Context-defined cancer co-dependency mapping identifies a functional interplay between PRC2 and MLL-MEN1 complex in lymphoma

Chen, Xiao; Li, Yinglu; Zhu, Fang; Xu, Xinjing; Estrella, Brian; Pazos, Manuel A.; McGuire, John T.; Karagiannis, Dimitris; Sahu, Varun; Mustafokulov, Mustafo; Scuoppo, Claudio; Sánchez-Rivera, Francisco J.; Soto-Feliciano, Yadira M.; Pasqualucci, Laura; Ciccia, Alberto; Amengual, Jennifer E.; Lu, Chao

doi:10.1038/s41467-023-39990-5

Download PDF

Article
Open access
Published: 17 July 2023

Context-defined cancer co-dependency mapping identifies a functional interplay between PRC2 and MLL-MEN1 complex in lymphoma

Nature Communications volume 14, Article number: 4259 (2023) Cite this article

5317 Accesses
5 Citations
19 Altmetric
Metrics details

Subjects

Abstract

Interplay between chromatin-associated complexes and modifications critically contribute to the partitioning of epigenome into stable and functionally distinct domains. Yet there is a lack of systematic identification of chromatin crosstalk mechanisms, limiting our understanding of the dynamic transition between chromatin states during development and disease. Here we perform co-dependency mapping of genes using CRISPR-Cas9-mediated fitness screens in pan-cancer cell lines to quantify gene-gene functional relationships. We identify 145 co-dependency modules and further define the molecular context underlying the essentiality of these modules by incorporating mutational, epigenome, gene expression and drug sensitivity profiles of cell lines. These analyses assign new protein complex composition and function, and predict new functional interactions, including an unexpected co-dependency between two transcriptionally counteracting chromatin complexes - polycomb repressive complex 2 (PRC2) and MLL-MEN1 complex. We show that PRC2-mediated H3K27 tri-methylation regulates the genome-wide distribution of MLL1 and MEN1. In lymphoma cells with EZH2 gain-of-function mutations, the re-localization of MLL-MEN1 complex drives oncogenic gene expression and results in a hypersensitivity to pharmacologic inhibition of MEN1. Together, our findings provide a resource for discovery of trans-regulatory interactions as mechanisms of chromatin regulation and potential targets of synthetic lethality.

Chromatin complex dependencies reveal targeting opportunities in leukemia

Article Open access 27 January 2023

EPIKOL, a chromatin-focused CRISPR/Cas9-based screening platform, to identify cancer-specific epigenetic vulnerabilities

Article Open access 16 August 2022

Synthetic lethality prediction in DNA damage repair, chromatin remodeling and the cell cycle using multi-omics data from cell lines and patients.

Article Open access 29 April 2023

Introduction

Chemical modifications of DNA and histones are important carriers of chromatin regulatory information that cooperate with transcription factors to integrate intrinsic and extracellular stimuli to control genome accessibility¹. Precise regulation of chromatin dynamics is essential for maintaining cellular and organismal phenotypes and is frequently perturbed in various human diseases including cancer². While modifiers (e.g. writer and eraser enzymes) and biological functions for many histone and DNA marks are well-defined individually, less is understood about how chromatin modifications can “communicate” with each other. Enzymatic activities of chromatin modifiers are known to be sensitive to the local chromatin environment such that various trans-regulatory interactions exist among epigenetic marks^3,4,5,6. These chromatin “crosstalk” are thought to contribute to the partitioning of the epigenome in a stable and sometimes heritable manner. For instance, studies have demonstrated that genome-wide patterns of DNA methylation result, at least in part, from the recruitment of de novo DNA methyltransferases guided by histone modifications^7,8,9,10,11. Similarly, the establishment and propagation of histone H3K27 methylation, which is catalyzed by Polycomb Repressive Complex 2 (PRC2) and associated with transcriptional silencing, is subject to complex regulation by various chromatin-associated activities¹². H3K36 methylation, histone marks of active transcription, allosterically inhibit PRC2’s nucleosome access and catalytic activity^13,14. The SWI/SNF chromatin remodeling complex also has an established role in antagonizing PRC2^15,16. Accordingly, tumors harboring inactivating mutations in H3K36 methyltransferases (e.g. NSD1, SETD2) or SWI/SNF complex members exhibit globally elevated levels of H3K27 methylation^17,18,19. Importantly, Tazemetostat, which inhibits the EZH2 catalytic subunit of PRC2, was recently approved to treat epithelioid sarcomas with loss of SWI/SNF complex member SMARCB1²⁰. Furthermore, diffuse large B-cell lymphoma growth can be blocked by inhibition of SETD2 in the case of EZH2 hypermorphic mutation-mediated drug addiction to PRC2 inhibitors²¹. These findings suggest that disease-associated defective chromatin crosstalk could represent targets of synthetic lethality for successful therapeutic intervention.

A limited number of crosstalk between chromatin regulators and modifications have been documented through conventional biochemical approaches^7,13,22,23. Yet for many epigenetic marks, it remains unknown if trans-regulatory pathways exist in addition to their direct modifiers. To date, there has been little effort to systematically survey the functional interactions among chromatin modifications and their associated complexes. Genetic interaction mapping is a powerful approach to discover gene–gene functional relationships²⁴. Indeed, previous high-throughput screens have quantified the impact of pairwise gene double-knockout on the fitness of yeast strains at the genome scale to identify genetic interactions^24,25,26. These efforts have successfully uncovered novel functional relationships between gene pairs, including a recent report that connects individual histone residues to chromatin-associated complexes and pathways²⁷. However, a similar approach in human cells is technically challenging due to the large genome size (~20,000 genes) and thus far the largest perturbation screen only covered a small number of genetic interactions (472 genes × 472 genes)²⁸.

As an alternative approach, co-dependency mapping (also known as co-essentiality mapping or parallel screening) has recently been applied to identify functionally interacting genes^29,30,31. Co-dependency mapping takes advantage of single gene perturbation screens in a large panel of diverse cell lines, with the assumption that two genes whose perturbations result in highly correlated phenotypes across genetically and biologically heterogeneous cell lines are likely to be functionally related. Over the past few years, several consortium-based efforts, such as DepMap³² and Project Score³³, have performed CRISPR-Cas9 genetic knockout fitness screens in hundreds of cell lines of diverse tumor origins, providing a catalog of their genetic dependencies. Furthermore, the molecular features of these pan-cancer cell lines have been extensively profiled, generating detailed information about their genetic makeup, transcriptome, protein expression, histone modifications, and metabolome³⁴. We reasoned that integrative analysis of co-dependency mapping and molecular phenotyping using these multidimensional datasets would enable systematic measurements of genetic interactions and provide a key resource to nominate new context-dependent chromatin crosstalk for mechanistic studies and translational application.

Here, we investigate the potential of combining large-scale CRISPR-Cas9 essentiality screens in >1000 pan-cancer cell lines with their molecular characteristics to define context-specific co-dependency of genes and reveal novel insights into chromatin complex composition and crosstalk. Our analysis uncovers a functional interplay between PRC2 and MLL–MEN1 complex that appears at odds with their roles in transcriptional regulation. We further investigate the impact of PRC2 and H3K27 methylation on regulating the binding of the MLL–MEN1 complex and determine the therapeutic implication of this crosstalk in the context of diffuse large B-cell lymphomas (DLBCLs) harboring gain-of-function EZH2 mutations. Together, this work highlights the utility of integrating genetic screens with molecular profiling of cancer cells for high-throughput genetic interaction discovery that will not only offer critical insight into basic regulatory mechanisms of cancer epigenome progression but also nominate potential therapeutic targets of chromatin-associated synthetic lethality.

Results

Developing the genetic dependency correlation network (DCN)

Knockout of functionally linked genes is expected to produce similar fitness effects across cancer cell lines of diverse tissue origins, mutational backgrounds, and gene expression profiles. We analyzed datasets of CRISPR-Cas9 genetic perturbation screens published by Broad’s Achilles and Sanger’s SCORE projects from the DepMap portal³¹, covering 17,386 genes across 1086 pan-cancer cell lines, to build a genetic co-dependency network. To enhance the specificity of our genetic interaction mapping, we filtered out 532 common essential genes (gene effect score < −1 in more than 90% of cell lines). For each pair of genes, we calculated the Pearson correlation coefficient for gene effect scores, which measure the effect size of gene knockout on cell fitness, across all cancer cell lines and generated the dependency correlation matrix (Fig. 1a). As a validation, we analyzed the correlation scores for genes coding for proteins with biochemical interactions documented in CORUM, a curated protein complex database³⁵. We found that their scores were significantly higher than randomly selected gene pairs in equal numbers (observed vs. simulated, Supplementary Fig. 1a, b). We also overlapped the physical interactions collected in the BioGRID database with the genetic interactions in our network and 39% (305 out of 781) of our genetic interactions can be identified in BioGRID datasets (Supplementary Data 1). These results demonstrate the utility of this approach to identify complex-level protein–protein interactions.

**Fig. 1: Development of dependency correlation network (DCN) and Deplink analysis.**

To better visualize and prioritize significant gene–gene functional interactions, we selected genes that have at least one strong interaction with other genes (matrix cutoff, Pearson’s r > 0.4). Distinct “modules”—groups of genes among which correlation coefficients are higher than the background—are readily discernable from the dependency correlation matrix (Supplementary Fig. 1c, Supplementary Data 2). To maximize the power of identifying functional modules, we examined the sensitivity and specificity to recover CORUM complex-level interactions using various thresholds and determined 0.34 as the optimized cutoff (Supplementary Fig. 1b). Based on this cutoff, we next generated the dependency correlation network (DCN) consisting of 145 modules (≤15 genes) (Supplementary Fig. 1d, Supplementary Data 3).

Deplink: an integrative analysis of context-specific genetic dependency

In addition to CRISPR-Cas9 genome-wide fitness screens, cancer cell lines in the DepMap project have been extensively profiled, with detailed annotations of their genetic mutations, gene and protein expression, histone modifications, and metabolome³⁴. Therefore, we sought to integrate these molecular features with our DCN analysis, in order to uncover the context underlying the essentiality of the co-dependency gene modules. We established a pipeline, named “Deplink”, to identify associations between module dependencies and molecular phenotypes, including cancer type, mutational profiles, drug sensitivity, and transcriptome/chromatin signatures (Fig. 1a).

Cancer type

Consistent with previous report²⁹, we observed that some modules are selectively essential to specific cancer types (Fig. 1b). Indeed, the dependency profiles of blood cancers—acute myeloid leukemia (LAML), multiple myeloma (MM), acute lymphoblastic leukemia (ALL) and diffuse large B cell lymphomas (DLBCL)—cluster closely and away from solid tumors (Fig. 1b, Supplementary Fig. 2a). Using a cutoff of FDR < 0.1, we identified 137 cancer type-specific modules (Supplementary Fig. 2b, Supplementary Data 4). Some of these represent well-studied lineage survival oncogenes, including RUNX1/MYB/IRF4/NFKB1/MEF2C (module #1) in LAML, ALL, MM and DLBC^36,37, and MITF/SOX10 (module #20) in cutaneous melanoma³⁸ (Supplementary Fig. 2b); whereas others are yet to be characterized, such as STXBP3/STX4/SNAP23 (module #15) in head and neck cancers (Supplementary Fig. 2b).

Mutational signature

We also identified 31 modules (p-value < 0.05) that are selectively essential to solid tumor cell lines carrying specific COSMIC mutational signatures³⁹ (Fig. 1c, Supplementary Data 5). As expected, COSMIC signature 7, predominantly found in UV exposure-linked skin cancers, predicted dependency on module #20 (BRAF/MAPK1/MITF/SOX10) for cell survival. These results suggest that other significantly enriched modules may be involved in the development and/or represent vulnerabilities of specific mutational processes. For example, module #79 (PRPF39/TRNAU1AP) is selectively essential for cells carrying mutational signatures of defective DNA mismatch repair (COSMIC 6, 15, and 26), suggesting that this poorly characterized gene pair may play a role in the etiology or maintenance of mismatch repair deficient tumors.

Chromatin modification

To integrate our DCN analysis with global chromatin profiles, we identified cell lines with high abundance (top 18%) of each of the 26 histone H3 modifications quantified by DepMap. Seventy-four (74) modules were significantly more essential to blood cancer cell lines carrying specific chromatin signatures (Fig. 1d, Supplementary Data 6). Notably, H3K14ac-, H3K18ac- and H3K23ac-high cell lines share a common dependency on module #24, which consists of genes coding for INO80 chromatin remodeling complex (Fig. 1d). To probe the potential underlying mechanism, we analyzed genome-wide binding of INO80 complex in relation to various chromatin features using publicly available datasets⁴⁰. INO80 complex localization positively correlated with open chromatin, histone acetylation, and the binding of various histone acetyltransferases (Supplementary Fig. 2c). Therefore, it is possible that cells with histone hyperacetylation recruit INO80 complex and require its nucleosome-remodeling activity to maintain genome access and/or integrity.

Transcriptome signature

By integrating the genetic co-dependency maps and gene expression profiles, we annotated modules that were preferentially essential to cell lines exhibiting various cancer hallmark transcriptional signatures (Fig. 1b, Supplementary Data 7). Several significant correlations emerged from this analysis that are consistent with the described functions of module genes. For example, module #51 (DCP2/XRN1/ADAR) and module #101 (FOXA1/SPDEF) are more essential to cells displaying transcriptional signatures of “Interferon alpha response” and “Estrogen Response”, respectively^41,42,43. We also found that functionally linked gene expression signatures, such as “Epithelial–Mesenchymal Transition” and “TGF beta signaling”, were closely clustered based on module effect scores (Fig. 1b). Therefore, gene expression signatures represent another important determinant for the specificity of module dependencies.

Drug response

We incorporated data from Genomics of Drug Sensitivity in Cancer (GDSC)⁴⁴ to determine the relationship between DCN module dependencies and drug response. For each DCN module, we quantified the differences in drug sensitivity (IC₅₀) between dependent vs. non-dependent cell lines. Using a cutoff of drug sensitivity Z-score > 0.2 and FDR < 0.1, we uncovered 1262 significant drug–module pairs (Supplementary Data 8). Importantly, cells that required module #20 (BRAF/MAPK1) for survival also displayed increased sensitivity to multiple inhibitors of the BRAF/MEK pathway, providing validation for our analysis (Supplementary Fig. 2d). Intriguingly, we found that cells dependent on module #53 (CCNE1/SKP2/CDK2) are highly resistant to CDK4/6 inhibitor Palbociclib (Fig. 1e). This finding is consistent with previous studies reporting the link between CCNE1 (cyclin E1) overexpression and Palbociclib resistance^45,46. Indeed, we confirmed that there was a significant positive correlation between CCNE1/SKP2 expression and dependency (Fig. 1e). These results suggest a potential synergy between inhibitors of cyclin E and CDK4/6 in cancer treatment.

DCN reveals novel insights into composition of chromatin complexes

We next examined chromatin-related DCN modules in further detail. Consistent with our global analysis (Supplementary Fig. 1), many modules correspond to chromatin biochemical complexes documented in the CORUM database, including EMSY complex (module #21), SIN3–HDAC complex (module #73), G9a/GLP H3K9 methyltransferase complex (module #49), ANCO1–HDAC3 complex (module #62), INO80 complex (module #24) and Integrator complex (module #35) (Fig. 2a). Interestingly, SIN3A and SIN3B were identified in distinct DCN modules (SIN3–HDAC and EMSY, respectively), highlighting a functional divergence between the two SIN3 family members⁴⁷.

**Fig. 2: Deplink reveals novel insights into the composition and interaction of chromatin complexes.**

Some modules contain genes that are not annotated as members of the corresponding chromatin complexes, such as SETD5 in module #62/ANCO1–HDAC3 complex^48,49,50 and C7ORF26 in module #35/Integrator complex (Fig. 2a). In particular, C7orf26 shows similar dependency landscape to INTS10/13/14, but not genes encoding other integrator complex subunits, across 1086 pan-cancer cell lines (Supplementary Fig. 3a). We tested if these genetic co-dependencies represent biochemical interactions. We ectopically expressed FLAG-tagged C7orf26 in HeLa cells and performed co-immunoprecipitation followed by mass spectrometry. A specific interaction between C7orf26 and INTS10/13/14, but not other Integrator complex members, was identified and validated using immunoblotting (Supplementary Fig. 3b–d, Supplementary Data 9). These data indicate that C7orf26 and INTS10/13/14 may form a biochemically distinct subcomplex that is functionally independent of the canonical Integrator complex. In agreement, Deplink analysis revealed that C7orf26-containing module #35 was preferentially essential to solid tumor cell lines exhibiting COSMIC mutational signature 3 (failure of DNA homologous recombination repair) (Fig. 1c and Supplementary Fig. 3e). Therefore, C7orf26/INTS10/13/14 subcomplex may participate in the DNA damage response independently of the canonical function of Integrator complex in transcriptional regulation.

Functional interplay between MLL–MEN1 complex and PRC2 complexes

We speculate that our DCN analysis may also reveal functional relationships between biochemically distinct chromatin complexes. To this end, we focused on module #8, which contains MLL(KMT2A)-MEN1, PRC1, PRC2, KAT7-JADE and MOZ/MORF (KAT6A/B) complexes (Fig. 2a, b). This co-dependency module is also intriguing since it contains well-characterized chromatin complexes with distinct and opposing functions in either transcriptional activation (MLL–MEN1, KAT7, and KAT6A/B) or repression (PRC1, PRC2). Deplink analysis found that genes within module #8 were preferentially essential to blood but not solid cancer cell lines with high levels of H3K27me3 (Fig. 2c, Supplementary Fig. 4a). While it is conceivable that H3K27me3-high cell lines, such as DLBCL cells carrying EZH2 gain-of-function mutations⁵¹, require PRC2 complex for survival⁵², the role of MLL–MEN1 complex in this setting is unclear. We confirmed that DLBCL cell lines with either mutations or overexpression of PRC2 complex members were significantly more dependent on MEN1 and DOT1L for survival (Fig. 2d, e).

EZH2 mutant DLBCL cells display preferential dependency on MLL-MEN1

To validate findings from our DCN and Deplink analysis, we independently performed a CRISPR genetic screen using a chromatin-focused gRNA library⁵³ in a Cas9-expressing EZH2 mutant DLBCL cell line (KARPAS422) (Fig. 3a). We found highly concordant effects of knocking out PRC2 complex members (EZH2/EED/SUZ12), MLL1 and MEN1 on decreasing cell competitive fitness between our screen and the DepMap dataset (Fig. 3b, c). As a comparison, knockout of Module #8 genes had minimal impact on cell fitness in K562 leukemia cell line that is EZH2 wildtype (Supplementary Fig. 4b). The detrimental effect of CRISPR-Cas9-mediated knockout of MEN1 on the proliferation of EZH2 mutant lymphoma lines KARPAS422 and SuDHL4 was also evident in competitive proliferation assays (Fig. 3d), and in xenograft experiments where depletion of MEN1 decreased the incidence of tumor formation (Fig. 3e). Furthermore, we subjected a panel of lymphoma cell lines to a small-molecule inhibitor of MEN1, MI-503⁵⁴, and observed that compared to EZH2 wildtype, EZH2 mutant cells were significantly more sensitive to MEN1 inhibition (Fig. 3f, Supplementary Fig. 4c).

**Fig. 3: CRISPR-Cas9 essentiality screens reveal preferential dependency of MLL–MEN1 in *EZH2*-mutated DLBCL.**

In addition to DLBCL, a subset of multiple myeloma (MM) tumors display globally elevated H3K27me3 levels due to either inactivation of H3K27 demethylases (UTX/KDM6A) or overexpression of PHF19^55,56. H3K27me3-high MM cell lines also show increased sensitivity to inhibition of EZH2^55,56,57. We treated a pair of MM cell lines that are either H3K27me3-high (RPMI-8226) or H3K27me3-low (MM.1S) (Supplementary Fig. 4d) with MEN1 inhibitor (VTP50469)⁵⁸. MEN1 inhibition significantly inhibited the viability of RPMI-8226 but not MM.1S cells in a dose-dependent manner (Fig. 3g). This finding further confirms our Deplink analysis result that the co-dependency between PRC2 and MLL–MEN1 applies broadly to hematopoietic tumors with high levels of H3K27me3 (Fig. 2c).

Expansion of H3K27me3 domains drives redistribution of MLL–MEN1 complex

To investigate how PRC2 and MLL–MEN1 complexes, which are involved in transcriptional repression and activation, respectively, demonstrate cooperative effects on lymphoma cell fitness, we performed CUT&Tag to profile genome-wide distribution of H3K27me3, MEN1, and MLL1 in both EZH2 wildtype (Farage) and mutant (KARPAS422) lymphoma lines. As expected, genome-wide bindings of MEN1 and MLL1 are strongly correlated and enriched at gene promoter regions (Supplementary Fig. 5a, b, Supplementary Data 10). Consistent with previous reports⁵⁹, there was an increase in the number of large (>100 kb) H3K27me3 domains in EZH2 mutant KARPAS422 cells compared to the wildtype Farage cells (Fig. 4a), suggesting that EZH2 gain-of-function mutation promotes the expansion of H3K27me3 domains. In contrast, the number of total or promoter-specific MEN1 and MLL1 peaks was markedly reduced in KARPAS422 cells compared to Farage cells (Fig. 4b, Supplementary Fig. 5c, d). By integrating these datasets, we found that the changes in H3K27me3 were significantly negatively correlated with changes in MLL1/MEN1 bindings (Fig. 4c, Supplementary Fig. 5e). Furthermore, these changes were associated with changes in gene expression, with promoters that gained H3K27me3 and lost MLL1/MEN1 bindings exhibited decreased expression of corresponding genes (Fig. 4c, Supplementary Fig. 5e). As representative examples, H3K27 was hypermethylated at genomic loci encompassing MYB and IRF5 genes in KARPAS422 cells, accompanied by abolished MEN1/MLL1 binding and silencing of these two genes (Fig. 4d). At the CD24 loci, on the other hand, loss of H3K27me3 from the promoter region was correlated with MEN1/MLL1 binding and CD24 transcription. These results prompted us to consider the possibility that the expansion of H3K27me3 by hyperactive mutant EZH2 prohibits MLL1/MEN1 binding, resulting in a genome-wide redistribution and “concentration” of MLL–MEN1 complex at a limited number of accessible gene promoters. Supporting this notion, MLL1/MEN1 promoter peaks present in both KARPAS422 and Farage cells showed markedly higher signal abundance (Fig. 4e) and higher expression of their associated genes (Fig. 4f) in KARPAS422 cells.

**Fig. 4: Expansion of H3K27me3 domains in *EZH2*-mutant cells drives redistribution of MLL–MEN1 complex.**

To determine if similar H3K27me3-driven redistribution of MLL–MEN1 complex binding can be observed in MM cells, we performed epigenomic profiling of H3K27me3-high RPMI-8226 and H3K27me3-low MM.1S cells. We found an increased number of H3K27me3 large domains in RPMI-8226 cells, which was accompanied by a decrease in the number of MEN1 but not MLL1 peaks (Supplementary Fig. 6a, b). Importantly, both genome-wide analysis and inspection at representative loci revealed higher signal abundance for MEN1/MLL1 promoter peaks in RPMI-8226 cells (Supplementary Fig. 6c, d), mirroring that in KARPAS422 cells.

We next determined if higher levels of MLL1/MEN1 binding in H3K27me3-high DLBCL or MM cells can be reversed with depletion of H3K27me3. To this end, we treated KARPAS422 cells with an EZH2 inhibitor (EPZ-6438). EZH2 inhibition effectively abolished genome-wide H3K27me3 enrichment, and regained MLL1/MEN1 bindings at H3K27me3-high regions in untreated cells (Supplementary Fig. 7a, b). Concomitantly, we observed decreased signal abundance of shared MLL1/MEN1 peaks (Fig. 4e, Supplementary Fig. 7a, b). A similar “titration” of MLL1/MEN1 binding was also observed in RPMI-8226 cells following EPZ-6438 treatment (Supplementary Fig. 6c). Together, these results suggest a causal function of H3K27me3 in regulating genome-wide patterns of MLL1/MEN1 binding.

EZH2 mutant lymphoma cells are addicted to MLL–MEN1 target genes

We reason that expression of genes bound by high levels of MEN1/MLL1 is also hypersensitive to MEN1 inhibition, which could underline the preferential toxicity of MEN1 inhibitor to EZH2 mutant lymphoma cell lines. We assessed MEN1/MLL1 and H3K27me3 enrichment and gene expression in Farage and KARPAS422 cells before and after the treatment of MEN1 inhibitors MI-503 and VTP50469 using CUT&Tag and RNA-seq, respectively. We observed a gain of H3K27me3 surrounding MLL1/MEN1 binding peaks following MEN1 inhibitor treatment in both KARPAS422 and Farage cells (Supplementary Fig. 7c–e), suggesting that the opposition between MLL-MEN1 complex and H3K27me3 is bidirectional. MEN1 inhibitor treatment resulted in a more pronounced loss of MEN1 and MLL1 bindings to gene promoters in KARPAS422 cells compared to Farage cells (Fig. 5a, Supplementary Fig. 8a, b). This preferential loss of MEN1/MLL1 binding was associated with significantly reduced transcription of MLL1/MEN1-bound genes and a larger impact on the transcriptome of KARPAS422 cells following MI-503 exposure (Fig. 5a, Supplementary Fig. 8c, Supplementary Data 11). We identified 60 genes that were bound by MEN1 and significantly downregulated after MI-503 treatment in KARPAS422 cells. Approximately one-third (21) of these genes were also downregulated by MI-503 in another EZH2 mutant lymphoma cell line DB. Importantly, genes that were downregulated following MEN1 inhibition showed significantly lower dependency scores in genome-wide knockout screens of KARPAS422 and DB cells, suggesting that MEN1-regulated genes are more essential to the fitness of EZH2 mutant DLBCL cell lines (Fig. 5b).

**Fig. 5: *EZH2* mutant lymphomas are addicted to MLL–MEN1-regulated oncogenic gene expression.**

Gene Ontology (GO) analysis of MEN1-regulated genes revealed a significant enrichment of mTORC1 signaling and unfolded protein response (UPR) pathways (Fig. 5c). Indeed, levels of phosphorylated S6, a marker of mTORC1 signaling, were higher in KARPAS422 cells and suppressed by MI-503 treatment (Fig. 5d). Closer examination of individual genes confirmed results of genome-wide analysis: compared to Farage cells, these genes exhibited decreased occupancy of H3K27me3, increased binding of MEN1 at their promoters, and increased transcription in KARPAS422 cells (Fig. 5e). Upon MEN1 inhibition, however, there was a more pronounced decrease in the MLL1/MEN1 binding at, and the expression of, these mTORC1 signaling pathway genes in EZH2 mutant KARPAS422 and DB cells (Fig. 5f, Supplementary Fig. 8d). Notably, two genes showing the largest decrease in expression upon MEN1 inhibition, PSAT1 and SLC1A5, were significantly associated with worse prognosis in a DLBCL patient cohort⁶⁰ (Supplementary Fig. 8e). Furthermore, compared to Farage cells, KARPAS422 cells were more sensitive to pharmacologic inhibition of mTORC1 by rapamycin (Fig. 5g). Conversely, knockout of TSC1, an upstream negative regulator of mTORC1 signaling, partially rescued the decreased viability of KARPAS422 cells following VTP50469 treatment (Supplementary Fig. 8f), supporting a functional role of mTORC1 signaling in EZH2 mutant DLBCL cells’ hypersensitivity to MEN1 inhibition. Taken together, it appears that the redistribution of the MLL–MEN1 complex by H3K27me3 expansion drives a higher and MEN1-dependent expression of genes involved in key oncogenic pathways (e.g. mTORC1 signaling) that are required for the growth of EZH2 mutant DLBCL cells (Fig. 5h).

MEN1 inhibitor synergizes with Tazemetostat in inhibiting growth of EZH2 mutant DLBCL

Tazemetostat (EPZ-6438) is an EZH2 inhibitor recently approved for relapsed or refractory follicular lymphomas carrying EZH2 gain-of-function mutations⁶¹. However, the clinical benefit of Tazemetostat as a single agent for EZH2 mutated DLBCL is modest⁶². We, therefore, tested if combined inhibition of MEN1 (MI-503 or VTP50469) and EZH2 (EPZ-6438) could result in synergistic effects on tumor transcriptome and growth in vitro and in vivo. RNA-seq analysis indicated that while VTP50469 and EPZ-6438 each had modest impact on gene expression of KARPAS422 cells, co-treatment of both drugs induced >2400 differentially expressed genes (914 upregulated and 1513 downregulated) which were functionally enriched for Myc targets, E2F targets as well as mTORC1 signaling (Fig. 6a, Supplementary Fig. 9a). This synergy on transcriptome remodeling was also evident in another EZH2 mutant DLBCL line SuDHL10, where VTP50469 had minimal impact as single treatment yet together with EPZ-6438 caused differential expression of >600 genes (Fig. 6b, Supplementary Fig. 9b). Consistently, a synergistic impact on decreasing cell viability between EZH2 and MEN1 inhibitors was observed in KARPAS422 and SuDHL10 but not Farage cells (Fig. 6c, Supplementary Fig. 9c). A similar trend was found in MM, where co-treatment of EPZ-6438 + VTP50469 decreased the viability of H3K27me3-high RPMI-8226 cells but not H3K27me3-low MM.1S cells (Supplementary Fig. 9d). We also tested the efficacy of combination therapy in DLBCL cells that have acquired resistance to EZH2 inhibition. We continuously exposed the EZH2-mutant lymphoma cell line DB to increasing concentrations of EPZ-6438 for 6 weeks to develop an EPZ-6438-resistant line (DB-r) (Supplementary Fig. 9e). Combined treatment of MI-503 and EPZ-6438 displayed a strong synergistic effect on inhibiting the proliferation of DB-r cells (Fig. 6d). Finally, to test the efficacy of this combination therapy in vivo, we assessed combined treatment of EPZ-6438 and the orally available VTP50469⁵⁸ on tumor growth using the EZH2 mutant SuDHL10 xenograft model. Compared to a single agent, combined treatment with EPZ-6438 and VTP50469 demonstrated markedly more potent efficacy in reducing tumor burden and extending animal survival (Fig. 6d, e). Taken together, these results suggest that the co-dependency between PRC2 and MLL–MEN1 complexes represents a potential therapeutic vulnerability for EZH2 mutant DLBCLs.

**Fig. 6: Inhibition of both MEN1 and EZH2 in *EZH2*-mutated DLBCL.**

Discussion

In this study, we combined CRISPR-Cas9 essentiality screen datasets with molecular profiles of cancer cell lines to enable the large-scale discovery of context-defined genetic interactions between chromatin pathways. Our approach uncovered functional intra- and inter-complex interactions. We also developed a computational platform for the identification of context-specific genetic dependency by integrating various histological and molecular features including cancer type, genetic mutation, copy number variation, chromatin modification, gene expression, and drug sensitivity (https://github.com/seanchen607/deplink). Collectively, this atlas provides a resource (http://www.chaolulab-database.com/) to the research community to facilitate hypothesis generation/testing and evaluate potential chromatin-based therapeutic strategies for cancer and other human diseases.

Our DCN analysis is based on the idea that knockout of functionally related genes would produce similar phenotypic effects (e.g. fitness) across multiple conditions (e.g. cell lines), thus enabling the identification of functional interactions by measuring the fitness effects of single-gene perturbations across a panel of genetically and phenotypically heterogeneous cancer cell lines. Compared to recent studies^{29,30,31,63,64,65}, the current work uses distinct approaches for module calling and network visualization, and builds the pan-cancer genetic co-dependency network from the largest CRISPR-Cas9 fitness screening dataset to date (Supplementary Data 12). We observed that direct physical interactions among complex subunits were often reflected by a strong correlation (Pearson’s r > 0.4), such as the co-dependency between C7orf26 and INTS10/13/14 (r = 0.52, 0.56, and 0.63, respectively). On the other hand, a moderate correlation (Pearson’s r between 0.34 and 0.4) may reflect an indirect interaction between two proteins functioning in the same pathway. This is the case for interactions between biochemically distinct chromatin complexes, such as MLL–MEN1 and PRC2 complex members. As proteins usually function in a complex and one complex could further functionally interact with other complexes, we came up with a complex-centric strategy using a two-step cutoff for correlation coefficient to enhance the power of discovering novel functional modules. We first use a more stringent cutoff (Pearson’s r > 0.4) to identify complex-level interactions, and then use a less stringent cutoff (Pearson’s |r | > 0.34) as the network cutoff to help capturing/visualizing the less strong genetic/functional interactions between members of distinct complexes. Our current pipeline has a good balance between sensitivity and specificity for predicting curated interactions in the BioGRID database, which collects experimentally validated biochemical and genetic interactions (Supplementary Fig. 10). The current co-dependency mapping approach likely lacks enough specificity to identify bona fide interactions among gene pairs with weak correlation (Pearson’s r < 0.34), as the number of interactions rises rapidly. This limitation may be circumvented with a secondary screen using complementary approaches, such as conventional pair-wise gene interaction analysis.

A unique feature of our study is the combination of co-dependency mapping with molecular features of the cell lines analyzed (Deplink). This integrative analysis allowed us to address why a module of co-dependent genes is specifically essential to a subset of cell lines but not others, thereby inferring the potential function of the module. For example, our DCN analysis predicted that an uncharacterized gene C7orf26 functionally interacts with INTS10/13/14 but not genes encoding other members of the Integrator complex (module #35). This genetic interaction was validated biochemically, as C7orf26 binds to INTS10/13/14 but not the rest of the Integrator complex. These findings are consistent with recent reports that INTS10/13/14 constitute a biochemically distinct subcomplex of Integrator^64,66,67,68. Our Deplink analysis shows that module #35 is selectively essential to solid tumor cell lines harboring COSMIC mutational signature 3 (failure of DNA homologous recombination repair) (Supplementary Fig. 3e). Therefore, we speculate that C7orf26/INTS10/13/14 subcomplex may participate in the DNA damage response independently of the canonical function of Integrator complex in transcriptional regulation. As another example, module #46, consisting of the INO80 chromatin remodeling complex, is required for the survival of blood cancer cell lines with histone hyperacetylation (Fig. 1d). While the causality of this correlation remains to be investigated, it is consistent with a recent study showing that INO80 complex is a central component of metabolic homeostasis that influences histone acetylation⁶⁹. Notably, we found that a number of modules showed correlations with copy numbers and/or expression levels of module genes (Supplementary Fig. 2e, f and Supplementary Data 13, 14). We speculate that these modules represent cases of oncogene addition, where module genes are amplified/overexpressed and required for tumor cell growth. Furthermore, while currently, it is statistically underpowered to construct co-dependency networks separately for each cancer type due to the limited number of cell lines, future efforts along this line will provide key insights into cancer type-specific gene co-functionalities.

Our DCN analysis also nominates a functional co-dependency between PRC2 and MLL-MEN1, two well-characterized and transcriptionally opposing chromatin-modifying complexes. Indeed, we demonstrated that genetic or pharmacologic inhibition of MEN1 could phenocopy the effects of PRC2 knockout in inhibiting the proliferation of EZH2 mutant and/or H3K27me3-high DLBCL and MM cells (Fig. 3). The therapeutic implication of this functional interplay between PRC2 and MLL–MEN1 is likely to extend beyond lymphoid malignancies^53,70. For example, MLL rearrangement is a common event in pediatric leukemias, and MLL-rearranged leukemias exquisitely depend on MLL1/MEN1 for survival. In several mouse models, PRC2 has also been demonstrated to be essential for the maintenance of MLL-rearranged leukemias^71,72. Furthermore, both MEN1 and PRC2 were identified as the top hits in a genome-wide screen for regulators of MHC-I expression as potential strategies to augment cancer immunotherapy⁷³. Importantly, a combination of highly specific and potent inhibitors of PRC2 and MEN1, which are either FDA-approved or in clinical trials to treat distinct cancer types, demonstrated remarkable efficacy in the treatment of DLBCL in xenograft study (Fig. 6). These encouraging preclinical results warrant future studies to design strategies of rationally repurposing and/or combining PRC2 and MEN1 inhibitors for synergistic anti-cancer effects.

MLL–MEN1 and PRC2 complexes are the mammalian homologs of the trithorax (TrxG) and polycomb group (PcG) proteins in Drosophila, respectively. TrxG and PcG regulate the developmental programs and the expression of Hox genes in an antagonistic manner⁷⁴. Consistently, we observed that regions with increased levels of H3K27me3 in EZH2 mutant cells showed reduced MLL1 and MEN1 binding (Fig. 4). Furthermore, inhibition of EZH2 can lead to increased MLL–MEN1 binding, while inhibition of MEN1 increases H3K27me3 at surrounding regions (Supplementary Fig. 7). However, perturbing the local opposition between PRC2 and MLL–MEN1 appears to induce a complex redistribution of these epigenetic regulators at the genome-wide level, as we found that the exclusion of MLL1/MEN1 from H3K27me3-high regions promoted a re-localization and strong enrichment of these proteins at a limited number of H3K27me3-low regions, thereby elevating the expression of corresponding genes. These genes are implicated in oncogenic pathways such as mTORC1 signaling, which has been linked to the development and drug resistance of EZH2 mutant DLBCL⁷⁵, and their overexpression predicted a worse prognosis for DLBCL patients. Therefore, we propose that in addition to the described function of silencing B cell differentiation genes⁵⁹, EZH2 gain-of-function mutations may also activate an MLL–MEN1-dependent oncogenic gene expression program to facilitate DLBCL pathogenesis (Fig. 5h). Notably, while this notion provides a potential mechanistic basis for the observed co-dependency between PRC2 and MLL–MEN1, it is also possible that these two complexes co-regulate the expression of bivalent genes such as MHC-I^73,76 in a cooperative manner.

In summary, our work highlights the potential of integrating parallel genetic screens with molecular phenotyping for high-throughput, context-defined genetic interaction discovery and analysis, which provides a powerful alternative to conventional biochemical approaches in the study of chromatin crosstalk. Our mechanistic investigation into the essentiality of the MLL–MEN1 complex in EZH2-mutated lymphomas provides one example of cancer-associated mutations in chromatin enzymes that creates a synthetic dependency on its interacting trans-regulatory pathway. We believe that further exploration of the resource presented here will uncover additional regulatory mechanisms involved in the progression of cancer epigenome and pave the way to new precision epigenetic therapeutic strategy.

Methods

Dependency correlation network analysis

The gene effect dataset of CRISPR-Cas9 essentiality screens in 1086 pan-cancer cell lines (gene effect scores derived from CRISPR knockout screens published by Broad’s Achilles and Sanger’s SCORE projects, release public 2022q2)^77,78,79 was downloaded from Cancer Dependency Map portal (DepMap, https://depmap.org/portal/). The Pearson correlation scores were calculated for gene effect scores in pan-cancer cell lines between each two of the genes using function ‘cor’ in R. To validate the power of identifying complex-level interactions, human core complexes information was downloaded from CORUM database (https://mips.helmholtz-muenchen.de/corum/)³⁵. The interactions mapped to reported protein interactions between CORUM core complex members and randomly selected interactions in equal numbers were generated as observed and simulated datasets, respectively. ROC curves were plotted using an R package ‘pROC’. The modularity of interactions from the observed and simulated datasets was assigned as 1 and 0, respectively. The sensitivity and specificity (AUC score) of identifying CORUM complex-level interactions using different cutoffs were determined by the p-values from T-test between the correlation scores from complex-level interactions and those from global interactions. The heatmap of the Pearson correlation matrix of genes that have at least one strong interaction with other genes (matrix cutoff: Pearson’s r > 0.4) was generated using the R package pheatmap (Pretty Heatmaps v1.0.10, parameters: clustering_method = ‘ward.D’, clustering_distance_cols = ‘euclidean’). Based on the correlation matrix, the dependency correlation network was generated (network cutoff: Pearson’s |r | > 0.34, only top 15 strongest co-dependency interactions for each gene were kept and singletons were removed) and visualized using GeNets (http://apps.broadinstitute.org/genets) and an R package ‘geNet’ (https://github.com/haneylab/geNet), which integrates a machine-learning algorithm Quack that is trained for comparing the global and local biological signal of networks and identifying the optimal network with which to interpret large genomic datasets such as cancer co-dependency relationships from project Achilles⁸⁰. To validate our prediction of genetic interactions, previously identified genetic or physical interaction information was acquired from the BioGRID database (https://downloads.thebiogrid.org/BioGRID)⁸¹.

Deplink

The genetic profile (gene expression, mutations, copy number variation, and chromatin modification) and drug sensitivity (GDSC and PRISM) datasets of pan-cancer cell lines were downloaded from the DepMap portal (https://depmap.org/portal/). Association of genetic dependency with various functional characterization of cell lines was performed by a custom R package ‘deplink’ (https://github.com/seanchen607/deplink, parameters: cutoff.freq = 10, cutoff.percentile = 0.18, cutoff.pvalue = 0.05, cutoff.qvalue = 0.1, cutoff.diff = 0.1, cutoff.fc = 2). For each module in the dependency correlation network, Deplink selects top and bottom cell lines based on their ranking of genetic dependency of the module members and compares their molecular features (chromatin modification, gene expression, genetic mutation, copy number variation, tumor mutation burden, microsatellite instability), various signatures (COSMIC, ISG, EMT, mRNAsi, GSEA hallmark), drug sensitivity and cancer types with those of the rest cell lines. A detailed description of the code’s functionality can be found on its tutorial page (https://seanchen607.github.io/deplink.html). The Stemness feature associated with oncogenic dedifferentiation was measured by mRNA stemness index (mRNAsi) which was calculated using the established approach⁸². Dependency score for each module was calculated as the opposite number of the mean value of the gene effect values of members in this module. The complete output of analysis using Deplink is provided in figshare (https://doi.org/10.6084/m9.figshare.21708425.v1) and can be queried via a searchable database (http://www.chaolulab-database.com/).

Cell lines and cell culture

Human DLBCL cell lines KARPAS-422, DB, SuDHL-10, Farage, HBL-1, SuDHL-4 and SuDHL-5 (a gift from Jennifer E. Amengual), human leukemia cell line K562 (ATCC, #CCL-243) and human MM cell line RPMI-8226 (a gift from High-Throughput Screening Facility at the JP Sulzberger Columbia Genome Center) were maintained in RPMI1640 Medium (Gibco, #21875034) with 10% FBS (Corning, #35-010-CV). Human MM cell lines MM.1S (a gift from Selina Chen-Kiang) was maintained in RPMI1640 Medium (Gibco, #21875034) with 10% FBS (Corning, #35-010-CV), 1% MEM Non-essential amino acids (Thermo Fisher Scientific, #11140-050), 1% Penn/Strep (Thermo Fisher Scientific, #15140-122), 0.4% 1 M HEPES (Thermo Fisher Scientific, #15630-080), 1% 200 mM l-glutamine (Thermo Fisher Scientific, #25030081), 0.00035% 2-Mercaptoethanol (SIGMA, #M3148), and 0.015% 10 N NaOH (SIGMA, #SX0607N). HEK293T (ATCC, #CRL-11268) and HeLa cells (a gift from Alberto Ciccia) were maintained in DMEM GlutaMAX from Gibco with 10% FBS (Corning, #35-010-CV), and 1% Penn/Strep (Thermo Fisher Scientific, #15140-122). None of the cell lines used were authenticated. All cells were supplied with 1X Penicillin-Streptomycin (Sigma-Aldrich) and kept at 37 °C in a 5% CO₂ atmosphere. All cell lines were routinely tested for mycoplasma contamination.

Lentivirus transduction

Lentivirus packaging was performed in HEK293T cells using Lipofectamine 2000 reagent (Invitrogen) in accordance with the manufacturer’s instructions. The medium-containing virus was concentrated using PEG-it Virus Precipitation Solution (System Biosciences). Spin infection was performed at 1500 rpm at 33 °C for 90 min and transduced cell populations were usually selected or sorted 48 h after infection.

Chromatin-focused CRISPR-Cas9 genetic screening and data analysis

Chromatin-focused CRISPR screening was performed as previously described⁵³. Barcoded PCR products were gel purified and sequenced using the Illumina NextSeq500 instrument. FASTQ files were processed and trimmed to retrieve sgRNA target sequences using cutadapt (v4.2). Sequencing reads were aligned to the reference sgRNA library file and counted at the gene level per sample using MAGeCK (v0.5.9.5). The beta score values were used for the final visualization.

CRISPR-Cas9-mediated knockout

To generate the indicated knockout cell lines, cells were first transduced with LentiCas9-Blast construct (Addgene #52962) and selected with blasticidin. Guide RNAs against MEN1 (gRNA sequences indicated in Supplementary Data 15) were cloned into pUiSEPR-puroR-RFP construct and introduced into target cells using lentiviral infection. Puromycin (1 μg/mL) was used to select gRNA+ cells.

Functional assays

For the competitive proliferation assays using sgRNAs, the percentage of sgRNA-expressing cells (RFP+) was measured over time using flow cytometry and normalized to the starting time point (3 days after infection). Data were acquired on LSR Fortessa (BD). Flow cytometry data were analyzed using FlowJo software (V10). For proliferation assay, cell lines were plated in triplicate in 96-well plates at low density (10,000 cells per well) for each condition. Cells were treated using EZH2 inhibitor EPZ-6438 (MedChemExpress, #HY-13803) and MEN1 inhibitor MI-503 (MedChemExpress, #HY-16925) solubilized in DMSO with indicated concentration for various time points. Following drug treatment, cell viability was determined using Cell Counting Kit-8 for Cell Proliferation and Cytotoxicity Assay (Dojindo Molecular Technologies, #DJDB4000X) according to manufacturer’s guidelines.

Immunoblotting and mass spectrometry

Western blot was performed as previously described⁸³. Antibodies used include anti-MEN1 (Bethyl, Cat#A300-105A, Lot#11, 1:2000 dilution), anti-MLL1 (Bethyl, Cat#A300-086A, Lot#6, 1:2000 dilution), anti-H3K27me3 (Cell Signaling, Cat#9733, Clone C36B11, Lot#19, 1:1000 dilution), anti-INTS13 (Bethyl, Cat#A303-575A, 1:1000 dilution), anti-INTS14 (Bethyl, Cat#A303-576A, 1:1000 dilution), anti-Phospho-S6 (Cell Signaling, Cat#4858, Clone D57.2.2E, 1:2000 dilution), anti-S6 (Cell Signaling, Cat#2317, Clone 54D2, 1:1000 dilution), anti-TSC1 (Cell Signaling, Cat#6935, Clone D43E2, 1:1000 dilution), anti-H3 (Abcam, Cat#ab1791, 1:10,000 dilution), and anti-Beta-actin (Abcam, Cat#ab8224, 1:2000 dilution). Antibodies were all validated by Western blot using the manufacturer’s data associated with antibodies, and their authentication data. For Immunoprecipitation-based mass spectrometry, 2*10⁷ parental HeLa cells or C7orf26-FLAG-expressing HeLa cells were lysed with HS lysis buffer (50 mM Tris–HCl pH 7.9, 500 mM NaCl, 1% NP-40, 20% Glycerol, 0.5 mM PMSF, 5 mM β-mercaptoethanol) on ice for 30 min followed by centrifugation at max g for 20 min in 4 °C centrifuge. Soluble extracts were diluted with BC-300 buffer (20 mM Tris–HCl pH 7.9, 300 mM KCl, 20% Glycerol, 0.1 mM EDTA, 0.5 mM PMSF, 5 mM β-mercaptoethanol) and incubated with 37 μL ANTI-FLAG® M2 Affinity Gel (Sigma, Cat#A2220) overnight at 4 °C with rotation. Beads were washed five times with BC-300 buffer + 0.1% NP-40 and one time with BC-100 buffer (20 mM Tris–HCl pH 7.9, 100 mM KCl, 20% glycerol, 0.1 mM EDTA, 0.5 mM PMSF, 5 mM β-mercaptoethanol) + 0.1% NP-40. 8 M Urea in 20 mM Tris–HCl pH 8.0 was used for elution. Mass spectrometry for protein identification was performed at Columbia University Herbert Irving Comprehensive Cancer Center (HICCC) Proteomics Core. Data analysis was performed with Scaffold Proteome Software.

Xenograft studies

Animal experiments were conducted in accordance with and with the approval of, the Columbia University Institutional Animal Care and Use Committee (IACUC). All mice were housed under specific-pathogen-free (SPF) conditions under controlled temperature (20–26 °C) and humidity (40–70%) with a 12 h/12 h light/dark cycle, following the guideline of the Columbia University animal facility. For the SuDHL-4 xenograft study, ten million SuDHL-4 cells expressing Cas9 and either control or MEN1 sgRNA were mixed with Matrigel and inoculated subcutaneously into the flank of 6–8 weeks old male athymic nu/nu mice purchased from Jackson Lab. The incidence of xenograft tumor formation was monitored and counted until animals were sacrificed at day 90 post-inoculation. For the SuDHL-10 xenograft study, 6-week-old female NOD-scid IL2Rgamma^null (NSG) mice were purchased from The Jackson Laboratory. Five million SuDHL-10 cells were subcutaneously injected into the right flank. Tumor growth was measured twice a week after tumor formation. Tumor volume was calculated by volume = (width² × length)/2. Vehicle, 125 mg/kg EPZ-6438 (0.5% NaCMC and 0.1% Tween 80), 20 mg/kg VTP50469 (5% DMSO, 40% PEG400, 5% Tween 80 and 50% Saline) or combo were administrated to mice via oral gavage twice a day for 28 days. Mice were sacrificed when tumor size reached 2000 mm³. All mice were humanely euthanized by carbon dioxide inhalation, followed by cervical dislocation. Tumor growth curves and survival curves were generated with GraphPad Prism 9. Log-rank (Mantel–Cox) test was used to calculate statistical significance.

RNA extraction and sequencing

RNA was harvested using a Trizol reagent. Illumina TruSeq RNA Sample Prep Kit (Cat#FC-122-1001) was used with 1 μg of total RNA for the construction of sequencing libraries. RNA libraries were prepared with ribosomal RNA depletion according to instructions from the manufacturer (NEB). Paired-end sequencing was performed on the Illumina NextSeq 550 sequencer.

CUT&Tag experiment and sequencing

CUT&Tag was performed as described previously⁸⁴. In brief, 1 × 10⁵ cells were washed with 1 ml of wash buffer (20 mM HEPES pH 7.5, 150 mM NaCl, 0.5 mM Spermidine (Sigma-Aldrich), 1× Protease inhibitor cocktail (Roche) once. Concanavalin A-coated magnetic beads (Bangs Laboratories) were washed twice with binding buffer (20 mM HEPES pH 7.5, 10 mM KCl, 1 mM MnCl₂, 1 mM CaCl₂). 10 μL/sample of beads were added to cells and incubated at room temperature for 15 min. Beads-bound cells were resuspended in 100 μL of antibody buffer (20 mM HEPES pH 7.5, 150 mM NaCl, 0.5 mM Spermidine, 0.05% Digitonin (Sigma-Aldrich), 2 mM EDTA, 0.1% BSA, 1× Protease inhibitor cocktail) and incubated with anti-MEN1 (Bethyl, Cat#A300-105A, Lot#11, 1:50 dilution), anti-MLL1 (Bethyl, Cat#A300-086A, Lot#6, 1:50 dilution), anti-H3K27me3 (Cell Signaling, Cat#9733, Clone C36B11, Lot#19, 1:100 dilution), or normal rabbit IgG (Cell Signaling, Cat#2792, Lot#9, 1:100 dilution) at 4 °C overnight on nutator. After being washed once with Dig-wash buffer (20 mM HEPES pH 7.5, 150 mM NaCl, 0.5 mM Spermidine, 0.05% Digitonin, 1× Protease inhibitor cocktail), beads-bound cells were incubated with 1 μL Guinea pig anti-rabbit secondary antibody (Antibodies Online) and 2 μL Hyperactive pA-Tn5 Transposase adapter complex made in-house in 100 μL Dig-300 buffer (20 mM HEPES pH 7.5, 0.5 mM Spermidine, 1× Protease inhibitor cocktail, 300 mM NaCl, 0.01% Digitonin) at room temperature for 1 h. Cells were washed three times with Dig-300 buffer to remove unbound antibody and Tn5 and then resuspended in 300 μL of tagmentation buffer (20 mM HEPES pH 7.5, 0.5 mM Spermidine, 1× Protease inhibitor cocktail, 300 mM NaCl, 0.01% Digitonin, 10 mM MgCl2) and incubated at 37 °C for 1 h. 10 μL of 0.5 M EDTA, 3 μL of 10% SDS, and 5 μL of 10 mg mL⁻¹ Proteinase K were added to each sample and incubated at 50 °C for 1 h to terminate tagmentation. DNA was purified using a PCR purification kit (QIAGEN) and eluted with 25 μL ddH2O. For library amplification, 21 μL of DNA was mixed with 2 µL of 10 µM Nextera i5 unique index primer (Illumina), 2 µL of 10 µM Illumina Nextera i7 unique index primer (Illumina), and 25 µL NEBNext 2× PCR mix (NEB) and subjected to the following PCR program with lid heat on in a Thermocycler: 72 °C, 5 min; 98 °C, 30 s; 13 cycles of 98 °C, 10 s and 63 °C, 10 s; 72 °C, 1 min and hold at 10 °C. To purify the PCR products, 1.1× volumes of pre-warmed Ampure XP beads (Beckman Coulter) were added and incubated at room temperature for 10 min. Libraries were washed twice with 80% ethanol and eluted in 20 μL of 10 mM Tris–HCl, pH 8. Paired-end sequencing was performed on the Illumina NextSeq 550 sequencer.

RNA-seq data analysis

RNA-seq reads were mapped to the human genome assembly hg38 using HISAT2 (v2.1.0). The mapped reads count of each gene was measured by featureCounts (v1.6.1). The differential gene expression was calculated and visualized by the R packages DESeq2 (v1.28.0) and ggplot2 (v3.2.1), respectively. We performed hierarchical clustering on gene expression profiles of samples using the R package pheatmap (Pretty Heatmaps v1.0.10, parameters: clustering_method = ‘ward.D’, clustering_distance_cols = ‘euclidean’, and cutree_cols = 2). Gene set enrichment analyses (GSEA) were performed by using GSEA software (v4.1.0). Pathway enrichment analysis based on The Molecular Signatures Database (MSigDB) hallmark gene set collection was performed using Enrichr (https://maayanlab.cloud/Enrichr).

CUT&Tag data analysis

CUT&Tag reads were mapped to the human genome assembly hg38 using HISAT2 (v2.1.0). Potential PCR duplicates were removed by the function “MarkDuplicates” (parameter: REMOVE_DUPLICATES = true) of Picard (v2.23.1). Broad H3K27me3 peaks were called using SICER2 (parameters: -w 10000 -g 30000 -fdr 0.01) with IgG input as control. Narrow MEN1 and MLL1 binding peaks were called using macs2 (parameters: -q 1e-4 --max-gap 2000 --keep-dup 1) and SICER2 (parameters: -w 100 -g 300). The peaks are annotated using an R package “ChIPseeker”. The overlap between called peaks was identified by the function “map” of bedtools (v2.27.1). The CUT&Tag read counts in promoter regions were measured by featureCounts (v1.6.1). Genomic enrichment of CUT&Tag signals was visualized using IGV. We performed hierarchical clustering on promoter signal enrichment of samples using the R package pheatmap (Pretty Heatmaps v1.0.10, parameters: clustering_method = ‘ward.D’, clustering_distance_cols = ‘euclidean’, and cutree_cols = 2). The signal distribution near promoter regions was measured and visualized by the functions “computeMatrix” and “plotProfile” of deepTools (v 3.3.2).

Statistical analysis

Statistical analysis was performed using Student’s t-test (two-sample equal variance; unpaired; two-tailed distribution) unless stated otherwise.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

The publicly available gene effect dataset of CRISPR-Cas9 essentiality screens in 1086 pan-cancer cell lines (gene effect scores derived from CRISPR knockout screens published by Broad’s Achilles and Sanger’s SCORE projects, release 2022q2) used in this study is available in the Cancer Dependency Map portal (DepMap) [https://depmap.org/portal]^77,78,79. The publicly available human core complexes data used in this study are available in the CORUM database [https://mips.helmholtz-muenchen.de/corum]³⁵. The publicly available genetic or physical interaction data used in this study are available in the BioGRID database [https://downloads.thebiogrid.org/BioGRID]⁸¹. The pan-cancer genetic co-dependency networks built from CRISPR-Cas9-based screening datasets in recent studies are publicly available in their supplementary materials and figshare [https://figshare.com/s/35a82ed1e48d0ec4e9e4]^29,30,31,65. The publicly available genome-wide binding data of INO80 complex and various chromatin features used in this study are available in the NCBI GEO database under accession code GSE97411⁴⁰. The publicly available survival information and gene expression data of a DLBCL patient cohort are available in the supplementary materials of a previous report and GDC Data Portal [https://gdc.cancer.gov/about-data/publications/DLBCL-2018]⁶⁰. The complete output of analysis using Deplink is provided in figshare [https://doi.org/10.6084/m9.figshare.21708425.v1] and can be queried via a searchable database (http://www.chaolulab-database.com). The raw sequencing data generated in this study (CUT&Tag, RNA-seq, and CRISPR-Cas9 genetic screening) are available in the NCBI GEO database under accession code GSE183487. The mass spectrometry raw data generated in this study are available in the ProteomeXchange member PRIDE database under accession code PXD033140. Source data are provided with this paper.

Code availability

A custom R package ‘deplink’ dedicated to the association of genetic dependency with various cell line molecular features, as well as example data are accessible via GitHub at the following address: https://github.com/seanchen607/deplink, including a readme.txt file providing instructions for installing and running the software. A detailed tutorial for software’s installing, running, and the expected output is exhibited on a GitHub page (https://seanchen607.github.io/deplink.html).

References

Allis, C. D. & Jenuwein, T. The molecular hallmarks of epigenetic control. Nat. Rev. Genet. 17, 487–500 (2016).
Article CAS PubMed Google Scholar
Valencia, A. M. & Kadoch, C. Chromatin regulatory mechanisms and therapeutic opportunities in cancer. Nat. Cell Biol. 21, 152 (2019).
Article CAS PubMed PubMed Central Google Scholar
Fischle, W., Wang, Y. & Allis, C. D. Histone and chromatin cross-talk. Curr. Opin. Cell Biol. 15, 172–183 (2003).
Article CAS PubMed Google Scholar
Suganuma, T. & Workman, J. L. Crosstalk among histone modifications. Cell 135, 604–607 (2008).
Article CAS PubMed Google Scholar
Du, J., Johnson, L. M., Jacobsen, S. E. & Patel, D. J. DNA methylation pathways and their crosstalk with histone methylation. Nat. Rev. Mol. Cell Biol. 16, 519–532 (2015).
Article CAS PubMed PubMed Central Google Scholar
Lee, J.-S., Smith, E. & Shilatifard, A. The language of histone crosstalk. Cell 142, 682–685 (2010).
Article CAS PubMed PubMed Central Google Scholar
Ooi, S. K. T. et al. DNMT3L connects unmethylated lysine 4 of histone H3 to de novo methylation of DNA. Nature 448, 714–717 (2007).
Article ADS CAS PubMed PubMed Central Google Scholar
Baubec, T. et al. Genomic profiling of DNA methyltransferases reveals a role for DNMT3B in genic methylation. Nature 520, 243–247 (2015).
Article ADS CAS PubMed Google Scholar
Weinberg, D. N. et al. The histone mark H3K36me2 recruits DNMT3A and shapes the intergenic DNA methylation landscape. Nature 573, 281–286 (2019).
Article ADS CAS PubMed PubMed Central Google Scholar
Weinberg, D. N. et al. Two competing mechanisms of DNMT3A recruitment regulate the dynamics of de novo DNA methylation at PRC1-targeted CpG islands. Nat. Genet. 53, 794–800 (2021).
Article CAS PubMed PubMed Central Google Scholar
Li, Y., Chen, X. & Lu, C. The interplay between DNA and histone methylation: molecular mechanisms and disease implications. EMBO Rep. 22, e51803 (2021).
Article CAS PubMed PubMed Central Google Scholar
Blackledge, N. P. & Klose, R. J. The molecular principles of gene regulation by Polycomb repressive complexes. Nat. Rev. Mol. Cell Biol. https://doi.org/10.1038/s41580-021-00398-y (2021).
Yuan, W. et al. H3K36 methylation antagonizes PRC2-mediated H3K27 methylation. J. Biol. Chem. 286, 7983–7989 (2011).
Article CAS PubMed PubMed Central Google Scholar
Schmitges, F. W. et al. Histone methylation by PRC2 is inhibited by active chromatin marks. Mol. Cell 42, 330–341 (2011).
Article CAS PubMed Google Scholar
Wilson, B. G. et al. Epigenetic antagonism between polycomb and SWI/SNF complexes during oncogenic transformation. Cancer Cell 18, 316–328 (2010).
Article CAS PubMed PubMed Central Google Scholar
Kia, S. K., Gorski, M. M., Giannakopoulos, S. & Verrijzer, C. P. SWI/SNF mediates polycomb eviction and epigenetic reprogramming of the INK4b-ARF-INK4a locus. Mol. Cell. Biol. 28, 3457–3464 (2008).
Article CAS PubMed PubMed Central Google Scholar
Papillon-Cavanagh, S. et al. Impaired H3K36 methylation defines a subset of head and neck squamous cell carcinomas. Nat. Genet. 49, 180–185 (2017).
Article CAS PubMed PubMed Central Google Scholar
Streubel, G. et al. The H3K36me2 methyltransferase Nsd1 demarcates PRC2-mediated H3K27me2 and H3K27me3 domains in embryonic stem cells. Mol. Cell 70, 371–379.e5 (2018).
Article CAS PubMed Google Scholar
Kim, K. H. et al. SWI/SNF-mutant cancers depend on catalytic and non-catalytic activity of EZH2. Nat. Med. 21, 1491–1496 (2015).
Article CAS PubMed PubMed Central Google Scholar
Gounder, M. et al. Tazemetostat in advanced epithelioid sarcoma with loss of INI1/SMARCB1: an international, open-label, phase 2 basket study. Lancet Oncol. 21, 1423–1432 (2020).
Article CAS PubMed Google Scholar
Kwok, H. S. et al. Drug addiction unveils a repressive methylation ceiling in EZH2-mutant lymphoma. Nat. Chem. Biol. (2023).
Briggs, S. D. et al. Trans-histone regulatory pathway in chromatin: gene silencing. Nature 418, 498–498 (2002).
Article ADS CAS PubMed Google Scholar
McGinty, R. K., Kim, J., Chatterjee, C., Roeder, R. G. & Muir, T. W. Chemically ubiquitylated histone H2B stimulates hDot1L-mediated intranucleosomal methylation. Nature 453, 812–816 (2008).
Article ADS CAS PubMed PubMed Central Google Scholar
Costanzo, M., Baryshnikova, A., Myers, C. L., Andrews, B. & Boone, C. Charting the genetic interaction map of a cell. Curr. Opin. Biotechnol. 22, 66–74 (2011).
Article CAS PubMed Google Scholar
Baryshnikova, A. et al. Quantitative analysis of fitness and genetic interactions in yeast on a genome scale. Nat. Methods 7, 1017–1024 (2010).
Article CAS PubMed PubMed Central Google Scholar
Costanzo, M. et al. A global genetic interaction network maps a wiring diagram of cellular function. Science 353, aaf1420 (2016).
Article PubMed PubMed Central Google Scholar
Braberg, H. et al. Genetic interaction mapping informs integrative structure determination of protein complexes. Science 370, eaaz4910 (2020).
Article CAS PubMed PubMed Central Google Scholar
Horlbeck, M. A. et al. Mapping the genetic landscape of human cells. Cell 174, 953–967.e22 (2018).
Article CAS PubMed PubMed Central Google Scholar
Wainberg, M. et al. A genome-wide atlas of co-essential modules assigns function to uncharacterized genes. Nat. Genet. 53, 638–649 (2021).
Article CAS PubMed PubMed Central Google Scholar
Pan, J. et al. Interrogation of mammalian protein complex structure, function, and membership using genome-scale fitness screens. Cell Syst. 6, 555–568.e7 (2018).
Article CAS PubMed PubMed Central Google Scholar
Kim, E. et al. A network of human functional gene interactions from knockout fitness screens in cancer cells. Life Sci. Alliance 2, e201800278 (2019).
Article PubMed PubMed Central Google Scholar
Tsherniak, A. et al. Defining a cancer dependency map. Cell 170, 564–576.e16 (2017).
Article CAS PubMed PubMed Central Google Scholar
Behan, F. M. et al. Prioritization of cancer therapeutic targets using CRISPR-Cas9 screens. Nature 568, 511–516 (2019).
Article ADS CAS PubMed Google Scholar
Ghandi, M. et al. Next-generation characterization of the Cancer Cell Line Encyclopedia. Nature 569, 503–508 (2019).
Article ADS CAS PubMed PubMed Central Google Scholar
Giurgiu, M. et al. CORUM: the comprehensive resource of mammalian protein complexes-2019. Nucleic Acids Res. 47, D559–D563 (2019).
Article CAS PubMed Google Scholar
Blyth, K., Cameron, E. R. & Neil, J. C. The RUNX genes: gain or loss of function in cancer. Nat. Rev. Cancer 5, 376–387 (2005).
Article CAS PubMed Google Scholar
Shaffer, A. L. et al. IRF4 addiction in multiple myeloma. Nature 454, 226–231 (2008).
Article ADS CAS PubMed PubMed Central Google Scholar
Garraway, L. A. et al. Integrative genomic analyses identify MITF as a lineage survival oncogene amplified in malignant melanoma. Nature 436, 117–122 (2005).
Article ADS CAS PubMed Google Scholar
Tate, J. G. et al. COSMIC: the catalogue of somatic mutations in cancer. Nucleic Acids Res. 47, D941–D947 (2019).
Article CAS PubMed Google Scholar
Runge, J. S., Raab, J. R. & Magnuson, T. Identification of two distinct classes of the human INO80 complex genome-wide. G3 (Bethesda) 8, 1095–1102 (2018).
Article CAS PubMed Google Scholar
Seachrist, D. D., Anstine, L. J. & Keri, R. A. FOXA1: a pioneer of nuclear receptor action in breast cancer. Cancers (Basel) 13, 5205 (2021).
Article CAS PubMed Google Scholar
Ng, C. S., Kasumba, D. M., Fujita, T. & Luo, H. Spatio-temporal characterization of the antiviral activity of the XRN1-DCP1/2 aggregation against cytoplasmic RNA viruses to prevent cell death. Cell Death Differ. 27, 2363–2382 (2020).
Article CAS PubMed PubMed Central Google Scholar
Lamers, M. M., van den Hoogen, B. G. & Haagmans, B. L. ADAR1: ‘Editor-in-Chief’ of cytoplasmic innate immunity. Front. Immunol. 10, 1763 (2019).
Article CAS PubMed PubMed Central Google Scholar
Yang, W. et al. Genomics of Drug Sensitivity in Cancer (GDSC): a resource for therapeutic biomarker discovery in cancer cells. Nucleic Acids Res. 41, D955–D961 (2013).
Article CAS PubMed Google Scholar
Guarducci, C. et al. Cyclin E1 and Rb modulation as common events at time of resistance to palbociclib in hormone receptor-positive breast cancer. NPJ Breast Cancer 4, 38 (2018).
Article PubMed PubMed Central Google Scholar
Min, A. et al. Cyclin E overexpression confers resistance to the CDK4/6 specific inhibitor palbociclib in gastric cancer cells. Cancer Lett. 430, 123–132 (2018).
Article CAS PubMed Google Scholar
Adams, M. K. et al. Differential complex formation via paralogs in the human Sin3 protein interaction network. Mol. Cell. Proteom. 19, 1468–1484 (2020).
Article CAS Google Scholar
Matsumura, Y. et al. Spatiotemporal dynamics of SETD5-containing NCoR-HDAC3 complex determines enhancer activation for adipogenesis. Nat. Commun. 12, 7045 (2021).
Article ADS CAS PubMed PubMed Central Google Scholar
Wang, Z. et al. SETD5-coordinated chromatin reprogramming regulates adaptive resistance to targeted pancreatic cancer therapy. Cancer Cell 37, 834–849.e13 (2020).
Article CAS PubMed PubMed Central Google Scholar
Nakagawa, T. et al. The autism-related protein SETD5 controls neural cell proliferation through epigenetic regulation of rDNA expression. iScience 23, 101030 (2020).
Article ADS CAS PubMed PubMed Central Google Scholar
Morin, R. D. et al. Somatic mutations altering EZH2 (Tyr641) in follicular and diffuse large B-cell lymphomas of germinal-center origin. Nat. Genet. 42, 181–185 (2010).
Article CAS PubMed PubMed Central Google Scholar
McCabe, M. T. et al. EZH2 inhibition as a therapeutic strategy for lymphoma with EZH2-activating mutations. Nature 492, 108–112 (2012).
Article ADS CAS PubMed Google Scholar
Soto-Feliciano, Y. M. et al. A molecular switch between mammalian MLL complexes dictates response to Menin–MLL inhibition. Cancer Discov. 13, 146–169 (2023).
Article CAS PubMed Google Scholar
Borkin, D. et al. Pharmacologic inhibition of the Menin–MLL interaction blocks progression of MLL leukemia in vivo. Cancer Cell 27, 589–602 (2015).
Article CAS PubMed PubMed Central Google Scholar
Ezponda, T. et al. UTX/KDM6A loss enhances the malignant phenotype of multiple myeloma and sensitizes cells to EZH2 inhibition. Cell Rep. 21, 628–640 (2017).
Article CAS PubMed PubMed Central Google Scholar
Ren, Z. et al. PHF19 promotes multiple myeloma tumorigenicity through PRC2 activation and broad H3K27me3 domain formation. Blood 134, 1176–1189 (2019).
Article CAS PubMed PubMed Central Google Scholar
Hernando, H. et al. EZH2 inhibition blocks multiple myeloma cell growth through upregulation of epithelial tumor suppressor genes. Mol. Cancer Ther. 15, 287–298 (2016).
Article CAS PubMed Google Scholar
Krivtsov, A. V. et al. A Menin–MLL inhibitor induces specific chromatin changes and eradicates disease in models of MLL-rearranged leukemia. Cancer Cell 36, 660–673.e11 (2019).
Article CAS PubMed PubMed Central Google Scholar
Béguelin, W. et al. EZH2 is required for germinal center formation and somatic EZH2 mutations promote lymphoid transformation. Cancer Cell 23, 677–692 (2013).
Article PubMed PubMed Central Google Scholar
Schmitz, R. et al. Genetics and pathogenesis of diffuse large b-cell lymphoma. N. Engl. J. Med. 378, 1396–1407 (2018).
Article CAS PubMed PubMed Central Google Scholar
Italiano, A. et al. Tazemetostat, an EZH2 inhibitor, in relapsed or refractory B-cell non-Hodgkin lymphoma and advanced solid tumours: a first-in-human, open-label, phase 1 study. Lancet Oncol. 19, 649–659 (2018).
Article CAS PubMed Google Scholar
Ribrag, V. et al. Interim results from an ongoing Phase 2 Multicenter Study of Tazemetostat, an EZH2 inhibitor, in patients with relapsed or refractory (R/R) diffuse large B-cell lymphoma (DLBCL). Blood 132, 4196–4196 (2018).
Article Google Scholar
Bayraktar, E. C. et al. Metabolic coessentiality mapping identifies C12orf49 as a regulator of SREBP processing and cholesterol metabolism. Nat. Metab. 2, 487–498 (2020).
Article CAS PubMed PubMed Central Google Scholar
Pan, J. et al. Sparse dictionary learning recovers pleiotropy from human cell fitness screens. Cell Syst. S2405-4712(21)00488–9 (2022) https://doi.org/10.1016/j.cels.2021.12.005.
Boyle, E. A., Pritchard, J. K. & Greenleaf, W. J. High-resolution mapping of cancer cell networks using co-functional interactions. Mol. Syst. Biol. 14, e8594 (2018).
Article PubMed PubMed Central Google Scholar
Sabath, K. et al. INTS10-INTS13-INTS14 form a functional module of Integrator that binds nucleic acids and the cleavage module. Nat. Commun. 11, 3422 (2020).
Article ADS CAS PubMed PubMed Central Google Scholar
Pfleiderer, M. M. & Galej, W. P. Structure of the catalytic core of the Integrator complex. Mol. Cell 81, 1246–1259.e8 (2021).
Article CAS PubMed PubMed Central Google Scholar
Replogle, J. M. et al. Mapping information-rich genotype-phenotype landscapes with genome-scale Perturb-seq. Cell 185, 2559–2575.e28 (2022).
Article CAS PubMed PubMed Central Google Scholar
Beckwith, S. L. et al. The INO80 chromatin remodeler sustains metabolic stability by promoting TOR signaling and regulating histone acetylation. PLoS Genet. 14, e1007216 (2018).
Article PubMed PubMed Central Google Scholar
Hemming, M. L. et al. MOZ and Menin–MLL complexes are complementary regulators of chromatin association and transcriptional output in gastrointestinal stromal tumor. Cancer Discov. 12, 1804–1823 (2022).
Article CAS PubMed PubMed Central Google Scholar
Neff, T. et al. Polycomb repressive complex 2 is required for MLL-AF9 leukemia. Proc. Natl Acad. Sci. USA 109, 5028–5033 (2012).
Article ADS CAS PubMed PubMed Central Google Scholar
Shi, J. et al. The Polycomb complex PRC2 supports aberrant self-renewal in a mouse model of MLL-AF9;Nras(G12D) acute myeloid leukemia. Oncogene 32, 930–938 (2013).
Article CAS PubMed Google Scholar
Burr, M. L. et al. An evolutionarily conserved function of polycomb silences the MHC Class I antigen presentation pathway and enables immune evasion in cancer. Cancer Cell 36, 385–401.e8 (2019).
Schuettengruber, B., Bourbon, H.-M., Di Croce, L. & Cavalli, G. Genome regulation by polycomb and trithorax: 70 years and counting. Cell 171, 34–57 (2017).
Article CAS PubMed Google Scholar
Oricchio, E. et al. Genetic and epigenetic inactivation of SESTRIN1 controls mTORC1 and response to EZH2 inhibition in follicular lymphoma. Sci. Transl. Med. 9, eaak9969 (2017).
Article PubMed PubMed Central Google Scholar
Sparbier, C. E. et al. Targeting Menin disrupts the KMT2A/B and polycomb balance to paradoxically activate bivalent genes. Nat. Cell Biol. 25, 258–272 (2023).
CAS PubMed PubMed Central Google Scholar
Meyers, R. M. et al. Computational correction of copy number effect improves specificity of CRISPR-Cas9 essentiality screens in cancer cells. Nat. Genet. 49, 1779–1784 (2017).
Article CAS PubMed PubMed Central Google Scholar
Pacini, C. et al. Integrated cross-study datasets of genetic dependencies in cancer. Nat. Commun. 12, 1661 (2021).
Article ADS CAS PubMed PubMed Central Google Scholar
Dempster, J. M. et al. Chronos: a cell population dynamics model of CRISPR experiments that improves inference of gene fitness effects. Genome Biol. 22, 343 (2021).
Article PubMed PubMed Central Google Scholar
Li, T. et al. GeNets: a unified web platform for network-based genomic analyses. Nat. Methods 15, 543–546 (2018).
Article CAS PubMed PubMed Central Google Scholar
Oughtred, R. et al. The BioGRID interaction database: 2019 update. Nucleic Acids Res. 47, D529–D541 (2019).
Article CAS PubMed Google Scholar
Malta, T. M. et al. Machine learning identifies stemness features associated with oncogenic dedifferentiation. Cell 173, 338–354.e15 (2018).
Article CAS PubMed PubMed Central Google Scholar
Lu, C. et al. Histone H3K36 mutations promote sarcomagenesis through altered histone methylation landscape. Science 352, 844–849 (2016).
Article ADS CAS PubMed PubMed Central Google Scholar
Kaya-Okur, H. S. et al. CUT&Tag for efficient epigenomic profiling of small samples and single cells. Nat. Commun. 10, 1930 (2019).
Article ADS PubMed PubMed Central Google Scholar

Download references

Acknowledgements

We thank Giuseppe Leuzzi, Lei Ding, Liling Wan, and Tahir Sheikh for technical help, and William Seong for experimental support. A schematic diagram of the current model was created using BioRender. These studies were supported by the Columbia Velocity Award (to J.E.A. and C.L.), the Pew-Stewart Scholars for Cancer Research (to C.L.), the V Scholar Grant funded by the V Foundation (to C.L.), and NIH grants (R35GM138181 and R01DE031873 to C.L.; R01CA197774 to A.C.; R01CA172492 to L.P.). F.J.S.R. is an HHMI Hanna Gray Fellow and was supported by the V Foundation for Cancer Research (V2022-028), NCI Cancer Center Support Grant P30-CA1405, the Ludwig Center at MIT (2036636), Koch Institute Frontier Awards (2036648 and 2036642), and the MIT Research Support Committee (3189800). These studies used the resources of the Columbia flow cytometry shared resource and the Judith P. Sulzberger Columbia Genome Center, funded in part through NCI Cancer Center Support Grant P30CA013696.

Author information

Xiao Chen
Present address: Marine College, Shandong University, 264209, Weihai, China
These authors contributed equally: Xiao Chen, Yinglu Li.

Authors and Affiliations

Department of Genetics and Development, Columbia University Irving Medical Center, New York, NY, 10032, USA
Xiao Chen, Yinglu Li, Fang Zhu, Xinjing Xu, John T. McGuire, Dimitris Karagiannis, Varun Sahu, Mustafo Mustafokulov, Alberto Ciccia & Chao Lu
Union Hospital Cancer Center, Tongji Medical College, Huazhong University of Science and Technology, 430022, Wuhan, China
Fang Zhu
Division of Hematology and Oncology, Department of Medicine, Columbia University Irving Medical Center, New York, NY, 10032, USA
Brian Estrella, Manuel A. Pazos II & Jennifer E. Amengual
Institute for Cancer Genetics, Columbia University Irving Medical Center, New York, NY, 10032, USA
Claudio Scuoppo, Laura Pasqualucci & Alberto Ciccia
Department of Pathology and Cell Biology, Columbia University Irving Medical Center, New York, NY, 10032, USA
Claudio Scuoppo & Laura Pasqualucci
David H. Koch Institute for Integrative Cancer Research, Massachusetts Institute of Technology, Cambridge, MA, 02142, USA
Francisco J. Sánchez-Rivera & Yadira M. Soto-Feliciano
Department of Biology, Massachusetts Institute of Technology, Cambridge, MA, 02142, USA
Francisco J. Sánchez-Rivera & Yadira M. Soto-Feliciano
Herbert Irving Comprehensive Cancer Center, Columbia University Irving Medical Center, New York, NY, 10032, USA
Laura Pasqualucci, Alberto Ciccia, Jennifer E. Amengual & Chao Lu

Authors

Xiao Chen
View author publications
You can also search for this author in PubMed Google Scholar
Yinglu Li
View author publications
You can also search for this author in PubMed Google Scholar
Fang Zhu
View author publications
You can also search for this author in PubMed Google Scholar
Xinjing Xu
View author publications
You can also search for this author in PubMed Google Scholar
Brian Estrella
View author publications
You can also search for this author in PubMed Google Scholar
Manuel A. Pazos II
View author publications
You can also search for this author in PubMed Google Scholar
John T. McGuire
View author publications
You can also search for this author in PubMed Google Scholar
Dimitris Karagiannis
View author publications
You can also search for this author in PubMed Google Scholar
Varun Sahu
View author publications
You can also search for this author in PubMed Google Scholar
Mustafo Mustafokulov
View author publications
You can also search for this author in PubMed Google Scholar
Claudio Scuoppo
View author publications
You can also search for this author in PubMed Google Scholar
Francisco J. Sánchez-Rivera
View author publications
You can also search for this author in PubMed Google Scholar
Yadira M. Soto-Feliciano
View author publications
You can also search for this author in PubMed Google Scholar
Laura Pasqualucci
View author publications
You can also search for this author in PubMed Google Scholar
Alberto Ciccia
View author publications
You can also search for this author in PubMed Google Scholar
Jennifer E. Amengual
View author publications
You can also search for this author in PubMed Google Scholar
Chao Lu
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

Conceptualization, X.C., A.C., and C.L.; Methodology, X.C., A.C., and C.L.; Formal Analyses, X.C., Y.L., X.X., F.Z., B.E., M.A.P., J.T.M., D.K., V.S., M.M., C.S., F.J.S.R., Y.M.S.; Investigation, X.C.; Writing—original draft, X.C. and C.L.; Writing—review & editing, X.C., Y.L., F.Z., D.K., A.C., J.E.A., L.P., and C.L.; Supervision, A.C., J.E.A., L.P., and C.L.; Funding acquisition, A.C., J.E.A., L.P., F.J.S.R., and C.L.

Corresponding author

Correspondence to Chao Lu.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Peer Review File

Description of Additional Supplementary Files

Supplementary Data 1-15

Reporting Summary

Source data

Source Data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Chen, X., Li, Y., Zhu, F. et al. Context-defined cancer co-dependency mapping identifies a functional interplay between PRC2 and MLL-MEN1 complex in lymphoma. Nat Commun 14, 4259 (2023). https://doi.org/10.1038/s41467-023-39990-5

Download citation

Received: 01 April 2022
Accepted: 06 July 2023
Published: 17 July 2023
DOI: https://doi.org/10.1038/s41467-023-39990-5

This article is cited by

Large-scale evaluation of the ability of RNA-binding proteins to activate exon inclusion
- Jonathan C. Schmok
- Manya Jain
- Gene W. Yeo
Nature Biotechnology (2024)
Generation of programmable splicing factors using RNA-binding proteins that activate exon inclusion

Nature Biotechnology (2024)

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.