Phosphorylation-dependent modulation of the Lamin A/C–EZH2 complex regulates epithelial–mesenchymal plasticity

Abstract

Epithelial-to-mesenchymal transition (EMT) is essential for normal development and cancer progression. However, how nuclear Lamins regulate EMT is unclear. Here, we show that Lamin A/C modulates the epithelial–mesenchymal (E-M) plasticity of cells through its interaction with the chromatin organizer, EZH2. The overexpression of Lamin A reinforces an epithelial identity, while its depletion promotes a mesenchymal phenotype. This positions Lamin A/C as a crucial modulator of Epithelial-Mesenchymal plasticity. Furthermore, CDK1-mediated phosphorylation of Lamin A/C (Ser22) and EZH2 (Thr345) disrupts Lamin A/C–EZH2 interaction, destabilizing EZH2, with a concomitant decrease in the occupancy of the heterochromatin mark (H3K27me3) on the SNAI1, TWIST1, and ZEB1 promoters, thereby facilitating a transition towards mesenchymal transcriptional programs. Conversely, phosphodeficient Lamin A/C (S22A) and EZH2 (T345A) mutants restore epithelial identity, highlighting a regulatory role of the Lamin A/C–EZH2 axis in maintaining epithelial homeostasis. In vivo, xenograft assays in NOD-SCID mice reveal that while phosphorylated Lamin A/C or EZH2 promote tumor growth and metastasis, phospho-deficient mutants markedly suppress it. Lamin A/C–EZH2 interaction regulates the expression of E–M-associated transcription factors, highlighting the role of this interaction in modulating transcriptional plasticity, thereby serving as a potential therapeutic target for regulating metastasis in breast cancers.

Graphical Abstract

Graphical Abstract.

Introduction

Epithelial-to-Mesenchymal Transition (EMT) and Mesenchymal-to-Epithelial Transition (MET) are fundamental biological processes that play critical roles in embryonic development, wound healing, and tissue homeostasis [1]. EMT is essential for gastrulation, somite formation and neural crest cell migration, while the converse of EMT i.e. MET is essential for organogenesis, such as kidney development and the formation of epithelial somites surrounding the core of mesenchymal cells [2, 3]. During EMT, epithelial cells show loss of polarity and cell–cell adhesion, which confer migratory and invasive characteristics that are typical of mesenchymal cells [4]. Conversely, MET induces enhanced cell–cell adhesion, decreases motility, restoring epithelial characteristics, enabling tissue differentiation and organogenesis [5]. While these processes are essential for normal physiology, they are subverted during cancer progression, fibrosis, and organ dysfunction [6]. EMT facilitates cells to invade the surrounding tissues and MET is crucial for the colonization of metastatic cells and integration into the secondary tissue. Rather than representing a binary switch, epithelial-to-mesenchymal plasticity (EMP) encompasses hybrid epithelial–mesenchymal (E-M) states that confer metastatic competence and survival advantages. Lineage-tracing and single-cell studies reveal that these hybrid states often underlie tumor dissemination, metastatic colonization, resistance to chemotherapy and recurrence, making it imperative to understand how tumor cells dynamically modulate EMP during cancer progression [7–10]. The transition between these phenotypic states (E and M) involves an intricate coordination among transcription factors, signalling pathways, and epigenetic mechanisms. However, the underlying mechanisms on how each of these processes crosstalk with one another are not fully understood [11].

During EMT, epithelial factors such as E-cadherin are downregulated, while mesenchymal markers—Vimentin, N-cadherin, and fibronectin—are upregulated, driven by transcription factors TWIST1, SNAI1, and ZEB1 [12]. Conversely, during MET, the activity of these transcription factors is attenuated, re-expressing epithelial factors, and suppressing mesenchymal properties [13]. However, the crosstalk between nuclear architecture and chromatin organization in regulating EM plasticity is unclear.

Lamin A/C, a core component of the nuclear lamina, provides structural support to the nucleus and maintains genome organization. It regulates chromatin architecture and gene expression, and its depletion increases nuclear deformability, alters chromatin accessibility, and enhances cellular plasticity [14–18]. Despite these broad nuclear functions, its direct role in EMT or MET remains unclear.

Lamin A/C regulates transcriptional programs through interactions with chromatin modifiers and transcriptional repressors. Notably, the Polycomb Repressive Complex 2 (PRC2), a histone methyltransferase, silences target genes via deposition of the repressive mark H3K27me3 [19, 20]. This connection suggests that Lamin A/C could modulate EM plasticity by coordinating with PRC2 components, such as EZH2 (the catalytic subunit of PRC2), thereby linking integrity of nuclear lamina to epigenetic regulation and transcriptional control. EZH2 is a well-established epigenetic regulator implicated in EMT [21]. However, the mechanisms by which Lamin A/C modulates EZH2 activity and if this interaction impacts E–M plasticity are largely unknown. Considering the central importance of epigenetic regulation in cancer initiation and progression, understanding the interplay between nuclear and chromatin organizers, namely Lamin A/C and EZH2 respectively, will uncover novel therapeutic targets to modulate EMT and MET dynamics in aggressive malignancies.

Here, we systematically investigate the role of Lamin A/C and EZH2 in EMT and MET using well-characterized paradigms of mammary breast epithelial cells and breast cancers. We demonstrate that depletion of Lamin A/C or EZH2 facilitates EMT, whereas its overexpression promotes MET, underscoring its role in maintaining epithelial identity. Furthermore, we identify that CDK1-mediated phosphorylation regulates the interaction between Lamin A/C-EZH2 altering their chromatin occupancy during EMT and MET. Integrative analyses involving transcriptomic, proteomic, and functional data, collectively provide crucial mechanistic insights into how nuclear Lamins contribute to EM plasticity. By elucidating the interplay between Lamin A/C and EZH2, this study elucidates the functional relevance of the interactome of nuclear Lamins in cancer progression. Our findings highlight the regulatory role of the phosphorylation of Lamin A/C in modulating EMT and implicate the Lamin A/C-EZH2 axis as a key target for novel therapeutic interventions in metastatic breast cancers.

Materials and methods

Cell culture

Immortalized human breast epithelial cell line MCF10A (CRL-10317) and cancer cell lines MCF7 (HTB-22), and MDAMB231 (HTB-26) were obtained from ATCC. MCF10A cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM)/F12 (1:1) supplemented with 5% horse serum, EGF (20 ng/ml), hydrocortisone (0.5 mg/ml), cholera toxin (0.1 mg/ml), insulin (10 mg/ml), and 1% penicillin–streptomycin. MCF7 and MDAMB231 cells were cultured in DMEM and Roswell Park Memorial Institute (RPMI), respectively, supplemented with 10% fetal bovine serum (FBS), 1% L-glutamine, and 1% penicillin–streptomycin. All cells were grown at 37°C with 5% CO₂. The cell types were validated based on morphology (Supplementary Fig. S1A), chromosome numbers (Supplementary Fig. S1B), and profile of key E and M markers (Supplementary Fig. S1C–G). Furthermore, 4′,6-diamidino-2-phenylindole (DAPI) staining did not show any visible extranuclear staining, negating mycoplasma contamination.

EMT induction in MCF7 upon TWIST1 overexpression

MCF7 cells (∼0.35 million) were seeded in six-well plates and allowed to adhere for ∼10–12 h, to attain a confluency of ∼50%. Cells were independently transfected with 2 μg of pEGFPN1 (Empty vector) or TWIST1 in Opti-MEM using polyethyleneimine (PEI; Sigma #408727-100ML) at a ratio of 1:3 plasmid: branched PEI. After ∼6 h post-transfection, Opti-MEM was replaced with DMEM containing 10% FBS. Transfected cells were selected with G418 (890 μg/ml, Roche #4727878001) for 24 h. Cells were harvested for reverse transcriptase-polymerase chain reaction (RT-PCR), immunofluorescence assay, or immunoblotting at ∼48 h post-transfection to assess for expression of E and M marks, respectively [22, 23].

TGF-β mediated EMT induction

MCF10A cells (∼0.15 million) were seeded at a confluency of (∼30%–40%) in each well of a six-well plate and allowed to adhere overnight (∼16 h). MCF10A cells were treated with recombinant human TGF-β1 (10 ng/ml, PeproTech, #100-21) to induce EMT in complete growth medium. Media containing fresh TGF-β1 was replenished every ∼48 h for a total duration of 7 days. Control cells were maintained in complete media without TGF-β [24, 25].

GRHL2 mediated MET induction

MET was induced by overexpression of GRHL2 using the lentiviral vector pLVX-TetOn-Puro-GRHL2 (gift from Dr Ruby Yun-Ju Huang, National Taiwan University). Lentiviral particles were generated by transfecting HEK293T cells (∼1 × 10⁶) with 8 µg plasmid DNA using branched PEI (1:3 DNA:PEI ratio). Viral supernatants collected at 48 and 72 h, were filtered using a 0.45 µm filter and used to transduce MDAMB231 cells (∼1 × 10⁶) in the presence of 8 µg/ml polybrene (Sigma #H9268). Cells were selected with puromycin (2 µg/ml, 2 days) and GRHL2 overexpression was validated by immunoblotting (Fig. 1I) [26, 27].

Figure 1.

Induction and characterization of EMT and MET in breast cancer cell lines. (A) Immunoblot analysis of EMT induction in MCF7 cells transiently overexpressing TWIST1-GFP for 48 h. Cells were transfected with 2 μg of pEGFPN1 or hTWIST1-GFP (B) Immunofluorescence analysis of EMT in MCF7 cells overexpressing TWIST1-GFP or control pEGFPN1. E-cadherin and Vimentin (red), nucleus (DAPI, blue). Scale bar ∼10 μm (C) Schematic representation of EMT induction in MCF7 cells (D) Scatter plot of immunofluorescence assay, showing relative changes in the integrated density of E-cadherin and Vimentin in MCF7 cells. Quantification for panel (B) (n = 200). Data represent mean ± standard deviation (SD) from N = 3 independent biological replicates. (E) Immunoblot analysis of EMT induction in MCF10A cells treated with 10 ng/ml TGF-β for 7 days (F) Immunofluorescence analysis of EMT in MCF10A cells treated with TGF-β. E-cadherin (green); Vimentin (red); nucleus (blue, DAPI). Scale bar ∼10 μm (G) Schematic representation of EMT induction in MCF10A cells (H) Scatter plot of immunofluorescence assay, showing relative changes in the integrated density of E-cadherin and Vimentin in MCF10A cells [data shown in panel (F)] following EMT induction (n = 200). Data represent mean ± SD from N = 3 independent biological replicates (I) Immunoblot analysis of MET induction in MDAMB231 cells following doxycycline-induced GRHL2 overexpression for 48 h (J) Immunofluorescence analysis of MET in MDAMB231 cells overexpressing GRHL2. E-cadherin and Vimentin (red), nucleus (blue, DAPI). Scale bar ∼10 μm (K) Schematic representation of MET induction in MDAMB231 cells (L) Scatter plot of immunofluorescence assay, showing relative changes in the integrated density of E-cadherin and Vimentin in MDAMB231 cells following MET induction [data shown in panel (J)] (n = 200). Data represent mean ± SD from three independent biological replicates. Unpaired Student’s t-test was used to compute the P-value.

small interfering RNA (siRNA) transfection

MCF7, MCF10A, and MDA-MB-231 cells (∼0.35 million) were seeded in six-well plates or 35 mm dishes ∼24 h prior to transfection. Cells were cultured in Opti-MEM and transfected with 50 nM siRNA using 4 μl of RNAiMAX (Invitrogen #13778150). After ∼6 h, Opti-MEM was replaced with DMEM with 10% FBS for a total duration of ∼48 h. Cells were harvested for immunofluorescence assay, RT-PCR, or immunoblotting post-transfection.

Retroviral transduction

HEK293T cells (∼1 million) were transfected with ∼8 μg of the retroviral plasmid of interest (Supplementary Table 4), 4 μg of packaging plasmid Gag-Pol (Plasmid #14887), and VSV-G (2 μg, Plasmid #14888) using PEI (∼42 μl), in the ratio of DNA:PEI = 1:3 (w/v). Viral supernatant (∼8 ml) was collected ∼48 h post-transfection from HEK293T cells and added to ∼0.35 million cells (∼1.5 ml; seeded at ∼0.35 million cells per well in a six-well plate). Polybrene (10 μg/ml ∼10 μl Sigma #H9268) was added to facilitate viral attachment and entry. This procedure was repeated at ∼72 h post-transfection. Cells were selected using puromycin (2 μg/ml, Gibco #A1113802; for pBabe-Puro constructs) for ∼48 h or hygromycin (50 μg/ml, Gibco #10687010; for MSCV-F-Hygro constructs) for ∼24 h [28].

Lentiviral transduction

HEK293T cells were co-transfected with the transfer vector (8 µg), packaging plasmid psPAX2 (4 µg; Addgene #12260), and envelope plasmid pMD2.G (2 µg; Addgene #12259) using PEI (DNA:PEI = 1:3, w/v). Viral supernatants were collected at 48 and 72 h post-transfection and used to transduce target cells (∼3.5 × 10⁵ cells/well in a six-well plate) in the presence of 10 µg/ml polybrene. Transduced cells were selected with 2 µg/ml puromycin for 2 days [29].

Immunofluorescence

Cells (∼3.5 × 10⁵) were seeded on 22 × 22 mm coverslips and fixed with 4% paraformaldehyde in 1× phosphate buffered saline (PBS; pH 7.4) for 15 min at room temperature (RT), followed by permeabilization with 0.5% Triton X-100 for 10 min. After blocking with 1% bovine serum albumin (BSA) in 1× PBS for 30 min, cells were incubated with primary antibodies for 90 min at RT. Cells were incubated with Alexa Fluor-conjugated secondary antibodies (488 or 568; Invitrogen, 1:1000) for 1 h, followed by DAPI counterstaining (0.05 µg/ml, 2 min). Coverslips were mounted in SlowFade Gold Antifade Reagent (Invitrogen, S36937) and stored at 4°C until imaging. Dilutions and Catalogue numbers of antibodies used are listed in Supplementary Table 4.

Imaging and acquisition parameters

Confocal microscopy was performed using a Zeiss LSM 710 microscope (Carl Zeiss, Thornwood, NJ, USA) equipped with a 63 × Plan-Apochromat 1.4 NA oil immersion objective and an AxioCam MRm Rev.3 charge-coupled device camera (Zeiss). Images were acquired at a zoom of 1.0 as Z-stacks at a resolution of 512 × 512 pixels per frame with an 8-bit pixel depth per channel. Voxel size was set to 0.105 μm × 0.105 μm × 0.34 μm, with line averaging of 4.0 in sequential three-channel mode. Fluorescence signals were detected appropriately with filters for DAPI, Alexa Fluor-488, and Alexa Fluor-568. Image processing and analysis were performed using ImageJ software.

Immunoblotting

Cells were lysed in Radioimmunoprecipitation Assay (RIPA) buffer [50 mM Tris–HCl, pH 7.4–8.0, 150 mM NaCl, 1% NP-40, 0.1% sodium dodecyl sulphate (SDS), 0.5% sodium deoxycholate, 1 mM ethylenediaminetetraacetic acid (EDTA)] supplemented with protease and phosphatase inhibitors (Roche) for 10 min at 4°C. Protein concentration was determined using the Bicinchoninic Acid (BCA) assay. Equal amounts of protein were resolved on 12.5% sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS–PAGE) and transferred to Polyvinylidene Fluoride (PVDF) membranes (0.45 µm; Millipore #IPVH00010) at 100 V for 2 h at 4°C. Membranes were blocked in 5% milk or 5% BSA in TBST (20 mM Tris–HCl, pH 7.5, 150 mM NaCl, 0.1% Tween-20) for 1 h and incubated overnight at 4°C with primary antibodies (1:1000 unless stated otherwise) against E-Cadherin, Vimentin, ZEB1, EZH2, pEZH2(T345), Lamin A/C, pLamin A/C(S22), pSmad3, Smad3, SUV39H1, Lamin B1, Lamin B2, Cyclin B1, Snail1, GRHL2, pCDK1, Twist1, GAPDH (1:7000) H3K27me3, H3K9me3, and H3 (1:3000). After washing, membranes were incubated with horseradish peroxidase-conjugated secondary antibodies (anti-rabbit #111-035-003; anti-mouse #515-035-003; Jackson Labs; 1:10 000) for 1 h at RT. Bands were visualized using ECL substrate (Bio-Rad #1705061) and imaged on LAS4000 (GE). Densitometric quantification was performed using ImageJ. Dilutions and Catalogue numbers of primary antibodies used are listed in Supplementary Table 4.

RT-PCR

RNA isolation

Cells (∼3 × 10⁵) were seeded in six-well plates and lysed in 1 ml TRIzol reagent (Invitrogen #15596018). Chloroform (100 µl per 500 µl TRIzol) was added, and samples were vortexed, incubated for 10 min at RT, and centrifuged at 12 000 × g for 15 min at 4°C. The aqueous phase was mixed with an equal volume of isopropanol, incubated for 15 min, and centrifuged at 12 000 × g for 15 min at 4°C. RNA pellets were washed with 70% ethanol, air-dried, and resuspended in nuclease-free water. RNA concentration and purity were measured using a NanoDrop spectrophotometer.

Complementary DNA synthesis and quantitative real-time PCR

Complementary DNA was synthesized from 1 µg total RNA using the PrimeScript RT Reagent Kit (Takara #6110B) with oligo(dT) and random hexamer primers. Quantitative PCR was performed using TB Green Premix Ex Taq II (Takara #RR820W) on a Bio-Rad CFX96 system under the following conditions: 95°C for 30 s, followed by 40 cycles of 95°C for 5 s and 60°C for 30 s. Melt-curve analysis (60–95°C, 0.3°C/s) confirmed amplicon specificity. Relative gene expression was calculated using the 2^−ΔΔCt method with GAPDH as the internal control and untreated samples as the reference.

Co-immunoprecipitation

Cells (∼1 × 10⁷) were lysed in co-immunoprecipitation (Co-IP) buffer (50 mM Tris–HCl, pH 8.0, 300 mM NaCl, 0.5% NP-40) supplemented with protease inhibitors (Roche #5892970001) and incubated on ice for 15 min. Lysates were cleared by centrifugation at 14 000 rpm for 15 min at 4°C, and protein concentration was determined by BCA assay. For each reaction, 2 µg of antibody or normal rabbit IgG (control) was pre-bound to Protein A–Sepharose beads (Roche #P3391) for 4 h at 4°C. Equal amounts of lysate (∼500 µg total protein) were added and incubated overnight at 4°C with rotation. Beads were washed 5–6 times with a cold Co-IP buffer and eluted in 2 × Laemmli buffer at 95°C for 10 min. Eluates were resolved by SDS–PAGE and analyzed by immunoblotting. Antibodies used: anti-Lamin A/C (Abcam #ab108595), anti-EZH2 (CST #5246), and normal rabbit IgG (Invitrogen #10500C).

Chromatin immunoprecipitation

Cells (∼1 × 10⁷) were crosslinked with 1% formaldehyde in 1× PBS (pH 7.5) for 10 min at RT, and the reaction was quenched with 125 mM glycine for 5 min. Cells were washed twice with cold PBS and lysed in a buffer containing 1% SDS, 10 mM EDTA, and 50 mM Tris–HCl (pH 8.1) supplemented with protease inhibitors. Chromatin was sonicated (Covaris S220; 60% duty cycle, 30 s on/off for 30 min) to ∼200–500 bp fragments, verified by agarose gel electrophoresis.

For each IP, 25 µg of chromatin was diluted in ChIP dilution buffer (0.01% SDS, 1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris–HCl, pH 8.1, 167 mM NaCl) and incubated overnight at 4°C with antibody-bound Protein A/G agarose beads. The following antibodies were used: Lamin A/C (Abcam #ab108595), EZH2 (CST #5246), H3K27me3 (Millipore #07-449), H3K36me3 (Abcam #ab9050), RNA Pol II (Ser2p; Covance #MMS-129R), and control IgG (Invitrogen #10500C). Beads were sequentially washed with low-salt, high-salt, LiCl, and TE buffers and eluted in 1% SDS, 0.1 M NaHCO₃. Eluates were treated with proteinase K (2 µl, 20 mg/ml, 55°C, 30 min), and crosslinks were reversed at 65°C for 4–6 h. DNA was purified using PCR purification columns (Promega #A9281) and analyzed by qPCR [30, 31].

Co-IP and mass spectrometry analysis

Endogenous Lamin A/C–interacting proteins were identified by Co-IP followed by Liquid Chromatography-Tandem Mass Spectrometry (LC–MS/MS) analysis [32, 33]. Briefly, MCF7−TWIST1(EV), MCF7+TWIST1, MDAMB231−Dox (−GRHL2), and MDAMB231+Dox (+GRHL2) cells (∼1 × 10⁷) were lysed in Co-IP buffer (50 mM Tris–HCl, pH 7.5; 150 mM NaCl; 1% NP-40) containing protease inhibitors. Clarified lysates (1 mg) were incubated overnight at 4°C with Protein A magnetic beads (SureBeads, Bio-Rad #161-4013) pre-bound to anti-Lamin A/C antibody (2 µg, Abcam #ab108595) or control IgG (Invitrogen #10500C). Immunocomplexes were washed, eluted in 2 × Laemmli buffer at 95°C, and resolved by SDS–PAGE.

Coomassie-stained gel bands (∼2 cm) were excised, reduced (10 mM Dithiothreitol (DTT)), alkylated (55 mM iodoacetamide), and digested with sequencing-grade trypsin (Promega #V5111). Peptides were extracted, desalted on C18 columns, and analyzed on a Sciex TripleTOF 6600 mass spectrometer coupled to an Eksigent nano-LC 425 system. Data were processed using ProteinPilot (v2.0.1, Sciex) and searched against the human UniProt database.

The Lamin A/C interactome under EMT and MET conditions is provided in Supplementary Tables 2 and 3, respectively. The mass spectrometry (MS) data are available in the PRIDE repository (Dataset ID: PXD060831).

RNA sequencing

MCF10A cells (∼0.35 million) were plated in six-well plates or 35 mm dishes around 24 h before transfection. The cells were maintained in Opti-MEM and transfected with 50 nM siRNA using 4 μl of RNAiMAX (Invitrogen #13778150). Transfection (was for ∼6 h), after which the Opti-MEM medium was replaced with complete media, and the cells were incubated for ∼48 h. Following transfection, MCF10A cells were collected in 1 ml TRIzol, and RNA sequencing (RNA-Seq) was outsourced to Medgenome, Bangalore. Quality control checks for the two independent biological replicates were confirmed through RNA Integrity Number (RIN) number analysis (RIN = 10, for all samples). RNA-Seq data are provided in Supplementary Table 1. The datasets have been deposited in Gene Expression Omnibus under accession number GSE289750.

Cell cycle analysis

Cells (∼0.3 million) were harvested by trypsinization, washed with 1× PBS (pH = 7.5), and fixed in 70% ice-cold ethanol at 4°C, overnight. The fixed cell pellets were resuspended in 1× PBS and treated with RNase A (10 μg; Sigma, R6513-10MG) at 37°C for 45 min. Subsequently, propidium iodide (PI; 10 μg; Sigma, P4170-10MG) was added to the samples for DNA staining. The cell suspensions were passed through a 70 μm cell strainer (Corning, CLS431751) and processed for further analysis. Cell cycle profiling was performed using a fluorescence-activated cell scanner (BD FACSCalibur™, BD Biosciences) with approximately 10,000 events recorded per sample. Data were analyzed using FlowJO (V10.10) software.

Transwell migration (Boyden-chamber assay)

Cells (∼3.5 × 10⁵) were seeded in six-well plates and subjected to overexpression of Lamin A or EZH2, or their phosphodeficient (S22A-Lamin A, T345A-EZH2) or phosphomimetic (S22D-Lamin A, T345D-EZH2) variants. Approximately 5 × 10⁴ cells were seeded in the upper chamber of 8 µm pore Transwell inserts (Corning #CLS3422) in serum-free media. Complete media (DMEM for MCF7 or RPMI for MDAMB231, 10% FBS) was added to the lower chamber and cells were allowed to migrate for ∼36 h. Nonmigrated cells were removed, and migrated cells were fixed with 4% paraformaldehyde for 10 min, washed, and stained with DAPI for 2 min. Nuclei were imaged at 10× on an EVOS microscope. Images were processed using the Threshold and Watershed functions, and nuclei were quantified with the “Analyze Particles” function (size 50 – ∞ pixels², circularity 0.5–1.0). Average nuclei per field were calculated and analyzed using imageJ.

Wound healing (scratch wound assay)

MCF10A cells (∼3.5 × 10⁵) were subjected to overexpression of Lamin A or EZH2, or their phosphodeficient (S22A-Lamin A, T345A-EZH2) or phosphomimetic (S22D-Lamin A, T345D-EZH2) variants. Cells were treated with 500 ng/ml doxycycline for ∼48 h to induce endogenous Lamin A/C or EZH2 depletion, and EMT was induced with 10 ng/ml TGF-β. After ∼48 h, ∼5 × 10⁴ cells were seeded per well in ibidi two-well inserts (ibidi #81176) in a 24-well plate overnight. Inserts were removed to create a wound gap, and migration into the wound area was monitored for 36 h (in assay media) using the Operetta High-Content Imaging System. Snapshots of wound closure was quantified with the integrated software in the Operetta Imaging System.

Animal studies

All animal procedures were conducted following approval from the Institutional Animal Ethics Committee (IISER-Pune/IAEC/2024–01/08) at the Indian Institute of Science Education and Research, Pune, in compliance with the Committee for the Purpose of Control and Supervision of Experiments on Animals (CPCSEA), Government of India. Mice were housed in the National Facility for Gene Function in Health and Disease (NFGFHD), IISER Pune, under standard laboratory conditions with unrestricted access to food and water.

In vivo tumor formation assay in NOD-SCID Mice

Female NOD-SCID mice (4–6 weeks old; Jackson Laboratory) were used for in vivo tumor formation studies. MDAMB231 cells expressing full-length (Lamin A or EZH2) or phosphodeficient (S22A-Lamin A, T345A-EZH2) or phosphomimetic (S22D-Lamin A, T345D-EZH2) variants were pre-treated with doxycycline for 48 h to deplete endogenous Lamin A/C or EZH2. Approximately 1 × 10⁶ cells in 100 µl PBS were injected into the fourth mammary fat pad of each mouse. Knockdown was maintained via drinking water containing 1% sucrose and 500 ng/ml doxycycline, replenished every 72 h. Tumor growth was monitored, and volumes were calculated as (width² × length)/2. Mice were euthanized after 5 weeks, and tumors were harvested for analysis.

Tumor dissociation and primary cell culture

Excised tumors were minced using a scalpel and enzymatically digested in Hank's Balanced Salt Solution (HBSS) (Gibco, #14025050) containing collagenase IV (1 mg/ml, Sigma #C5138-1G), hyaluronidase V (0.1 mg/ml, Sigma #H6254-500MG), and DNase IV (20 mg/ml, Sigma #D5025-15KU) at RT. The suspension was filtered through a 70 μm mesh, centrifuged (1200 rpm, 5 min), and washed twice with HBSS. Cells were either lysed for Co-IP or plated in RPMI (Gibco, #11875093) for primary culture.

Proximity ligation assay

PLA was performed using the Duolink in Situ Detection Kit (Sigma #DUO92008) following the manufacturer’s protocol. Cells were treated with cytoskeletal (CSK) buffer (10 mM PIPES, pH 6.8, 100 mM NaCl, 300 mM sucrose, 3 mM MgCl₂, 1 mM Ethylene Glycol Tetraacetic Acid (EGTA)), fixed in 4% Paraformaldehyde (PFA) (pH 7.4, 12 min), permeabilized with 0.5% Triton X-100 (10 min), and blocked with Duolink Blocking Solution (30 min, RT). Primary antibodies—Mouse anti-Lamin A/C (Jol2, Abcam #ab40567, 1:50), Rabbit anti-EZH2 (D2C9, CST #5246, 1:1000), and Rabbit anti-Phospho-cdc2 (Thr161, CST #9114, 1:500)—were applied overnight at 4°C.

PLA probes (anti-Mouse PLUS and anti-Rabbit MINUS, 1:5) were added and incubated at 37°C for 1 h, followed by ligation (Duolink ligase, 1 U/µl, 1:40) and amplification (Duolink polymerase, 10 U/µl, 1:80). Nuclei were counterstained with DAPI using Duolink Mounting Medium (#DUO82040).

Imaging and acquisition parameters

PLA images were acquired on a Zeiss anisotropy microscope (Carl Zeiss, Thornwood, NJ, USA) with a 63×/1.4 NA oil objective and AxioCam MRm Rev.3 camera using ZEN software. Z-stacks (512 × 512 pixels, 8-bit) were captured with a voxel size of 0.105 × 0.105 × 0.34 μm and line averaging of 4 in sequential two-channel mode. Fluorescence was detected for DAPI and Alexa Fluor-568. Images were processed and analyzed using ImageJ.

Statistical analysis

All statistical analyses were performed using GraphPad Prism version 8.4 (GraphPad Software, San Diego, CA, USA). Data are presented as mean ± standard deviation (SD) unless otherwise specified. Comparisons between the two groups were conducted using an unpaired two-tailed Student’s t-test. For multiple-group comparisons, one-way or two-way analysis of variance (ANOVA) followed by Tukey’s or Sidak’s post hoc test was used, as appropriate. Statistical significance was defined as P <0.05. Sample sizes and replicates for each experiment are detailed in the figure legends.

Results

Induction and reversal of EMT

EMT is a dynamic and reversible cellular process that plays a pivotal role in cancer progression [34]. EMT is involved in ∼90% of cancers of epithelial origin, which increases cell migration while MET facilitates colonization and metastasis [35].

To elucidate the mechanisms that regulate EMT and MET, we selected three well-established cell types derived from the human breast, namely MCF7, MCF10A, and MDAMB231 [36]. The authenticity of these cell lines was validated by evaluating modal chromosome numbers (Supplementary Fig. S1A and B). We also performed phalloidin staining, followed by confocal imaging (Supplementary Fig. S1C), which showed an increased aspect ratio (AR) of mesenchymal [MDAMB231 cells (AR: 7.19±1.80)], as compared to epithelial cells [MCF7 (AR: 1.06±0.62) and MCF10A (AR: 1.09±0.59); Supplementary Fig. S1D]. Furthermore, immunoblotting and immunofluorescence assays at the single-cell level underscore the epithelial nature of MCF7 cells, characterized by high E-cadherin expression and low Vimentin levels. In sharp contrast, MDA-MB-231 cells are mesenchymal, with relatively higher Vimentin and lower E-cadherin levels (Supplementary Fig. S1C and E–G).

We induced EMT in MCF7 cells by overexpressing hTWIST1-GFP for ∼48 h. Immunoblotting analysis revealed a ∼75% reduction in E-cadherin and a ∼90% increase in Vimentin levels. Additionally, key mesenchymal transcription factors—Snail1, Twist1, and Zeb1—were significantly upregulated at the protein level (Fig. 1A). We corroborated these findings with immunofluorescence staining, which showed a decrease in E-cadherin and an increase in Vimentin expression in MCF7 cells overexpressing hTWIST1-GFP (Fig. 1B and D, methodological illustration in Fig. 1C), consistent with the immunoblot data. We treated MCF10A cells with TGF-β (10 ng/ml) for 7 days, which resulted in a striking reduction in E-cadherin expression (>70%) and a concomitant increase in Vimentin (∼60%; Fig. 1E), consistent with EMT induction [24]. Immunofluorescence staining of these cells also showed a significant upregulation of Vimentin and downregulation of E-cadherin (Fig. 1F and H, methodological illustration in Fig. 1G). Furthermore, transcription factors that mark mesenchymal cells (Snail1, Twist1, and Zeb1) were significantly upregulated at the protein level (Fig. 1E).

We next overexpressed the transcription factor GRHL2 in the mesenchymal MDAMB231 cells for ∼48 h [26, 27]. GRHL2 is a transcription factor that suppresses EMT by directly activating E-cadherin and repressing mesenchymal programs, thereby promoting MET, which makes it a useful model to study the reversibility of EMT [37, 38]. Overexpression of GRHL2 showed a striking increase in E-cadherin (>95%), decrease in Vimentin and mesenchymal transcription factors levels, leading to a transition in epithelial phenotype (Fig. 1I–L). Taken together, these representative cell lines derived from the E–M spectrum, serve as suitable models for studying EMT and its reversibility MET, offering a framework to investigate their regulatory mechanisms in the context of mammary epithelial and breast cancer respectively.

Impact of EMT induction on nuclear topology and Lamin A/C expression

EMT is accompanied by significant changes in cellular morphology and cytoskeletal organization. Altered cytoskeletal organization directly impacts nuclear morphology [39]. We examined the extent of changes in nuclear morphology upon EMT by overexpressing TWIST1 in MCF7 cells. This showed a significant decrease in the circularity and sphericity [from 0.94 ± 0.03 to 0.50 ± 0.09 in MCF7-TWIST1 cells (P < 0.0001)] of the nucleus, with no significant changes in either surface area or volume (Supplementary Fig. S2A–D). In contrast, GRHL2-mediated MET induction, restored nuclear shapes consistent with epithelial-like cells, with increased sphericity (from 0.45 ± 0.08 to 0.94 ± 0.03 in MDAMB231 +GRHL2 cells; Supplementary Fig. S2E and F). Taken together, these data reveal significant changes in nuclear morphology upon EMT or MET induction.

We next assessed the protein levels of Lamin A/C and B-type lamins, as lamins define nuclear structure and function [39, 40]. While EMT induction in the epithelial cells (MCF7 and MCF10A) showed a marked reduction in Lamin A/C levels (Fig. 2A–F), the levels of B-type Lamins were relatively unaltered (Supplementary Fig. S2G, H). Conversely, Lamin A/C levels were upregulated upon GRHL2-driven MET in MDAMB231 cells (Fig. 2G–I), without any significant changes in the expression levels of B-type lamins (Supplementary Fig. S2I and J).

Figure 2.

Impact of EMT induction on Lamin A/C expression. (A, D, G) Immunofluorescence analysis of Lamin A/C (red) in MCF7 (A), MCF10A (D), and MDAMB231 (G) cells upon EMT (A, D) or MET (G) nucleus (DAPI, blue). Scale bar ∼10 μm. (B, E, H) Mean fluorescence intensity of Lamin A/C quantified by line scan analysis across the nucleus in MCF7 (B), MCF10A (E), and MDAMB231 (H) cells. Data represent mean ± SD from N = 3 independent biological replicates (n = 250). Unpaired Student’s t-test was used to calculate P-values. (C, F, I) Immunoblot analysis of total Lamin A/C protein levels in MCF7 (C), MCF10A (F), and MDAMB231 (I) cells upon EMT (C, F) or MET (I) induction. GAPDH (C, F) and HSP70 (I) are loading controls. (J) Immunoblot analysis of Lamin A/C, Lamin B1, and Lamin B2 levels across 11 cell lines of breast origin with increasing mesenchymal characteristics. Loading control: Histone H3 (K) RT-qPCR analysis of LMNA transcript levels in MCF7 and MCF10A upon EMT and MET in MDAMB231 cells. Data represent mean ± SD (N = 3, n = 9). Unpaired Student’s t-test was used to compute the P-values. Means are compared between (B) −Twist1 (control) and +Twist1; (E) −TGFβ (control) and (H) +TGFβ; −GRHL2 (control) and +GRHL2, statistical significance, P-value <0.05.

We quantified the levels of A- and B-type lamins across a panel of 11 breast cancer cell lines spanning the EM spectrum [36]. Interestingly, Lamin A/C showed a progressive decline with increasing mesenchymal characteristics, whereas B-type lamin levels did not display such a trend (Fig. 2J), suggesting a selective downregulation of Lamin A/C during EMT.

Importantly, transcript levels of LMNA remained unchanged upon both EMT and MET (Fig. 2K), indicating that Lamin A/C is post-transcriptionally regulated [41]. To further determine if EMT impacts the stability of Lamin A/C, we performed a time-course cycloheximide (CHX) chase assay in MCF7 cells. TWIST1 overexpression, followed by CHX treatment (10 μg/ml) showed a temporal reduction in Lamin A/C protein levels, whereas Lamin A/C remained unaltered in control cells (Empty vector) over the same duration (Supplementary Fig. S3A and B). Furthermore, co-treatment with 1 μM MG132 and CHX restored the stability of Lamin A/C protein after ∼8 h upon TWIST1 overexpression (Supplementary Fig. S3C and D), suggesting that EMT induces Lamin A/C degradation via a proteasome-dependent pathway.

Collectively, these results demonstrate that EM plasticity is associated with changes in Lamin A/C expression and nuclear morphology, with Lamin A/C being regulated by post-translational degradation during EMT.

Lamin A/C determines EM identity

To investigate the role of Lamin A/C in EM plasticity, we performed RNA-Seq analysis upon Lamin A/C depletion in MCF10A cells. Differential gene expression analysis (DESeq2) revealed a robust upregulation of mesenchymal-associated genes including VIM, SNAI1, TWIST1, and ZEB1, which were among the top 50 upregulated genes, accompanied by downregulation of epithelial markers such as CLDN1, OCLN, and CDH1 (Fig. 3A and B) [42]. Consistent with these observations, Gene Ontology (GO) analysis showed a significant enrichment (P <.05) of genes associated with EMT regulation, cell migration, secondary neoplasms, cell–cell adhesion, and regulation of mitotic nuclear division (Fig. 3C) [43]. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis revealed modulation of signalling pathways implicated in ECM–receptor interaction, TGF-β signalling pathway and regulation of actin cytoskeleton, while GO Cellular Component analysis highlighted ECM organization and cell migration (Supplementary Fig. S4A and B) [44]. Furthermore, Gene Set Enrichment Analysis (GSEA) enriched for cell adhesion and migration-related gene sets, both hallmarks of EMT (Fig. 3D; Principal Component Analysis (PCA) analyses of RNA seq data between independent replicates Supplementary Fig. S4C).

Figure 3.

Effect of Lamin A/C perturbation on EMT and MET. (A) Volcano plot showing differentially expressed genes in MCF10A cells upon Lamin A/C knockdown. Downregulated, upregulated, and nonsignificant genes. (B) Heatmap of the top 50 [P <0.05; log₂ Fold Change (FC) > 2] differentially expressed genes in MCF10A cells upon Lamin A/C knockdown. Downregulated and upregulated genes, respectively (N = 2 biological replicates). (C) GO enrichment analysis of differentially expressed genes (P ≤0.05), showing the most enriched biological processes. (D) GSEA plot showing enrichment for EMT upon Lamin A/C knockdown (normalized enrichment score = 3.337). (E, F) Immunofluorescence analysis of MCF7 (E) and MCF10A (F) cells upon Lamin A/C knockdown. Lamin A/C, E-cadherin or Vimentin, and phalloidin. Nucleus (DAPI). Scale bars, ∼10 μm (G) Immunofluorescence analysis of MDAMB231 cells overexpressing Lamin A*-GFP upon endogenous Lamin A/C depletion. E-cadherin (top panel) or Vimentin (bottom panel), and Lamin A/C (−Dox and +Dox panels only). Nucleus (DAPI). Scale bar ∼10 μm. (H–J) Immunoblot analysis of EMT markers in MCF7 (H), MCF10A (I) cells upon Lamin A/C knockdown, and MDAMB231 (J) cells upon Lamin A overexpression. RNA-Seq was performed in N = 2 independent biological replicates. Lamin A* denotes a full-length Lamin A construct resistant to doxycycline-induced depletion of endogenous Lamin A/C. Statistical significance, P-value <0.05.

To validate these results in an independent epithelial context, Lamin A/C was depleted in MCF7 cells. Lamin A/C depletion showed an increase in the AR of MCF7 cells [si Lamin A/C: mean ± SD = 5.49 ± 2.09, si LacZ (control) mean ± SD = 1.02 ± 0.58] [Supplementary Fig. S5A, A(i), and A(ii)]. In contrast, overexpression of Lamin A in MDAMB231 cells, revealed the acquisition of cobblestone-like epithelial morphology. This was corroborated by quantifications of AR, which showed a shift towards an epithelial-like architecture of cells [Supplementary Fig. S5B, B(i), and B(ii)].

Immunofluorescence analyses in both MCF10A and MCF7 cells showed a ∼70% reduction in E-cadherin and a ∼65% increase in Vimentin upon Lamin A/C loss (Fig. 3E and F; quantification: Supplementary Fig. S5C and D). Transcriptional upregulation of mesenchymal factors (VIM, SNAI1, TWIST1, and ZEB1) and decreased CDH1, was consistent with the mesenchymal morphology in MCF7 cells upon Lamin A/C depletion (Supplementary Fig. S5F). Immunoblotting showed a decrease in E-cadherin and increase in the levels of Vimentin, Snail1, Twist1, and Zeb1 in Lamin A/C–depleted cells (Fig. 3H and I).

In contrast, Lamin A overexpression in MDAMB231 cells (upon endogenous Lamin A/C depletion) showed a significant upregulation in the transcript levels of E-cadherin (∼95% increase), and downregulation of mesenchymal regulators (VIM, SNAI1, TWIST1, and ZEB1; Supplementary Fig. S5G). Immunoblotting and immunofluorescence analyses reiterate these effects, showing enhanced E-cadherin expression and suppression of mesenchymal markers (Fig. 3G and J; quantification: Supplementary Fig. S5E).

We also determined the EM status in MCF7 and MDAMB231 cells upon depleting B-type Lamins (Lamin B1 and Lamin B2). Depletion of B-type Lamins did not significantly alter the expression of EM markers underscoring a specific role of Lamin A/C in regulating EM plasticity (Supplementary Fig. S5H and I).

Impact of EMT or MET induction on the lamin interactome

Nuclear Lamins play a critical role in the maintenance of nuclear integrity, chromatin organization and gene regulation [45]. Lamina-associated domains (LADs) are genomic regions enriched for nuclear Lamins and are marked by histone marks H3K9me3 and H3K27me3 associated predominantly with heterochromatin [46, 47]. We next performed immunoblotting upon EMT induction, which showed a decrease (∼50%) in the total levels of H3K9me3 and H3K27me3, suggesting a marked decrease in inactive histone marks [Supplementary Fig. S6A(i) and B(i)]. In contrast, MET induction increased levels of these heterochromatic marks by ∼60% [Supplementary Fig. S6C(i)]. Notably, Lamin A/C knockdown in the epithelial cells (MCF7 and MCF10A) also showed a significant reduction (∼50%) in H3K27me3 and H3K9me3, reinforcing the functional association between Lamin A/C and chromatin compaction [Supplementary Fig. S6A(ii) and B(ii)]. Furthermore, Lamin A overexpression, showed an increase in H3K27me3 and H3K9me3 in MDAMB231 cells [Supplementary Figure S6C(ii)], consistent with the role of Lamin A/C in the maintenance of heterochromatin [48].

Given the dynamic changes in heterochromatin markers during EMT and MET, we next examined whether the interactors of Lamin A/C are differentially enriched across E and M states [19, 49, 50]. We performed Co-IP followed by MS to identify the sub-interactome of Lamin A/C in the epithelial and mesenchymal states. While, MS analyses identified 44 unique protein-protein interactors, including EZH2, in Lamin A/C pull-downs from epithelial MCF7 cells, 90 unique interactors including CCNB1 and pCDK1, among others were detected upon EMT induction in MCF7-TWIST1 cells. A total of 49 interactions (LMNB1, LMNB2, EHMT2, SUV39H1, PRRX1) were common to both the control and Twist1 overexpressing MCF7 cells (Fig. 4A; PCA analysis Supplementary Fig. S6D). In contrast, MET induction in MDA-MB-231 cells markedly reconfigures the Lamin A/C interactome. In control cells, 103 unique interactors were identified involving (CCNB1 and pCDK1). In contrast, GRHL2 mediated MET-induction revealed 71 distinct interactors, including (EZH2). Notably, 72 proteins, including (LMNB1, LMNB2, CBX5, EHMT2, SUV39H1) were consistently associated with Lamin A/C in both control and GRHL2-overexpressing MDAMB231 cells (Fig. 4B; PCA analysis Supplementary Fig. S6E).

Figure 4.

Dynamic Remodeling of the Lamin A/C Interactome During EMT and MET. (A, B) Venn diagrams showing unique and common interactors of Lamin A/C identified by (Immunoprecipitation - Mass Spectroscopy) IP-MS in MCF7 versus MCF7-TWIST1 (A) and MDAMB231 versus MDAMB231-GRHL2 (B). (C, D) Representative STRING network analysis of Lamin A/C interactors in MCF7 versus MCF7-TWIST1 (C) and MDAMB231 versus MDAMB231-GRHL2 (D). (E–G) Co-IP of Lamin A/C in MCF7 (E), MCF10A (F), and MDAMB231 (G) cells upon EMT (E, F) or MET (G) induction, followed by immunoblotting for EZH2 and Lamin A/C. IgG: isotype control, an approximately equal amount of antibody is used for immunoprecipitation. (H, I) Proximity ligation assay (PLA) detects Lamin A/C–EZH2 interaction in MCF7 (H) and MDAMB231 (I) cells upon EMT (H) or MET (I) induction. PLA signal (red), nucleus (blue, DAPI). Scale bar: ∼10 μm. (J, K) Quantification of PLA signal in MCF7 (J) and MDAMB231 (K) cells. Data represent mean ± SD from N = 3, independent biological replicates, and P-values calculated by one-way ANOVA and means are compared between pBp-EV and pBP-Twist (J) and −Dox (GRHL2) and +Dox (GRHL2) (K). (L) Time-course analysis of Lamin A/C–EZH2 interaction by immunoprecipitation of Lamin A/C in MCF10A cells during EMT progression [∼12 to ∼168 h (∼7 days) post-TGF-β treatment] and MET recovery [5 days post-TGF-β withdrawal (WD)], assessed by Co-IP and immunoblotting. IgG: isotype control, statistical significance, P-value <0.05.

GO pathway enrichment analysis of the Lamin A/C interactome in epithelial MCF7 cells revealed that its interacting partners are primarily involved in the regulation of gene expression, messenger RNA (mRNA) metabolic processes, ribonucleoprotein complex assembly, and regulation of mRNA stability. In contrast, Twist1-mediated EMT shows a shift in the Lamin A/C interactome, enriching for pathways related to cell cycle regulation, nuclear envelope disassembly and DNA damage repair (Supplementary Fig. S7A and B). Remarkably, GRHL2-mediated MET re-established interactions between Lamin A/C and proteins involved in gene expression regulation, such as EZH2 (Supplementary Fig. S7C and D). These findings suggest that the Lamin A/C interactome is dynamically reprogrammed during EMT and MET, reflecting the functional requirements of epithelial versus mesenchymal cell states [51].

A key finding from the mass spectroscopy studies unravelled dynamic Lamin A/C interactome as it showed a reversible association with histone methyltransferase EZH2, which was enriched in MCF7 and upon MET induction in MDAMB231 cells. Interestingly, Lamin A/C-EZH2 interaction was undetectable in (i) MDAMB231 cells (ii) upon Twist1 induction in MCF7 cells. In contrast, Lamin A/C interacts with phosphorylated CDK1 (pCDK1) predominantly in the mesenchymal cells (MCF7-TWIST1 and control: MDAMB231). However, this interaction was undetectable upon MET induction (Fig. 4C and D). Taken together, these findings suggest a switch in the interactions of Lamin A/C, consistent with the cellular state, with Lamin A/C involved in chromatin-modifying functions via EZH2 in epithelial cells, while in mesenchymal states, Lamin A/C facilitates cell cycle regulation through its interaction with pCDK1.

To validate these findings, we performed Co-IP of Lamin A/C followed by immunoblotting for EZH2 upon EMT and MET induction. Co-IP results showed a loss of the Lamin A/C–EZH2 interaction upon EMT induction and its restoration during MET (Fig. 4E–G). Furthermore, immunoprecipitation with either of the B-type Lamins (Lamin B1 or B2), did not show an interaction with EZH2, indicating that this association with EZH2 is specific to Lamin A/C. However, both the B-type Lamins interact with the methyltransferases G9a and SUV39H1, consistent with previous studies [52]. Of note, these interactions with the B-type Lamins were unaltered upon EMT induction in MCF10A cells (Supplementary Fig. S7E and F).

To further determine if Lamin A/C interacts with PRC2 components other than EZH2, we performed Co-IP of Lamin A/C, followed by immunoblotting for PRC2 components—EED and Suz12—in MCF10A cells. This showed a specific association with EZH2, but not with other PRC2 components such as SUZ12 or EED (Supplementary Fig. S7G). To assess the integrity of the PRC2 complex upon EMT, EZH2 was immunoprecipitated from TGF-β-treated MCF10A cells, where it retained its association with SUZ12 and EED, indicating that the Lamin A/C–EZH2 interaction is independent of the PRC2 complex (Supplementary Fig. S7H).

To further substantiate the interaction between Lamin A/C and EZH2, we performed PLA upon EMT and MET induction. We detected a striking decrease (∼10-fold) in the number of PLA foci upon EMT induction, suggesting a loss of Lamin A/C–EZH2 interaction (Fig. 4H and J). In contrast, GRHL2-mediated MET induction re-established the interaction between Lamin A/C and EZH2, as demonstrated by a significantly higher number (∼10-fold) of PLA foci (Fig. 4I and K).

To investigate the temporal dynamics of the Lamin A/C–EZH2 interaction, we performed a time-course assay following EMT induction via TGF-β treatment and subsequent MET induction by withdrawal of TGF-β. Immunoprecipitation of Lamin A/C at different time points [∼12–168 h (∼7 days)] revealed a progressive loss of interaction with EZH2, with complete dissociation at ∼72 h post-EMT induction. Conversely, withdrawal of TGF-β [∼12–120 h (∼5 days)] restored Lamin A/C–EZH2 interaction after ∼72 h (Fig. 4L). These findings further corroborate a dynamic and reversible association between Lamin A/C and EZH2. In summary, these studies reveal that EMT induction disrupts the Lamin A/C–EZH2 interaction, with a concomitant decrease in heterochromatin marks, while MET promotes this interaction and re-establishes heterochromatin marks.

CDK1-mediated phosphorylation regulates Lamin A/C–EZH2 interaction and EMT progression

To address the molecular basis of Lamin A/C–EZH2 interaction, we performed HADDOCK, which predicted an interface between the head domain of Lamin A/C (Arg25, Arg28) and EZH2 (Asp136, Asp140; Supplementary Fig. S8A) [53]. We co-overexpressed GFP-tagged Lamin A and FLAG-tagged EZH2 deletion constructs into HEK293T cells. Co-IP revealed that Lamin A binds to EZH2 at the Ala301–Ala500 region, whereas reverse Co-IP (using Lamin A deletion constructs) shows that EZH2 interacts with the head domain of Lamin A (Fig. 5A and B).

Figure 5.

CDK1-mediated phosphorylation regulates Lamin A/C–EZH2 interaction and EMT progression. (A) Co-IP of FLAG in HEK293T cells co-transfected with full-length EZH2-FLAG and Lamin A-GFP deletion mutants (ΔHead 1–29, ΔRod 31–387, ΔIgG 428–549, ΔTail 550–664 of Lamin A). (B) Co-IP of GFP in HEK293T cells co-transfected with full-length Lamin A-GFP and EZH2-FLAG deletion mutants (Δ1–300, Δ301–500, Δ501–746 of EZH2). (C) PLA detects Lamin A/C–pCDK1(T161) interaction in MCF10A cells treated with 10 ng/ml TGF-β for ∼7 days. Nucleus (DAPI), PLA signal in red. Scale bar ∼10 μm (D) PLA detects Lamin A/C– EZH2 interaction in MCF10A cells treated with DMSO or 10 μM RO3306 for ∼18 h in the ± TGF-β for ∼7 days, nucleus (DAPI). PLA signal in red. Scale bar ∼10 μm. (E) Schematic representation of RO3306 and MG132 treatment in MCF7 cells (F) Quantification of PLA foci/nucleus of the data in (C). P-values were calculated using ANOVA. Means are compared between +TGFβ and –TGFβ (control) conditions with the single antibody control. (G) Quantification of PLA foci/nucleus of the data in (D). Statistical significance was determined using unpaired Student’s t-tests. Means are compared between +TGFβ and –TGFβ (control) conditions within each group (DMSO and RO-3306). (H) Immunofluorescence of MCF7 cells transiently transfected with pEGFP-N1 or Twist1-GFP and treated with 10 μM RO3306 for 18 h. nucleus (DAPI). Scale bar ∼10 μm (I) Immunofluorescence and quantification of colocalized voxels and Mander’s coefficient for Lamin A/C and EZH2 in MCF10A cells ± TGF-β and 1 μM MG132. Nucleus (DAPI). Scale bar ∼10 μm. Statistical significance was determined using unpaired Student’s t-tests. Means are compared between +TGFβ and –TGFβ (control) conditions within each treatment group (DMSO and RO-3306). (J) Immunofluorescence of MCF7 cells treated with MG132 and transient overexpression of hTWIST1-GFP and stained for E-cadherin and Vimentin ; nucleus (DAPI). Scale bar ∼10 μm. (K) Immunoblotting for EMT markers in MCF7 treated with RO3306 and Twist1-GFP in MCF7 (L) Immunoblotting for EMT markers in MCF7 cells treated with MG132 and Twist1-GFP in MCF7. For all experiments, data are represented as mean ± SD from N = 3 three independent biological replicates, statistical significance, P-value <0.05.

Putative CDK1 phosphorylation sites were identified to be the region of interaction (through deletion experiments) between Lamin A/C and EZH2 (Supplementary Fig. S8B), suggesting CDK1 regulation. IP-MS also showed CDK1 enrichment with Lamin A/C in Twist1-overexpressing MCF7 and untreated MDAMB231 cells. PLA demonstrated that EMT induction in MCF10A cells enhanced the interaction of Lamin A/C with phosphorylated CDK1 (pCDK1-T161; Fig. 5C and F), which was further corroborated by Co-IP. In contrast, induction of MET in MDAMB231 cells abolished this interaction (Supplementary Fig. S8C–E).

Since CDK1 phosphorylates Lamin A/C and EZH2 independently, immunoblotting revealed increased ratios of pLMNA(S22)/LMNA and pEZH2(T345)/EZH2 along with elevated pCDK1-T161 levels during EMT, whereas MET induction markedly decreased these phosphorylation events (Supplementary Fig. S8F–I) [54, 55]. To address the effect of altered CDK1 activity during EMT/MET on cell cycle, we performed FACS profiling, which showed an increase in cell numbers in G2/M upon EMT (in MCF7 and MCF10A) and decreased cell numbers in G2/M upon MET (in MDAMB231; Supplementary Fig. S8J and K). We surmised that increased CDK1 activity could destabilize Lamin A/C–EZH2 binding during EMT. To validate this finding, we performed PLA, which showed a loss of Lamin A/C–EZH2 interaction upon EMT induction, which was restored upon CDK1 inhibition with RO-3306 (∼10-fold increase in PLA signal; Fig. 5D and G).

MCF10A cells treated with TGF-β showed an increase in mesenchymal markers (ZEB1, SNAI1, TWIST1, Vimentin) and reduction in E-cadherin. To determine if CDK1 inhibition alters the dynamics of EMT markers, EMT-induced cells were treated with the CDK1 inhibitor (RO-3306). This showed the attenuation of mesenchymal markers, while E-cadherin expression was sustained (Supplementary Fig. S9A and C–E). A similar effect was observed in Twist1-overexpressing MCF7 cells, where CDK1 inhibition suppressed EMT induction (Fig. 5H and K, and Supplementary Fig. S9I).

Since CDK1 phosphorylates EZH2 at Thr345, leading to its proteasomal degradation, we asked if phosphorylation contributes to the dissociation of EZH2 from Lamin A/C. Upon EMT induction, Lamin A/C levels were markedly reduced, resulting in the loss of colocalization with EZH2. However, this interaction was restored when cells were treated with the proteasomal inhibitor MG132 (Fig. 5I). Interestingly, MG132 treatment also attenuated EMT-associated phenotypic changes (Fig. 5J and L, and Supplementary Fig. S9B, F–H, and J).

To further assess the consequence of increased CDK1 activity in regulating cell cycle and EM status of cells, we synchronized MCF10A cells by double-thymidine block followed by release and treated cells with TGF-β. While Cyclin D1 levels remained unaltered, Cyclin B1 was upregulated at both mRNA (∼25-fold) and protein (∼15-fold) levels in mesenchymal cells (Supplementary Fig. S10A and B). Flow cytometry showed a ∼3-fold increase in the G2/M sub-population (Supplementary Fig. S10C and D). BrdU incorporation assays showed no changes in the intensity of BrdU, negating a G1 arrest (Supplementary Fig. S10E).

To determine the state of cells at G2/M, we examined the levels of H3S10p - a marker of late mitosis [56]. Immunofluorescence showed a marked reduction of H3S10p-positive cells in TGF-β-treated cells (∼0.67 ± 0.08%; control: ∼19 ± 7.8%; Supplementary Fig. S10F). Consistently, immunoblotting of synchronized, and FACS-sorted G2/M cells revealed reduced H3S10p in EMT-induced cells (Supplementary Fig. S10G). These results collectively indicate that the cells undergoing EMT are arrested at the late S-to-early G2 transition.

We next asked whether CDK1 activity per se, rather than cell cycle arrest, is essential for EMT. In Twist1-overexpressing MCF7 cells, nocodazole-induced prometaphase arrest did not impair EMT, whereas CDK1 inhibition with RO-3306 markedly suppressed EMT (Supplementary Fig. S11A). Similarly, TGF-β did not induce EMT in thymidine-arrested MCF10A cells, but ectopic overexpression of a phosphomimetic CDK1 mutant (T161D) restored EMT progression (Supplementary Fig. S11B).

To directly test whether CDK1 activity regulates Lamin A/C–EZH2 stability, we performed Co-IP assays in thymidine-arrested MCF10A cells expressing CDK1 phosphomutants. The Lamin A/C–EZH2 interaction was disrupted upon expression of the phosphomimetic mutant (T161D; Supplementary Fig. S11C).

In summary, these findings suggest that EMT progression is driven by elevated CDK1 activity leading to late S-to-early G2 arrest. CDK1 phosphorylates Lamin A/C and EZH2, disrupts the Lamin A/C–EZH2 complex and promotes EZH2 degradation, thereby facilitating EMT [57–59].

Regulatory role of phosphorylation in the interaction of Lamin A/C–EZH2

Phosphorylation of nuclear lamins regulate nuclear organization and chromatin dynamics [60, 61]. Since Lamin A/C and EZH2 are involved in chromatin organization and transcriptional regulation, we surmised that perturbing their phosphorylation sites could modulate EMT/MET. Furthermore, CDK1-mediated phosphorylation impacts lamin organization, prompting an in-depth analysis of how CDK1 phosphorylation affects Lamin A/C–EZH2 binding. Specifically, Serine 22 (S22) in Lamin A/C and Threonine 345 (T345) in EZH2—both established CDK1 phosphorylation sites with roles in protein–protein interactions and nuclear dynamics—were selected for site-directed mutagenesis to mimic either a phosphorylated (phosphomimetic: Lamin A–S22D, EZH2–T345D) or nonphosphorylated (phospho-deficient: Lamin A–S22A, EZH2–T345A) state [62].

Lamin A/C knockdown lines (doxycycline-inducible) of MCF7 cells were generated, followed by ectopic overexpression of (i) GFP–Lamin A (shRNA resistant), (ii) S22A, and (iii) S22D constructs. To ensure homogeneity, GFP-positive cells were FACS-sorted. Additionally, TWIST1-GFP was transiently overexpressed (∼48 h) and selected with G418 to enrich TWIST1-expressing cells [(Fig. 6A), schematic for generation of EZH2 mutants (Fig. 6B)]. Immunoprecipitation of Lamin A/C followed by immunoblotting for EZH2 showed that full-length Lamin A and S22A retained their interaction with EZH2, independent of TWIST1-induced EMT, whereas the S22D variant did not bind to EZH2, irrespective of EMT status. These results indicate that phosphorylation at S22 disrupts Lamin A/C–EZH2 interaction (Fig. 6C).

Figure 6.

Phosphorylation-dependent regulation of Lamin A/C–EZH2 binding in EMT and MET. (A, B) Schematic representation of the workflow for generating stable cell lines with inducible knockdown of Lamin A (A) or EZH2 (B), followed by rescue with full-length, phosphodeficient, or phosphomimetic mutants. (C, D) Co-IP of Lamin A in MCF7 and MDAMB231 cells after doxycycline-induced Lamin A/C depletion and rescue with full-length, phosphodeficient (S22A), or phosphomimetic (S22D) Lamin A. TWIST1-GFP was transiently overexpressed in MCF7 cells, and GRHL2-GFP was stably overexpressed in MDAMB231 cells. (E, F) Coimmunoprecipitation of EZH2 in MCF7 and MDAMB231 cells after doxycycline-induced EZH2 depletion and rescue with full-length, phosphodeficient (T345A), or phosphomimetic (T345D) EZH2. TWIST1-GFP was transiently overexpressed in MCF7 cells, and GRHL2-GFP was stably overexpressed in MDAMB231 cells. (G) Immunofluorescence images of MCF10A cells showing the extent of colocalization between Lamin A [full-length, phosphodeficient (S22A), or phosphomimetic (S22D)] and EZH2 ± TGF-β. Nucleus (DAPI). Scale bar ∼10 μm. (H) Immunofluorescence images of MCF10A cells showing the extent of colocalization between EZH2 [full-length, phosphodeficient (T345A), or phosphomimetic (T345D)] nucleus (DAPI). Scale bar ∼10 μm. (I, J) Quantification of Lamin A and EZH2 colocalization in MCF10A cells using Mander’s coefficient. Unpaired Student’s t-test was used to compute the P-value. Means are compared between (I) LMNA-GFP (UT; control) versus LMNA-S22D (UT) and LMNA-GFP (TGFβ; control) versus LMNA-S22D (TGFβ). (J) EZH2-FLAG (UT; control) versus EZH2-T345D (UT) and EZH2-FLAG (TGFβ; control) versus EZH2-T345D (TGFβ). Statistical significance, P-value <0.05.

We adopted a similar strategy in MDAMB231 cells stably expressing GRHL2 to induce MET. Consistent with findings in MCF7, full-length Lamin A and S22A interacted with EZH2, while the S22D variant did not (Fig. 6D). Thus, phosphorylation at Ser22 consistently inhibits Lamin A/C–EZH2 interaction during both EMT and MET.

To further delineate the role of EZH2 phosphorylation, T345A and T345D mutants were tested in MCF7 (±TWIST1) and MDAMB231 (±GRHL2) cells. In both contexts, full-length EZH2 and T345A retained their interaction with Lamin A/C, whereas the T345D variant failed to bind Lamin A/C (Fig. 6E and F).

Taken together, these data demonstrate that phosphorylation at either S22 in Lamin A/C or T345 in EZH2 prevents complex formation. We further validated the interaction between the mutants of Lamin A/C and EZH2 through immunofluorescence studies and found that the full length (control) and phosphodeficient mutants interact with each other (Lamin A/C and EZH2), while phosphomimetic mutants lost the interaction irrespective of the presence of TGF-β (Fig. 6G–J).

To determine if there is a requirement for dual phosphorylation of Lamin A and EZH2, we co-transfected HEK293T cells with combinations of Lamin A–GFP and EZH2–FLAG (wild-type, phospho-deficient, or phospho-mimetic). Co-IP revealed robust binding only when both proteins are in their unphosphorylated or phospho-deficient forms. Notably, phosphorylation of either Lamin A or EZH2 alone was sufficient to weaken or abolish this interaction. Lamin A/C–EZH2 complex is sensitive to phosphorylation of either partner. Therefore, the disruption of this interaction does not require both interacting partners to be phosphorylated (Supplementary Fig. S12A). Importantly, the localization of the Lamin A and EZH2 mutants was validated by immunofluorescence assays, which predominantly showed a nuclear localization (Supplementary Fig. S12B and C).

Since phosphorylation impacts cell cycle dynamics, we next examined how phosphomutants of Lamin A/C and EZH2 affect cell cycle progression. MCF10A cells expressing Lamin A or EZH2 mutants were synchronized by double-thymidine block and released (for ∼18 h), followed by FACS profiling. Notably, phosphomimetic mutants (∼25%–30%) showed an increased number of cells in G2/M compared with full-length and phospho-deficient forms (∼6%–9%; Supplementary Fig. S13A and B). To further examine the effect of phosphomutants of Lamin A and EZH2 on cell cycle, we performed PI-FACS, BrdU incorporation, and H3S10p staining. BrdU incorporation was comparable across full-length, phospho-deficient, and phospho-mimetic constructs (∼85% BrdU positive cells), which excluded G1 arrest (Supplementary Fig. S13C and D). However, H3S10p staining revealed that full-length and phospho-deficient mutants progressed normally through mitosis (∼10% H3S10p-positive cells), whereas phospho-mimetic mutants (S22D, T345D) stalled at G2/M as indicated by PI-FACS, but displayed markedly reduced H3S10p positivity (<2%), suggesting an arrest in late S-to early G2 phase of the cell cycle (Supplementary Fig. S13E and F).

Lamin A and EZH2 phosphorylation determines the dynamics of EMT

We next sought to determine the effect of Lamin A-EZH2 phosphorylation on EM plasticity. To address this question, we generated cell lines expressing either phosphodeficient or phosphomimetic mutants of Lamin A (S22A or S22D) and EZH2 (T345A or T345D) in the background of endogenous protein depletion.

Remarkably, the overexpression of the phosphomimetic Lamin A mutant (S22D) disrupts the Lamin A–EZH2 interaction, showing a pronounced mesenchymal phenotype, characterized by the upregulation of Vimentin, TWIST1, SNAI1, and ZEB1, and downregulation of E-cadherin, notwithstanding the absence of EMT inducers. In contrast, cells expressing either full-length Lamin A or the phosphodeficient mutant (S22A) retain their interaction with EZH2 and their epithelial features, even under EMT-inducing conditions such as TGF-β or TWIST1 overexpression (Fig. 7A, E, and F; quantification in Supplementary Fig. S14A–E).

Figure 7.

Phosphorylation-dependent regulation of EZH2 and Lamin A/C during EMT and MET. Representative mid-optical sections of immunofluorescence images showing the effect of Lamin A and EZH2 mutants on EMT (in MCF10A) and MET (in MDAMB231). MCF10A (A, C) and MDAMB231 (B, D). Cells were transduced with full-length, phospho-deficient, or phospho-mimetic constructs of Lamin A (A, B) or EZH2 (C, D) following doxycycline-induced knockdown (0.5 μg/ml, 48 h) of endogenous Lamin A/C or EZH2. EMT was induced in MCF10A cells by TGF-β treatment (10 ng/ml, ∼7 days), while MET was induced in MDAMB231 cells by stable, constitutive overexpression of GRHL2. EZH2 was immunostained in green; E-cadherin and Vimentin were immunostained in red. Lamin A constructs were GFP-tagged. Nucleus (blue, DAPI). Scale bar ∼10 μm. (E, F) Immunoblot analysis of EMT marker expression in MCF7 and MCF10A cells upon Lamin A/C knockdown and rescue with full-length, phospho-deficient (S22A), or phospho-mimetic (S22D) Lamin A-GFP. EMT was induced by TWIST1 overexpression (∼48 h) in MCF7 cells or by 10ng/ml TGF-β (∼7 days) in MCF10A cells. (G) EM marker expression in cells with Lamin A/C knockdown was rescued with full-length, phospho-deficient (S22A) or phospho-mimetic (S22D) Lamin A/C. MET was induced by GRHL2 overexpression. (H, I) Immunoblot analysis of EMT marker expression in MCF7 and MCF10A cells upon EZH2 knockdown and rescue with full-length, phospho-deficient (T345A), or phospho-mimetic (T345D) EZH2-FLAG. EMT was induced by TGF-β (∼7 days) in MCF10A cells or by TWIST1 overexpression (∼48 h) in MCF7 cells. (J) EM marker expression in cells with EZH2 knockdown rescued with full-length, phospho-deficient (T345A), or phospho-mimetic (T345D) EZH2. MET was induced by GRHL2 overexpression.

Similarly, overexpression of full length or the phosphodeficient Lamin A (S22A) mutant (both capable of binding EZH2), was sufficient to induce a robust epithelial state, marked by high E-cadherin levels and reduced mesenchymal markers, even without GRHL2-mediated MET induction. In contrast, knockdown of Lamin A/C or overexpression of Lamin A (S22D)—which does not interact with EZH2—resulted in the persistence of mesenchymal characteristics (Fig. 7B and G). Notably, MDAMB231 cells exhibited partial MET upon co-overexpression of GRHL2 and Lamin A (S22D), or GRHL2 overexpression combined with Lamin A/C depletion (Supplementary Fig. S14F–J). This reinforces the significance of the Lamin A–EZH2 interaction in facilitating MET and repressing mesenchymal genes.

To delineate the role of EZH2 phosphorylation in this regulatory axis, we introduced phosphodeficient (T345A) and phosphomimetic (T345D) EZH2 mutants in the background of EZH2 depletion in epithelial (MCF7, MCF10A) and mesenchymal (MDAMB231) cells. In MCF7 and MCF10A cells, overexpression of the phosphomimetic EZH2 mutant (T345D)—deficient in Lamin A/C binding—shows an EMT-like phenotype, including loss of E-cadherin and induction of mesenchymal markers, independent of external EMT induction. Conversely, overexpression of full length EZH2 or the phosphodeficient EZH2 mutant (T345A), which retains its interaction with Lamin A/C, maintained epithelial properties and resists EMT induction, even in the presence of TWIST1 or TGF-β (Fig. 7C, H, and I; quantification: Supplementary Fig. S14K–N).

Overexpression of full length or phosphodeficient EZH2 (T345A) promoted MET in MDAMB231 cells, consistent with elevated E-cadherin and suppression of mesenchymal markers, independent of GRHL2 overexpression. However, overexpression of the phosphomimetic EZH2 (T345D) or EZH2 knockdown (both of which abrogate Lamin A/C interaction), exacerbated mesenchymal properties. Overexpression of EZH2 (full length) or EZH2 (T345A) in the background of GRHL2 overexpression, facilitates transition of MDAMB231 cells to an epithelial-like phenotype [EZH2(OE) and EZH2(T345A)—E-cadherin (integrated density): 2108 ± 516 and 3560 ± 532, respectively], whereas EZH2 (T345D) or EZH2 depletion only partially reversed cells to an epithelial phenotype [EZH2(T345D)—E-cadherin (integrated density): 320 ± 58] as it maintains mesenchymal marker expression and a hybrid phenotype [Fig. 7D and J; quantification: Supplementary Fig. S14O–R).

Collectively, these data demonstrate that phosphorylation at Lamin A (S22) or EZH2 (T345) disrupts Lamin A/C–EZH2 interaction, activating mesenchymal transcription factors, thereby promoting EMT. In contrast, phosphodeficient mutants that preserve this interaction sustain epithelial identity and enable MET.

Lamin A/C promotes chromatin compaction suppressing EMT

To investigate the functional consequences of Lamin A/C–EZH2 interaction, we examined the effects of CDK1-mediated phosphorylation of Lamin A/C or EZH2 on the expression of mesenchymal transcription factors SNAI1, TWIST1, and ZEB1. RNA-Seq following Lamin A/C depletion in MCF10A cells showed a significant upregulation of these transcription factors, indicating a role of Lamin A/C in their transcriptional repression. This indicates that Lamin A/C and repressive chromatin factors work in coordination, which may arise from spatial–epigenetic coupling, whereby Lamin A/C anchors PRC2-enriched chromatin at the nuclear periphery, promoting robust H3K27me3 deposition and enforcing transcriptional repression at key EMT-regulatory loci.

Analysis of publicly available ChIP-seq data sets from ChIP-Atlas revealed an enrichment of Lamin A, EZH2, and H3K27me3 at the −1 kb regions of the SNAI1, TWIST1, and ZEB1 promoters in epithelial MCF7 cells, and a loss of this occupancy in mesenchymal MDAMB231 cells (Lamin A/C: Supplementary Fig. S15A–C; EZH2: Supplementary Fig. S16A–C, H3K27me3: Supplementary Fig. S17A) [63]. We overexpressed phosphodeficient or phosphomimetic mutants of Lamin A or EZH2 in MCF7 cells and performed ChIP-qPCR at the −1 kb regions of SNAI1, TWIST1, and ZEB1 (Fig. 8A and B). Twist mediated EMT induction showed a marked reduction of Lamin A/C, EZH2, and H3K27me3 occupancy on SNAI1, TWIST1, and ZEB1 promoters, but were otherwise enriched on these promoters in untreated MCF7 cells (control). Lamin A/C depletion markedly reduced EZH2 and H3K27me3 occupancy, while control regions (MYT1, positive control; GAPDH, negative control) showed no differences in their occupancy (Fig. 8C and D, and Supplementary Fig. S17B).

Figure 8.

Lamin A/C promotes chromatin compaction suppressing EMT. (A) Schematic representation of chromatin immunoprecipitation (ChIP) experiments performed in MCF7 cells overexpressing TWIST1-GFP, following knockdown (KD) of Lamin A/C or overexpression (OE) of full-length Lamin A or phospho-mutant Lamin A (S22A: phospho-deficient; S22D: phospho-mimetic). ChIP was performed using antibodies against Lamin A/C, EZH2, H3K27me3 [−1 kb of transcription start sites (TSS)], and RNA Polymerase II (Ser2P) and H3K36me3 (+1 kb of TSS) at the promoters of key mesenchymal transcription factors SNAI1, TWIST1, and ZEB1. (B) Schematic representation of ChIP analysis in MCF7 cells with EZH2 knockdown or overexpression of full-length or phospho-mutant EZH2 (T345A: phospho-deficient; T345D: phospho-mimetic) in the context of TWIST1-GFP induction. Promoter occupancy of Lamin A/C, EZH2, and associated histone modifications (H3K27me3 and H3K36me3), as well as RNA Pol II (Ser2P), was examined at the TSS of SNAI1, TWIST1, and ZEB1. Green and red arrows indicate primers for the qPCR experiment. (C–E) ChIP PCR analysis in MCF7 cells stably transduced with full-length Lamin A, phosphodeficient (S22A), or phosphomimetic (S22D) variants, followed by transient transfection with pEGFP-N1 (control) or TWIST1-GFP for 48 h. ChIP occupancy of (C) Lamin A, (D) EZH2 at the −1 kb regions, and (E) H3K36me3 at the +1 kb regions of SNAI1, TWIST1, and ZEB1 were assessed. DEFA3 (C), MYT1 (D), and GAPDH (E) served as positive controls, while Jun (C), GAPDH +1 kb (D), and MYT1 (E) were negative controls for Lamin A/C, EZH2, and H3K36me3 respectively. (F–H) ChIP PCR analysis in MCF7 cells stably transduced with full-length EZH2, phosphodeficient (T345A), or phosphomimetic (T345D) variants, followed by transient transfection with pEGFP-N1 or TWIST1-GFP for 48 h. ChIP occupancy of (F) Lamin A, (G) EZH2 at the −1 kb regions, and (H) H3K36me3 at the +1 kb regions of SNAI1, TWIST1, and ZEB1 was assessed. DEFA3 (F), MYT1 (G), and GAPDH (H) served as positive controls, while Jun (F), GAPDH +1 kb (G), and MYT1 (H) were negative controls for Lamin A/C, EZH2, and H3K36me3 respectively. Data represent mean ± SD from three independent experiments.

ChIP-qPCR for transcriptional elongation markers revealed reduced occupancy of H3K36me3 and RNA polymerase II Ser2P at the +1 kb regions of SNAI1, TWIST1, and ZEB1 in control MCF7 cells, consistent with a reduced expression of mesenchymal genes in epithelial cells. However, Lamin A/C depletion and TWIST1 overexpression significantly increased the enrichment of H3K36me3 and RNA Pol II Ser2P on the promoters of mesenchymal transcription factors (Fig. 8E and Supplementary Fig. S17C and D).

We next analyzed the effect of Lamin A phosphorylation. Overexpression of full-length Lamin A or the phosphodeficient mutant S22A preserved Lamin A/C, EZH2, and H3K27me3 occupancy at the −1 kb regions of SNAI1, TWIST1, and ZEB1, even under EMT-inducing conditions. In contrast, the phosphomimetic mutant S22D showed reduced enrichment of Lamin A (Fig. 8C), EZH2 (Fig. 8D), and H3K27me3 (Supplementary Fig. S17B) at these loci, accompanied by increased H3K36me3 (Fig. 8E) and RNA Pol II Ser2P (Supplementary Fig. S17D) at the +1 kb regions of SNAI1, TWIST1, and ZEB1. A similar regulatory mechanism was observed for EZH2 phosphorylation. Overexpression of full-length EZH2 or the phosphodeficient mutant T345A enhanced Lamin A/C, EZH2, and H3K27me3 occupancy at the −1 kb regions of SNAI1, TWIST1, and ZEB1. Conversely, the phosphomimetic mutant T345D reduced this enrichment, with a corresponding increase in H3K36me3 and RNA Pol II Ser2P occupancy at the +1 kb regions (Fig. 8F–H, and Supplementary Fig. S18A and B). This was corroborated by analysis of primer efficiency profiles (Supplementary Fig. S18C–F).

Taken together, these findings demonstrate that Lamin A/C facilitates EZH2 mediated deposition of H3K27me3 on mesenchymal gene promoters. Furthermore, CDK1-mediated phosphorylation of either protein disrupts this repression, thereby facilitating EMT.

Impact of Lamin A/C–EZH2 phosphorylation on cancer progression

To investigate the functional role of Lamin A/C–EZH2 interaction in EM plasticity, we performed cell migration, and tumor induction assays. We performed Boyden-chamber assay (transwell to quantify migration) for MCF7 and MDAMB231 cells. In MCF7 cells, transient overexpression of hTwist1 followed by depletion of Lamin A/C or EZH2 increased migration by ∼30%. Similarly, overexpression of the phosphomimetic mutants Lamin A (S22D) or EZH2 (T345D) enhanced migration, whereas overexpression of the full-length proteins or their phosphodeficient mutants (Lamin A S22A, EZH2 T345A) reduced migration by ∼20% (Supplementary Fig. S19A and B). In MDAMB231 cells, GRHL2 overexpression reduced migration, while depletion of Lamin A or EZH2 increased migration. When GRHL2 was overexpressed in cells lacking Lamin A or EZH2, migration was further elevated compared to GRHL2 overexpression alone. Consistently, the phosphomimetic Lamin A (S22D) and EZH2 (T345D) variants enhanced migration than their full-length or phosphodeficient counterparts (Supplementary Fig. S19C and D). We performed wound healing assays to determine the migratory ability of MCF10A cells. Depletion of Lamin A/C or EZH2 increased cell migration, comparable to that induced by TGF-β. While, overexpression of full-length Lamin A or EZH2, as well as their phosphodeficient mutants (S22A, T345A), suppressed migration in both untreated and TGF-β-treated cells. In contrast, the phosphomimetic mutants (S22D, T345D) enhanced migration, even in the absence of TGF-β (Supplementary Fig. S20A and B).

To investigate the in vivo implications of Lamin A (S22A and S22D) and EZH2 (T345A and T345D) mutants to form tumors, we orthotopically injected MDA-MB-231 cells (∼1 × 10⁶), overexpressing (i) full length (Lamin A or EZH2 – control), (ii) phospho-deficient (Lamin A S22A; EZH2 T345A) (iii) phospho-mimetic (Lamin A S22D; EZH2 T345D) forms into the mammary fat pad of female NOD-SCID mice (∼6 weeks old). Tumour volume was monitored, and the experiment was terminated after ∼5 weeks (Validation of Lamin A and EZH2 knockdown and overexpression are shown in Supplementary Fig. S21A and B). Both phospho-mimetic Lamin A (S22D) and EZH2 (T345D) exhibited a marked increase in tumor growth, as reflected by both tumor volume and final tumor weight (Fig. 9A–F), as compared to their full-length and phospho-deficient counterparts (full length control in Supplementary Fig. S21C–H).

Figure 9.

Impact of Lamin A/C–EZH2 phosphorylation on tumor progression. (A, B) Representative images of excised tumors from NOD-SCID female mice injected with MDAMB231 cells expressing either phospho-mutant Lamin A (A) or EZH2 (B). N = 5 mice per group. (C, D) Quantification of tumor volume (C) and tumor weight (D) from Lamin A phospho-mutant xenografts. (E, F) Quantification of tumor volume (E) and tumor weight (F) from EZH2 phosphomutant xenografts. (G) Representative Hematoxylin and Eosin (H&E) images (2×, 50×) of lung metastases from MDA-MB-231 cells expressing phosphodeficient (S22A) or phosphomimetic (S22D) Lamin A. (H) Representative H&E images (2×, 50×) of lung metastases from MDA-MB-231 cells expressing phosphodeficient (T345A) or phosphomimetic (T345D) EZH2. Quantification of the number of nodules per lung is given alongside the H&E images. (I) Immunoprecipitation of Lamin A/C from tumor lysates followed by immunoblotting for EZH2 and phospho-CDK1 (J) Immunoprecipitation of EZH2 from tumor lysates followed by immunoblotting for Lamin A/C and phospho-CDK1 (K) Immunofluorescence of tumor-derived MDAMB231 cells stained for E-cadherin, Vimentin (red), and Lamin A (GFP). Nucleus (blue, DAPI). Scale bar ∼10 μm. (L) Quantification of the integrated density of E-Cadherin and Vimentin upon overexpression of phosphomutants of Lamin A. Data represent mean ± SD. Statistical significance was determined using an unpaired two-tailed Student’s t-test, P-value <0.05 (M) Immunofluorescence of tumor-derived MDAMB231 cells stained for E-cadherin, Vimentin and EZH2. Nucleus (DAPI). Scale bar ∼10 μm. (N) Quantification of the integrated density of E-Cadherin and Vimentin upon overexpression of phosphomutants of EZH2. Data represent mean ± SD. Statistical significance was determined using an unpaired two-tailed Student’s t-test. Means are compared between phosphodeficient (control) and phosphomimetic mutants.

Previous studies have consistently shown that the lungs are a frequent site of metastasis in breast cancers. We therefore performed H&E staining of lung sections and quantified metastatic nodules to determine whether Lamin A and EZH2 phosphomutants (with respect to the full-length control) affect metastatic dissemination [64, 65]. Mice injected with phospho-mimetic mutants (S22D-Lamin A and T345D-EZH2) showed a significantly higher metastatic burden, whereas full length (LMNA-WT and EZH2-WT) and phospho-deficient mutants (S22A-Lamin A and T345A-EZH2) did not form detectable lung metastases. These findings demonstrate that CDK1-dependent phosphorylation of Lamin A/C and EZH2 promotes both primary tumor growth and in vivo metastatic dissemination of breast cancer cells (Fig. 9G and H; full length control in Supplementary Fig. S21I–L).

To further address the significance of the Lamin A/C-EZH2 interaction, we performed triple enzyme digestion followed by immunoprecipitation on primary tumor-derived cells. Phosphodeficient Lamin A (S22A) maintained its interaction with EZH2, whereas the phosphomimetic Lamin A (S22D) preferentially interacted with phosphorylated CDK1 (pCDK1-T161; Fig. 9I). Similarly, immunoprecipitation of EZH2 revealed that the phosphodeficient EZH2 (T345A) interacted with Lamin A/C, while phosphomimetic EZH2 (T345D) interacted with pCDK1 (Fig. 9J).

We performed immunofluorescence assays in cells (cultured for ∼72 h) derived from the tumors (the EM state could not be examined in situ, as the subsequent assays necessitated culturing and expansion of cells). These assays also reveal morphological characteristics since the cells expressing phosphodeficient Lamin A or EZH2 mutants showed an epithelial-like morphology, whereas those expressing phosphomimetic mutants retained a mesenchymal phenotype (Fig. 9K–N).

In summary, these findings underscore a crucial role for Lamin A and EZH2 phosphorylation in modulating EM plasticity. Phosphomimetic modifications favour mesenchymal properties, whereas phosphodeficient or full-length variants reinforce epithelial characteristics.

Discussion

Summary of key results

In this study, we uncover a phosphorylation-sensitive interaction between the inner nuclear envelope protein Lamin A/C and the histone methyltransferase EZH2, revealing its pivotal role in regulating EM plasticity. Using a combination of models of EMT and its reversal (MET), we show that Lamin A/C interacts with EZH2 in epithelial cells, which renders a repressive chromatin environment on SNAI1, TWIST1, and ZEB1 loci that encode for mesenchymal transcription factors (EMT-TF). To ensure comparable expression of empty vector and TWIST1 (in MCF7) or GRHL2 overexpression (in MDAMB231 cells; EGFP panel is included in Supplementary Fig. S25A–D). This interaction is disrupted upon EMT induction through CDK1-mediated phosphorylation of both Lamin A/C and EZH2, facilitating EMT-TF activation (Fig. 10). Phosphomimetic and phosphodeficient variants of Lamin A (S22D/S22A) and EZH2 (T345D/T345A) reveal their regulatory role of phosphorylation in controlling EMT. In vivo, phosphomimetic forms promote tumor growth and metastasis, whereas phosphodeficient forms maintain epithelial characteristics, decreasing tumor growth potential. These findings implicate Lamin A–EZH2 interaction as a critical epigenetic checkpoint modulated by phosphorylation in cancer progression.

Figure 10.

Lamin A/C-EZH2 interaction modulates EM plasticity. This schematic illustrates the dynamic regulation of EMT and MET by Lamin A/C–EZH2 signaling. In epithelial cells (top), high Lamin A/C levels promote the sequestration of EZH2 at the nuclear lamina, diminishing the transcriptional activation and expression of key EMT transcription factors (SNAI1, TWIST1, ZEB1) by enforcing H3K27me3-mediated repression at their promoters. Reduced CDK1 activity leads to reduced phosphorylation on both Lamin A/C (Ser22) and EZH2 (Thr345), limiting EZH2 degradation and further stabilizing chromatin repression. During EMT, increased CDK1 activity and phosphorylation of Lamin A/C (Ser22) and EZH2 (Thr345), results in proteasomal degradation of EZH2 which releases the repressive chromatin state, recruits active RNA polymerase II to EMT gene promoters, and upregulates EMT transcription factors, promoting mesenchymal properties. Created in BioRender. Sengupta, K. (Schematic illustration of EMT and MET regulation by Lamin A/C-EZH2).

Lamins and EMT: contributions of A- and B-type lamins

Nuclear lamins regulate nuclear architecture with genome organization and transcriptional control, and their altered expression or modification profoundly impacts cancer cell behavior by modulating nuclear mechanics and chromatin accessibility [66].

Lamin A/C restricts EMT by reinforcing nuclear stiffness and limiting chromatin openness. Its reduction decreases nuclear rigidity, increases chromatin accessibility, and promotes invasion by enhancing mesenchymal gene expression [67, 68].

Lamin B1, in contrast, acts as a context-dependent EMT regulator. Its loss in lung cancer cells suppresses E-cadherin, elevates fibronectin and N-cadherin, and activates rearranged during transfection proto-oncogene and MAPK/PI3K–AKT signaling, suggesting an EMT-restraining role [69]. Lamin B1 enrichment at TAD boundaries supports EMT programs, and its depletion disrupts them in NMuMG cells [70]. However, in our study, Lamin B1 levels remained stable, and depletion of Lamin B1 or B2 had no effect on EMT, indicating cell-type-specific variability. While Lamin B2 contributes to nuclear integrity and metastasis in other cancers, it did not impact EMT in breast cancer [45, 71].

Overall, Lamin A/C consistently functions as a repressive scaffold maintaining epithelial identity, whereas B-type lamins show context-dependent roles. Loss or modification of Lamin A/C represents a critical step enabling EMT and invasiveness of cancer cells.

Coordination of Lamin A/C–PRC2 Interactions regulate EM Plasticity

Previous studies have reported divergent patterns of EZH2 expression during EMT, suggesting that its regulation may be context dependent [72]. While some reports did not observe a reduction of EZH2 in mesenchymal-like cells, our data reveal a modest but consistent decrease in EZH2 upon EMT induction and restoration during MET. This aligns with recent studies indicating that attenuation of PRC2/EZH2 function accompanies mesenchymal transition. Specifically, loss or dysfunction of PRC2 promotes a quasi-mesenchymal trajectory and enhance metastatic traits, implicating PRC2 as an organizer of the epithelial state [73]. In lung carcinoma cells, inhibition or knockout of EZH2 induces mesenchymal gene expression and a shift toward the mesenchymal phenotype [74], while in breast cancer models, EZH2 sustains epithelial identity by repressing mesenchymal programs [75]. Reduction in EZH2 levels, was undetected in other studies [76], underscoring that the relationship between EZH2 expression and EMT may vary across cellular contexts and experimental systems.

Since EZH2 levels are modulated during EMT, we next explored whether changes in the sub-interactome of Lamin A/C could explain these dynamics, focusing on Lamin A/C as a potential scaffold coordinating PRC2 activity and chromatin organization.

Lamin A/C–mediated chromatin reorganization during EMT and MET

Beyond its structural role, Lamin A/C serves as a dynamic scaffold that coordinates chromatin modifiers and transcriptional regulators to control EM plasticity. In epithelial MCF7 cells, Lamin A/C interacts with EZH2—the catalytic subunit of PRC2 responsible for H3K27 trimethylation—as well as other chromatin organizers such as NUDT21, LIMA1, and EHMT2 [67, 68]. During EMT, this interaction is lost, coinciding with Lamin A/C association with phosphorylated CDK1 and CCNB1, indicating a potential shift in the functional role of Lamin from chromatin organization to proliferative signaling [67].

A-type lamins associate with both euchromatin and heterochromatin, and loss of LAP2α redistributes Lamin A/C toward heterochromatin with altered histone modifications [77]. Similarly, reduced Lamin A/C decreases H3K27me3 levels and reorganizes, activating mesenchymal genes. Thus, Lamin A/C preserves a repressive chromatin environment, and its loss promotes chromatin relaxation and mesenchymal gene activation.

The Lamin A/C–EZH2 interaction is re-established upon MET, restoring PRC2 scaffolding and epithelial identity. [78]. Unlike A-type lamins, B-type lamins interact with G9a and SUV39H1, unaffected during EMT, emphasizing their distinct functional roles [45, 71].

Overall, Lamin A/C functions as a phosphorylation-dependent hub linking chromatin organization and cell cycle signaling. Its dynamic interaction with PRC2 and post-translationally modified EZH2 integrates nuclear architecture and epigenetic regulation to control EM plasticity, with MET-driven restoration of the Lamin A/C–EZH2 complex reinstating epithelial chromatin states.

pCDK1, Lamin A/C, and EMT–MET dynamics

During EMT, Lamin A/C associates with pCDK1 (T161), coinciding with loss of Lamin A/C–EZH2 interaction. BrdU incorporation in MCF10A cells undergoing EMT did not show G1 arrest, while decreased H3S10 phosphorylation indicated EMT-associated cell cycle arrest occurs between late S and early G2 phase.

Mechanistically, pCDK1 phosphorylates Lamin A/C and EZH2, disrupting their interaction, relieving PRC2-mediated H3K27me3 repression at mesenchymal loci, and activating EMT transcriptional programs. Upon MET, Lamin A/C–EZH2 interactions are restored, reinstating epithelial chromatin and transcriptional repression [67, 76].

These results are consistent with the finding that CDK1 phosphorylation modulates EMT, and frequently correlates with cell cycle stalling at late S/early G2 transition, coupling proliferative signalling with transcriptional reprogramming [57–59]. In line with recent studies highlighting the role of EMT-inducing transcription factors in late S/early G2-phase regulation, replication stress, and DNA damage responses [79–83], certain partial or metastatic EMT states are associated with transient G1 arrest or reduced proliferation, underscoring the temporal and contextual plasticity of EMT–cell cycle coupling [84, 85].

Collectively, our data suggest that cells undergoing EMT show cell cycle arrest at late S or early G2 phase, enabling pCDK1-dependent remodeling of Lamin A/C scaffolds, dissociation and degradation of EZH2, leading to activation of mesenchymal programs. Restoration of the Lamin A/C–EZH2 complex during MET underscores the reversible coupling between cell-cycle state, nuclear organization, and EM plasticity.

Decoding EMT plasticity through Lamin A and EZH2 phosphorylation

EMT and MET occur along a continuum, with hybrid epithelial/mesenchymal (E/M) states serving as key intermediates in tumor progression and metastasis. This hybrid population, marked by CD104 and CD44 co-expression [86].

During TWIST1-induced EMT, cells acquire a mesenchymal-like (CD104⁺CD44⁺) phenotype (Supplementary Fig. S22A and B), whereas overexpression of Lamin A or EZH2, with or without TWIST1, restores epithelial identity (CD104⁺CD44⁻), bypassing the hybrid state. In contrast, phospho-mimetic mutants (Lamin A S22D, EZH2 T345D) enrich hybrid populations in MCF7 cells, indicating that phosphorylation disrupts chromatin binding and repression of EMT-TFs such as SNAI1, TWIST1, and ZEB1. Under GRHL2-driven MET in MDA-MB-231 cells, these mutants maintain hybrid populations despite epithelial cues (GRHL2 overexpression).

Rescue experiments reveal that a complete commitment to an epithelial state requires unphosphorylated Lamin A and EZH2. Phospho-deficient mutants (S22A, T345A) restore epithelial identity (CD104⁺CD44⁻), while phospho-mimetic forms sustain hybrid states (Supplementary Figs S22C and D, and S23A–D), consistent with Lamin A phosphorylation promoting nuclear softening and chromatin decondensation [87].

Overall, the hybrid E/M state is a regulated phenotype governed by post-translational modifications of Lamin A and EZH2. Phosphorylation-dependent disruption of their interaction enhances EM plasticity and sustains hybrid populations linked to metastasis.

Implications for diagnosis and treatment

The Lamin A–EZH2 interaction and their phosphorylation states offer potential diagnostic and therapeutic avenues. Phosphorylated Lamin A (S22) and EZH2 (T345) may serve as markers of mesenchymal transition and metastatic potential, detectable by mass spectrometry or immunohistochemistry to refine EMT-based prognosis.

CPTAC proteomics showed increased pLMNA(Ser22) and decreased total LMNA with tumor progression (Supplementary Fig. S24A), paralleled by elevated pEZH2(T345) and reduced total EZH2 (Supplementary Fig. S24B). TCGA data further revealed progressive CCNB1 upregulation and increased pCDK1(T161) across tumor stages, linking CDK1–CCNB1 activity to malignancy (Supplementary Fig. S24C and D) [88, 89].

Therapeutically, inhibiting CDK1 or stabilizing the Lamin A–EZH2 complex could restore heterochromatin and suppress EMT-TFs, promoting a less aggressive phenotype. However, broad PRC2 inhibition may disrupt Lamin A–EZH2 tumor-suppressive functions, emphasizing patient-tailored approaches. The reversibility of this interaction highlights tumor plasticity and its potential as a target to drive cells toward a more differentiated, less malignant state.

Limitations and future perspectives

This study identifies a phosphorylation-dependent Lamin A–EZH2 mechanism regulating EMT, but several aspects remain unresolved. The universality of this pathway beyond breast and liver carcinoma (HepG2; Supplementary Fig. S24E) needs validation, as tissue-specific factors may influence its function.

Although in vivo data support a phosphorylation-dependent Lamin A–EZH2 interaction in primary tumors, its role during metastasis and under therapeutic stress remains unclear. Determining whether EMT involves cell-cycle arrest specifically in late S or early G2 phases will further refine our understanding of this mechanism. Future studies employing high-resolution cell-cycle mapping approaches, such as short BrdU/EdU pulse labeling, combined DNA-content analysis with replication and mitotic markers, or single-cell cell-cycle indexing will be essential to definitively distinguish late S from early G2 states and clarify their contribution to EMT regulation.

Additionally, the phosphatases responsible for Lamin A and EZH2 dephosphorylation are yet to be identified. These enzymes are likely to mediate complex disassembly and chromatin reorganization, influencing EMT reversibility. Elucidating these regulatory layers will be key to fully defining the Lamin A–EZH2–CDK1 axis and its therapeutic relevance in tumor progression and chemoresistance.

Supplementary data

Supplementary data is available at NAR online.

Conflict of interest

None declared.

Funding

We gratefully acknowledge funding support from ANRF, Department of Science and Technology, Science and Engineering Research Board (grant #CRG/2020/002563), Ministry of Education-STARS (grant #MoE-STARS/STARS-2/2023-0603), Department of Biotechnology (DBT) MedDevice (grant #BT/PR49745/MED/32/925/2023), Department of Biotechnology (BT/PR52973/MED/30/2535/2024), Intermediate Fellowship of the Wellcome Trust–DBT India Alliance (grant #500164/Z/09/Z) and Intramural funding from IISER-Pune.

Data availability

The RNA sequencing data generated in this study have been deposited in the Gene Expression Omnibus under accession number GSE289750.

The mass spectrometry proteomics data are available via the PRIDE repository with the dataset identifier PXD060831.