Probing the Cancer Mutational Landscape of KMT2 Regulatory Subunits

ABSTRACT

Members of the Lysine MethylTransferase 2 (KMT2) family are often abnormally expressed and mutated in many cancers. Similarly, several mutations listed in cancer databases map to key functional regions of KMT2 regulatory subunits, such as WD repeat domain 5 (WDR5), Retinoblastoma binding protein 5 (RbBP5), absent‐small‐homeotic‐2‐like (ASH2L), and DumPY‐30 (DPY‐30). In this study, we report the systematic characterization of cancer‐associated mutations that map to regions important for the WDR5/RbBP5/ASH2L/DPY‐30 (WRAD) complex formation. Both binding and thermal stability assays show that several cancer‐related mutations do not affect ASH2L binding to DPY‐30 or RbBP5. A subset of gain‐of‐function mutants highlights the role of long‐range networks of interactions underlying RbBP5 binding by ASH2L. Parallel analysis of RbBP5 mutations shows additional variants that weaken its interactions with WDR5. Finally, systematic mapping of RbBP5 residues interacting with WDR5 defines the optimal WDR5‐binding motif and shows that introducing hydrophobic residues beyond the central VDV sequence increases binding affinity. Overall, these findings reveal surprising gain‐of‐function mutations in ASH2L and provide a framework for targeting this epigenetic hub therapeutically.

Pipeline to characterize the KMT2 cancer mutational landscape starts with querying genomic databases (top left) for mutations mapping to ASH2L, RbBP5, WDR5, and DPY‐30. Subsequent steps include the expression of these mutants in cell (bottom left), their systemic purification and biochemical characterization using isothermal titration calorimetry (bottom centre) and stability by different scanning fluorometry (bottom right).

1. Introduction

Histone methylation is a key epigenetic modification contributing to stem cell maintenance, cell lineage development, mitosis, genomic stability, DNA methylation, and gene expression regulation [1, 2]. This modification occurs primarily on lysine or arginine residues of histones H3 and H4. Among these, methylation of histone H3 on lysine 4 (H3K4) is notably enriched at the transcriptional start sites and is a well‐established marker of active transcription, playing a particularly important role in development and hematopoiesis [3, 4]. The predominant group of enzymes that methylate H3K4 belongs to the Lysine MethylTransferase 2 (KMT2) family. In humans, KMT2 comprises six members, including KMT2A‐F. Each member is composed of a catalytic Su(var)3–9, Enhancer‐of‐zeste and Trithorax (SET) domain, and depending on the KMT2 family members, these enzymes will preferentially mono‐, di‐, or tri‐methylate H3K4 [5]. For several members, their ability to carry out methylation requires the presence of regulatory subunits that include WDR5 (tryptophan‐aspartate [WD] repeat protein 5), RbBP5 (retinoblastoma‐binding protein 5), ASH2L (absent‐small‐homeotic‐2‐like), and DPY‐30 (DumPY‐30), collectively referred to as the WRAD complex (Figure 1A,B) [5, 6, 7, 8, 9]. For example, while KMT2A weakly mono‐methylates H3K4, its association with WRAD stimulates its methyltransferase activity by 600‐fold and enables the formation of H3K4 di‐ and tri‐methylated products [10, 11]. WDR5, RbBP5, and ASH2L constitute the minimal WRA complex required for methylation activity, whereas DPY‐30 increases the catalytic efficiency and specificity of WRAD and KMT2 enzymes for the nucleosome [12]. Therefore, disruption of any subunit results in severe loss of KMT2 activity and broad transcriptional misregulation in cells [11, 13, 14, 15].

FIGURE 1.

Cancer‐related mutations mapping to ASH2L impact complex formation. (A) Schematic representation of KMT2 subunits and their interactions within the complex. Created with BioRender. (B) Cartoon representation of ASH2L^SPRY (gray) bound to the stick representation of RbBP5^ABM (teal) and cartoon of ASH2L^SDI (gray) bound to DPY‐30^DD domain (beige and yellow). Carbon, oxygen, and nitrogen atoms are colored in yellow, red, and blue, respectively. (C) Lollipop plot showcasing the frequency and distribution of specific missense mutations within the SPRY (amino acids 276–510) and SDI (amino acids 510–534) domains of ASH2L Isoform 3. Data were sourced and cross‐referenced from the TCGA and COSMIC databases. The plot highlights mutational hotspots, with the blue circle indicating mutation frequency according to the y‐axis. (D) Bar plot showing the fold change in dissociation constant normalized to the wild‐type value, with decreases in binding affinity shown in red and increases in green. Dotted line highlights a fold‐change of ≥ 1.7. Bars reaching the bottom of the y‐axis are considered to have a ≥ 25‐fold change difference in K _D. (E) Binding affinity (K _D ) of WT ASH2L^SPRY and number of sites (N) calculated from the ITC experiment displayed in (F). (F) ITC measurements of ASH2L^SPRY WT with RbBP5, showing the raw heat data signals (top) and the binding isotherm from the integration of the heat peaks (bottom). The line of best fit is based on a 1:1 binding model. (G) Western blot of Co‐IP of ASH2L^SPRY mutants shows that some mutations negatively impact the binding between ASH2L and RbBP5 (red). Input and Elution samples were probed with anti‐Flag (ASH2L) and anti–RbBP5 antibodies. (H) Western blot of Co‐IP of ASH2L^SPRY‐SDI mutants, where input and elution samples were probed with anti‐flag (ASH2L) and anti‐Myc (DPY‐30) antibodies.

Each WRAD subunit plays a specific role in complex assembly and regulation. ASH2L is composed of a C4 zinc finger and a helix‐wing‐helix (WH) domain that binds DNA, linking methyltransferase activity to chromatin binding [16]. Just after its nuclear localization signal, ASH2L harbors a SPIa and Ryanodinine receptor (SPRY) domain and a Sdc1‐DPY‐30‐Interacting (SDI) motif [17, 18]. ASH2L SPRY domain (ASH2L^SPRY) binds to RbBP5 [17], and its SDI (ASH2L^SDI) motif interacts with DPY‐30 [19]. Another region in ASH2L^SPRY contacts DNA in the nucleosome core particle [20]. The significance of these interactions for complex assembly and KMT2 methyltransferase activity is demonstrated by studies showing that mutations weakening the interaction between RbBP5 and ASH2L greatly reduce the methyltransferase activity of KMT2 enzymes in vitro, in cells, and in vivo [21]. Furthermore, mutations of residues located at the ASH2L‐DPY‐30 interface disrupt complex assembly and reduce methyltransferase activity on nucleosome substrates [19, 22].

WDR5 belongs to the WD40‐repeat protein family, whose members are important scaffolding components of multi‐subunit complexes [23]. In KMT2 complexes, WDR5 binds a short fragment located on the N‐terminus of KMT2 catalytic domains, referred to as WDR5 INteracting motif (WIN) and a shallow hydrophobic cleft on the opposite face of WDR5 directly interacts with the WDR5 binding motif of RbBP5 (RbBP5^WBM) [9, 15, 24, 25]. Together, these interactions ensure proper H3K4 methylation and transcriptional regulation as kinetic studies show that the absence of WDR5 causes a loss of catalytic activity for selected KMT2 enzymes [15, 26].

In this study, we systematically investigated reported cancer‐associated mutations mapping to the interface between WRAD subunits, focusing on how they alter subunit interactions and how they impact complex formation. Our findings reveal both loss‐ and gain‐of‐function mutations within WRAD, outlining the mutational landscape of the WDR5‐RbBP5‐ASH2L‐DPY‐30 axis in cancer and establish a framework for how this epigenetic hub can be targeted therapeutically.

2. Materials and Methods

2.1. Expression and Protein Purification

WDR5 (a.a 22–334) was overexpressed and purified as previously described [27]. ASH2L^SPRY domain (a.a. 286–504 with residues 403–444 replaced with ISGRG) was overexpressed in fusion with a SUMO and a hexahistidine tag [21] (pSMT3‐Ash2L^SPRY). Missense mutations mapping to evolutionarily conserved residues reported in the cancer atlas database were introduced in pSMT3‐ASH2L^SPRY using PCR site‐directed mutagenesis (Agilent Technologies). Briefly, recombinant ASH2L^SPRY WT and mutants were overexpressed in E. coli Rosetta cells (Novagen) using 0.1 mM IPTG for 16 h at 18°C. His‐Sumo tagged proteins were purified by Talon metal affinity chromatography followed by ULP1 cleavage and the cleaved protein was purified by size exclusion chromatography using Superdex 75 (GE Healthcare) in 25 mM Tris pH 7.0, 150 mM NaCl, and 5 mM BME buffer [21].

2.2. RbBP5^ABM Peptide Synthesis

Synthesis for peptide RbBP5^ABM (H‐EYEERESEFDIEDED‐OH) was carried out using solid‐state peptide synthesis under protocols reported by our team [28, 29]. Briefly, Fmoc‐protected amino acids and low‐loading wang resin were purchased from CEM. All peptides were synthesized using microwave‐assisted Fmoc solid‐phase peptide synthesis in a Liberty Blue automated system. The required amount of resin was swelled in DMF for 5 min. Next, Fmoc deprotection was carried out with 20% piperidine at 90°C for 60s. Standard coupling cycles using DIC/Oxyma Pure were run at 90°C for 240 s in each amino acid. Peptides were cleaved from the resin and deprotected with TFA/TIS/EDT/H₂O (92.5/2.5/2.5/2.5% v/v) at 42°C for 30 min and then precipitated in −20°C diethyl ether. Peptide crude products were then dried under vacuum overnight and purified by Agilent 1290 Infinity II Autoscale Preparative LC System with a 30.0 × 100 mm Agilent 5 Prep‐C18 column at 50 mL/min. Peptide analysis was performed in an Agilent 1290 Infinity II Autoscale Preparative LC System with column Kinetex 2.6 μm 100 Å 50 × 2.10 mm at a flow of 0.8 mL/min. Phase A is water/acetonitrile 95/4.9%, and phase B is acetonitrile/water 98/1.9%, both with modifiers of formic acid 0.1% and trifluoroacetic acid 0.01%. The gradient of B starts at 0% and is held isocratic for 0.25 min, then ramps %B to 50% until 14 min, then ramps to %B to 99% at 18.25 min, and held isocratic at this percentage until minute 19. Peptide purity was determined to be over 98%.

2.3. Co‐Immunoprecipitation

Mutations were introduced using PCR site‐directed mutagenesis (Agilent Technologies) using a plasmid designed for the expression of FLAG‐tagged full‐length ASH2L (ASH2L^FL) [16]. WT and mutant constructs were transfected into HEK293 cells and maintained at 37°C with 5% CO₂. After 24 h, the cells were harvested and lysed in RIPA buffer (50 mM Tris pH 7.4, 1% NP‐40, 0.25% sodium deoxycholate, 150 mM NaCl, 1 mM EDTA). 300 μg of each protein extract was incubated with ANTI‐FLAG M2 magnetic beads (Sigma) at 4°C overnight. Beads were washed with RIPA buffer and eluted using 3× FLAG peptide (Sigma). Input and eluted proteins were resolved on a 4%–15% gradient gel, transferred to a PVDF membrane, and blotted with anti‐RbBP5, anti‐FLAG (for ASH2L), or anti‐Myc (for DPY‐30) antibodies.

2.4. Isothermal Titration Calorimetry

ITC experiments were conducted using a VP‐ITC calorimeter (MicroCal) by injecting the WDR5 binding motif (WBM) of RbBP5 (0.7 mM; amino acids 371–380; RbBP5^WBM) into a solution containing WDR5 (0.05 mM) in 20 mM sodium phosphate (pH 7.0), 150 mM sodium chloride, 5 mM B‐mercaptoethanol (BME), and 1 mM EDTA at 19°C. ASH2L binding motif (ABM) of RbBP5 (0.5 mM; amino acids 344–357; RbBP5^ABM) was injected into a solution containing ASH2L^SPRY (0.04 mM) in 20 mM Tris, pH 7.0, with 150 mM NaCl and 5 mM BME at 19°C. The RbBP5 peptides bind with binding stoichiometries (N values) of 0.8–1.0 molecules of peptide per monomer of WDR5 or ASH2L^SPRY. Results for WT were repeated three times, while the mutant results were replicated twice. The Advanced Peptide Synthesis Core Facility synthesized the RbBP5^ABM peptide at the University of Ottawa Heart Institute (See Section 2.2 in the Section 2). RbBP5^WBM peptides were purchased from Genscript USA Inc.

2.5. Differential Scanning Fluorimetry

A mixture of 7.5X SYPRO Orange (Invitrogen), ASH2L^SPRY WT or mutants (30 μM) was preequilibrated in the gel filtration buffer by incubating on ice for 20 min. A melting curve protocol was run on a BioRad CFX96 Touch Real‐Time PCR machine, with temperatures ranging from 10°C to 95°C at a heating rate of 0.5°C per cycle, each lasting 1 s. Data were analyzed using CFX Maestro Software, and the second derivative of the fluorescence signal was used to determine the melting temperature (T_m). Tₘ values are mean ± SD from n = 3 independent experiments.

2.6. Crystallization, Data Collection, and Structure Determination

Purified WDR5 was mixed in equal molar ratio with RbBP5^WBM‐D376N peptide and incubated on ice for 60 min. Diffraction quality crystals were grown in 0.15 M Ammonium citrate and 18% PEG3350. Crystals were sequentially soaked in the mother liquor supplemented with 20% glycerol, harvested, and flash frozen in liquid nitrogen. A full data set was collected on a HomeSource MicroMax‐007 (Rigaku) equipped with an Raxis IV image plate detector. Diffraction data were indexed and scaled using HKL2000 [30]. A molecular replacement solution was found using Phaser and apo‐WDR5 as a starting model [27, 31]. The structure was further refined using iterative cycles of energy minimization and modeling using phenix.refine [32] and Coot [33], respectively. The quality of the structure was assessed using Molprobity [34] (Table 1).

TABLE 1.

Data collection and refinement statistics for the WDR5/RbBP5^D376N complex.

Data collection
Wavelength (Å)	1.54
Resolution range (Å)	30.0–2.2 (2.24–2.20) a
Space group	P21
Unit cell (Å)	78.5; 105.7; 80.5 β = 90
Total reflections	188 599
Unique reflections	65 288 (3248)
Completeness (%)	97.9 (96.5)
Redundancy	2.9 (2.7)
R‐meas	0.08 (0.38)
Refinement
Reflections used in refinement	65 241
R‐work/R‐free	18.8/24.3
Number of non‐hydrogen atoms
WDR5 (A/C/E/G)	2345/2337/2338/2321
RbBP5D376N (B/D/F/H)	59/63/40/29
Solvent	567
RMS (bonds) (Å)	0.01
RMS (angles) (o)	0.893
Molprobity score	1.98
Average B‐factor (Å2)
WDR5 (A/C/E/G)	22.8/23.1/24.9/24.7
RbBP5D376N(B/D/F/H)	35.5/36.2/48.4/41.2
Solvent	29.5

^a Highest‐resolution shell is shown in parentheses.

2.7. Mutational Analysis

To analyze the mutational landscape mapping to WRAD subunit, we downloaded datasets from the Catalogue of Somatic Mutations in Cancer (COSMIC) database which integrates mutation data from major cancer genomics initiatives including TCGA, ICGC, and targeted sequencing projects. Mutations were aggregated by primary tissue type and mutation classification, excluding entries with unknown or unclassified mutation types. The bar plot showing the prevalence of mutations was generated using R (version 4.1.2) [35] with the ggplot2 package [36]. All mutations mapping to the SPRY domain (a.a. 27–510) and SDI motif (a.a. 510–534) of ASH2L Isoform 3 were filtered by type of mutations and mutation frequency. Similarly, we filtered all mutations mapping to the WBM motif of RbBP5‐201 (a.a. 371–380). The lollipop plots were generated in R (version 4.5) programing language using ggplot2 [36] and ggrepel [37] packages.

2.8. Statistical Analysis

Differences in ITC‐derived entropy (ΔS) and enthalpy (ΔH) values between each mutant and WT were considered meaningful when they exceeded the combined fitting uncertainty of the compared values, calculated as √(SEM_mutant² + SEM_WT²), where SEM represents the fitting error derived from nonlinear regression of the binding isotherms.

3. Results and Discussion

3.1. Cancer‐Related Mutations Mapping to ASH2L Impact Complex Formation

Defined as an oncogene, ASH2L protein levels are elevated in many tumors and tumor cell lines [38, 39] and its overexpression promotes tumor proliferation, while its depletion suppresses tumorigenesis [40]. In vitro, loss of ASH2L or RbBP5 decreases KMT2 complex activity by ~600‐fold [10, 21], a function partly mediated by the ASH2L^SPRY [21, 24]. ASH2L also interacts with DPY‐30 dimerizing domain through its SDI amphipathic α‐helix [18, 19], and disruption of this interaction impairs β‐globin transcription [19, 21]. Therefore, to determine whether cancer‐associated mutations in ASH2L alter complex formation and potentially KMT2 activity, we analyzed cancer‐related missense mutations reported in the Catalogue of Somatic Mutations in Cancer (COSMIC) databases. Interestingly, 32 cancer‐related mutations map to the SPRY or SDI domains of ASH2L and are implicated in one or more forms of cancer (Figure 1C) (Figure S1).

To first test the impact of the identified cancer‐related mutations mapping to ASH2L^SPRY, we introduced the mutations in a construct corresponding to the ASH2L^SPRY domain and purified the mutants to homogeneity followed by measuring the binding affinities using Isothermal Titration Calorimetry (ITC) experiments (Figure 1D–F, Figure S3). While some mutants could not be purified and therefore could not be subjected to binding experiments (e.g., G345D, S365Y), the D293Y substitution result in a ≥ 45‐fold loss in binding to RbBP5 compared to WT ASH2L^SPRY (Figure 1D). Intriguingly, several mutations increase binding for RbBP5 and are referred to here as gain‐of‐function mutations (GoF) (highlighted in green in Figure 1D). To confirm whether we see the same effect with full‐length ASH2L in cells, we introduced some of the mutations (R286W, D293Y, R294Q, P296S) by site‐directed mutagenesis and expressed as Flag‐tagged proteins in HEK293T cells. Flag co‐immunoprecipitation (co‐IP) confirms the ITC results by showing a similar trend where D293Y completely disrupt binding and R286W, R294Q, and P296S retain similar binding for RbBP5 (Figure 1G, Figure S2A). Mutations mapping to the ASH2L^SDI were also introduced in a construct corresponding to ASH2L^SPRY‐SDI, but we failed to obtain an appropriate ITC measurement and therefore only performed co‐IP to assess its impact on binding to DPY‐30. We demonstrate that all ASH2L^SDI variants retain comparable interaction with DPY‐30 compared to the WT ASH2L^SDI (Figure 1H, Figure S2B).

Binding assays of ASH2L^SPRY mutants also reveal mutation‐dependent shifts in conformational dynamics, showing significant changes in both entropy (ΔS) and enthalpy (ΔH) (Table S1). Interestingly, GoF mutants that enhance binding by ≥ 1.7‐fold, corresponding to the average fold change for GoF mutants, result in a gain in entropy (highlighted in light green in Table S1). This indicates increased conformational flexibility upon binding, resulting in a reduced entropic penalty and stronger binding. In contrast, the R286W, Y328C, S383F, and L474M mutants, which exhibited reduced binding, demonstrate a significant decrease in both entropy and enthalpy (highlighted in light orange in Table S1). This reduction implies a more rigid and energetically unfavorable interaction. Together, these energetic signatures highlight how distinct mutations modulate the conformational landscape of the SPRY domain and indicate that cancer‐associated mutations can either weaken or enhance the ASH2L–RbBP5 interaction, both of which may negatively impact KMT2 activity and epigenetic regulation.

3.2. Mutations in ASH2L Impact RbBP5 Binding Through Long‐Range Structural Effects

To better understand how patient‐derived mutations influence ASH2L function, we first examined whether these substitutions alter the intrinsic stability of the ASH2L^SPRY domain. To assess this, we measured the melting temperatures (T_m) of each mutant using differential scanning fluorimetry (DSF) (Figure S4). This analysis reveals that 12 of 26 ASH2L^SPRY mutants exhibit a modest decrease in Tₘ (≥ 3°C), while I396V and L474M cause the largest destabilization, with a T_m of 7.3°C and 8.5°C lower than WT ASH2L, respectively (Figure 2A). A decrease in Tₘ was considered significant when the mutant Tₘ was less than one standard deviation below the WT mean (WT mean ± SD). Consistent with this, L474M also disrupts complex assembly, exhibiting a 1.3‐fold reduction in binding affinity. Therefore, we used AlphaFold to predict how these mutations might affect protein folding (Figure 2B,C). Positioned within the protein core, I396V and L474M models suggest reduced Van der Waal interactions and loss of hydrophobic packing within ASH2L^SPRY (Figure 2B). Interestingly, despite destabilizing the ASH2L^SPRY fold by potentially weakening the domain's tertiary structure, I396V mutations did not reduce binding to RbBP5, suggesting that the ASH2L‐RbBP5 interface remains largely intact and that binding is tolerant to modest changes in domain stability. Conversely, D293Y decreases the melting temperature of ASH2L^SPRY by ~5°C and completely disrupts binding. AlphaFold predictions suggest that D293 contributes to ASH2L structural stability by forming hydrogen bonds with R317 implicated in β‐sheet formation (Figure 2B). Substituting D293 with tyrosine abolishes these interactions, likely weakening the ASH2L tertiary structure and its ability to bind RbBP5. Most mutations that enhance binding affinity exhibit thermal denaturation profiles comparable to the WT, except for D502G. The substitution of aspartate with glycine disrupts hydrogen bonding between R498 and P499 (Figure 2B). This likely affects protein stability, but in this case, it may increase RbBP5's binding affinity for ASH2L by introducing greater flexibility. Interestingly, the other GoF mutations with a binding increase of ≥ 1.7‐fold appear to arise from changes in Van der Waal interactions and in hydrogen bonding with neighboring residues but lie, on average, ~22 Å away from the RbbP5 binding site. This suggests that the increased binding affinity may arise from long‐range structural reorganization that creates new contacts with RbBP5. Specifically, from these GoF mutations, T479M, E335G, and E310D, are closest with a distance of ~14, ~17, and ~14 Å away from the RbBP5 binding site, respectively.

FIGURE 2.

Cancer‐related mutations in ASH2L impact RbBP5 binding through long‐range structural effects. (A) Average melting temperatures (T_m) determined using the second derivative of the normalized relative fluorescence units (RFU) acquired by differential scanning fluorimetry. Error bars represent the SD of two biological replicates performed in triplicates. Highlighted in blue is the range of T_m of the WT. (B) Close‐up view of I396V, E335G, T479M, D502G, D293Y, L474M, E310D and their interactions with neighboring residues represented as stick model. AlphaFold predictions (cyan) were overlayed on the WT structure (gray). Sulfur, oxygen, and nitrogen atoms are colored in yellow, red, and blue, respectively. (C) Cartoon representation of ASH2L^SPRY domain with mutations shown as spheres. RbBP5 peptide is represented as a stick model (yellow) with oxygen and nitrogen atoms colored in red, and blue, respectively. Green spheres highlight mutants with an increase in binding (≥ 1.7‐fold change difference in K _D), yellow spheres highlight mutants with a ≥ 25‐fold decrease in K _D, red spheres highlight mutants with a ≥ 1.5‐fold decrease in K _D.

3.3. RbBP5 Mutations Mapping to the WBM Motif Impair Binding to WDR5 V‐Shaped Cleft

RbBP5 is linked to poor prognosis in hepatocellular carcinoma and promotes epithelial‐mesenchymal transition and metastasis in breast cancer by maintaining the transcription of invasion‐related genes [41, 42]. Cryo‐EM studies have shown that RbBP5 interacts with several subunits of the complex, with the RbBP5‐WDR5 interface being crucial for MLL catalytic activity [9]. Consequently, we hypothesized that multiple mutations could disrupt complex assembly and possibly impair epigenetic signaling [9, 15]. WDR5 features a seven‐bladed β‐propeller domain formed by tryptophan‐aspartate (WD) repeats, a structure well known for mediating protein–protein interactions [27] (Figure 3A). The crystal structure of WDR5 bound to the WDR5 binding motif (WBM) of RbBP5 (RbBP5^WBM) shows that the valine‐aspartate‐valine (VDV) motif is anchored within two distinct hydrophobic pockets that each accommodate residues V375 and V377. The first pocket is formed by Tyr228, Leu240, and Lys250. The second pocket interacts with V377 through residues Val268, Leu288, and Phe266 (Figure 3F). Notably, the COSMIC database identifies three mutations within RbBP5^WBM (Figure 3B). To examine how RbBP5^WBM mutants affect binding to WDR5, we performed ITC with WDR5 using a short RbBP5 peptide corresponding to the RbBP5^WBM (amino acids 371–380) (Figure S5). Of the cancer‐related mutations, V375M and V377I map to the valine residues of the VDV motif and reduce binding by 15‐ and 3‐fold, respectively (Figure 3C). Remarkably, V380M, located at the C‐terminus of RbBP5^WBM, completely abolishes binding. The replacement of valine with the bulkier hydrophobic residue methionine likely disrupts interactions with the aliphatic region of Arg181 in WDR5 by introducing steric clashes (Figure 3F). This underscores that residues beyond the central VDV motif significantly contribute to the interaction.

FIGURE 3.

RbBP5 mutations mapping to the WBM motif impair binding to WDR5 V‐shaped cleft. (A) Schematic representation of RbBP5 domains. (B) Lollipop plot showcasing the frequency and distribution of specific missense mutations within the WBM motif of RbBP5 (amino acids 371–380). (C) Bar plot showing the fold change in dissociation constant (K _D ) normalized to the wild‐type value, with decreases in binding affinity shown in red and increases in green. Asterisks denote RbBP5 mutants identified in the COSMIC database. Bars reaching the bottom of the y‐axis are considered to have a ≥ 25‐fold change difference in K _D. (D) Schematic representation of RbBP5 peptide interactions with WDR5 (pink labels) and the effects of different substitutions on binding affinity (purple = similar to WT, red = decreased binding, green = complete loss of binding, yellow = increased binding). (E) Overall structure of WDR5 (gray) showing the relative orientations of WT (orange) and D376N (cyan) RbBP5 peptides. (F) Zoomed view of the WDR5 V‐shaped cleft in which WDR5 (WT), WDR5 (D376N), RbBP5_371‐380 WT, and D376N carbon atoms are shown in light orange, light cyan, orange, and cyan, respectively. Intramolecular hydrogen bonds are shown as dashed gray lines, and the red arrow indicates a hydrogen bond lost in the D376N complex.

Owing that WDR5 is a major hub in protein complex formation, we sought to expand our mutational analysis by examining the role of RbBP5^WBM residues. We performed a systematic mapping of the RbBP5^WBM‐WDR5 interface by incorporating different substitutions in RbBP5^WBM. ITC results with RbBP5^WBM mutants reveal that residues with small hydrophobic side chains, such as valine or leucine, fit optimally into the first hydrophobic pocket of WDR5, yielding similar binding affinities, while smaller (alanine) or bulkier (phenylalanine) residues reduce binding (Figure 3C,F). This observation is consistent with evolutionary conservation, as the yeast homolog Cps50 (also referred to as Swd3) contains an isoleucine at this position, suggesting that while the first pocket can accommodate some variation, it exhibits selective size constraints for the hydrophobic side chain. In the VDV motif, the central aspartate residue (D376) forms an intramolecular H‐bond with S379, which likely helps position the adjacent valine side chains within their respective hydrophobic pockets of the WDR5 V‐shaped cleft. This central residue is important for binding as substitutions to asparagine or glutamate, which retain the side chain (Asn) or charge (Glu), significantly reduce or abolish binding, respectively (Figure 3C). To further understand the molecular underpinnings of this interaction, we solved the structure of WDR5 bound to the RbBP5^D376N (PDB: 10TC) (Table 1). Structural analysis indicates that D376 helps stabilize a kink at the adjacent V377 residue through an intramolecular hydrogen bond. Substituting D376 with asparagine disrupts this interaction by shifting the S379 side chain from 3 Å to 4.9 Å away from N376 causing the loss of the intramolecular H‐bond (Figure 3D,E). In the second pocket, the valine is the optimal residue as its substitution to a leucine, phenylalanine, glutamate, and glycine abolishes binding but tolerates, albeit with reduced affinity, a smaller hydrophobic residue (V377A) or a more flexible residue (V377I). Interestingly, the WDR5‐RbBP5 interface is not strictly dependent on the VDV, as mutations outside of the central VDV increase binding affinity toward WDR5. Specifically, T378V, T378C, and S379A result in a 6.5‐, 5‐, and 2‐fold gain in binding, respectively (Figure 3C). This enhancement can be explained by the replacement of polar threonine and serine residues with less polar cysteine or more hydrophobic residues (valine and alanine), which potentially facilitates desolvation by making it easier to displace water molecules at the interface leading to a tighter binding.

4. Discussion

Mapping of the ASH2L mutational landscape highlighted the impact residues located more than 10 Å from the RbBP5 binding site. Distal mutations are increasingly recognized as capable of modulating binding affinity via long‐range conformational/dynamic changes. A previous study employing statistical‐mechanical modeling of CheY, a bacterial response regulator, Cyclophilin A (CypA), and a PDZ domain of the scaffolding protein PSD‐95 demonstrated that the effects of mutations decay with distance but can propagate up to 15–20 Å from the mutation site [43]. Moreover, several examples have shown that mutations can alter the dynamics of distant regions, such as in transketolase [44], and in viral spike proteins, where the receptor‐binding domain (RBD) mutations propagated allosteric effects to distal domains altering RBD binding to human angiotensin‐converting enzyme‐2 (ACE2) [45]. Specifically, the K417N mutation in RBD of the spike protein showed the strongest allosteric coupling with the N‐terminal domain (NTD). This mutation increases the number of intermolecular contacts, including salt bridges, and induces significant changes in ΔG and a noticeable 4000× fold increase in ACE2 binding affinity [45]. These studies support our findings that the observed GoF ASH2L^SPRY mutants may act by distal perturbation of the domain fold, which primes the SPRY domain for increased flexibility and promotes a conformation that can better accommodate RbBP5. In our case, these GoF mutations could remodel the local water network or induce a domino‐like propagation of minor energetic changes (ΔΔG) that ultimately reach and affect the RbBP5 binding site. Furthermore, while we examined single mutations individually, it is possible that those showing no substantial effect on protein stability and function could still influence RbBP5 interaction through a cooperative mechanism, where one mutation alone is insufficient to shift the conformational equilibrium. This epistatic effect of multiple mutations on binding affinity has been described previously, where a silent “permissive” mutation in DANCER‐3 reshapes the protein's energy landscape, enabling access to previously inaccessible conformational states, and a subsequent “destabilizing” mutation shifts the equilibrium to favor this new state [46]. These observations highlight that even “silent” mutations can remodel the energy landscape and influence binding, underscoring the structural plasticity of ASH2L and reinforcing its role as a key regulatory node within KMT2 complex. Together, these findings position ASH2L as a therapeutically actionable epigenetic hub, with insights into its cancer mutational landscape guiding the rational design of targeted interventions.

Interestingly, previous studies mapping the WDR5 interactome upon binding of small‐molecule inhibitors to the WIN site revealed that WIN‐site inhibition induces bidirectional changes in WDR5 interactions [47]. Notably, CHD8, a known WDR5 interactor through its WBM motif, is enriched upon WIN inhibition. This observation suggests that disruption at the WBM site may likewise trigger structural rearrangements that enhance WDR5's association with KMT2 catalytic subunits through its WIN site [47, 48]. Therefore, having identified RbBP5^WBM mutations that enhance binding to WDR5, this information could be leveraged to design WBM‐mimicking peptides or small molecules with increased affinity, providing effective inhibitors targeting the RbBP5 binding pocket of WDR5. Given that WDR5 has emerged as a key therapeutic target in cancer and directly engages the oncogenic transcription factor MYC through its WBM site, these findings provide a molecular framework for selectively and/or synergistically disrupting the WBM and WIN interaction and potentially inhibiting aberrant gene expression in cancer [49, 50, 51].

Overall, our mutational analysis revealed that while WDR5 binding remains highly constrained, it can tolerate certain substitutions in the RbBP5^WBM. From this, we defined a consensus motif for RbBP5^WBM recognition by WDR5 as X‐Ω‐[D/N]‐[VIA]‐[VC > T]‐[A>SC] (Ω denotes a hydrophobic residue). While previous studies of the RbBP5‐WDR5 interface have identified a similar pattern, our findings reveal a distinct preference for hydrophobic residues at the C‐terminal end of the RbBP5^WBM [52]. To this end, we performed a ProSite search using the consensus motif and identified ZZZ3, PTHD1, CPMD8, and HIPL1 as candidates. Notably, ZZ‐type zinc finger‐containing protein 3 (ZZZ3) is a histone H3 reader essential for maintaining histone acetylation through the ATAC complex. Several studies have reported WDR5 as an interactor of ATAC, and network analyses suggest that WDR5 and ZZZ3 co‐purify within the same complex [48, 53]. As such, these observations further indicate that WDR5 may carry out an important role within the ATAC complex.

5. Conclusion

Collectively, our findings indicate that ASH2L can absorb a significant mutational burden, as most cancer‐associated mutations did not disrupt ASH2L^SPRY binding or stability. This robustness likely reflects evolutionary pressure to maintain essential subunit interactions despite destabilizing mutations, while more subtle or allosteric mutations may fine‐tune binding affinity in oncogenic contexts. For instance, gain‐of‐function mutations that enhance RbBP5 binding could hyperactivate KMT2 enzymes, leading to elevated H3K4me3 deposition and aberrant gene expression. Conversely, destabilizing mutations may weaken complex integrity and impair epigenetic regulation. Beyond the analysis of cancer‐associated variants, our systematic approach also provided a detailed characterization of the WDR5‐RbBP5 binding motif. We acknowledge that many of the mutations analyzed may represent passenger mutations that arise through clonal expansion in tumors with a mutator phenotype, without directly contributing to tumorigenesis. Future studies assessing recurrence and functional impact across larger patient cohorts will be necessary to distinguish potential driver mutations from incidental passenger variants mapping to the WRAD. Nonetheless, our results serve as a foundation for comprehensive mapping of mutations affecting KMT2 regulatory subunits and offer a valuable resource for investigating how cancer mutations perturb epigenetic mechanisms.

Author Contributions

Participated in research design: J.‐F. Couture, S. Grégoire, S. Chow, M. Joshi, P. Zhang, A. Janna, and V. Tremblay; Performed in vitro experiments: S. Chow, S. Grégoire, M. Joshi, P. Zhang, A. Janna, and V. Tremblay; Participated in data analysis: J.‐F. Couture, S. Grégoire, S. Chow, M. Joshi, P. Zhang, and V. Tremblay; Conducted bioinformatics analyses for the mutational study: A. Mer and A.A. Ahmad; Performed peptide synthesis: M. Muñoz; Participated in manuscript preparation: Writing (original draft): S. Grégoire, J.‐F. Couture, and S. Chow; Writing – review and editing: all authors.

Funding

This work was supported by the Canadian Institutes of Health Research (CIHR), 517542. Natural Sciences and Engineering Research Council of Canada (NSERC) RGPIN‐2023‐05555.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Prevalence of Mutations in RBBP5 (Q15291‐1) (A) and ASH2L (B). (A, B) The bar plot above showcases the prevalence of mutations across primary tissue types, categorized by mutation types, excluding those of unknown classification. Mutation frequencies are depicted as a stacked bar plot, with the x‐axis representing various primary tissues and the y‐axis indicating the mutation frequency. Mutation types are color‐coded, including frameshift deletions, in‐frame deletions, frameshift insertions, coding silent substitutions, missense substitutions, and nonsense substitutions. The mutational data are sourced from the COSMIC database.

Figure S2: Second replicate of the Co‐IPs of ASH2L^SPRY (A) and ASH2L^SPRY‐SDI mutants (B). (A) Western blot of Co‐IP of ASH2L^SPRY mutants, where input and elution samples were probed with anti‐ASH2L and anti‐RbBP5 antibodies. (B) Western blot of Co‐IP of ASH2L^SPRY‐SDI mutants, where input and elution samples were probed with anti‐flag (ASH2L) and anti‐Myc (DPY‐30) antibodies.

Figure S3: ITC trace of ASH2L^SPRY WT and mutants, showing the raw heat data signals (top) and the binding isotherm from the integration of the heat peaks (bottom). The line of best fit is based on a 1:1 binding model. Each ITC trace corresponds to one of two replicates.

Figure S4: First‐derivative plots of fluorescence as a function of temperature for all ASH2L^SPRY mutants obtained by differential scanning fluorimetry. The melting temperature (T_m) was determined from the temperature corresponding to the minimum of the derivative curve (peak of –d(RFU)/dT) for each sample performed in triplicates.

Figure S5: ITC trace of RbBP5^WBM WT and mutants with WDR5, showing the raw heat data signals (top) and the binding isotherm from the integration of the heat peaks (bottom). The line of best fit is based on a 1:1 binding model. Each ITC trace corresponds to one of two replicates.

Table S1: Thermodynamic parameters (K_D, ΔH, and ΔS) obtained from ITC measurements for ASH2L^SPRY mutants. Errors indicate the standard deviation. Mutants highlighted in green represent gain‐of‐function mutations (≥ 1.7 fold change) that have a significant gain in entropy. Mutants highlighted in orange decrease binding and have a significant decrease in entropy and enthalpy.

Data Availability Statement

All data supporting the findings of this study are found within the paper and its supporting information. The coordinates of the WDR5/RbBP5^D376N complex structure have been deposited in the protein databank under accession number 10TC and are publicly available as of the date of publication.