A SETD2-CDK1-Lamin axis maintains nuclear morphology and genome stability

Abstract

Histone methyltransferases regulate chromatin organization and are frequently mutated in human diseases, including cancer. One such often mutated methyltransferase, SETD2, associates with transcribing RNA polymerase II and catalyzes H3K36me3 – a modification that contributes to gene transcription, splicing, and DNA repair. While its catalytic function is well-characterized, non-catalytic roles remain unclear. Here, we reveal a catalysis-independent function of SETD2 in nuclear lamina stability and genome integrity. Through its intrinsically disordered N-terminus, SETD2 associates with lamina-associated proteins including lamin A/C, lamin B1, and emerin. Loss of SETD2 or its N-terminus leads to severe nuclear morphology defects and genome instability, mirroring lamina dysfunction. Mechanistically, the N-terminus of SETD2 serves as a scaffold for the mitotic kinase CDK1 and lamins, facilitating lamin phosphorylation and depolymerization during mitosis. Restoring N-terminal regions required for interaction with CDK1 and lamins rescues nuclear morphology and suppresses tumorigenic growth in a clear cell renal cell carcinoma model with SETD2 haploinsufficiency. These findings reveal a previously unrecognized role of SETD2 in nuclear lamina organization and genome maintenance that likely extends to its role as a tumor suppressor.

Introduction

SETD2 is an evolutionarily conserved histone methyltransferase that associates with elongating RNA Polymerase II (RNAPII) to catalyze H3K36me3 in transcribed regions^{1, 2}. Most studies of SETD2 have focused on how H3K36me3 contributes to gene expression, splicing and DNA repair^3-6. Like other histone methylation marks, H3K36me3 functions by recruiting ‘reader’ proteins equipped with specialized domains^{3, 7}. While SETD2's catalytic roles are well-studied, its non-catalytic functions remain largely unexplored. In addition to transcriptional regulation and DNA repair, SETD2 is also frequently mutated in cancers^{8, 9}, particularly in clear cell renal cell carcinomas (ccRCCs) where it acts as a tumor suppressor¹⁰. An early hallmark of ccRCCs is loss of the 3p chromosomal arm, deleting one copy of VHL, PBRM1, BAP1, and SETD2, resulting in haploinsufficiency of these genes^11-15. Despite retaining H3K36 methylation, these haploinsufficient tumors exhibit severe genome instability¹⁶, suggesting SETD2 performs important functions distinct from its catalytic activity. Here, we reveal a non-catalytic role of SETD2 in regulating nuclear lamina disassembly during mitosis, a process critical for nuclear stability and genome integrity.

Results

SETD2 interacts with nuclear lamina proteins

All eukaryotic SETD2’s share core functional domains, including a conserved SET domain for H3K36me3 and a C-terminal SRI domain that interacts with RNAPII (Extended Data Fig. 1a). In mammals, SETD2 enzymes possess an extended N-terminal domain that is poorly characterized and enriched with intrinsically disordered regions (IDRs), particularly within the first 500 amino acids. Genetic and biochemical evidence shows this N-terminus is dispensable for H3K36me3^{2, 17}, which have led most studies to be focused on the C-terminal half. Yet, studies by Bhattacharya et al. indicate the N-terminus regulates SETD2 stability, implying this region is endowed with important regulatory functions^{18, 19}. Along with studies identifying new protein-protein interaction motifs in the C-terminus of SETD2²⁰, we surmised other important regulatory functions of SETD2 existed.

To uncover new regulatory mechanisms of SETD2, we mapped its protein-protein interactome using proximity-based biotinylation with APEX2, which captures both stable and transient interactions²¹. A doxycycline-inducible, full-length SETD2-APEX2 fusion was integrated into immortalized human kidney epithelial (HKC) cells and proximity labeling was performed, followed by mass spectrometry (Fig. 1a). This analysis revealed 1876 significantly enriched proteins over control (P < 0.05, Supplementary Table 1), including known interactors of SETD2 such as RNAPII (e.g., POLR2A, POLR2B). Gene ontology classification revealed additional expected proteins including those involved in mRNA processing, transcription, and DNA repair (log2FC > 2, P < 0.05) (Fig. 1c, Supplementary Table 2). Unexpectedly, we found a significant enrichment of proteins involved in nuclear envelope functions, including lamin A/C, lamin B1, lamin B2, and emerin (Fig. 1a,b). SETD2-lamina association seemed paradoxical because lamins are located at the nuclear periphery where they typically tether heterochromatin to the lamina²², whereas SETD2 is known to associate with actively transcribed genes that are generally located towards the nuclear interior²³. Nonetheless, co-immunoprecipitation of endogenous or Halo-Flag tagged SETD2 (Fig. 1e) for endogenous lamina-associated proteins confirmed these interactions (Fig. 1d-1f).

Fig. 1 ∣. SETD2 interacts with nuclear lamina proteins.

a, Schematic of SETD2-APEX2 and the experimental approach of the proximity biotinylation method. Schematic created with BioRender. b, Volcano plot of SETD2-APEX2 mass spectrometry analysis, with nuclear lamina-associated proteins highlighted in red and RNAPII subunits in green. The x axis shows enrichment (Log2FC) of proteins in SETD2-APEX2-expressing (DoxON) cells compared with control cells (DoxOFF). The y-axis shows significance of enrichment. Two-tailed unpaired t-test was performed from three independent biological replicates. c, DAVID Gene Ontology (GO) analysis of mass-spec hits from 1b, defined as proteins with Log2FC>1 and p-value<0.05. Nuclear envelope is highlighted in pink. X-axis is fold enrichment, and y-axis is significance (−logFDR). d, Immunoblot analyses for lamin proteins following immunoprecipitation of endogenous SETD2 with IgG as a control. e, Immunoblot analyses for endogenous lamin proteins following immunoprecipitation of exogenous Flag-Halo tagged SETD2 using anti-Flag antibodies with Halo-Flag as control. f, Immunoblot analyses for endogenous SETD2 upon lamin A/C immunoprecipitation with IgG as a control. g, Co-immunofluorescence of cells expressing exogenous SETD2 immunostained with anti-SETD2 (green) and anti-Lamin A/C (red) antibodies at the indicated cell cycle phase. h, Proximity ligation assay (PLA) with cells expressing exogenous Flag-Halo tagged SETD2 using anti-Flag and anti-Lamin B1 antibodies. PLA pairs are in green; SETD2 (red) visualized using fluorescent Halo ligand JFX-549. Control cells lack Flag-Halo-SETD2. i, Quantitative analysis of PLAs between Flag and Lamin B1 shown in h. Two-tailed unpaired t-test was performed from three independent biological replicates, with n (left to right) = 62, 42, 13 cells. Data are mean ± s.e.m. DNA was counterstained with DAPI. All scale bars are 10 μm. Source numerical data and unprocessed blots are available in source data.

The discovery of SETD2-lamina associations prompted us to determine their spatiotemporal context. Co-immunofluorescence experiments revealed little overlap between SETD2 and lamins during interphase but strong co-localization in mitosis, when the nuclear envelope disassembles and lamins depolymerize²⁴ (Fig. 1g and Extended Data Fig. 1b). We next performed proximityligation assays (PLA) using a lamin B1 antibody and a Flag antibody to detect Halo-Flag tagged SETD2 and found a significant increase in frequency of SETD2-lamin B1 interaction during mitosis relative to interphase (Fig. 1h, i). Similar trends were observed when using antibodies toward endogenous SETD2 and lamin A/C (Extended Data Fig. 1c, d). The smaller interphasemitosis difference for SETD2-lamin A/C pairs compared with SETD2-lamin B1 pairs may be due to lamin A/C reported occupancy at active genes during interphase²⁵. These findings reveal a previously unrecognized link between SETD2 and lamina proteins during mitosis that prompted further investigation.

SETD2 loss leads to nuclear lamina defects

The nuclear lamina is a filamentous meshwork of lamin proteins beneath the nuclear envelope. Lamin proteins homo-polymerize in a head-to-tail orientation to form filaments²⁴, which contact large heterochromatin regions (lamina associated domains, LADs)²². By acting as a cushion to absorb mechanical forces, the lamina protects LADs, maintaining genome stability^26-28. Tissue stiffness is also directly proportional to lamin expression, indicating that lamins provide tensile strength to the nuclei^{29, 30}. Consistently, mutations or alterations in lamins increase nuclear dysmorphia and genome instability, contributing to diseases like progeria and various cancers^{24, 31}. Notably, nuclear dysmorphia and a compromised nuclear structure are hallmark features of cancer^31-34.

To investigate the functional significance of SETD2-lamin interactions, we depleted SETD2 in HKC cells using a doxycycline-inducible shRNA and performed confocal microscopy for lamin-stained cells. Knockdown of SETD2 (SETD2-KD) led to widespread nuclear abnormalities, including lamina invaginations, micronuclei, and blebbing – all hallmarks of a defective nuclear lamina (Fig. 2a). Quantification revealed 20.53% of SETD2-KD cells contained micronuclei versus 6.2% in controls (Fig. 2b). Similarly, 20.3% SETD2-KD cells exhibited blebs versus 7.12% in controls (Fig. 2d). A smaller subset (1.7%) exhibited interphase bridges (Fig. 2f).

Fig. 2 ∣. SETD2 loss disrupts nuclear-lamina architecture.

a, Representative HKC images immunostained with anti-Lamin A/C (red) and anti-Lamin B1 (green) antibodies in control and SETD2-depleted cells; indicated phenotypes are shown and wider field examples are in Extended Fig. 2a. b, Quantification of cells having one or more micronuclei in control (n = 277) and SETD2-depleted (n = 174) cells. Two-tailed unpaired t-test of three independent biological replicates; mean ± s.d. c, Quantification of invagination of lamina as internal Lamin B1 area per nucleus in control (n = 127) and SETD2-depleted cells (n = 146). Two-tailed non-parametric Mann-Whitney U test was performed; medians in red. d, Quantification of cells exhibiting blebbing from three independent replicates. n = 277 (control), n = 149 (SETD2-KD). Two-tailed unpaired t-test was performed from three independent biological replicates; means ± s.d. e, Quantification of cells with interphase bridges from three independent replicates. n = 624 (control), n = 585 (SETD2-KD) cells. Two-tailed unpaired t-test was performed from three independent biological replicates; means ± s.d. f, Quantification of nuclei Roundness (4πArea/perimeter²) in control (n = 272) and SETD2-depleted cells (n = 272). Two-tailed non-parametric Mann-Whitney U test was performed; medians in red. g, Quantification of all nuclear defects from three independent replicates. Two-tailed unpaired t-test was performed from three independent biological replicates; means ± s.d. n = 669 (control), 592 (SETD2-KD) cells. h, Representative MEF images showing immunostaining with anti-lamin B1 (green) antibody in vehicle or tamoxifen-induced Setd2 KO. Additional wider fields in Extended Data Fig. 2b. i, Quantification of cells having one or more micronuclei in control and Setd2 KO cells. Two-tailed unpaired t-test with data from three independent biological replicates; means ± s.d. j, Quantification of cells exhibiting invagination of lamina in control and Setd2 KO cells. Two-tailed unpaired t-test with data from three independent biological replicates; means ± s.d. k, Quantification of cells exhibiting blebbing in control and Setd2 KO cells. Two-tailed unpaired t-test with three independent biological replicates; means ± s.d. l, Quantification as percent of cells having interphase bridges in control and Setd2 KO cells. Two-tailed unpaired t-test on data from three independent biological replicates; means ± s.d. m, Quantification of binucleated control or Setd2 KO cells. Two-tailed unpaired t-test on data from three independent biological replicates; means ± s.d. n, Quantification of all nuclear defects from control or Setd2 KO cells. Two-tailed unpaired t-test on data from three independent biological replicates; means ± s.d. For 2i – 2n, n = 388 (Control) and 170 (Setd2 KO) cells. Scale bars,10 μm. DNA counterstained with DAPI. Source numerical data are available in source data.

To quantify nuclear abnormalities, we adapted the CellProfiler pipeline to measure invagination and nuclear shape³⁵. Automated image analyses revealed a significant increase in the internal lamin B1 area, indicating lamina invagination (Fig. 2c), and a reduction in nuclear “roundness” (4πArea/perimeter²), reflecting greater nuclear deformation in SETD2-KD cells (Fig. 2e). Notably, these phenotypes were often concurrent, as many SETD2-KD cells exhibited multiple abnormalities (Extended Data Fig. 2a). Overall, nearly 40% of SETD2-KD cells exhibited at least one nuclear defect (Fig. 2g).

The unexpected link between SETD2 and nuclear lamina stability prompted us to examine other cell types. Using mouse embryonic fibroblasts (MEFs) with floxed Setd2 alleles for conditional knock-out upon 4-hydroxytamoxifen treatment¹⁶, we observed even more severe nuclear abnormalities than those observed in HKC cells (Fig. 2h). The severity of the phenotypes prevented us from using automated image analysis; manual tabulation showed that 55.28% of Setd2-KO cells had one or more micronuclei versus 5.25% of control cells (Fig. 2i). Additionally, nearly 21% of Setd2-KO cells exhibited invaginations and 27% had blebs, compared to 1.4% and 2.8% of control cells, respectively (Fig. 2j, k). Interphase bridges (12% Setd2-KO cells versus 0.7% in controls, Fig 2I) and binucleation (12.3% Setd2-KO cells versus 2% of controls, Fig. 2m) were also elevated. In total, 86.47% of Setd2-KO cells exhibited at least one lamina defect (Fig. 2n). Together, these data underscore a conserved role for SETD2 in the maintenance of the nuclear lamina and in genome stability.

Given SETD2-lamin interaction predominated in mitosis (Fig. 1g, h, i), we hypothesized that this interaction would be most critical during M phase. If correct, we would expect that the loss of SETD2 only during mitosis would result in nuclear lamina defects. To test this, we generated an endogenous SETD2-FKBP (dTAG) fusion in HKC cells using CRISPR-Cas9, allowing temporal control of SETD2 loss at mitosis. We synchronized these cells at the G1/S boundary using the double-thymidine block method and released them back into the cell cycle (Extended Data Fig. 2c). These cells enter G2/M around 9h, peaking at 12h and finally exit mitosis by 15h, as evidenced by H3S10ph levels (Extended Data Fig. 2d). Treating cells with dTAG-47 at 6h post release from G1/S block resulted in an almost complete depletion of SETD2 by 9h, i.e., at the onset of G2 (Extended Data Fig. 2d). While nuclear morphology appeared normal at 9h, a slight increase in defects emerged by 12h. In contrast, a much greater increase nuclear morphology defects emerged by 15 h when most cells exited mitosis (9% at G2 vs 26% at G1) (Extended Data Fig. 2e, 2f). These data support the idea that nuclear lamina phenotypes observed in interphase cells emerge as a consequence of SETD2 loss during the preceding round of mitosis.

Given the temporal relationship between SETD2 loss and nuclear defects in HKC cells, we examined whether removing SETD2 during G2/M in floxed Setd2 MEFs would produce similar phenotypes (Fig. 2h). MEFs were serum-starved for 48 hours, then released into media with or without tamoxifen (to induce Setd2-KO) (Extended Data Fig. 2g). Immunoblot analysis of the mitotic marker H3S10ph revealed that cells entered mitosis at ~24h post-release (Extended Data Fig. 2h). At 24 hours, tamoxifen-treated and control cells appeared similar, whereas 24h of tamoxifen treatment in asynchronous cells was sufficient to cause widespread nuclear morphology defects. Remarkably, just 6 hours later, at the 30h timepoint (i.e., post mitosis), nuclear abnormalities surged from11.3% vs ~60%, rising further to 80.3% by 42 hours (Extended Data Fig. 2i, 2j). These data reinforce that SETD2’s mitotic function is critical for maintaining nuclear lamin stability, with defects manifesting in the subsequent cell cycle.

We next sought to test one last permutation that our data implied, namely, loss of SETD2 post-mitotically or in non-cycling cells should not cause nuclear morphology defects. To test this, we used immortalized retinal pigment epithelial (RPE) cells that exit the cell cycle into a G0/G1 phase upon contact inhibition. We generated a doxycycline-inducible SETD2 shRNA system in RPE cells that were subjected to contact inhibition for 7 days to drive post-mitotic arrest prior to SETD2 depletion for an additional 3 days (Extended Data Fig 3a). Flow cytometry analysis confirmed that 95% of cells arrested in a G0/G1 phase, and SETD2 depletion did not affect the arrest (Extended Data Fig. 3e). As a control, we also depleted SETD2 in asynchronous cells and observed a slight increase in G1 population upon SETD2 knockdown (Extended Data Fig. 3d). Importantly, immunofluorescence analysis revealed that SETD2-KD in G0/G1 arrested cells did not have any impact on nuclear morphology, whereas asynchronous cells depleted for SETD2 showed significant nuclear morphology defects (Extended Data Fig. 3b, 3c). These data confirm that the function of SETD2 in regulating nuclear morphology and genome stability is tied to mitosis, underscoring its direct role in these processes.

Although our data suggest a mitotic role forSETD2 in maintaining nuclear morphology via lamin interactions, other SETD2 activities could also contribute to the nuclear defect phenotypes. For example, SETD2 and H3K36me3 are important for DNA damage repair, a process that suppresses genome instability^{3, 16, 36}. Additionally, other heterochromatin-associated histone marks, such as H3K27me3 and H3K9me3, promote genome stability and histone-lamin anchoring^{37, 38}. However, our immunoblot analyses showed no changes in the levels of H3K27me3 and H3K9me3 upon SETD2 depletion, suggesting that while these modifications are relevant to lamin organization, their bulk levels remain unaffected by SETD2 loss in our assays (Extended Data Fig. 3f).

To assess whether DNA damage in SETD2 loss contributed to nuclear defects, we synchronized SETD2-FKBP (dTAG) cells via double-thymidine block and measured γH2AX levels during G2/M. Quantitative immunofluorescence did not reveal a significant difference between control and dTAG-47-treated cells at 9h and 12h timepoints; on the contrary, immunofluorescence showed a decrease in SETD2-depleted cells at 15h, indicating that SETD2 loss does not lead to intrinsic DNA damage that alone could account for the observed nuclear defects (Extended Data Fig. 3g). Notably, both control and SETD2-depleted cells showed elevated γH2AX levels at 9 h, likely reflecting synchronization stress rather than a result from the absence of SETD2 (Extended Data Fig. 2e, 2f). Overall, although DNA damage can contribute to nuclear abnormalities, SETD2 loss alone does not induce DNA damage at levels that could explain the severe nuclear defects observed.

Absence of SETD2 impairs lamin phosphorylation during G2/M

To clarify how SETD2 loss impacts nuclear morphology during mitosis, we next conducted performed live-cell confocal microscopy on control and SETD2-depleted cells stably expressing emerald-lamin A. Cells were synchronized using the double-thymidine block and release method, then imaged over time. In control cells, emerald-lamin A was completely depolymerized by metaphase, resulting in a homogenous signal. In contrast, a significant fraction (19.43%) of SETD2-depleted cells retained partially depolymerized lamin filaments throughout mitosis (Fig. 3a, b and Extended Fig. Data 4a). Additionally, lamina reassembly post-mitosis was slightly faster in SETD2-depleted cells compared with control cells (Fig. 3c), indicating that SETD2 absence also alters nuclear envelope breakdown and reassembly kinetics during the G2-M.

Fig. 3 ∣. SETD2 absence impairs lamin phosphorylation during G2/M.

a, Representative time-lapse images of control and SETD2-depleted HKC expressing Emerald-lamin A (green). DNA was labeled with SiR-Hoechst (red). Arrowheads point to partially depolymerized lamin filaments observed throughout mitosis. b, Quantification of cells having partially depolymerized Emerald-lamin A filaments from three independent replicates; n = 91 (Control) and 86 (SETD2-KD) cells. Two-tailed parametric t-test was performed; data shown are mean ± s.d. c, Quantitative analysis of lamina reassembly from live-cell microscopy of Emerald-lamin A-expressing cells. Time taken to reach maximum intensity from anaphase onset was determined in control (n = 10) and SETD2-KD (n = 12) cells from two independent experiments. Two-tailed non-parametric Mann-Whitney U test was performed; medians in red. d, Mass spectrometry analysis of immunoprecipitated lamin A/C and lamin B1 phosphorylation from three independent replicates. The x-axis shows position of amino acids of the indicated protein. Blue lollipop shows relative changes in phosphorylation level in SETD2-depleted cells compared with control. The y-axis is log2FC of shSETD2 over control. Dashed line and red circle indicate phosphorylation of serine 23 in lamin B1. Phospho-peptides of S23-lamin B1 were detected in all replicates of control but in neither of shSETD2 samples. e, Immunoblot analysis of vehicle and dTAG-47-treated SETD2-FKBP cells that were synchronized by the double-thymidine block method and then released for indicated time points. f, g, Representative images of S22-phos-lamin A/C (green, f), S392-phos-lamin A/C (green, g) and pan lamin A/C (red) from control and SETD2 depleted cells. DNA counterstained with DAPI. h, i, Quantitative analysis of immunofluorescence data from f, g, respectively. Integrated fluorescence intensity of either S22ph- or S392ph-lamin A signal from mitotic cells normalized to integrated intensity of pan lamin A/C signal in control (n = 252, S22ph; n = 256, S392ph) and shSETD2 (n = 175, S22ph; n = 222, S392ph) cells from two independent biological replicates. Two-tailed non-parametric Mann-Whitney U test was performed; medians in red. j, Representative images of indicated phenotypes from LMNA KO human fibroblasts cells expressing either WT, S->A (S22A; S392A) or S->D (S22D; S392D) mutant lamin A. Cells were immunostained with lamin A/C antibody (red); DNA counterstained with DAPI. k, Quantitative analysis of phenotypes shown in j. Data from two independent experiments is shown. Phenotypes were scored manually from n = 219 (WT), 242 (S—A) and 236 (S—D) cells; means ± s.d. All scale bars are 10 μm. Source numerical data and unprocessed blots are available in source data.

Nuclear envelope disassembly involves microtubule-dependent tearing of the nuclear envelope and subsequent lamin depolymerization²⁴. Lamin hyperphosphorylation at several conserved residues (i.e., S22 and S392 in lamin A/C and S23 and S393 in lamin B1) at late G2 drives depolymerization of lamin filaments and their detachment from chromatin^{24, 39}. At anaphase onset, rapid lamin dephosphorylation by mitotic phosphatases (PP1 and PP2A) facilitates lamin repolymerization, restoring the nuclear lamina and chromatin contacts^{24, 39}. This cycle of phosphorylation-dependent depolymerization and re-polymerization is critical to the maintenance of genome integrity and proper nuclear morphology. In agreement with this, studies have shown that serine-to-alanine mutations of S22/S392 in lamin A or S23/S393 in lamin B1 lead to incomplete breakdown of lamin filaments and persistent chromosome attachment during mitosis^{40, 41}.

Observing that SETD2 influenced the dynamics of lamin disassembly and reassembly, we hypothesized SETD2 might regulate lamina function through controlling lamin phosphorylation. To test this, we immunoprecipitated lamin A/C and lamin B1 from control and SETD2-KD cells and performed quantitative mass spectrometry to measure lamin phosphorylation. These analyses revealed a significant decrease in S22 and S390 phosphorylation with SETD2 depletion and nearly abolished phosphorylation at S23 in lamin B1 (Fig. 3d). Phosphorylation at some sites remained unchanged (e.g., T25 in lamin B1), and we did not detect certain phospho-sites in lamin A/C and lamin B1 (S392 and S393, respectively) in these analyses. Nonetheless, these data demonstrated that loss of SETD2 caused a decrease in lamin phosphorylation at specific sites that mediate lamin disassembly and reassembly.

Because lamin phosphorylation predominantly occurs in G2/M, we asked whether the decrease in lamin A/C phosphorylation in SETD2 depleted cells was due simply to changes in the G2/M population of cells. However, flow cytometry did not reveal drastic changes in G2/M population that could explain the decrease in lamin A/C S22 phosphorylation in SETD2-depleted cells (Extended Data Fig. 4b, c). Consistent with our previous findings⁴², we observed a subtle increase in the G1 population of cells and an equal decrease in the S-phase population.

To exclude the possibility that cell cycle alterations by SETD2 loss might contribute towards the observed decrease in lamin A/C phosphorylation, we examined lamin phosphorylation in cells acutely depleted of SETD2 during G2/M. To accomplish this, we synchronized our SETD2-dTAG cell line again using the double-thymidine block method and monitored the levels of lamin A/C phosphorylation at two key residues (S22 and S392) as cells progressed through G2/M. Immunoblot analyses revealed that lamin A/C phosphorylation rose from 9h post release and peaked at 12h, coinciding with mitotic entry (H3S10p). Strikingly, SETD2 depletion by dTAG-47-treated resulted in a significant decrease in phosphorylation levels of both lamin A/C S22 and S392 compared to control cells (Fig. 3e). Quantitative immunofluorescence of nocodazole-arrested mitotic cells further confirmed decreased lamin S22 and S392 phosphorylation in SETD2-depleted cells (Fig. 3f-3i). Similar results were observed in RPE and MEFs (Extended Data Fig. 4d, 4e), establishing SETD2 as a key regulator of lamin phosphorylation dynamics during G2/M.

Finally, we tested whether decreased lamin phosphorylation explains the nuclear morphology defects observed under SETD2 loss. We reasoned that lamin mutations that prevent its phosphorylation should phenocopy the nuclear morphology defects observed upon depletion of SETD2. Using LMNA-KO human fibroblasts stably expressing doxycycline-inducible wild-type lamin A or lamin A mutated to S22A/S392A or S22D/S392D⁴³, we observed that the S22A/S392A lamin A mutant fibroblasts exhibited nuclear abnormalities highly similar to those seen in SETD2-deficient HKC, RPE and MEFs (Fig. 3j, 3k and Extended Data Fig. 4f). Thus, decreased lamin phosphorylation at G2/M underlies the nuclear morphology and genome stability defects triggered by SETD2 loss.

A non-catalytic role for SETD2 in nuclear lamina regulation

ccRCCs often contain a heterozygous deletion of SETD2 due to the loss of chromosome arm 3p, yet these cells retain significant levels of H3K36 methylation^{12, 16}. Cell-based models of SETD2 haploinsufficiency also retain H3K36me3 in addition to harboring a variety of genome integrity defects that include micronuclei and chromosomal bridges¹⁶. Notably, studies suggest that SETD2 methylation of α-tubulin may be responsible for these genome stability defects^{16, 44}. However, given our findings that SETD2 associates with lamins and that lamin phosphorylation defects phenocopy SETD2 loss, we asked whether SETD2’s catalytic activity underlies its lamina-regulatory role. Surprisingly, treating HKC cells with the highly potent SETD2 methyltransferase inhibitor (EPZ-719; SETD2i)⁴⁵ erased H3K36me3 but did not lead to any significant nuclear defects (Fig. 4a-c). In complete contrast, acute depletion of SETD2 consistently resulted in both nuclear abnormalities and a decrease in lamin phosphorylation (Fig. 4a-c).

Fig. 4 ∣. Nonenzymatic function of SETD2 in maintenance of nuclear morphology.

a, Immunoblot analyses of indicated proteins from control, SETD2-KD and SETD2i (EPZ-719)-treated cells for 48 hr. b, Representative images of control, SETD2-KD and SETD2i-treated HKC cells immunostained with lamin B1 antibody. c, Quantitative analysis of nuclear defects from b. Two-tailed unpaired t-test was performed from two independent biological replicates; data are mean ± s.d. d, Schematic of doxycycline inducible shRNA and genetic complementation system in HKC cells. e, Immunoblot analyses of indicated proteins in control, SETD2-KD or SETD2-KD cells expressing either WT or mutant SETD2. f, Representative images of control or SETD2-KD cells expressing the indicated transgene, immunostained with lamin A/C and lamin B1 antibodies. DNA was counterstained with DAPI. Scale bar is 10 μm. g, Quantification of cells containing one or more micronuclei from n (left to right) = 154, 188, 204, 210, 204, 157 cells. One-way ANOVA with Dunnett’s multiple comparisons test was performed on three independent biological replicates. Data shown are mean ± s.d. h, Quantification of cells with one or more blebs in the indicated samples from n (left to right) = 153, 193, 294, 233, 204, 157 cells. One-way ANOVA with Dunnett’s multiple comparisons test was performed on three independent biological replicates. Data shown are mean ± s.d. i, Quantification of nuclei Roundness (4πArea/perimeter²) in the indicated samples. Kruskal-Wallis with Dunn’s multiple comparisons test was performed on n (left to right) = 65, 102, 81, 104, 106, 96 cells from three biological replicates; red bar indicates median. j, Quantification of Invaginations in the indicated samples. Kruskal-Wallis with Dunn’s multiple comparisons test was performed on n (left to right) = 91, 125, 122, 114, 132, 120 cells from three biological replicates; red bar indicates median. k, Quantification of total nuclear defects in control, SETD2-KD or SETD2i-treated cells for 2d or 6d. Two-tailed unpaired t-test was performed from two independent biological replicates; means ± s.d. l, Quantification of cells with micronuclei in control, SETD2-KD or SETD2i-treated cells for 2d or 6d. Two-tailed unpaired t-test was performed from two independent biological replicates. n (for k and l, left to right) = 387, 413, 404, 300, 357, 355 cells; means ± s.d. Source numerical data and unprocessed blots are available in source data.

To analyze the foregoing findings further, we used a MEF model in which one of the two Setd2 alleles could be conditionally deleted, creating a haploinsufficiency scenario. As expected, losing one Setd2 allele did not significantly alter H3K36me3 levels (Extended Data Fig. 5a). In striking contrast, haploinsufficiency resulted in nuclear lamina abnormalities and genome stability defects similar to those observed in Setd2-KO MEFs (Extended Data Fig. 5b, 5c). By contrast, inhibiting SETD2 by EPZ-719 erased H3K36me3 but did not result in any nuclear morphology or genome stability defects (Extended Data Fig 5a-c). These data indicated two crucial facts. First, the methyltransferase activity of SETD2 is dispensable for its role in regulating the nuclear lamina as well as for the prevention of a host of genome integrity defects previously thought to be associated with SETD2’s catalytic function. Second, both alleles of SETD2 are required to maintain nuclear lamina shape and function. Thus, precise intracellular concentrations of SETD2, and not its catalytic activity, is vital to the stability of the nucleus and integrity of the genome.

The N-terminus of SETD2 regulates nuclear lamina stability

Having established that the catalytic activity of SETD2 is dispensable for nuclear lamina maintenance, we next asked which region of SETD2 contributes to this function. Accordingly, we generated a doxycycline-inducible HKC cell system, wherein endogenous SETD2 was depleted with an inducible shRNA expression construct while simultaneously inducing the expression of either WT or various SETD2 mutant forms (Fig. 4da). Immunoblot analyses confirmed previous reports that expression of SETD2 lacking the N-terminus of SETD2 (tSETD2) restores H3K36me3^{18, 46}. Additionally, and as expected, intact WT and the SRI domain mutant (SRI_mut, R2510H), but not a catalytically inactive SET domain mutant (SET_mut, R1625C), were also able to rescue H3K36 methylation. Significantly, immunoblot analysis of lamin A/C phosphorylation revealed that expression of either the WT, SET_mut or SRI_mut forms of SETD2 rescued lamin phosphorylation in SETD2-KD cells. By contrast, however, cells expressing tSETD2 were deficient in lamin phosphorylation, similar to SETD2-KD cells (Fig. 4e), indicating that the N-terminus of SETD2 is required for lamin phosphorylation.

To confirm that the nuclear defects in SETD2-KD cells mirrored those lacking the N-terminus, we performed immunofluorescence imaging of lamin proteins in cells expressing WT or various SETD2 mutants (Fig. 4f-j). Both WT SETD2 and the SRI_mut restored nuclear morphology in SETD2-depleted cells, while tSETD2 failed to rescue blebbing, invaginations, or roundness defects. We note, however, that tSETD2 did lead to a partial rescue of micronuclei generation. Collectively, these data demonstrate that the N-terminus of SETD2 is required for lamin phosphorylation and essential to maintain nuclear lamina shape and genome stability.

Our studies with the WT and mutant SETD2 forms also revealed a curious finding. Although SET_mut rescued nuclear invaginations and blebbing, it did not completely rescue nuclear roundness or prevent micronuclei generation (Fig. 4g, 4id, f). This finding seemed at odds with our earlier studies showing that acute inhibition SETD2’s activity did not cause nuclear morphology defects (Fig. 4a-c and Extended Data Fig. 5a-c). However, the complementation experiments required ~6 days to allow for full expression and rescue of H3K36me3, whereas prior knockdowns and chemical inhibition studies were performed at 2 days, when SETD2 and H3K36me3 were first significantly depleted. To test whether extended inhibition of SETD2 activity leads to increased micronuclei generation, we compared 2- vs. 6-day treatments with EPZ-719 and examined if these cells for nuclear morphology defects. At the 2-day time point, inhibitor treated cells showed no nuclear defects, but by 6 days, a small yet significant increase in micronuclei were detected (Fig. 4k, 4l; Extended Data Fig. 5d). This finding is consistent with what was observed for SETD2 depletion at 2 or 6 days, which also showed an increase in the overall percent of nuclear defects (from 37% [2d] to 53% [6d]) (Fig. 4k, 4l)We surmise that increased micronuclei found with prolonged SETD2 inhibition is likely a secondary effect over prolonged SEDT2 and H3K36me3 loss.

SETD2 N-terminus interacts with lamins

Because N-terminally truncated tSETD2 did not rescue the nuclear and lamin phosphorylation abnormalities associated with SETD2 depletion, we examined how the N-terminus of SETD2 contributes to nuclear lamina stability. We employed APEX2-mediated proximity labeling to determine the protein-protein interaction differences between WT and tSETD2 in HKCs stably expressing SETD2-APEX2 or tSETD2-APEX2, followed by mass spectrometry analysis. WT SETD2 showed 1068 significantly enriched interacting proteins compared with tSETD2 that showed 458 significantly enriched interacting proteins (log2FC > 1, p<0.05) (Fig. 5a; Supplementary Table 1). Notably, nuclear lamina-associated proteins (e.g., lamin A/C, lamin B1, and emerin) were enriched with WT SETD2 compared with tSETD2 (Fig. 5a). Immunoprecipitation of Halo-Flag tagged WT SETD2 or tSETD2, followed by immunoblotting, confirmed these results: WT SETD2 robustly co-precipitated lamin A/C, lamin B1, and emerin, whereas tSETD2 did not associate with either lamin B1 or emerin. Intriguingly, tSETD2 selectively associated with lamin C, whereas WT SETD2 associated with both lamin A and C. Because lamin C is localized at the nuclear interior during the early G1 phase of the cell cycle prior to its assembly into the nuclear envelope⁴⁷, we suspect the differential preferences of tSETD2 and SETD2 for different lamins may reflect distinct SETD2-lamin functions (e.g., transcription vs nuclear lamina formation).

Fig. 5 ∣. SETD2 N-terminus associates with the mitotic lamin kinase CDK1.

a, Mass-spectrometry analysis of SETD2-APEX2 and tSETD2-APEX2 interactome. Y-axis is significance, x-axis is log2FC of tSETD2/WT normalized data. Two-tailed unpaired t-test was performed from three independent biological replicates. Nuclear lamina proteins highlighted in red. Blue dots are significant hits (p<0.05). b, Immunoblot analysis of the indicated proteins following immunoprecipitation of Flag-Halo-tagged full-length WT-SETD2 or tSETD2. Flag-Halo served as control. c, Gene Ontology analysis of APEX2 mass-spec data from a. Relative enrichment between WT-SETD2 and tSETD2 mass-spec datasets of the proteins in the indicated gene ontologies. d, Immunoblot analysis of the indicated proteins following immunoprecipitation of endogenous CDK1 from the indicated samples. e, Proximity ligation assay (PLA) using anti-CDK1 and anti-lamin B1 antibodies in SETD2-depleted cells expressing either WT SETD2 (n = 164) or tSETD2 (n = 179). PLA pairs are in green; DNA was counterstained with DAPI. Scale bar is 10μm. f, Quantitative analysis of PLA pairs between CDK1 and lamin B1 shown in e; means ± s.e.m. g, Schematic of SETD2 with indicated regions (A, B, or C) to illustrate regions deleted to create individual N-terminal mutants. h, Representative confocal images of SETD2-depleted cells expressing either WT SETD2 or the indicated N-terminal mutants, immunostained with lamin B1 (green). SETD2 (red) visualized using Halo ligand JFX-549. DNA counterstained with DAPI. Scale bar is 10 μm. i, Quantification of total nuclear defects in HKC cells expressing either WT SETD2 or the indicated N-terminal mutants. One-way ANOVA with Dunnett’s multiple comparisons test was performed on three independent biological experiments with n = 131, 152, 143, 256, 297, 199 (left to right) cells; means ± s.d. Source numerical data and unprocessed blots are available in source data.

SETD2 coordinates mitotic kinase CDK1-lamin phosphorylation

To clarify how SETD2’s N-terminus controls nuclear lamina organization, we further examined the protein-protein interactions differences between WT SETD2 and tSETD2. Gene Ontology analysis revealed that WT SETD2 preferentially interacts with proteins related to nuclear envelope, mRNA processing, and mitosis. Notably, the full-length form of SETD2 also preferentially interacted with cyclin dependent kinase 1 (CDK1) (Fig. 5c). This observation was intriguing because CDK1 is the master kinase that regulates the G2-to-M transition by phosphorylating multiple substrates, including lamins⁴⁸. Importantly, CDK1 phosphorylation of lamin A/C at S22, S390, S92 and lamin B1 at S23, S391, S393 during G2-to-M transition is a required step in disassembly of lamins and the nuclear envelope^{24, 39}. The finding that this kinase also associates with the N-terminus of SETD2, a result we confirmed by co-immunoprecipitation experiments (Fig. 5d and Extended Data Fig. 6a), deepened the connection between SETD2 and lamina regulation.

SETD2’s association with lamins and CDK1 suggested that its N-terminus may function as a scaffold to locally coordinate CDK1-mediated lamin phosphorylation. If correct, we predicted that deletion of the SETD2 N-terminus would reduce CDK1-lamin association during mitosis. To test this, we performed PLA experiments using antibodies against CDK1 and lamin B1 in SETD2-depleted cells that expressed either full-length SETD2 or tSETD2 (Fig 5e, 5f and Extended Data Fig. 6b). Cells expressing tSETD2 showed a significant decrease in CDK1-lamin B1 interaction compared to those expressing full-length SETD2. These findings indicate that the SETD2 N-terminus is important for bringing CDK1 and lamins together, explaining how the absence of SETD2 or its N-terminus diminishes lamin phosphorylation and disrupts nuclear lamina dynamics.

The N-terminal half of SETD2 (amino acids 1-1322) contains multiple regions of low complexity, intrinsically disordered regions (IDRs), with the highest concentration being within the first 500 N-terminal amino acids (Fig. 5g). Because IDRs can facilitate liquid-liquid phase separation⁴⁹, we asked whether these IDRs or N-terminal regions are important for its nuclear lamina maintenance function. Using a dox-inducible system, we expressed various N-terminal deletion mutants (region A: 1-503, region B, 504-816, region C, 817-1322, and region B+C, 504-1322) (Fig. 5g) in SETD2-depleted cells to identify regions necessary to rescue the nuclear morphology defects. Intriguingly, SETD2 lacking region A fully rescued the nuclear lamina and genome integrity defects (Fig. 5h, 5i and Extended Data Fig. 6c), indicating that the N-terminal IDRs are dispensable for this activity of SETD2. In stark contrast, while deletion of either region B or region C alone produced minimal effects on nuclear morphology, the deletion of both the B and C regions phenocopied the loss of SETD2 or absence of the entire N-terminus (Fig. 5h, 5i and Extended Data Fig. 6c). Thus, regions B and C (a.a. 504-1322) are required for SETD2’s role in nuclear lamina organization.

SETD2 N-terminus enhances CDK1 phosphorylation of lamin A in vitro

Having determined that the N-terminal B/C regions in SETD2 are critical for nuclear lamina stability, we next investigated how these regions biochemically contribute to this function. First, using recombinant proteins, we asked whether regions B and/or C could interact directly with lamin A and/or CDK1. However, given the disordered nature of the SETD2 N-terminal regions, their expression and purification from several sources proved to be extremely difficult. We overcame this challenge by fusing the SETD2 B or C regions to 6xHis-SUMO on the N-terminus and GB1 (B1 domain of Protein G) on the C-terminus, which greatly increased their expression and solubility⁵⁰. A similar problem was encountered with expression of full-length lamin A, and therefore we resorted to expressing it as two continuous fragments, namely, GST-lamin A-NT (1-386 a.a) and GST-lamin A-CT (387-664 a.a.). In GST pulldown assays with these fragments, the SETD2-C region (His-SUMO-C-GB1) showed strong binding to the C-terminus of lamin A and weaker interaction with its N-terminus (see higher exposure in Extended Data Fig. 7a). This was in contrast to results with the SETD2-B region, which showed only weak interaction with both lamin fragments (Fig. 6a and Extended Data Fig. 7a).

Fig. 6 ∣. SETD2 N-terminus enhances CDK1-mediated lamin phosphorylation and is required to suppress ccRCC tumor growth.

a, Western blot analyses with indicated antibodies following in vitro GST pulldown assays using either GST-tagged lamin A N-terminus (NT) or C-terminus (CT) or free GST (control). b, Immunoblot analyses using indicated antibodies following GST pulldown assays using either GST-tagged CDK1/Cyclin B1 complex or free GST (control). In both, a and b, anti-SUMO antibody was used to detect either SUMO-tagged SETD2 or a control His-SUMO-GB1 protein. c, Immunoblot analysis of indicated proteins following an in vitro kinase assay using lamin A-NT and CT as substrates and CDK1/Cyclin B1 as kinase in the presence of either SETD2-C or a control His-SUMO-GB1 protein. Asterisk indicates cross-reacting band. d, Representative images of UMRC2 cells expressing either Halo (control) or Halo-tagged WT SETD2 or tSETD2 or BC-tSETD2, immunostained with pan-lamin A/C antibody. DNA counterstained with DAPI; scale bar is 10 μm. e, Quantification of nuclei shape (solidity) in the indicated samples from d. Kruskal-Wallis with Dunn’s multiple comparisons test was performed on n (left to right) = 441, 342, 396, 367 cells from three biological replicates; medians in red. f, Quantification of Invaginations in the indicated samples from d. Kruskal-Wallis with Dunn’s multiple comparisons test was performed on n (left to right) = 202, 191, 238, 198 cells from three biological replicates; medians in red. g, Quantification of cells from d containing one or more micronuclei. One-way ANOVA with Dunnett’s multiple comparisons test was performed on n (left to right) = 385, 356, 408, 423 cells from three independent biological replicates; means ± s.d. h, Representative images of soft agar colony formation assays showing 3D colony growth of UMRC2 cells expressing the indicated transgenes. Scale bar is 0.5mm. i, Quantification of number of colonies from data in h, normalized to Halo (control) from three independent biological replicates. One-way ANOVA with Dunnett’s multiple comparisons test was performed; means ± s.d. j, Image of tumors derived from sub-cutaneous xenografts of UMRC2 cells expressing the indicated transgenes. k, Quantification of tumor weight from j. One-way ANOVA was performed for multiple comparison; means ± s.d. l, Model depicting SETD2’s molecular functions in interphase (left) and G2/M (right). Source numerical data and unprocessed blots are available in source data.

Given that proximity labeling and co-immunoprecipitation identified the SETD2 N-terminus associates with CDK1, we next asked whether this interaction was direct in vitro. Free GST or GST-tagged CDK1/Cyclin B1 complex was pulled down using glutathione beads in the presence of either His-SUMO-B-GB1, His-SUMO-C-GB1, or the His-SUMO-GB1 control followed by immunoblot analyses using anti-SUMO antibodies. Again, we found that both B and C regions could interact with CDK1/Cyclin B1 complex, with the SETD2-C region showing a much stronger interaction than the SETD2-B region (Fig. 6b). Since CDK1 exists in complex with either Cyclin B1 or Cyclin A2 in cells, we asked if CDK1/Cyclin A2 complex could also interact with the SETD2 N-terminus. Similar to their interaction with the CDK1/Cyclin B1 complex, the SETD2 B and C regions also directly interacted with a CDK1/Cyclin A2 complex (Extended Data Fig. 7b).

Given the SETD2 N-terminus regulates nuclear morphology and lamin phosphorylation, and its B and C regions directly interact with lamin A and CDK1, we asked whether these regions of SETD2 directly influence lamin phosphorylation by CDK1/Cyclin B1 complex in vitro. We combined lamin A-NT and lamin A-CT with CDK1/Cyclin B1 in the presence of His-SUMO-B-GB1, His-SUMO-C-GB1, or a His-SUMO-GB1 control, then monitored lamin A S22 and S392 phosphorylation over a 40-minute time course (Fig. 6c). Strikingly, the SETD2-C region significantly increased lamin phosphorylation compared to the control, while the B region resulted in only a minor increase in lamin phosphorylation (Extended Data Fig. 7c). These data suggest that the B and C regions of SETD2 control nuclear lamina organization by directly promoting efficient lamin phosphorylation by CDK1.

The SETD2 N-terminus inhibits ccRCC tumor growth in a xenograft model

Our findings reveal new connections between SETD2, CDK1, and lamins that impinge on nuclear lamina function and genome stability. To explore the disease significance of these findings, we turned to ccRCC as a model given the important function of SETD2 in suppressing these cancers. As mentioned above, ccRCC typically exhibits 3p arm loss, resulting in concurrent one-copy loss of VHL, PBRM1, BAP1 and SETD2, rendering these four genes haploinsufficient^13-15. Since SETD2 haploinsufficiency in MEFs led to profound nuclear morphology defects, we reasoned that ccRCCs bearing SETD2 haploinsufficiency would show similar nuclear morphological defects that could potentially be rescued by adding back full-length SETD2. Indeed, patient-derived UMRC2 cells (which harbor a heterozygous loss of the 3p arm) displayed severe nuclear morphology defects reminiscent of the range of defects observed in all of our cell line examples that lacked SETD2. Remarkably, re-expression of WT SETD2 rescued these defects and restored round nuclear shape (Fig. 6d and Extended Data Fig 7d). Further analyses of the tSETD2 and BC-tSETD2 re-expressed forms confirmed again that the ability to rescue the morphology defects in UMRC2 was due to the N-terminus and specially the B and C regions. (Fig. 6d-6g and Extended Data Fig. 7e). Interestingly, tSETD2 partially rescued the micronuclei phenotypes observed in control UMRC2 cells, confirming our earlier finding that micronuclei are partially generated through H3K36me3-dependent and independent mechanisms.

Lastly, we asked whether SETD2’s tumor suppressor activity in ccRCC is also driven by its N-terminus. Because the B and C regions of SETD2 restored normal nuclear morphology in UMRC2 cells, we hypothesized that these same regions of SETD2 might similarly be important to UMRC2 growth and proliferation in vitro and in vivo. Using soft agar assays to assess anchorage-independent growth, re-expression of WT SETD2 and BC-tSETD2 in UMRC2 cells inhibited 3D colony formation compared with control (Halo) cells and tSETD2-expressing cells (Fig. 6h, 6i). Next, we performed subcutaneous xenografts of these cell lines to ask whether growth inhibition by B/C regions was true for in vivo tumor growth. Strikingly, the UMRC2-derived xenografts phenocopied the 3D colony soft agar assays and showed that WT SETD2 and BC-tSETD2, but not tSETD2, suppressed the tumor growth usually observed with control UMRC2 cells (Fig. 6j, 6k and Extended Data Fig. 7g). Collectively these data link the B and C regions of SETD2 – essential for nuclear morphology and genome stability – to its tumor suppressing function.

Discussion

Our studies uncover an unexpected link between SETD2, the nuclear lamina, and CDK1, highlighting a crucial role for SEDT2 in the organization and function of the nuclear lamina. Importantly, we provide evidence that this non-catalytic function of SETD2 may underlie its tumor suppressing activities. Although the precise molecular details of how the N-terminus of SETD2 regulates the enhancement of CDK1-mediated lamin phosphorylation remain to be elucidated, we propose a model in which SETD2’s N-terminus serves as a scaffold, bringing CDK1 into proximity with lamins to facilitate their phosphorylation during mitosis (Fig. 6l). Supporting this scaffolding concept, studies have shown that the super elongation complex member AFF4 enhances P-TEFb (CDK9/Cyclin T1) activity and interaction with the HIV protein Tat by directly binding to P-TEFb and Tat⁵¹. Future studies will be needed to unravel the mechanistic basis of this regulation. Nonetheless, the connection demonstrated here between SETD2 and nuclear lamina proteins not only extends SETD2’s functional repertoire but also positions it as a critical regulator of mitotic processes previously unimagined to involve chromatin modifiers.

To date, most phenotypes associated with SETD2 have been attributed to its methylation activity. However, using chemical inhibition, targeted protein degradation, and genetic complementation experiments, we demonstrated that SETD2 participates in a fundamental aspect of the cell cycle that is independent of its catalytic activity. Many other chromatin regulators also contain large, disordered regions similar to SETD2; thus, we predict that some of these regions may have critical noncatalytic and non-nucleosomal activities. Interestingly, a study by Schibler et al. has implicated many chromatin regulators to be involved in nuclear morphology regulation³⁸; it will be exciting to determine whether protein regions outside of the enzymatic activities of these chromatin regulators are also involved in regulation of nuclear morphology.

An intriguing observation from these studies is the delineation of distinct functions within the SETD2 N-terminus. Despite the high concentration of IDRs within the first 500 amino acids (region A) and their role in LLPS mediated protein sequestration, our initial hypothesis was that these IDRs would be required for SETD2’s function in the maintenance of nuclear morphology by bringing CDK1 and lamin filaments in close proximity to facilitate lamin phosphorylation. However, this region was completely dispensable for this function. It remains to be determined how the A region contributes to the overall functions of SETD2, although we hypothesize that these IDRs may play important role in transcription and splicing. By contrast, we defined cell cycle-related functions in the B and C regions, indicating an interesting functional diversity within the N-terminus. The idea that SETD2 would have a cell cycle function in mitosis is supported by studies that show SETD2 regulates the timing of the cell cycle and that its protein level is stabilized during the G2/M phase⁴². Additionally, mammalian SETD2 is directly targeted for phosphorylation by CDK1⁵². In all, these findings paint a vital function for SETD2 that is not expected from its previously defined activities in transcription. Interestingly, the extended SETD2 N-terminus required for the nuclear lamina function we describe here is missing in less complex organisms such as yeast^{1, 7}, implying an evolutionary expansion of the SETD2 N-terminus to fulfill multiple functions as the complexity of eukaryotic cells increased.

Intriguingly, our findings also showed that SETD2 level rises at G2/M, coinciding with lamin phosphorylation (Fig. 3e and Extended Data Fig. 2d, 2h). This temporal regulation is consistent with a key G2/M function for SETD2 wherein SETD2 tethers CDK1 and lamins together to promote lamin phosphorylation (Fig. 6l). Indeed, cells lacking SETD2 or its N-terminus exhibit defects matching this idea. However, the mechanism by which SETD2 stabilization occurs in G2/M remains unclear. As the SETD2 N-terminus is directly targeted by CDK1⁵², and our data show a slight increase in SETD2 mRNA (Extended Fig. 4g), both protein modification and transcriptional upregulation may contribute to its mitotic stability.

Another key finding from this study is the critical importance of precise SETD2 protein levels. This was best illustrated in our model MEF cells, where Setd2 haploinsufficiency resulted in severe defects in nuclear morphology and genome stability, even though H3K36me3 levels were mostly unaltered (Fig. 4d, e). Because greater than 90% ccRCCs have heterozygous deletions of SETD2 in addition to VHL, PBRM1 and BAP1^{13, 15}, we hypothesize that SETD2 loss becomes a pivotal driving event in the initiation of these tumors through its impact on nuclear lamina stability and genome integrity. Indeed, UMRC2 cells that harbor a 3p loss exhibit profound nuclear morphology defects that are rescued by SETD2 reintroduction – an effect tied to suppressed tumor growth (Fig. 6j, k). However, a limitation of these studies is the challenges of determining if other regions in SETD2 also contribute with the B and C-regions in tumor growth, and the ability to know if the rescue of morphological defects by re-expressing SETD2 is causal in this process.

Beyond heterozygous 3p loss, approximately 8-30% of ccRCCs harbor a variety of additional SETD2 mutations that serve as secondary events^{8, 13-15}. Analysis of SETD2 mutations specific to ccRCC in the COSMIC database revealed that the majority of these mutations (~15% of total) were either nonsense or frameshift mutations, resulting in truncated proteins that were distributed across the entire SETD2 coding region without localized “hot spots”. In contrast, missense mutations showed enrichment within the SET domain, accounting for 57% of all the missense mutations, with a weaker enrichment in the SRI domain (Fig. 1b). Among these, the recurrent SET domain mutation, R1625C, is thought to create a null allele because it renders the enzyme inactive⁸. However, we previously showed that this mutation destabilizes SETD2 protein⁴⁶, thereby creating a hypomorph allele in addition to its catalytic inactivation. Taken together, while some ccRCCs appear to select for loss of SETD2’s catalytic activity, the predominance of truncating mutations and widespread heterozygous loss of SETD2 strongly suggests that overall SETD2 protein levels are critical to its cellular functions.

We note that mutations in the SET or SRI domains that abolish H3K36me3 often arise later in ccRCC and are associated with metastasis rather than tumor growth⁵³. Coupled with our findings, this suggests two distinct roles for SETD2 in ccRCs: one, driven early by SETD2 haploinsufficiency via 3p arm loss that affects nuclear lamina stability that promotes tumor initiation, and a second, more advanced phase involving the loss of catalytic activity that promotes metastasis (~15% of cases). Finally, mutations or altered expression of lamins are known to cause nuclear morphological defects, which, like SETD2 mutations, can promote tumorigenesis through increased aneuploidy and genome instability^{29, 31-33, 54}. Given SETD2 is frequently mutated across many cancers – particularly in ccRCCs characterized by genome stability defects – we propose that the function of SETD2 as a tumor suppressor may be mediated, at least in part, by its role in regulating nuclear envelop disassembly during the cell cycle. Overall, our findings redefines the role of SETD2, shifting its primary significance from a histone methyltransferase to a multifaceted regulator of nuclear morphology and genome stability. They also provide a framework for exploring the molecular underpinnings of tumor suppression regulated by SETD2.

Methods

Chemistry

EPZ-719 was synthesized using a modification of the published route (Scheme S1)⁵⁵. Compound 1 was reacted with MsCl to afford 2, which was deprotected with HCl, followed by reductive amination with tert-butyl (R)-(3-oxocyclohexyl)carbamate to give 3. The N-deprotection of 3 and amide coupling with 4-fluoro-7-methyl-1H-indole-2-carboxylic acid to afford EPZ-719. The ¹H and ¹³C NMR spectra match the reported values.³⁰

Scheme S1. Modified synthesis of EPZ-719.

(A) MsCl, diisopropylethylamine, DCM, 91%; (B) (1) HCl/DCM/ether, then (2) tert-butyl (R)-(3-oxocyclohexyl)carbamate, NaBH₄, MeOH/NaOMe, 38% over 2 steps; (C) (1) HCl/DCM/ether, then (2) 4-fluoro-7-methyl-1H-indole-2-carboxylic acid, DIC, HOBt, DMAP, diisopropylethylamine, DCM, 43% over 2 steps.

Antibodies

SETD2 (in-house generated, 1:1000), GAPDH (Cell Signaling, 14C10, 1:5000), H3K36Me3 (Abcam, ab9050 and in-house generated, 1:5000), CDK1 (Millipore, 06-9230, 1:2000), Lamin A/C (Sigma, SAB4200236, 1:2000), Ser22 Lamin A/C Phos (Cell Signaling, 13448S, 1:1000), S392 Lamin A/C Phos (Invitrogen, PA5-104731, 1:1000), RNAPII RPB1 (Cell Signaling, 14958S, 1:1000), M2 FLAG (Sigma, F1804, 1:2000), Emerin (Thermo, MA5-31328, 1:2000), Lamin B1 (Abcam, ab16048, 1:2000), Lamin B1 (ProteinTech, 66095-1-lg, 1:2000), SRSF1 (ThermoFisher, 32-4600, 1:2000). CDK1 (clone A17.1.1 Millipore-Sigma, MAB8878, 1: 1000).

Cell culture and treatment

Human kidney tubule epithelial (HKC) were obtained from Dr. Lorraine Racusen, (Johns Hopkins Hospital, Baltimore, MD) and Bj5a LMNA mutant fibroblast cells were a gift from Dr. Kohta Ikegami (Cincinnati Children’s Hospital). HKC, RPE, UMRC2 and Bj5a were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 100 U ml⁻¹ penicillin, 100 μg ml⁻¹ streptomycin and 10% fetal calf serum. Setd2^fl/fl and Setd2^fl/wt mouse embryonic fibroblasts (MEFs), were maintained in phenol-red free DMEM supplemented with 100 U ml⁻¹ penicillin, 100 μg ml⁻¹ streptomycin and 10% fetal calf serum. All cultures were maintained at 37°C and 5% CO₂. MEF cells were serum starved in medium containing 0.1% FBS for 48h prior to refeeding (release) with normal medium. To induce Setd2 deletion, MEFs were treated with 3 μM 4-hydroxytamoxifen (4-OHT) for 48 h or indicated time points. HKC cells were synchronized using double-thymidine block method by treating cells with 2 mM thymidine for 16 h, followed by release for 8 h and re-treatment with 2 mM thymidine for 16 h to bring cells to G1/S boundary. Cells were then released for appropriate times as indicated. For mitotic arrest, cells were treated with 50 ng/ml of nocodazole for 4 h (for IF) and 12 h (for co-IP).

Cloning/Plasmid generation

An shRNA against SETD2 (TRCN0000237839) was cloned into Tet-pLKO-puro vector (Addgene, 21915). Dox-inducible full-length SETD2 constructs were generated by Gibson assembly of the N-terminus (1-1322 a.a), the C-terminus (1323-2564 a.a) and Halo-3xFLag (gBlock, IDT) into Xlone-GFP (Addgene, 96930) following KpnI/SpeI digestion to excise EGFP cDNA. Generation of the C-terminus (also referred to as tSETD2) and its mutant derivatives (R1625C, R2510H) were previously described⁵⁶. Mutants consisting of N-terminal truncations or deletions of regions A, B and/or C (Fig. 5g) were also generated by Gibson assembly. For protein expression, SETD2 N-terminus regions B or C were cloned by Gibson assembly together with SUMO and GB1 fragments into pET28a to create 6xHis-SUMO-SETD2(B/C)-GB1. Lamin A NT (1-386) and lamin A-CT (387-664) were PCR amplified with a 6xHis on C-terminus from plasmid containing human pre-lamin A and cloned into pGEX-6p1 by restriction cloning.

Recombinant protein purification

BL21(DE3) E. coli cells were transformed with plasmids and grown on LB plates with the relevant antibiotic. Starter cultures were transferred to 1L of LB + antibiotic for each construct and grown to OD₆₀₀ ~0.6 at 37°C and 200 RPM (approx. 4 hours). Recombinant protein expression was induced with 1 mM Isopropyl β-D-1-thiogalactopyranoside (Sigma; 16°C, 200 RPM, 20 hours or 4 hours at 30°C). Cells were harvested by centrifugation (5000 RPM, 4°C) and flash frozen in liquid nitrogen.

Frozen cell pellets were resuspended in 50 mL of lysis buffer (50 mM HEPES pH 8.0, 300 mM NaCl, 30 mM imidazole, and 1 mM TCEP) containing 0.5 mg/mL lysozyme (Sigma), 250 U Pierce Universal Nuclease (ThermoFisher), and an EDTA-free protease inhibitor cocktail (Roche) and rotated at 4°C for 1 hour. This crude lysate was then sonicated [five cycles of 30 sec on/30 sec off at 50% output] and centrifuged at 4°C, 15,000 RPM for 35 min. The clarified lysate was flowed over a 5 mL HisTrap nickel column (Cytiva) using an AKTA Pure FPLC (Cytiva) at 0.5 mL/min; all FPLC steps were conducted at 4°C. Unbound species were removed with 20 column volumes of wash buffer (WB: 50 mM HEPES pH 8.0, 300 mM NaCl, 30 mM imidazole, and 1 mM TCEP at 2 mL/min). His-tagged proteins were eluted in a 15-column volume linear gradient from WB to Elution Buffer (WB + 500 mM imidazole) at 2 mL/min. Fractions containing His-tagged protein were identified by SDS-PAGE, pooled, and dialyzed against 50 mM HEPES pH 8.0, 300 mM NaCl, and 1 mM TCEP to remove imidazole. His-Lamin A-NT was concentrated using a 30 kDa MWCO centrifugal filter (EMD Millipore) and aliquoted for storage. SETD2 constructs were also concentrated in an identical fashion, but down to > 2 mL each and further purified through gel filtration using a Superdex 75pg column (Cytiva). Fractions containing pure SETD2 constructs were pooled and concentrated again. Protein purity was assessed via SDS-PAGE.

Generation of stable cell lines

HKC shSETD2-TetOn cells were generated by infecting cells with lentivirus containing the dox-inducible shSETD2 construct, followed by selection with puromycin (1 μg/ml). To induce SETD2 knockdown (KD), HKC-shSETD2-TetOn cells were treated with 500 ng/ml of doxycycline for 48-72 h. To generate SETD2 complementation cell lines, HKC-shSETD2-TetOn cells were nucleofected with a transposase expression vector (System Biosciences, PB210PA) and Xlone-piggyBac plasmid containing dox-inducible SETD2 transgene (WT or mutant), followed by selection with increasing concentration of blasticidin (2-10 μg/ml) over during a period of two weeks. Cells were treated with doxycycline at 500 ng/ml for 5-6 days to allow for simultaneous knockdown of endogenous SETD2 and expression of the exogenous SETD2 transgene before downstream analyses. UMRC2 stable cell lines were generated similarly.

Generation of SETD2-FKBP (dTAG) knock-in cell line

Generation of SETD2-FKBP knock-in cell line was performed as described previously with brief modification⁵⁷. HKC cells were transduced with adeno-associated virus containing homology-directed repair templates (Vectorbuilder). After 6 h, cells were transduced with lentivirus containing Cas9 and SETD2-targeting guide RNA (GACTCACGGTGTTATGAATAAGG). Cells were expanded and selected with Hygromycin for 2 weeks. Hygromycin colonies were screened via PCR and immunoblot for homozygous insertion.

Co-immunoprecipitation

Cells were resuspended in ice-cold lysis buffer containing 50 mM Tris-HCl pH 7.5, 150 mM NaCl, 10% glycerol, 0.25% Triton-X100, 1 mM MgCl₂, 1 mM PMSF, 1 U/μl Universal Nuclease (Pierce) supplemented with protease and phosphatase inhibitors. Lysates were incubated on rotation for 2 hours followed by brief sonication and centrifugation at 13,000g for 20 min. Lysates were pre-cleared using equilibrated Protein-G Sepharose beads for 1 h. Pre-cleared lysates were incubated with 2-4 μg of indicated antibody conjugated Protein-G Dynabeads (Life Technologies) for 4 hours or overnight. Beads were gently centrifuged, separated on magnet and washed five times with lysis buffer (with 1mM EDTA) for 10 min each wash. Immunoprecipitated proteins were eluted by heating the beads at 95°C for 5 min in SDS-loading buffer.

Immunofluorescence

Cells grown on glass coverslips were fixed in 4% formaldehyde (methanol-free) in PBS for 15 min at room temperature. Cells were washed three times with PBS for 5 min each, followed by permeabilization with 0.5% Triton in PBS for 10 min. Cells were blocked in 2% BSA, 1% normal goat serum (NGS) in PBS for 30 min at RT prior to antibody incubation for 1 hour at RT. Cells were washed twice with PBS for 5 minutes each, followed by secondary antibody incubation (Alexa Fluor, Life Technologies) at 1:1000 for 1 hour at RT. Cells were washed with PBS followed by DAPI staining and mounting on glass slides with ProLong Gold Antifade mounting media (Life Technologies). Halo-tagged SETD2 constructs were visualized by using 100 nM JFX-549 (Janelia Farms) ligand for 20 min before fixing.

Confocal microscopy

Images were acquired using Zeiss LSM 800 confocal microscope using either a 40x/1.3NA oil or 63x/1.4NA oil objective. Z-stacks of 0.5 μm slices were collected and maximum intensity projections were used for analysis and display. For colocalization analyses, fluorescence intensity profiles were extracted for all fluorophores from the same Z-axis plane using the Zen Blue software and graphed using Prism 9. Images were processed and analyzed with ImageJ (version 2.3.0) and CellProfiler (version 4.2.4).

Live cell microscopy

HKC-shSETD2-TetON cells expressing Emerald-lamin A were plated in an 8-well chambered cover glass (Cellvis). Cells were treated with dox one day prior to synchronization using double thymidine block and released for 4 hours in the presence of 100 nM SiR-Hoechst (Spirochrome) prior to imaging. Images were acquired using Zeiss LSM 880 confocal microscope with a 40x/1.3NA oil immersion objective. Z-stacks of 1-μm slices were collected every 4 or 10 min intervals for 6-8 hours. Images were processed and analyzed using ImageJ (version 2.3.0).

Proximity Ligation Assay (PLA)

PLA was performed according to the manufacturer’s protocol (Duolink PLA kit, Sigma) with minor modifications. Briefly, after fixation and permeabilization, cells were treated with previously described blocking solution⁵⁸ (10% NGS, 2% BSA, 5% sucrose in PBS) for 45 min, followed by primary antibody incubation in blocking solution for 1 hour at RT. Cells were washed and treated with plus and minus probes diluted in blocking solution for 1 hour at 37°C. Probes were ligated for 30 min at 37°C followed by in situ PCR amplification using fluorescent oligonucleotides (green or red) for 120 min at 37°C. Cells were counterstained with DAPI and mounted using ProLong Gold Antifade mounting media (Life Technologies). Images were acquired on a Zeiss 880 confocal microscope with 63x/1.4 oil immersion objective. Images were processed using ImageJ and PLA pairs were quantified using CellProfiler (version 4.2.4)⁵⁹.

Image analysis

For PLA analyses, a published pipeline for foci counting from cellprofiler.org was adapted to quantify PLA spots. Briefly, nuclei were segmented based on DAPI signal. Next, PLA foci were detected with 4-30 pixels limit and global thresholding. PLA objects were related to parent nuclei objects, and per object counts were calculated and exported as spreadsheet. To detect SETD2 expressing cells, an additional step was added to detect SETD2 intensity as secondary object per nucleus. For live-cell image analysis in Fig. 3c, cells were manually tracked and fluorescent intensities were extracted using ImageJ. Briefly, ROIs were drawn encompassing Emerald-lamin A signal for every frame from the last metaphase frame immediately preceding anaphase onset to the end of mitosis (typically 10 frames of 4 min intervals) and mean fluorescence intensity (MFI) was extracted for Emerald-lamin A. MFI of every frame was normalized to the first frame (last metaphase frame, set as 0 min) and the time taken to reach maximum MFI was determined for each cell analyzed. Maximum MFI of Emerald-lamin A was indicative of complete lamina reassembly. Quantification of phos-lamin A/C was performed using CellProfiler. Briefly, a manual threshold was applied to all images of S22p/S392p-lamin A/C from a given experiment such that only mitotic cells were identified. Integrated intensities from S22p/S392p-lamin A/C and total lamin A/C channels were exported. Phos-lamin A/C intensities were normalized to total lamin A/C intensities per cell and data were plotted on Prism.

Cell cycle analysis

Cells were treated with 10 μM EdU for 20 min before collection. Cells were trypsinized, washed with PBS and pelleted. Cells were fixed in 4% formaldehyde in PBS for 15 min. After permeabilization and blocking in 1% BSA, click reaction was performed to detect EdU incorporation. Briefly, cells were incubated in PBS with 1 mM CuSO₄, 100 mM ascorbic acid and 1 μM Alexa-647-azide for 30 min at RT in the dark. After washes, cells were resuspended in PBS-BSA containing DAPI and 100 μg/ml RNase A overnight. Flow cytometry was performed on Attune Nxt flow cytometer (Thermo Fisher) and data were processed and analyzed using FlowJo (BD Biosciences). See Extended Data Fig. 8 for gating strategy.

Western blot

Cells were lysed in lysis buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 10% glycerol, 0.5% NP40, 0.1 mM EDTA, protease and phosphatase inhibitors). SDS-loading buffer was added to lysates heated for 10 min at 95°C. Proteins were separated on a 4-15% gradient SDS-PAGE gel (BioRad) and proteins were subsequently transferred onto a PVDF membrane. Membranes were blocked in 5% non-fat milk in TBST and incubated with antibodies overnight in milk. Membranes were washed and incubated with secondary antibodies for 1 h prior to ECL treatment and signal detection using ChemiDoc imaging system (BioRad).

In vitro kinase assay

Approximately 1 μM of lamin A-NT and lamin A-CT were mixed with 10 nM CDK1/Cyclin B1 complex in kinase buffer (20 mM Tris-HCl pH 7.5, 10 mM MgCl₂, 100 mM NaCl, 10 mM β-glycerophosphate, 2 mM DTT, 50 ng/μl BSA, protease and phosphatase inhibitors). This mixture was then divided equally into three tubes and 0.5 μM of either SETD2-B or SETD2-C or His-SUMO-GB1 (control) proteins were added to each tube. Samples were incubated at 30°C for 30 min for equilibration prior to initiating the reaction by adding 100 μM ATP. Samples were incubated further for a total of 40 min at 30°C. Aliquots from each sample were drawn at 0, 10, 20, 40 min and added to tubes containing SDS-loading buffer to stop the reaction. Samples were analyzed by immunoblotting.

In vitro interaction assay

Recombinant proteins were diluted in binding buffer (20 mM Tris-HCl, 150 mM NaCl, 2 mM DTT, 10 mM MgCl₂, 0.1% Tween-20, 100 ng/μl BSA and protease inhibitors) and incubated at 30°C for 30 min. Samples were pre-cleared using protein A agarose beads for 30 min in cold room, followed by centrifugation at 13000 rpm for 10 min. Supernatants were added to glutathione agarose beads (Pierce, 16101) that were pre-washed in binding buffer. Samples were rotated for 4 h to overnight at 4°C. Beads were gently centrifuged and washed 5 times in binding buffer prior to heating in SDS-loading buffer for immunoblotting.

Proximity biotinylation and affinity purification

HKC cells stably integrated with dox-inducible WT-SETD2-APEX2 or tSETD2-APEX2 were induced with dox for three days prior to proximity labeling and affinity purification. Corresponding cells without dox treatment were used at controls. Proximity-labeling and streptavidin AP were performed as described⁶⁰ with a few modifications. Briefly, cells were pulsed with 500 μM biotinphenol for 30 min at 37°C. Biotinylation was performed with 1 mM H₂O₂ for 1 min and reaction was stopped by rinsing cells with quencher solution (10 mM sodium ascorbate, 5 mM Trolox, and 10 mM sodium azide in 1xPBS). Cells were resuspended in lysis buffer (50 mM Tris-HCl pH 7.5, 200 mM NaCl, 10% glycerol, 0.25% Triton-X100, 1 mM MgCl₂, 1 mM PMSF, 1 U/μl Universal Nuclease (Pierce) supplemented with protease and phosphatase inhibitors). Lysates were rotated for 2 h followed by brief sonication and centrifugation at 13,000 x g for 20 min. Protein lysates were then dialyzed against lysis buffer overnight with two changes of buffer to remove excess free biotin-phenol. Lysates were centrifuged at 13,000 x g for 20 min and incubated with equilibrated streptavidin magnetic beads (Pierce) for 4 h. Beads were washed and proteins were eluted off as described⁶⁰. Eluted proteins were separated on 4-15% gradient SDS-PAGE gel for 1 cm or until the 25 kDa band was visible. Gel was stained with colloidal blue (Thermo, LC6025) and lanes were excised and the proteins were reduced with 5 mM DTT, alkylated with 15 mM iodoacetamide, and in-gel digested with sequencing grade trypsin (Promega) overnight at 37°C. Peptides were extracted, desalted with C18 Desalting Spin Columns (Thermo) and dried via vacuum centrifugation. Peptide samples were stored at −80°C until analysis.

Mass-spectrometry analysis

SETD2 interactome analysis:

Streptavidin pulldown samples (n=3) were analyzed by LC/MS/MS using an Easy nLC 1200 coupled to a QExactive HF mass spectrometer (Thermo Scientific). Samples were injected onto an Easy Spray PepMap C18 column (75 μm id × 25 cm, 2 μm particle size) (Thermo Scientific) and separated over a 90-minute method. The gradient for separation consisted of 5–45% mobile phase B at a 250 nl/min flow rate, where mobile phase A was 0.1% formic acid in water and mobile phase B consisted of 0.1% formic acid in 80% ACN. The QExactive HF was operated in data-dependent mode (DDA) where the 15 most intense precursors were selected for subsequent fragmentation. Resolution for the precursor scan (m/z 350–1700) was set to 60,000, while MS/MS scans resolution was set to 15,000. The normalized collision energy was set to 27% for HCD. Peptide match was set to preferred, and precursors with unknown charge or a charge state of 1 and ≥ 7 were excluded.

Lamin A/C and Lamin B1 phosphorylation analysis:

Lamin A/C and lamin B1 were immunoprecipitated from control and shSETD2 HKC cells as described above. Immunoprecipitated proteins were eluted by heating and separated on 4-15% gradient gel. Gel bands corresponding to Lamin A/C or Lamin B1 were excised and reduced, alkylated, and digested as described above. Peptide samples (n=3) were analyzed by LC/MS/MS using an Ultimate 3000 coupled to an Exploris 480 mass spectrometer (Thermo Scientific). Samples were injected onto an IonOpticks Aurora series 2 C18 column (75 μm id × 15 cm, 1.6 μm particle size) and separated over a 65-minute method. The gradient for separation consisted of 2-40% mobile phase B at a 250 nl/min flow rate, where mobile phase A was 0.1% formic acid in water and mobile phase B consisted of 0.1% formic acid in 80% ACN. The Exploris 480 was operated in data-dependent mode (DDA) where duty cycle was set to 1.5 sec. Resolution for the precursor scan (m/z 375–1500) was set to 120,000, while MS/MS scans resolution was set to 15,000. The normalized collision energy was set to 30% for higher collision dissociate (HCD). Monoisotopic peak determination was set to ‘peptide’ and precursors with unknown charge or a charge state of 1 and >5 were excluded. For targeted analysis, Lamin A/C samples were analyzed on an Easy nLC 1200 coupled to a QExactive HF mass spectrometer (Thermo Scientific). Samples were injected onto an Easy Spray PepMap C18 column (75 μm id × 25 cm, 2 μm particle size) (Thermo Scientific) and separated over a 60-minute method. The gradient for separation consisted of 5-40% mobile phase B at a 250 nl/min flow rate, where mobile phase A was 0.1% formic acid in water and mobile phase B consisted of 0.1% formic acid in 80% ACN. Targeted parallel reaction monitoring (PRM) method was developed for a specific Lamin A/C peptide spanning S22 (unphosphorylated and phosphorylated versions). The QExactive HF was operated in MS1 and PRM mode. Resolution for the precursor scan (m/z 400–1200) was set to 60,000 with an AGC target set to 1e⁶ and a maximum injection time set to 200 ms. PRM scans (15,000 resolution) consisted of HCD set to 27; AGC target set to 1e⁶; maximum injection time set to 100 ms; isolation window of 2 Da.

Mass-spectrometry Data Analysis

SETD2 interaction analysis:

Raw data files were processed using MaxQuant version 1.6.15.0 and searched against the reviewed human database (containing 20,350 entries), appended with a contaminants database, using Andromeda within MaxQuant. Enzyme specificity was set to trypsin, up to two missed cleavage sites were allowed, and methionine oxidation and N-terminus acetylation were set as variable modifications, with carbomidomethylated cysteines set as a static modification. A 1% FDR was used to filter all data. Match between runs was enabled (5 min match time window, 20 min alignment window), and a minimum of two unique peptides was required for label-free quantitation using the LFQ intensities. Perseus was used for further processing⁶¹. Only proteins with >1 unique+razor peptide were used for LFQ analysis. Proteins with 50% missing values were removed and missing values were imputed from normal distribution within Perseus. Log2 fold change (FC) ratios were calculated using the averaged Log2 LFQ intensities of $\frac{W{T}_{DoxON}}{W{T}_{DoxOFF}}$ (for WT-SETD2-APEX2, Fig.1b). For Fig. 6a, difference between WT-SETD2-APEX2 and tSETD2-APEX2 was calculated as such $log2(\frac{tSETD{2}_{DoxON}}{tSETD{2}_{DOxOFF}})-log(\frac{W{T}_{DoxON}}{W{T}_{DoxOFF}})$. Students t-test performed for each pairwise comparison, with p-values calculated. Proteins with significant p-values (<0.05) and Log2 FC >1 were considered biological interactors.

Gene Ontology analyses were conducted using DAVID⁶². Genes were searched against UP_KW_BIOLOGICAL_PROCESS, UP_KW_CELLULAR_COMPONENT, UP_KW_MOLECULAR_FUNCTION, GOTERM_BP_DIRECT, GOTERM_CC_DIRECT and GOTERM_MF_DIRECT. GO terms used are described in Table 2. Figures 1c, Fig. 6c were created using tidyverse⁶³ and ggplot2 packages in R (version 4.0.4).

Table 2:

GO terms from Fig. 1c

GO Term	Figure Label
KW-0498~Mitosis	Mitosis
GO:0005635~nuclear envelope	Nuclear Envelope
GO:0006281~DNA repair	DNA repair
KW-0507~mRNA processing	mRNA processing
GO:0006260~DNA replication	DNA replication
GO:0000785~chromatin	Chromatin
KW-0804~Transcription	Transcription

Lamin phosphorylation analysis:

DDA and PRM raw data files were processed using Proteome Discoverer version 2.5 (Thermo Scientific). Peak lists were searched against a reviewed Uniprot human database, appended with a common contaminants database, using Sequest. The following parameters were used to identify tryptic peptides for protein identification: 10 ppm precursor ion mass tolerance; 0.02 Da product ion mass tolerance; up to two missed trypsin cleavage sites; (C) carbamidomethylation was set as a fixed modification; (M) oxidation, (S,T,Y) phosphorylation were set as variable modifications. Peptide false discovery rates (FDR) were calculated by the Percolator node using a decoy database search and data were filtered using a 1% FDR cutoff. The PhosphoRS node was used to localize phosphorylation sites within the peptides. Peak areas were extracted using the Minora node, and log2 fold change (FC) ratios of each peptide were calculated (shSETD2/Control).

Statistical analysis and reproducibility

All experiments were conducted in N = 3 independent biological replicates except a few where N = 2 independent biological replicates were performed as indicated in the figure legends. For experiments involving quantitative per-cell measurements, statistics were performed with ‘n’ (number of cells). In experiments involving quantification of percentages of a population, statistical tests were performed on number of replicates (N). Statistical tests were chosen based on normality of the data. Data were tested for normality prior to statistical analyses. For normally distributed data, unpaired, two-tailed t-tests were performed with mean ± s.d indicated in the figure. For data without Gaussian distribution, two-tailed non-parametric Mann-Whitney U test was performed with median indicated in the figure. Similarly, for comparison of multiple groups with non-normal data, Kruskal-Wallis with Dunn’s or Sidak’s multiple comparisons test was performed as indicated in the figure legends. For comparison of multiple groups with normal data, one-way ANOVA with Dunnett’s multiple comparisons test was performed. No statistical methods were used to predetermine sample size. Cell numbers are provided in the figure legends and are based on what is generally used in the field. All non-parametric tests were performed on independent biological samples in aggregate. All statistical analyses were performed on GraphPad PRISM 9.

3D colony formation assay

The 3D soft agar colony formation assays were performed as described previously⁶⁴. Briefly, 0.4% agar top layer containing 30,000 UMRC2 cells was applied to 1% agar bottom layer. Add 0.5 mL medium after the top layer becomes solid then add 3 drops of complete media every 4 days. After incubating for 2 weeks, cell colonies were stained with 100 mg/mL iodonitrotetrazoliuim chloride (Sigma-Aldrich, I8377) in a complete medium overnight. The foci numbers were counted by ImageJ software.

Tumor xenograft

Subcutaneous xenograft tumor growth was performed with 6-8 weeks-old NSG mice (Jackson Lab, 6M+3F per group; male: female=2:1 to mirror the human scenario where men are twice as likely as women to develop kidney cancer and also exhibit a higher mortality rate) by injecting with UMRC2 cells subcutaneously. Approximately 1x10⁶ UMRC2 cells were resuspended in 100mL Matrigel that was diluted with the medium by 1:1 and then injected into the left flank of each mouse. To induce target gene expression, one week after injection, mice were fed with Purina rodent chow #5001 with 2000ppm doxycycline (Research Diets Inc.). Tumor size was measured with calipers once a week and tumor volume was calculated as V = 0.5 x L x S². L and S indicate long measurement and short measurement. After the stated number of weeks, mice were euthanized, and tumors were taken out and weighed. All animal procedures were performed in accordance with the National Institutes of Health guidelines and were approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Texas Southwestern Medical Center (Protocol #: 2019-102794, Qing Zhang).

Extended Data

Extended Data Fig. 1 ∣. SETD2 interacts with nuclear lamina proteins.

a, Schematic of SETD2 with its indicated domains. b, Fluorescence intensity profiles of SETD2, lamin A/C and DAPI in interphase or mitotic cells. Scale bar, 10 μm. c, Proximity ligation assay (PLA) on SETD2-KD cells expressing Halo-Flag-tagged SETD2 using anti-SETD2 and anti-lamin A/C antibodies. PLA pairs are in green; SETD2 (red) visualized using fluorescent Halo ligand JFX-549. DNA was counterstained with DAPI. Scale bar, 10 μm. d, Quantitative analysis of PLAs between SETD2 and lamin A/C shown in c. Two-tailed unpaired Student’s t-test was performed from two independent biological replicates, with n (left to right) = 55, 41, 7 cells; mean ± s.e.m.

Extended Data Fig. 2 ∣. SETD2 loss during G2/M leads to nuclear lamina defects.

a, Representative confocal microscopy images of control and SETD2-KD HKC cells immunostained with anti-lamin A/C and lamin B1 antibodies. DNA counterstained with DAPI. b, Representative confocal microscopy images of control and Setd2-KO MEF cells immunostained with anti-lamin B1 antibody. DNA counterstained with DAPI. c, Schematic of cell synchronization by double thymidine block and release method used in d–f. d, Immunoblot analysis of indicated proteins in untreated and dTAG-47 treated (at 6 h post release) cells. e, Representative images of either untreated or dTAG-47 treated SETD2–FKBP cells, fixed at the indicated timepoints post release from thymidine block and immunostained with anti-lamin A/C antibody. DNA counterstained with DAPI. f, Quantification of nuclear defects observed in e. Ordinary one-way ANOVA with Sidak’s multiple comparison test was performed on data from two independent biological replicates, with n (left to right) = 614, 558, 735, 654, 754, 693 cells; mean ± s.e.m. g, Schematic of cell synchronization by serum starvation of Setd2 (w/flox) MEFs. h, Immunoblot analysis of indicated proteins in untreated and tamoxifen treated (Setd2-KO) MEF cells upon release from G0/G1 block. i, Representative images of either control or tamoxifen-treated MEF cells at the indicated timepoints post release from serum starvation, immunostained with anti-lamin A/C antibody. j, Quantification of nuclear defects from data in i. Two-way ANOVA with Sidak’s multiple comparison test was performed on data from three independent replicates with n (left to right) = 184, 166, 247, 223, 256, 247, 387, 348 cells; mean ± s.d. All scale bars are 10 μm.

Extended Data Fig. 3 ∣. Nuclear lamina defects caused by SETD2 loss arise from events dysregulated during G2/M.

a, Western blot analysis of indicated proteins from control and SETD2-KD RPE cells that are either arrested in G0/G1 or growing asynchronously (Async). b, Representative images of control and SETD2-KD RPE cells that are either proliferating (Async) or arrested in G0/G1, immunostained with anti-lamin A/C antibody. DNA counterstained with DAPI. Scale bars, 10 μm. c, Quantification of nuclear defects observed in b. Two-tailed unpaired Student’s t-test was performed from two independent biological replicates with n (left to right) = 357, 318, 395, 424 cells; mean ± s.d. d, Representative flow cytometry analysis of control and SETD2-KD RPE cells growing asynchronously. e, Representative flow cytometry analysis of control and SETD2-KD RPE cells arrested in G0/G1. Y-axis is EdU intensity and x-axis DNA content. f, Immunoblot analysis of indicated proteins in two independent replicates of control and SETD2-KD HKC cells. g, Quantitative analysis of γH2AX intensity in untreated (control) and dTAG-47 treated SETD2–FKBP HKC cells that were synchronized by the double thymidine block method and released for indicated timepoints. Integrated intensity per nucleus was measured from n (left to right) = 2,349, 2,180, 2,044, 2,061, 3,188, 3,191 cells. Kruskal–Wallis with Dunn’s multiple comparisons test was performed on data from two independent biological replicates; medians in red.

Extended Data Fig. 4 ∣. SETD2 loss impairs lamin phosphorylation during G2/M.

a, Representative time-lapse images of control and SETD2-depleted HKC expressing Emerald-lamin A (green). DNA was labelled with SiR-Hoechst (red). Arrowheads point to partially depolymerized lamin filaments observed throughout mitosis. b, Representative flow cytometry analysis of control and SETD2-KD cells to determine cell-cycle distribution. Y-axis is EdU intensity and x-axis DNA content. c, Quantitation of flow cytometry analysis data showing cell-cycle distribution of control and SETD2-KD cells. Two-tailed unpaired Student’s t-test was performed from three independent biological replicates; mean ± s.d. d, Quantitative analysis of immunofluorescence data from control and SETD2-KD RPE cells. Integrated fluorescence intensity of either S22ph- or S392ph-lamin A signal from mitotic cells normalized to integrated intensity of pan lamin A/C signal in control (n = 505, S22ph; n = 390, S392ph) and shSETD2 (n = 490, S22ph; n = 382, S392ph) cells from two independent biological replicates. Two-tailed non-parametric Mann–Whitney U-test was performed; medians in red. e, Quantitative analysis of immunofluorescence data from control and Setd2-KO MEF cells. Integrated fluorescence intensity of either S22ph- or S392ph-lamin A signal from mitotic cells normalized to integrated intensity of pan lamin A/C signal in control (n = 515, S22ph; n = 407, S392ph) and Setd2-KO (n = 457, S22ph; n = 327, S392ph) cells from two independent biological replicates. Two-tailed non-parametric Mann–Whitney U-test was performed; medians in red. f, Representative images of LMNA KO human fibroblasts cells expressing either WT, S- > A (S22A; S392A) or S- > D (S22D; S392D) mutant lamin A. Cells were immunostained with lamin A/C antibody (red) and DNA counterstained with DAPI. All scale bars are 10 μm. g, qRT-PCR analysis of SETD2 and Cyclin B1 mRNA levels during G2/M, normalized to Actin mRNA. X-axis is time (h) post release from a double thymidine block, similar to Fig. 3e. Y-axis is fold change relative to 0 hr timepoint. Paired Student’s t-test comparing different timepoints for each gene was performed from two independent biological replicates.

Extended Data Fig. 5 ∣. Non-enzymatic function of SETD2 in maintenance of nuclear morphology.

a, Immunoblot analyses of indicated proteins from control, Tamoxifen (Setd2-KO) and SETD2i (EPZ-719)-treated MEF cells for 48 h. b, Representative images of control, Setd2-KO and SETD2i-treated MEF cells immunostained with lamin B1 antibody. c, Quantitative analysis of nuclear defects from b. Two-tailed unpaired Student’s t-test was performed from two independent biological replicates; mean ± s.d.; DNA counterstained with DAPI. d, Representative images of control, SETD2-KD or SETD2i (EPZ-719) treated HKC cells for 2 d or 6 d, immunostained with lamin B1 antibody. DNA counterstained with DAPI. Scale bars, 10 μm.

Extended Data Fig. 6 ∣. The N terminus of SETD2 regulates nuclear lamina stability.

a, Immunoblot analysis for the indicated proteins following immunoprecipitation of endogenous SETD2 from asynchronous or nocodazole-arrested mitotic cells, showing SETD2 associates with CDK1 more robustly during mitosis. b, Representative images of Proximity Ligation Assay (PLA) analysis using anti-lamin B1 and anti-CDK1 antibodies in SETD2-KD cells expressing either WT-SETD2 or tSETD2. PLA pairs are green and DNA was counterstained with DAPI. c, Representative images of SETD2-depleted cells expressing either WT SETD2 or the indicated N-terminal mutants, immunostained with lamin B1 (green). SETD2 (red) visualized using Halo ligand JFX-549. DNA was counterstained with DAPI. Scale bars, 10 μm.

Extended Data Fig. 7 ∣. SETD2 N terminus enhances CDK1-mediated lamin phosphorylation and is required to suppress ccRCC tumour growth.

a, Long exposure of western blot analyses (from Fig. 6a) with anti-SUMO antibody following in vitro GST pulldown assays using either GST-tagged lamin A N-terminus (NT) or C-terminus (CT) or free GST (control). b, Immunoblot analyses using indicated antibodies following GST pulldown assays using either GST-tagged CDK1/Cyclin A2 complex or free GST (control). Anti-SUMO antibody was used to detect SUMO-tagged SETD2 (B/C) proteins. Asterisk indicates cross-reacting band of His–SUMO-B–GB1 in the anti-GST immunoblot. This is because the GB1 (B1 domain of Protein G) exhibits significant affinity towards most immunoglobulin IgG. c, Immunoblot analysis of indicated proteins following an in vitro kinase assay using lamin A-NT as substrate and CDK1/Cyclin B1 as kinase in the presence of either SETD2-B or SETD2-C or a control His–SUMO-GB1 protein. Red asterisk highlights cross-reacting band of His–SUMO-B–GB1 due to aforementioned reasons. d, Western blot with indicated antibodies from UMRC2 cells expressing either Halo (control) or Halo-3xFlag tagged WT-SETD2 or tSTED2 or BC-tSETD2. e, Representative images of UMRC2 cells expressing either Halo (control) or Halo-tagged WT SETD2 or tSETD2 or BC-tSETD2, immunostained with pan-lamin A/C antibody. DNA counterstained with DAPI; scale bar, 10 μm. f, Representative images of the full 35 mm cell culture well with 3D colony growth assay of UMRC2 cells expressing the indicated transgenes; scale bar is 10 mm. g, Xenograft tumour growth measurements (volume) of UMRC2 cells expressing the indicated transgene (n = 9). Related to Fig. 6j; mean ± s.e.m.

Extended Data Fig. 8 ∣.

Flow analysis gating strategy.

Data Availability

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD040771. Source numerical data and unprocessed blots are available in source data.