The SETD2 L1609P mutation found in leukemia disrupts methyltransferase activity and reduces histone H3K36 trimethylation

TL;DR

A mutation in the SETD2 protein found in leukemia disrupts its ability to modify histones, which may contribute to cancer development.

Contribution

This study provides the first structural and mechanistic insights into how a specific SETD2 mutation affects histone methylation in cancer.

Findings

The L1609P mutation in SETD2 reduces H3K36 methyltransferase activity and protein stability.

The crystal structure of the SETD2 L1609P mutant reveals active site remodeling affecting substrate recognition.

SETD2 mutations in leukemia disrupt H3K36me3 deposition, potentially driving oncogenesis.

Abstract

SET-domain containing protein 2 (SETD2) is the primary methyltransferase responsible for generating H3K36me3, an epigenetic mark that is essential for transcriptional regulation and chromatin integrity. SETD2 mutations are frequently observed in various cancers and tend to cluster within its catalytic SET domain. Despite the clinical relevance of SETD2 missense mutations in cancer, their biochemical and structural consequences remain insufficiently characterized. Here, we present the enzymatic and structural characterization of the SETD2 L1609P mutant enzyme identified in leukemia. The L1609 residue is located in the SET domain within a conserved hydrophobic pocket that is involved in substrate H3K36 recognition. Interestingly, site-directed mutagenesis of residues within this hydrophobic pocket leads to SETD2 enzyme variants with either decreased or increased H3K36me3 methyltransferase activity, suggesting that cancer mutations affecting the L1609 residue could result in a loss- or gain-of-function enzyme variant. Using molecular and cellular approaches, we show that the SETD2 L1609P mutant exhibits reduced H3K36 methyltransferase activity, decreased protein stability, and poor cellular expression. Consistently, the crystal structure of the SETD2 L1609P in complex with a H3K36M peptide shows remodeling of the active site. These findings support the pivotal role of SETD2 inactivation and subsequent disruption of H3K36me3 deposition in oncogenesis, particularly in hematologic malignancies. Our study provides the first mechanistic and three-dimensional protein structure information on how SETD2-associated cancer mutations can lead to altered H3K36 methyltransferase activity.

Figures

Full text

SET-domain containing protein 2 (SETD2) is a critical histone methyltransferase that is responsible for catalyzing the trimethylation of histone H3 at lysine 36, thereby generating the epigenetic mark H3K36me3 in higher eukaryotes (1, 2, 3, 4). This pivotal chromatin modification is linked to several key biological processes, including transcriptional activation, DNA damage repair, alternative RNA splicing, and DNA methylation (3, 5, 6, 7, 8). Although SETD2 has been shown to catalyze mono- and dimethylation of H3K36 in vitro, silencing the enzyme results in a global loss of H3K36me3, but not of H3K36me1 or H3K36me2 marks (1, 2, 9). This evidence confirms that SETD2 is the primary histone methyltransferase responsible for cellular H3K36 trimethylation, using H3K36me2 as a substrate (3, 10). In cells, specific mono- and dimethylation of H3K36 are catalyzed by other members of the SET family of histone methyltransferases, notably NSD1/2/3 or ASH1L (3). While the capacity of SETD2 to regulate fundamental cellular processes is predominantly associated with its H3K36 trimethylation activity and the H3K36me3 epigenetic mark, recent evidence has demonstrated that SETD2 can also methylate non-histone proteins (11, 12, 13). For instance, SETD2-dependent methylation of α-tubulin is involved in mitosis and cytokinesis, while methylation of STAT1 by the enzyme affects interferon signaling (11, 14, 15). Furthermore, SETD2 has been shown to methylate the catalytic subunit of the Polycomb complex, EZH2, promoting its degradation and leading to decreased levels of the H3K27me3 chromatin modification (16).

Consistent with the important cellular functions of SETD2, knockout of the enzyme results in embryonic lethality (17, 18). The pivotal biological functions of SETD2 are further supported by the growing identification of pathogenic genetic alterations in SETD2 in neurodevelopmental disorders and, most notably, in cancers (3, 19, 20, 21). Numerous recurrent mutations in SETD2 (both monoallelic and biallelic) and aberrant H3K36me3 deposition have been identified in a wide range of pediatric and adult malignancies including clear cell renal carcinoma, gliomas, and hematologic malignancies (22, 23, 24, 25, 26, 27, 28). The high frequency of SETD2 mutations across numerous cancer types is reminiscent of classical tumor-suppressor genes. In line with this, recent studies have provided clear evidence that the loss of SETD2 drives tumor progression, supporting its classification as a “cancer driver” gene (3, 29, 30). Mutations in SETD2 are distributed throughout the gene. Missense and nonsense mutations occur at roughly equal proportions, though a clustering of missense mutations is particularly evident in the catalytic SET domain of the enzyme (2, 21). Notably, pathogenic missense mutations in the SET domain of other SET family of methyltransferases, particularly the H3K36 methyltransferases NSD1, NSD2, and NSD3, have been shown to result in either decreased or increased histone methyltransferase activity (31, 32, 33). In addition, mutational studies of residues within the SET domain of SETD2 have been shown to result in enzyme variants with decreased or increased H3K36 methyltransferase activity (34). Despite the important pathological implications of SETD2 cancer mutations, their enzymatic and structural impacts remain poorly documented (3, 7, 35, 36). To date, the SETD2 R1625C mutation, identified in renal cell carcinomas, is the only variant with rigorous biochemical data showing functional impairment has been obtained (35). Here, we present a biochemical, cellular, and structural characterization of the SETD2 L1609P mutant enzyme. The L1609P mutation is located in the SET domain of the enzyme and has been previously identified in a cohort of acute leukemia patients (24). Enzymatic analysis using purified recombinant proteins revealed that this mutation results in strongly decreased methyltransferase activity and reduced intrinsic protein stability. Consistent with these findings, experiments conducted in HEK293T SETD2 KO cells transfected with WT or SETD2 L1609P plasmids, as well as in HEK293T CRISPR/Cas9-edited cells expressing SETD2 L1609P, revealed a loss-of-function phenotype of this SETD2 mutant. Finally, the X-ray structure of the SETD2 L1609P mutant in complex with an H3K36M peptide revealed that the mutation results in the remodeling of the active site around the H3K36M peptide substrate. Taken together, our findings are consistent with the critical role of altered SETD2 methyltransferase activity and aberrant H3K36me3 deposition in cancer development, particularly in hematologic malignancies (23, 26, 37, 38, 39). More broadly, this study provides mechanistic and structural insights into how SETD2-specific cancer mutations contribute to the loss of enzyme activity, and sheds light on the role of certain key residues/domains in SETD2 structure and function.

Results and discussion

The SETD2 L1609P mutant exhibit reduced lysine methyltransferase activity and protein stability in vitro

The SETD2 L1609P mutation, previously identified in pediatric patients with acute leukemia, is a missense mutation located in the SET domain of the human SETD2 catalytic core (24, 34). Residue L1609 is highly conserved in the SET domain protein methyltransferase superfamily and is located at the end of the β5-strand in human SETD2 (34) (Fig. 1A). This strand and its flanking residues are part of the hydrophobic pocket that recognizes residue H3K36 and its surrounding residues in H3 histone (such as H3V35 or H3K37) (34, 42). Mutational analyses have confirmed the essential role of residues flanking or within the SETD2 β5-strand in H3K36 substrate recognition (34). Interestingly, these studies have also shown that certain mutations, such as Y1604A or M1607A, result in SETD2 variants with either decreased or increased H3K36 methyltransferase activity, suggesting that mutations affecting residues within or flanking the β5-strand may result in loss- or gain-of-function enzyme variants (34, 43).Figure 1The L1609P mutation decreases methyltransferase activity and intrinsic protein stability of SETD2 catalytic core in vitro. A, upper panel: schematic representation of the SETD2 domains. The SETD2 L1609P mutation is located in the SET domain within the SETD2 catalytic core (composed of the AWS, SET, and post-SET domains). Lower left panel: Structural representation of the SETD2 active site (PDB entry: 5JJY) with a zoomed-in view of the substrate (H3K36M peptide) and cofactor (SAH) binding sites. Lower right panel: Sequence alignment of residues 1603 to 1619 of the SET domain of human SETD2 with the equivalent sequences of human G9A, EZH2, NSD1, NSD2, SETD8, MLL1, MLL2, SETD8, ASH1 (sequence retrieved from the UniProt database). Conserved residues are highlighted in blue. The secondary structure of the SETD2 residues (deduced from PDB entry: 5JJY) is shown above the alignment. The SETD2 residue L1609 and the equivalent residues in the other SET domain-containing enzymes are highlighted in orange. B, in vitro methylation of recombinant histone H3, core histones (purified from HEK293T SETD2-KO cells) or recombinant nucleosomes. SETD2-dependent H3K36me3 methylation was detected using an anti-H3K36me3 antibody. Ponceau Red staining of histones is shown. The purified catalytic core of SETD2 WT and SETD2 L1609P mutant used in the assays were detected using an anti-6xHis-tag antibody. C, SETD2 mono-methylation, dimethylation, or trimethylation activities were determined by UFLC assays using H3K36 fluorescent peptides as previously described (40, 41). Bar graphs and error bars represent the mean and SD of three independent experiments. D, automethylation of SETD2 and methylation of α-tubulin detected by autoradiography using ^3^H-SAM. Coomassie Blue staining was used as loading control. E, determination of the intrinsic protein stability of SETD2 WT or SETD2 L1609P by thermal shift assay (TSA). Left panel: T_m_ values were determined by the minimum of the first derivative of the fluorescence emission as a function of temperature (dFluo/dT). Right panel: Bar graphs and error bars represent the mean and SD of nine experiments. SETD2, SET-domain containing protein 2; UFLC, ultrafast liquid chromatography.

To investigate the effects of the L1609P mutation on the SETD2 enzyme in vitro, we expressed and purified the catalytic core (residues 1433–1711, corresponding to the AWS, SET, and post-SET domains) of WT and SETD2 L1609P mutant in Escherichia coli. This recombinant catalytic core of human SETD2 is easier to express and purify and has been used in several structure-function studies of human SETD2 (12, 34, 35, 44). As shown in Figure S1A, the WT and L1609P SETD2 constructs yielded soluble and highly pure proteins. Notably, a slight migration shift of the SETD2 L1609P protein was observed, suggesting possible conformational changes due to the mutation. To test the in vitro H3K36 methyltransferase activity of the WT and SETD2 L1609P enzymes, we performed assays using recombinant histone H3, core histones (extracted from HEK293T SETD2 KO cells (35)) and recombinant nucleosomes as substrates. We found that the SETD2 L1609P mutant exhibited significantly reduced H3K36me3 activity with all substrates compared to the WT form (Fig. 1B). Although SETD2 is the primary cellular histone methyltransferase responsible for generating the H3K36me3 mark from H3K36me2, the enzyme is known to catalyze all H3K36 methylation states (H3K36me1, H3K36me2, and H3K36me3) (4, 45). Therefore, we analyzed the mono-methyltransferase, dimethyltransferase and trimethyltransferase activity of WT and SETD2 L1609P mutant by performing reverse-phase ultrafast liquid chromatography (UFLC) assays using H3K36-derived peptides as previously described (13, 40, 41). To this end, unmodified H3K36me1 and H3K36me2 peptides were used as substrates in separate reactions. Consistent with the results reported above, the SETD2 L1609P mutant was found to exhibit significantly reduced and comparable levels (∼15–20% of WT activity) of mono-methyltransferase, dimethyltransferase, and trimethyltransferase activity compared to SETD2 WT (Fig. 1C). The decreased H3K36 methyltransferase activity of SETD2 L1609P was further confirmed by steady-state kinetic analyses, which showed that the mutant enzyme was 2.5-fold less efficient than the WT form in catalyzing H3K36 peptide methylation (k_cat_/K_m_ = 66 ± 12 M^-1^ min^-1^ and 171 ± 25 M^-1^ min^-1^ for SETD2 L1609P and WT, respectively). The difference in catalytic efficiency was mainly due to altered K_m_ values (K_m_ = 350 ± 50 μM and 756 ± 120 μM for SETD2 WT and SETD2 L1609P, respectively), suggesting that the mutation may affect substrate binding, which is consistent with the location of the L1609 residue and the role of β5-strand residues in H3 substrate binding/recognition (34, 43). Moreover, nucleosome pull-down experiments demonstrated that the SETD2 L1609P mutant exhibited reduced association with nucleosomes compared with SETD2 WT, thereby further confirming that the L1609P mutation weakens substrate binding (Fig. S1B). In addition to H3K36 methylation, SETD2 is known to catalyze its own methylation (automethylation) and to methylate non-histone substrates, particularly α-tubulin (11, 15, 46). As shown in Figure 1D, radiometric methyltransferase assays confirmed that, in contrast to SETD2 WT, the L1609P mutant had significantly reduced auto and α-tubulin methyltransferase activity. Taken together, these results clearly show that the L1609P mutation results in a SETD2 variant with intrinsically low lysine methyltransferase activity leading, notably, to decreased levels of the H3K36me3 mark in vitro.

To test whether the L1609P mutation could affect the intrinsic protein stability of SETD2, we performed thermal shift assays as described previously (42, 47). We found that the SETD2 L1609P mutant exhibited significantly decreased thermal stability when compared to the WT enzyme (temperature of melting, Tm, equal to 35.8 ± 0.1 °C and 42.3 ± 0.02 °C for SETD2 L1609P and SETD2 WT, respectively) (Fig. 1E). This indicates that the L1609P mutation confers a lower intrinsic protein stability to the SETD2 catalytic core and further suggests that the mutation may affect, at least locally, its conformation.

The SETD2 L1609P mutation results in low cellular levels of the H3K36me3 mark and reduced SETD2 expression in CRISPR-edited HEK293T cells

To better understand the impact of the L1609P mutation on SETD2 and H3K36me3 mark in a cellular context, we CRISPR-engineered HEK293T cells to generate isogenic WT SETD2 and homozygous SETD2 L1609P mutant. We first examined the levels of H3K36me3 in these cells by immunofluorescence and Western blot analysis (Fig. 2A). Consistent with the in vitro data reported above, we found that the levels of the H3K36me3 mark were significantly reduced in cells expressing SETD2 L1609P compared to cells expressing SETD2 WT enzyme (Fig. 2A). As expected, no significant changes in H3K36me1 or H3K26me2 levels were observed in SETD2 WT or SETD2 L1609P cells (Fig. S2). These results are consistent with previous reports indicating that SETD2 is the primary H3K36me3 methyltransferase in mammalian cells and that ablation of SETD2 alone does not affect H3K36me1 or H3K36me2 marks (2, 35). We interrogated the effects of the L1609P mutation on SETD2 protein levels by fluorescence microscopy and Western blot as previously reported (6, 35). As shown in Figure 2B, SETD2 protein levels were significantly decreased in CRISPR-engineered HEK293T cells expressing the SETD2 L1609P mutant compared to cells expressing the WT enzyme. Similar results to those reported above were obtained using other independent CRISPR-edited SETD2 WT and L1609P clones (Fig. S3). We next asked whether expression of SETD2 WT in cells expressing the SETD2 L1609P mutant could restore H3K36me3 levels. To this end, the SETD2 L1609P engineered cells were transfected with GFP-tagged plasmids encoding the catalytic core of SETD2 WT or SETD2 L1609P. As expected, transfection of SETD2 WT in the SETD2 L1609P-engineered cells restored H3K36me3 levels, whereas transfection of the SETD2 L1609P construct did not (Fig. 2C). We performed similar transfection studies in a model of HEK293T-SETD2 KO cells (displaying very low levels of the H3K36me3 mark), which has been established and used to study SETD2 functions (6, 35) (Fig. S4A). Consistent with the results reported above with the CRISPR-engineered HEK293T cells, we found that GFP-SETD2 WT was well expressed and resulted in H3K36me3 deposition in SETD2-deficient cells, whereas GFP-SETD2 L1609P protein was poorly expressed and resulted in low H3K36me3 levels similar to those in nontransfected cells (Fig. S4A). Following cycloheximide treatment, we observed decreased protein levels of the SETD2 L1609P mutant relative to the SETD2 WT suggesting that the SETD2 L1609P variant is less stable in cells than the WT enzyme (Fig. S4B). These results are consistent with the reduced intrinsic protein stability of the SETD2 L1609P mutant that we observed in vitro (Fig. 1E). Interestingly, similar data were reported for the SETD2 R1625C mutant which was found to display shortened half-life relative to SETD2 WT in HEK293T transfected cells (35). Mutations/truncations affecting SETD2 have been shown to affect protein stability and expression in cells (6, 35). Moreover, SETD2 is known to be a rather unstable protein whose intracellular amount is tightly regulated notably through to proteolytic degradation (6, 48, 49). As shown in Figure 2D, the proteasome inhibitor MG132 was able to partially restore SETD2 L1609P expression in CRISPR-engineered cells, confirming that the SETD2 L1609P mutant is susceptible to proteasomal degradation, as also observed for other mutated or truncated SETD2 proteins (6, 35). As shown previously for the SETD2 R1625C, we found that the L1609P mutant mRNA levels were less abundant in SETD2 L1609P cells compared to WT (Fig. S5) (35). Collectively, these data suggest that the L1609P mutation results in a loss-of-function SETD2 enzyme that is poorly active and poorly expressed in vivo.Figure 2The L1609P mutation results in low levels of the H3K36me3 mark and in low expression of SETD2 in CRISPR/Cas9-engineered HEK293T cells and in transfected HEK293T-SETD2 KO cells. A, endogenous H3K36me3 levels in CRISPR/Cas9-engineered HEK293T cells expressing SETD2 WT or L1609P mutant. Left panel: the H3K36me3 mark was detected by immunofluorescence using an anti-H3K36me3 antibody. DAPI staining was used for nuclei localization. Optical sections are shown with 10 μm scale bars. Right panel: Histones from CRISPR/Cas9-engineered HEK293T cells expressing SETD2 WT or L1609P mutant were extracted and H3K36me3 levels were determined by Western blotting using a an anti-H3K36me3 antibody. Ponceau Red staining of extracted histones is shown. B, endogenous SETD2 levels in CRISPR/Cas9-engineered HEK293T cells expressing SETD2 WT or L1609P mutant. Left panel: Cells were fixed and SETD2 was detected using an anti-SETD2 antibody. DAPI staining was used for nuclei localization. Optical sections are shown with scale bars of 10 μm. Right panel: SETD2 was detected in cell extracts by Western blot using an anti-SETD2 antibody. Ponceau Red staining of the cell extracts is shown. C, CRISPR/Cas9-engineered HEK293T cells expressing SETD2 L1609P were transfected with GFP-SETD2 WT or GFP-SETD2 L1609P plasmids. Nontransfected CRISPR/Cas9-engineered HEK293T cells expressing SETD2 WT or SETD2 L1609P were used as controls. Ectopic GFP-SETD2 expression and H3K36me3 mark levels were detected by Western blot using anti-GFP or anti-H3K36me3 antibodies, respectively. Ponceau Red staining of cellular histones or extracts on membranes are shown. D, CRISPR/Cas9-engineered HEK293T cells expressing SETD2 WT or SETD2 L1609P were treated with MG132 or DMSO. Endogenous SETD2 WT and SETD2 L1609P expression levels were detected by Western blotting using an anti-SETD2 antibody. Ponceau Red staining of the cell extracts is shown. SETD2, SET-domain containing protein 2.

Determination of the crystal structure of the SETD2 L1609P mutant in complex with a H3K36M peptide and SAM cofactor

To understand the structural basis underlying the loss of function of the SETD2 L1609P mutant, we solved its crystal structure in complex with SAM and a H3K36M peptide (derived from H3.3 sequence and harboring a K36M mutation) as previously used by Yang et al. for the crystallization of the catalytic core of SETD2 WT (34). To date, no cancer-associated mutant form of SETD2 has been described at the structural level (12, 34, 42, 50). Despite the lower protein stability of the SETD2 L1609P mutant, we were able to obtain crystals and to solve the ternary crystal structure of the mutant form (catalytic core, residues 1433–1711) in complex with the natural cofactor SAM and an H3K36M peptide (residues 29–39 of histone H3.3) at 2.2 Å (Table 1). The overall structure of the SETD2 L1609P mutant was found to be very similar (RMSD value of 0.3 Å over 230 Cα) to previously reported structures of SETD2 WT in complex with SAM/SAH and H3K36M peptides (12, 34, 42, 50) (Fig. 3A). As observed in the SETD2 WT ternary complexes, the SETD2 L1609P mutant structure contained the N-terminal AWS domain (residues 1494–1549), which coordinates the two zinc ions and the SET (residues 1550–1667) and post-SET (residues 1674–1690) domains, which together coordinate the third zinc ion. Furthermore, the H3K36M peptide and the SAM cofactor were found at positions similar to those of the SAM/SAH and H3K36M peptides in previously reported SETD2 WT structures indicating that the L1609P mutation does not abolish cofactor and peptide substrate binding (12, 34, 42, 50) (Fig. 3A). However, while the entire H3K36M peptide (residues A29–R42) was observed in SETD2 WT structures, only residues A29 to H39 (with no clear electron density observed for the H39 residue side chain) could be traced in the SET domain peptide-binding channel in the SETD2 L1609P mutant, suggesting that the peptide substrate-binding mode might be affected by the L1609P mutation (34, 42) (Fig. 3A). The structure of the mutant clearly shows the replacement of residue L1609 by a proline within the β5-strand, with residues L1609 and P1609 being found at similar positions in the SETD2 WT and mutant structures, close to but not directly interacting with the H3K36M peptide (12, 34, 42, 50) (Figs. 3A and S4). Importantly, the structure of the SETD2 L1609P mutant clearly shows that the replacement of the leucine residue with a proline residue disrupts the structure of the β5-strand, which adopts a loop conformation. This structural effect is consistent with the properties of proline, which has a strong destabilizing effect on the conformation of regular secondary structures such as β-strand due to its fixed backbone dihedral ϕ angle and lack of an NH group for participating in H-bonding (51, 52). As expected, the conformation of the adjacent anti-parallel β6-strand, which interacts with β5 is also found to be partially affected by the L1609P mutation (Figs. 3B and 4A). Interestingly, the β4-β5-β6 strands form a discrete β-sheet, which together with two other β-sheets (β1-β2 and β3-β8-β7), is arranged in a conserved triangular shape that is characteristic of the SET domain and is critical for substrate binding and enzyme activity (34) (Fig. 3B). A more detailed comparison of the β5β6 region further reveals important rearrangements of the main and side chain conformations of residues K1610, N1611, D1612, and E1613 in the SETD2 L1609P structure, as evidenced by the distance between the Cα of residues K1610 and E1613, which is almost doubled (5 and 8.3 Å in SETD2 WT and L1609P structures, respectively), and the 140° flip in the opposite direction of residue K1610 toward residue H3K37 (Fig. 4, A and B). Furthermore, to accommodate the altered conformation of residue K1610 and to avoid steric clashes with it, the conformations of the side chains of residues K1639 in SETD2 and K37 in the peptide substrate are rearranged in the SETD2 mutant complex (Fig. 4B). As expected, the conformational rearrangement of these SETD2 residues in the mutant leads to a reshaping of the substrate-binding pocket (Fig. 4C). Interestingly, although we used a H3K36M peptide similar to the one previously used in the structure of ternary complexes of WT SETD2, we did not observe the last C-terminal residues of the H3K36M peptide (R40-Y41-R42) in the substrate pocket, probably due to the altered or absent binding of these residues (resulting in increased flexibility and very low electron density) (Figs. 4C and 5A). These observations are consistent with our enzymatic studies reported above showing an altered affinity (higher K_m_) of SETD2 L1609P for the H3K36 peptide substrate and further support that the mutation affects the substrate-binding mode and subsequently the activity of the enzyme. As shown in Figure 5A, the H3K36M peptide (residues A29–R42 in SETD2 WT and residues A29–H39 in SETD2 L1609P) adopts a similar extended conformation within the substrate binding site in the WT and SETD2 L1609P structures. However, slight differences were observed for the main chains of residues T32, G33, P38, and H39 (Fig. 5B). Furthermore, as mentioned above, the side chain of residue K37 in the H3 peptide adopts a different orientation in the SETD2 L1609P structure that avoids a steric clash with K1610 (Figs. 4B and 5B). Interaction analysis by LIGPLOT confirmed that although several SETD2–H3K36M interactions involving key residues such as Y1666-M36, Y1666-K37, F1668-G34, or Y1670-G33 are conserved between SETD2 WT and L1609P structures, a number of interactions occurring in the WT complex such as M1607-P38, K1639-P38, or T1637-H39 are not observed in the SETD2 L1609P structure (Figs. 5 and S4). Conversely, specific interactions such as A1699-K37, A1699-P38, K1610-P38, and K1610-H39 are observed in the SETD2 L1609P structure (Figs. 5 and S4). These observations suggest that the L1609P mutation, by disrupting the conformations of the β5-β6 strands, leads to remodeling of part of the substrate binding pocket, resulting in an altered binding mode of the H3K36M peptide. To better understand the structural effects of the L1609P mutant, molecular dynamics (MD) simulations were performed. As shown in Figure S5A, the RMSD values for the SETD2 WT backbone Cα atoms fluctuated mainly between 1.3 Å and 2.4 Å (average RMSD = 1.6 ± 0.02 Å), whereas the values for the L1609P mutant were higher, with fluctuations mainly between 1.9 Å and 3 Å (average RMSD = 2.3 ± 0.03 Å), suggesting that the structure of the mutant is intrinsically less stable. This is consistent with our experimental data showing that the SETD2 L1609P catalytic core has a significantly lower thermal stability than WT (Tm WT=42.3 ± 0.02 °C, Tm L1609 p = 35. 8 ± 0.1 °C) (Fig. 1E). In addition, root mean square fluctuation analysis of Cα atoms showed that the L1609P mutation mainly results in increased flexibility and dynamics of the SETD2 catalytic core residues, especially residues in the SET domain involved in H3K36M peptide binding (especially residues 1630–1670), but also residues in the flanking AWS and post-SET domains, both of which are required for the structure and catalytic activity of the enzyme (4, 34, 42, 43) (Fig. S4B). These results support that, in addition to local conformational remodeling of the substrate binding site, the SETD2 L1609P has also a broader structural effect that alters the intrinsic stability of the catalytic domain of the enzyme. SETD2 is a rather unstable protein with a rapid turnover that is at least partially dependent on the proteasome (6, 48, 49). While the N-terminal region of SETD2 has been shown to be the major contributor to SETD2 protein instability, the catalytic core domain is considered to be the most stable domain of the protein (6, 49). Thus, the altered stability of the catalytic core domain in the SETD2 L1609P mutant is likely to further decrease the overall stability of SETD2, resulting in robust cellular degradation and low expression as observed in the CRISPR-engineered cells expressing SETD2 L1609P (Fig. 2).Table 1. Data collection and refinement statisticsSETD2 L1609P (PDB entry: 8RZU)Wavelength0.97857 ÅResolution range47.283–2.186 (2.223–2.186)Space groupP 21 21 21Unit cell59.718 77.408 77.884 90 90 90Total reflections258,323 (13,274)Unique reflections19,287 (961)Multiplicity13.4 (13.8)Completeness (%)100.0 (100.0)Mean I/sigma(I)19.3 (2.2)R-merge0.075 (1.215)R-meas0.078 (1.262)R-pim0.029 (0.471)CC1/20.999 (0.911)Reflections used in refinement19,130Reflections used for R-free964R-work0.2212R-free0.2644Number of nonhydrogen atoms2151 Macromolecules2106 Ligands3 Solvent42Protein residues262RMS (bonds)0.008RMS (angles)1.01Ramachandran favored (%)97.25Ramachandran allowed (%)2.75Ramachandran outliers (%)0.00Rotamer outliers (%)0.44Clash score6.04Average B-factor63.95 Macromolecules64.41 Ligands46.67 Solvent53.41Statistics for the highest-resolution shell are shown in parentheses.Figure 3Overall structure of the ternary complex of SETD2 L1609P mutant bound to H3K36M peptide and SAM cofactor. A, left panel: cartoon representation of SETD2 WT (PDB: 5JJY) (cyan) bound to H3K36M peptide (orange) and the SAH cofactor (gray sticks). The protein surface is shown as transparent. The side chains of the SETD2 L1609 and H3M36 residues are represented by yellow and orange sticks, respectively. The close-up view shows the region around residue L1609 with the H3K36M peptide (residues 29–42, orange) and the SAH cofactor (black sticks). Zinc atoms are shown in gray. Right panel: cartoon representation of the SETD2 L1609P mutant (PDB: 8RZU) (salmon) bound to the H3K36M peptide (green) and the SAM cofactor (gray sticks). The protein surface is shown as transparent. The side chains of the SETD2 P1609 and H3M36 residues are shown as yellow and green sticks, respectively. The close-up view shows the region around the residue P1609 with the H3K36M peptide (residues 29–39, green) and the SAM cofactor (black sticks). B, left panel: cartoon representation of the characteristic triangular shape of the SET domain formed by 3 β-sheets (β1-β2; β3-β8-β7; β4-β6-β5 strands) of SETD2 WT in complex with the H3K36M peptide (residues 29–42 in orange) (PDB: 5JJY). The β-sheet composed of β4-β6-β5 strands is boxed and the SETD2 L1609 residue is shown in yellow. Right panel: cartoon representation of the triangular β-sheet structure of the SET domain of the SETD2 L1609P mutant (salmon) in complex with the H3K36M peptide (residues 29–39, green) (PDB: 8RZU). The β5-strand in SETD2 WT adopts a loop conformation in the structure of the SETD2 L1609P mutant (boxed). The P1609 residue in mutant SETD2 is shown in yellow. SETD2, SET-domain containing protein 2.Figure 4Effects of the SETD2 L1609P mutation on the conformations of neighboring residues of SETD2 and the H3K36M peptide.A, the left panel shows a cartoon overlay of the β5-β6 hairpin of SETD2 WT (PDB: 5JJY) (cyan) and SETD2 L1609P mutant (salmon) structures. The H3K36M peptide is shown in orange and green for SETD2 WT and SETD2 L1609P, respectively. The side chains of residues L1609 and P1609 residues are shown as sticks (yellow CPK). The middle panel shows a close-up view of the hairpin residues (1609–1613) of SETD2 WT (cyan) and SETD2 L1609P (salmon). The side chains are shown in CPK sticks. The right panel shows the β5-β6 hairpin residues of SETD2 WT (top) and SETD2 L1609P (bottom) in sticks. Dashes represent the distance between Cα of residues K1610 and E1613 residues. B, conformational remodeling of residues K1610 and K1639 of SETD2 and residue K37 of H3 induced by the L1609P mutation. Left panel shows residues SETD2 L1609 (yellow), K1610 (cyan), K1639 (cyan), and H3K37 (orange) in spheres and sticks in the SETD2 WT structure (PDB: 5JJY). Middle panel shows residues SETD2 P1609 (yellow), K1610 (salmon), K1639 (salmon), and H3K37 (green) in spheres and sticks in the SETD2 L1609P structure. The right panel shows residues P1609 (yellow) and K1610 (salmon) from the SETD2 L1609P structure and residues K1639 (cyan) and H3K37 (orange) from the SETD2 WT structure. Steric clashes between side chains are shown in boxes. The orientations are the same in all three panels and were obtained by superimposing the SETD2 WT and L1609P main chains. C, surface representation of the SETD2 substrate-binding region. H3K36M peptides are shown as sticks. The left panel shows the SETD2 WT structure (PDB: 5JJY) in light cyan. The SETD2 L1609 residue is shown in yellow. The SETD2 K1610 and K1639 residues are shown in blue. H3K36M peptide residues diffracting in both WT and L1609P structures (residues A29–H39) are shown in green. H3K36M peptide residues observed only in the SETD2 WT structure (residues R40-R42) are shown in transparent orange. The right panel shows the SETD2 L1609P structure in light pink. The SETD2 P1609 residue is shown in yellow. The K1610 and K1639 residues are shown in purple. H3K36M peptide residues observed in the SETD2 L1609P structure (A29–H39) are shown in green. SETD2, SET-domain containing protein 2.Figure 5Details of H3K36M peptide recognition by SETD2 L1609P mutant. A, the left panel shows a clipped surface representation of the SETD2 WT-H3K36M peptide complex (PDB: 5JJY). Peptide residues (residues A29–R42) are represented by sticks. The right panel shows a clipped surface representation of the SETD2 L1609P-H3K36M peptide complex. Peptide residues (A29–H39) are represented by sticks. The structures of the SETD2-H3K36M peptide complexes are shown in the same orientation after superimposition of the main chains. B, upper panel: Structural alignment of H3K36M peptides (residues A29–H39) in SETD2 WT (PDB: 5JJY) (orange) and SETD2 L1609P (green) structures. Lower panel: Differences between SETD2-H3K36M peptide interactions in SETD2 WT and SETD2 L1609P complexes. Residue interactions across the binding interface of SETD2 WT or SETD2 L1609P mutant with H3K36M peptide were determined using LIGPLOT (53). Residues are represented by sticks. Residues involved in SETD2-H3K36M peptide interactions (nonbonded and hydrogen bonds) are represented by sticks and spheres. Dashes represent hydrogen bond. The lower left panel shows the SETD2 WT (cyan)-H3M36 (orange) interacting residues that are specific for the SETD2 WT complex and not present in the SETD2 L1609P complex. These interactions are listed in a table (bottom left). The lower right panel shows SETD2 L1609P (salmon)-H3K36M (green) peptide interacting residues that are specific for the SETD2 L1609P complex and not present in the SETD2 WT complex. These interactions are listed in a table (bottom right). SETD2, SET-domain containing protein 2.

It is known that certain pathogenic missense mutations in the SET domain of H3K36 methyltransferase enzymes such as NSD2 or NSD3 can result in variants with increased or reduced methyltransferase activity (31, 32). In addition, mutational studies of residues within the SET domain of SETD2 have shown that mutations of specific residues (such as Y1604 or M1607) can result in SETD2 enzymes with increased or decreased H3K36 methyltransferase activity (34, 43). In this study, we present molecular, enzymatic, and structural evidence indicating that the L1609P mutation, located in the SET domain of SETD2 and identified in patients with acute leukemia, results in a loss-of-function enzyme with decreased H3K36 methyltransferase activity and poor cellular expression that lead to low H3K36me3 levels. Our findings are consistent with numerous studies showing that loss of SETD2 activity and aberrant deposition of the H3K36me3 mark play a key role in cancer development, particularly in leukemia (23, 24, 37, 39). Moreover, our study provides the first mechanistic and structural insights into how SETD2-associated cancer mutations lead to altered methyltransferase activity.

Experimental procedures

Alignment of human SET domain sequences

Sequences of the SET domain of human histone methyltransferases SETD2, G9A, EZH2, NSD1, NSD2, SETD8, MLL1, MLL2, SETD7, ASH1 (UniProt/Swiss-Prot database Q9BYW2, Q96KQ7, Q15910, Q96L73, O96028, Q9NQR1, Q03164, O14686, Q8WTS6, Q9NR48, respectively) were aligned using the Clustal Ω server (EMBL-EBI) and analyzed by the ESPript 3 software.

SETD2 cDNA constructs and site-directed mutagenesis

A pet28a-MHL plasmid containing the cDNA coding for the human SETD2 catalytic domain (Addgene #25348, residues 1433–1711) was used in order to produce 6xHis-tagged WT SETD2 in BL21 HI-control (DE3) E. coli. SETD2 L1609P mutation was introduced using the QuickChange XL site-directed mutagenesis kit (Agilent) with the following primers:

His_SETD2_L1609P_F: CACTATTACTTTATGGCTCCGAAAAATGATGAAATTATCGACGCG.

His_SETD2_L1609P_R: CGCGTCGATAATTTCATCATTTTTCGGAGCCATAAAGTAATAGTG.

SETD2 catalytic domain (residues 1433–1711) was subcloned onto a pEGFP-SETD2 vector, (kindly given by Dr Sergio de Almeida) between BglII and PstI sites using the following primers (54):

GFP_catalytic_SETD2_F: CGAGATCTCAGGGAGAGACATCAGTGC.

GFP_catalytic_SETD2_R: GCCTGCAGTCACTTACGAGATCGTTCCTTCTTCATTTTC.

SETD2 L1609P mutation was introduced using the Agilent QuickChange XL site-directed mutagenesis kit with the following primers:

GFP_SETD2_L1609P_F: CATCCATTACTATTTCATGGCCCCGAAGAATGATGAGATAATAG.

GFP_SETD2_L1609P_R: CTATTATCTCATCATTCTTCGGGGCCATGAAATAGTAATGGATG.

Successful cloning and mutagenesis were verified by sequencing.

Recombinant protein expression and purification

Recombinant SETD2 WT or L1609P proteins were expressed as previously described (41). Briefly, E. coli bacteria were cultured in Luria-Broth medium at 37 °C under agitation until an absorbance (600 nm) of 0.6. Protein expression was induced overnight (ON) at 16 °C by adding 500 μM IPTG. Bacteria were harvested by centrifugation and washed with PBS. Pellets were lysed in lysis buffer (PBS, 300 mM NaCl, 1 mg/ml lysozyme, 1% Triton X-100, and Complete protease inhibitors (Sigma)) for 30 min at 4 °C. Lysate was sonicated (10 min, 20% power, 10 s pulse on, 10 s pulse off) and cleared by centrifugation (30 min, 17,000g, 4 °C). Supernatant was collected and incubated for 2 h in presence of Ni-NTA beads and 10 mM imidazole. Beads were loaded on a column and washed with PBS, NaCl 300 mM, pH 8, with and without Triton X-100 0.2%. Proteins were eluted with 300 mM imidazole and incubated with 10 mM DTT for 20 min. Purified proteins were buffer-exchanged using PD-10 desalting columns to methyltransferase buffer (50 mM Tris, 50 mM NaCl, pH8) and centrifuged for 15 min at 15,000g at 4 °C. Protein purity was determined by SDS-PAGE and Coomassie Blue staining and aliquots were stored to −80 °C until further use. Before use, proteins were centrifuged for 15 min at 15,000g at 4 °C and concentration was assessed by Nanodrop.

Crystallization of SETD2 L1609P in complex with H3K36M peptide and SAM

Purified SETD2 L1609P protein was buffer-exchanged to a crystallization buffer (20 mM Pipes, 250 mM NaCl, pH 6,5). The sample was then injected into a Superdex 200 16/60 exclusion chromatography column (Cytiva), previously equilibrated with the crystallization buffer. The fractions corresponding to the protein of interest were pooled and concentrated using Vivaspin 10 MWO concentrators (Sartorius), by centrifugation at 3700g at 4 °C. Samples were reduced with 10 mM DTT and kept on ice until screening. Protein purity was determined by SDS-PAGE and Coomassie Blue staining. Crystallization screening trials were done using the experimental procedures and equipment available in the crystallography core facility of the Institut Pasteur (Paris) (55). A sample containing 10 mg/ml SETD2 L1609P, 3.2 mM SAM, 1.6 mM H3K36M peptide (Histone H3.3, A_29_PSTGGVM36KPHRYR_42_) (molar ratio between protein, peptide, cofactor at 1:5:10, respectively) was used for an initial screening of seven multiwell plates (Greiner Bio-One) containing a total of 672 crystallogenesis conditions. The H3K36M peptide was synthesized on demand by ProteoGenix. The screening was carried out using a MosquitoTM nanoliter (TPP Labtech), by the sitting drop vapour diffusion method. Two hundred nanoliters of sample was mixed with an equivalent volume of the reservoir solution from the multiwell plates. The plates were stored at 18 °C in the RockImager system (Formulatrix) and crystal growth was monitored using the automated imaging system. Crystals of SETD2 L1609P were obtained in 32% w/v PEG 4K, 100 mM Tris, 800 mM lithium chloride, pH 8.5, and reproduced manually using the hanging-drop method. Crystals were cryoprotected with a solution containing oil and crystallization solution (50%/50% v/v), and flash-frozen in liquid nitrogen until they were diffracted.

Diffraction data collection and crystal structure determination

X-ray diffraction data were collected at beamline PROXIMA-1 of the SOLEIL synchrotron. The data were processed using AutoPROC (56). The structure of the catalytic domain of SETD2 L1609P was solved by molecular replacement with Phaser (57) using as a template the structure of the SETD2 WT in complex with SAH and the substrate peptide H3K36M (PDB ID: 5JJY). The final model was obtained through iterative cycles of manual rebuilding with Coot (58) and reciprocal space refinement using BUSTER (59). Crystallographic model coordinates and structure factors were deposited in the RCSB Protein Data Bank under accession code 8RZU. Structural analyses were performed using PyMol software (The PyMOL Molecular Graphics System, Version 2.5.7 Schrödinger, LLC, https://www.pymol.org/) and Chimera 1.14 (60). Residue interactions across the binding interface of SETD2 WT (PDB: 5JJY) or SETD2 L1609P mutant with H3K36M peptide substrate were analyzed using LIGPLOT and interaction diagrams generated with PDBsum (53).

MD simulations

We used 100-ns MD simulations to study the dynamic behavior of the SETD2 WT (PDB: 5JJY) and SETD2 L1609P mutant (PDB: 8RZU) structures. All crystallographic water molecules were removed from the coordinate set. MD simulations were performed using the OpenMM software package version 7.7.0 with the ff19sb force field (61, 62). A 1 fs integration time step was used for the equilibration phase, while a 4 fs time step was applied during the production phase using hydrogen mass repartitioning (63). All atom positions and velocities were collected at every 1000 steps or 4 ps in the trajectory file. To simulate the aqueous environment, the complex was solvated in a cubic TIP4P water box, maintaining a minimum distance of 10 Å from the outer residues to the box boundaries to prevent interaction with itself. Periodic boundary conditions were used in the MD simulations. To achieve charge neutrality, sodium and chloride ions were introduced using the Monte Carlo method. Restrained MD simulations were first performed for system equilibration (e.g., with harmonic restraints on protein heavy atoms), followed by unrestrained production runs to sample conformational dynamics. VMD software (https://www.ks.uiuc.edu/Research/vmd/) was used to calculate various properties such as RMSD, root mean square fluctuation, and distance between relevant amino acids (64).

Determination of SETD2 H3K36-methyltransferase activity using recombinant H3 histone, HEK293T SETD2 KO purified core-histones, or recombinant nucleosome substrates

Five micromolars of recombinant SETD2 (WT, L1609P, or T1663M) were incubated ON at room temperature with 1 μg of recombinant human histone H3.3 (NEB) or 1 μg of recombinant human nucleosomes (16–0009, EpiCypher) or 2 μg core histones purified from HEK293T SETD2 KO cells (35), 100 μM SAM, and 1 mM DTT. Samples were separated by SDS-PAGE (18% gel) and transferred onto a nitrocellulose membrane. Ponceau Red staining was used as loading control. H3K36me3 mark was determined using an anti-H3K36me3 antibody (Ab9050, Abcam) and SETD2 protein was detected using an anti-6xHis antibody (H1029, Sigma-Aldrich).

UFLC-Mediated SETD2 H3K36-methyltransferase activity and enzyme kinetics determination

Methyltransferase activity of SETD2 was determined by UFLC as previously described (40, 41). Peptides were synthesized on demand by ProteoGenix. Briefly, SETD2 (5 μM) was incubated in presence of 75 μM fluorescent peptide substrate (H3K36: FAM-TGGVKRPHR-NH_2_, H3K36me1: FAM-TGGVKme1RPHR-NH_2_, H3K36me2: FAM-TGGVKme2RPHR-NH_2_), 100 μM SAM, and 1 mM DTT, in methyltransferase buffer (50 mM Tris, 50 mM NaCl, pH 8). To ensure that only methylation events on K36 are detected, the peptide used harbor a K37R mutation (40). Michaelis–Menten kinetics were determined by varying H3K36 peptide substrate concentrations (0, 25, 50, 100, 200, 400 μM) or SAM concentrations (0, 6.25, 12.5, 25, 50, 100 μM SAM). The enzymatic reaction was carried out at room temperature for 4 h and stopped by 15% v/v perchloric acid. Ten microliters of sample mixture was injected in a reverse-phase UFLC system (Shimadzu). Separation was performed under isocratic conditions, 80% A (water with 0.1% perchloric acid)/20% B (acetonitrile with 0.1% TFA), with a flow rate of 1 ml/min at 40 °C for 27 min. Peptides (substrate and SETD2 methylation product) were separated using a Kromasil column (100-5-C18, 4.6 × 250 mm, particle size 5 μm, Phenomenex) and detected by fluorescence (λ_ex_ = 485 nm, λ_em_ = 530 nm). Quantification and data analysis were performed using LabSolutions software (Shimadzu, https://www.shimadzu.fr/products/software-informatics/labsolutions-series/index.html). H3K36 peptides (substrates and products) were quantified by integration of the peak fluorescence areas as described previously (40). Michaelis–Menten equation was applied to determine enzyme kinetics using GraphPad Prism 8.0.2 (https://www.graphpad.com/).

Determination of SETD2 automethylation and SETD2-dependent methylation of tubulin using a radiometric assay

SETD2 WT or SETD2 L1609P (3 μM) was incubated with ^3^H radioactive SAM (16.7 kBq, PerkinElmer), 1 mM DTT, and 1 μg tubulin (purified from pig brain as previously described (65)) in methyltransferase buffer (50 mM Tris, 50 mM NaCl, pH 8) at room temperature ON. Samples were migrated onto a 4 to 12% SDS-PAGE gel. The radioactive gel was blocked in a solution containing 50% ethanol, 5% acetic acid, and 5% PEG 300 for 1 h under agitation, and then stained with Coomassie Blue. Incorporation of ^3^H-methyl group was detected with an autoradiography film.

Thermal shift assay

Intrinsic protein stability of SETD2 was measured in an opaque 384-well PCR plate (Roche) at room temperature. Ten micromolars SETD2 WT or SETD2 L1609P were incubated in the presence of SYPRO dye (5X, Sigma-Aldrich). Fluorescence emission was detected on LightCycler 480 (Roche) unit, according to SYPRO properties (λ_ex_ = 465 nm, λ_em_ 580 nm). The thermocycler was programmed for an initial incubation at 20 °C for 15 s, followed by a gradual increase of temperature to 95 °C (rise of 0.05 °C/s and 12 readings/degree) and a final incubation at 20 °C for 15 s (66). Quantification and data analysis were performed using LightCycler 480 SW 1.5 software (Roche, https://diagnostics.roche.com/ch/fr/products/instruments/lightcycler-480-ins-445.html). Temperature of melting (T_M_) values was determined by the minimum of the first derivative of fluorescence emission as a function of temperature (dFluo/dT).

Nucleosome pulldown

Nucleosome pull-down assay was adapted from Mack et al. (67). Briefly, 12.5 pmol of biotinylated nucleosomes (H3K36me2, EpiCypher) were incubated with 10 pmol of SETD2 WT or SETD2 L1609P in nucleosome binding buffer. Ten microliters of streptavidin magnetic beads (Dynabeads) were added to the sample and further incubated at 4 °C. An input was kept as loading control. Samples were migrated onto a 4 to 12% SDS-PAGE gel and transferred onto a nitrocellulose membrane. Nucleosomes were detected using an anti-H3 antibody and SETD2 protein (pulldown and input) was detected using an anti-6xHis antibody.

HEK293T CRISPR/Cas9 modification

HEK293T cells (ATCC CRL-3216) were modified by electroporation of Cas9-crRNA-tracrRNA ribonucleoprotein (RNP). Specifically, crRNA (IDT) targeting SETD2 was annealed with Atto550-labeled tracrRNA (1075927, IDT) and subsequently formed RNP with Alt-R S.p. Cas9 Nuclease V3 (1081058, IDT) prior to electroporation. A total of 5 × 10^4^ cells were mixed with 1.5 μM RNP and 4 μM single-stranded oligonucleotide donor template, followed by electroporation with Neon Transfection System (MPK1096, Thermo Fisher Scientific) according to the manufacturer’s instructions. Single PE-positive cells were sorted 2 days after transfection, and colonies expanded and validated by Sanger sequencing of the amplified targeted region. All cells were maintained in Dulbecco's modified Eagle's medium (10013CV, Corning) with 10% FBS (35-010-CV, Corning) and penicillin/streptomycin (30–002-CI, Corning).

crRNA sequence

SETD2 L1609P crRNA1: CATGGCCCTGAAGAATGATGAGG; SETD2 L1609P crRNA2: GCTTACCTCATCATTCTTCAGGG.

Single-stranded oligonucleotide donor template sequence

SETD2 L1609P

GTCCTAGAATATTGTGGAGAGGTACTCGATCATAAAGAGTTTAAAGCTCGAGTGAAGGAGTATGCACGAAACAAAAACATACATTACTATTTCATGGCACCGAAGAATGATGAAGTAAGCAGCTTGGATATGTTTGCCTTGAAACAGATCTGTTAGGTTTGAAGGAGCCATTGATACTGATTTGATTTTTTTTTT.

Primers to validate point mutations

FW: CCCTACTGCTCCTAGGTAGGAAGT; RV: AACGTGGTGAAACCCCGT.

Cell culture, transfection, and MG132/cycloheximide cell treatment

CRISPR-engineered HEK293T cells expressing WT or SETD2 L1609P mutant, and HEK293T SETD2 KO cells (kind gift of Pr. W. Kimryn Rathmell) were cultured in RPMI 1640 medium supplemented with 10% FBS and 1 mM L-glutamine at 37 °C under 5% CO_2_ in 100 cm^2^ Petri dishes. In MG132 treatments, cells were treated with 20 μM MG132 (Euromedex) or DMSO for 16 h prior to cell lysis and Western blot analysis (35). For transfections studies, 10^6^ cells were seeded in 6-well plates and transfected with a solution consisting of 2 μg pEGFP-SETD2 WT (catalytic core, residues 1433–1711) or pEGFP-SETD2 L1609P (catalytic core, residues 1433–1711) plasmid and 4 μl METAFECTENE (Biontex) in Dulbecco's modified Eagle's medium. Transfected cells were cultured for 48 h. In cycloheximide experiments, transfected HEK293T SETD2 KO cells were treated with 30 μg/ml cycloheximide (Euromedex) for 8 h prior to cell lysis and Western blot as described previously (35).

Cell extracts and acid extraction of endogenous histones

Cells were harvested and lysed using PBS pH 7.5, 1% Triton X-100 and protease inhibitors or using SDS-sample buffer for 30 min at 4 °C. Cell lysates were sonicated twice for 2 s (10% power) and centrifuged for 15 min at 15,000g (4 °C). For histone extractions, the pellets of PBS-Triton cell extracts were further incubated ON with 0.2 N HCl at 4 °C. Extracted histones were centrifuged for 15 min at 15,000g (4 °C). Cell extracts (40 μg) were separated by SDS-PAGE (4–15% gradient gel, Bio-Rad), transferred onto a nitrocellulose membrane and SETD2 expression was detected by Western blotting using an anti-SETD2 antibody (PA5-43071, Thermo Fisher Scientific). GFP SETD2 expression in cell extracts was detected by Western blotting using an anti-GFP antibody (sc-9996, Santa Cruz). Extracted histones (3 μg) were separated by SDS-PAGE (18% gel), transferred onto a nitrocellulose membrane, and H3K36me3 levels were detected by Western blotting using an anti-H3K36me3 antibody (Ab9050, Abcam). H3K36me1 or H3K36me2 levels were detected by Western blotting using specific antibodies (C15410089 or C15310127, Diagenode). Ponceau Red staining was used as a loading control.

Immunofluorescence detection of H3K36me3 and SETD2 in CRISPR-edited HEK293T cells

HEK293T CRISPR SETD2 WT and HEK293T CRISPR SETD2 L1609P cells (1 × 10^5^ cells) were fixed with methanol at −20 °C for 20 min. The slides were then dried for 15 min and washed 3 times with PBS before being incubated ON with an anti-SETD2 (PA5-43071, Thermo Fisher Scientific) or anti-H3K36me3 (Ab9050, Abcam) antibody in PBS at 4 °C. Cells were washed 3 times with PBS and incubated with an anti-rabbit secondary antibody (GaR IgG-Alexa-Fluor-488 for SETD2 or GaR IgG-Alexa-Fluor-594 for H3K36me3, Immunoresearch Laboratories) for 45 min at room temperature. All slides were mounted in DAPI-containing mounting medium (Fluoroshield, Sigma-Aldrich). Cells were imaged using a confocal microscope Leica SP5 (objective 40X, 4 × 0.5 μm sections), images were recorded using Leica Application Suite Advanced Fluorescence Software, and then processed using ImageJ software (Fiji, https//imagej.net/ij/).

mRNA extraction, reverse transcription, and qPCR

mRNA of CRISPR-engineered HEK293T cells expressing WT or SETD2 L1609P mutant were extracted by tri-reagent. Two micrograms of mRNA were reverse transcribed with M-MLV reverse transcriptase. Twenty nanograms of cDNA were finally mixed with SYBR Green (Roche) and primers targeting endogenous SETD2 or RPL13A (control). qPCR was conducted using LightCycler 480 (Roche), and results were analyzed with LightCycler 480 SW 15 software. Results are presented as normalized expression of SETD2 toward to RPL13A housekeeping gene. The primers used were:

SETD2_FW: GGACAACTGAGGGTTGGGTTT; SETD2_RV: AAACCCAACCCTCAGTTGTCC.

RPL13A_FW: TGTTTGACGGCATCCCAC; RPL13A_RV: GTGGGATGCCGTCAAACA.

Statistical analysis

Values are presented as means ± SD of a minimum of three independent experiments and analyzed by either Welch’s t test or two-way ANOVA followed by two-tailed Bonferroni’s test using Prism GraphPad (8.0.2) and Jamovi (2.5.6). p < 0.05 was used to consider differences as statistically significant.

Data availability

The data produced by this study are available in this article and in the supporting information. The atomic coordinates and structure factors have been deposited in the Protein Data Bank (PDB entry: 8RZU).

Supporting information

This article contains supporting information (53).

Conflict of interest

M. R. G. reports research funding from Sanofi, Kite/Gilead, Abbvie and Allogene; consulting for Abbvie, Allogene, Bristol Myers Squibb; Johnson & Johnson, and Arvinas. The other authors declare that they have no conflicts of interest with the contents of this article.