2, is critical for AML1/ETO’s activity. Cancer Cell 2006;9:249–60. [DOI] [PubMed] [Google Scholar]
  • 52.Falkenberg KJ, Johnstone RW. Histone deacetylases and their inhibitors in cancer, neurological diseases and immune disorders. Nat Rev Drug Discov 2014;13:673–91. [DOI] [PubMed] [Google Scholar]
  • 53.Liang H, Wang Q, Wang D, et al. RGFP966, a histone deacetylase 3 inhibitor, promotes glioma stem cell differentiation by blocking TGF-beta signaling via SMAD7. Biochem Pharmacol 2020;180:114118. [DOI] [PubMed] [Google Scholar]
  • 54.Oehme I, Deubzer HE, Wegener D, et al. Histone deacetylase 8 in neuroblastoma tumorigenesis. Clin Cancer Res 2009;15:91–9. [DOI] [PubMed] [Google Scholar]
  • 55.Rettig I, Koeneke E, Trippel F, et al. Selective inhibition of HDAC8 decreases neuroblastoma growth in vitro and in vivo and enhances retinoic acid-mediated differentiation. Cell Death Dis 2015;6:e1657. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.Frumm SM, Fan ZP, Ross KN, et al. Selective HDAC1/HDAC2 inhibitors induce neuroblastoma differentiation. Chem Biol 2013;20:713–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57.Benard A, Goossens-Beumer IJ, van Hoesel AQ, et al. Nuclear expression of histone deacetylases and their histone modifications predicts clinical outcome in colorectal cancer. Histopathology 2015;66:270–82. [DOI] [PubMed] [Google Scholar]
  • 58.Della Chiara G, Gervasoni F, Fakiola M, et al. Epigenomic landscape of human colorectal cancer unveils an aberrant core of pan-cancer enhancers orchestrated by YAP/TAZ. Nat Commun 2021;12:2340. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 59.Brognard J, Zhang YW, Puto LA, et al. Cancer-associated loss-of-function mutations implicate DAPK3 as a tumor-suppressing kinase. Cancer Res 2011;71:3152–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 60.Cai Z, Cao R, Zhang K, et al. Oncogenic miR-17/20a Forms a Positive Feed-forward Loop with the p53 Kinase DAPK3 to Promote Tumorigenesis. J Biol Chem 2015;290:19967–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 61.Hontecillas-Prieto L, Flores-Campos R, Silver A, et al. Synergistic Enhancement of Cancer Therapy Using HDAC Inhibitors: Opportunity for Clinical Trials. Front Genet 2020;11:578011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 62.Na YS, Jung KA, Kim SM, et al. The histone deacetylase inhibitor PXD101 increases the efficacy of irinotecan in in vitro and in vivo colon cancer models. Cancer Chemother Pharmacol 2011;68:389–98. [DOI] [PubMed] [Google Scholar]
  • 63.Loi P, Schade AE, Rodriguez CL, et al. Epigenetic and oncogenic inhibitors cooperatively drive differentiation and kill KRAS-mutant colorectal cancers. Cancer Discov 2024. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 64.Wang F, Jin Y, Wang M, et al. Combined anti-PD-1, HDAC inhibitor and anti-VEGF for MSS/pMMR colorectal cancer: a randomized phase 2 trial. Nat Med 2024;30:1035–1043. [DOI] [PubMed] [Google Scholar]
  • 65.Choudhary C, Kumar C, Gnad F, et al. Lysine acetylation targets protein complexes and co-regulates major cellular functions. Science 2009;325:834–40. [DOI] [PubMed] [Google Scholar]
  • 66.Goldman MJ, Craft B, Hastie M, et al. Visualizing and interpreting cancer genomics data via the Xena platform. Nat Biotechnol 2020;38:675–678. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material
NIHMS2158553-supplement-Supplementary_Material.pdf (4.2MB, pdf)
Supplemental Table 1
NIHMS2158553-supplement-Supplemental_Table_1.xlsx (11.3KB, xlsx)
Supplemental Table 2
NIHMS2158553-supplement-Supplemental_Table_2.xlsx (10.5KB, xlsx)
Supplemental Table 3
NIHMS2158553-supplement-Supplemental_Table_3.xlsx (78.7KB, xlsx)
Supplemental Table 4
NIHMS2158553-supplement-Supplemental_Table_4.xlsx (12.2KB, xlsx)
Supplemental Table 5
NIHMS2158553-supplement-Supplemental_Table_5.xlsx (481.7KB, xlsx)

Data Availability Statement

All data that support the findings of this study are available within the paper and its supplementary files. Raw and processed sequencing data are available in the Sequence Read Archive (SRA) at the NCBI Center with the accession number PRJNA1142083 and Gene Expression Omnibus (GEO) with the accession number GSE273626. All processed data are also available at https://doi.org/10.5281/zenodo.13145655. Source data are provided with this paper. All other data and bioinformatics code supporting the findings of this study are available from the corresponding author on reasonable request.

RESOURCES