KMT1/Suv39 methyltransferase family regulates peripheral heterochromatin tethering via histone and non-histone protein methylations

Euchromatic histone methyltransferases (EHMTs), members of the KMT1 family, methylate histone and non-histone proteins. Here we uncover a novel role for EHMTs in regulating heterochromatin anchorage to the nuclear periphery (NP) via non-histone (LaminB1) methylations. We show that EHMTs methylates and stabilizes LaminB1 (LMNB1), which associates with the H3K9me2-marked peripheral heterochromatin. Loss of LMNB1 methylation or EHMTs abrogates the heterochromatin anchorage from the NP. We further demonstrate that the loss of EHMTs induces many hallmarks of aging including global reduction of H3K27methyl marks along with altered nuclear-morphology. Consistent with this, we observed a gradual depletion of EHMTs, which correlates with loss of methylated LMNB1 and peripheral heterochromatin in aging human fibroblasts. Restoration of EHMT expression reverts peripheral heterochromatin defect in aged cells. Collectively our work elucidates a new mechanism by which EHMTs regulate heterochromatin domain organization and reveals their impact on fundamental changes associated with the intrinsic aging process.


Introduction
The Euchromatic histone lysine methyltransferases G9a, encoded by EHMT2, and GLP, encoded by EHMT1 (KMT1/Suv39 methyltransferase family), are present as heteromeric complex and negatively regulate gene transcription. The SET domain of EHMT catalyzes mono and dimethylation of lysine residues at histone3 (H3) in vitro and in vivo [1][2][3][4][5]. H3K9me2 deposited by EHMT1/2 complex demarcates heterochromatin, particularly non-genic regions and is prevalent in gene deserts, pericentromeric and subtelomeric regions, with little being observed at individual active or silent genes. Non-coding and gene containing DNA present at the NP are also marked by the presence of H3K9me2 which spans several megabases [6,7].
Specifically, these domains are strongly correlated with binding of LMNB1 and are depleted of H3K4me3 and RNA Polymerase II activity [6]. These data suggest that H3K9me2 domains are critical determinants of higher-order chromosome structure in association with the nuclear lamina (NL).
In mammalian cells, the NL acts as a hub for multiple cellular functions including chromatin organization [8][9][10][11]. NL is composed of A and B type lamins along with inner nuclear membrane (INM) proteins [12], and together with mediator proteins such as Barrier-to-Autointegration Factor (BAF) and Heterochromatin Protein 1 (HP1), facilitate attachment of chromatin to NL [13,14]. Additionally, these interactions have been proposed to form specific chromatin organization that opposes transcriptional activity [15]. The association between LMNB receptor and LMNA/C mediates peripheral heterochromatin attachment in a wide variety of mammalian cells [16]. Any perturbation in such organization leads to a complete loss of peripheral heterochromatin and developmental abnormalities [17]. 4 Recent studies demonstrated that the Lamina associated domains (LADs) enriched in H3K9 methyl (me2/me3) marks contact NL via association with LMNB1 [7,[18][19][20].
These interactions are highly stochastic in nature and are dependent on H3K9me2 activity governed by G9a/EHMT2. Accordingly, G9a/EHMT2 promotes LAD formation and its loss leads to the opposite effect [18]. Similar to humans, H3K9 methylation is important for heterochromatin positioning in C. elegans [21], as depletion of H3K9 methyltransferases Met2 and Set-25 (mammalian SETDB1 and G9a /EHMT2 homolog) leads to detachment of large gene-array from peripheral heterochromatin. Altogether, loss of lamins and INM proteins or H3K9me2 activity leads to peripheral heterochromatin defects, however, whether there is a link between these common consequences remains unknown. Here we establish EHMT proteins as a common module that governs heterochromatin tethering via histone dependent (H3K9me2) and independent mechanisms (by directly regulating LMNB1 methylation).

EHMTs associates with LMNB1
To identify the novel non-histone interactors of EHMT proteins, endogenous EHMT1 was immunoprecipitated (IP) and unique bands in the EHMT1 pull down were subjected to LC/MS analysis. We found LMNB1 and histone proteins as interactors of EHMT1 ( Fig 1A). Mass spectrometry data was validated by sequential IP reactions of endogenous EHMT1 and LMNB1 proteins (Fig 1B). This interaction was also found using nuclear extracts from human dermal fibroblasts (HDFs) (Fetal derived unless mentioned otherwise), suggesting that it is not cell type specific interaction ( Fig   Supplementary Fig 1A). Consistent with previously published reports, we also detected HP1 in association with EHMT and LMNB1 (Supplementary Fig 1A). The absence of Ash2l (a member of H3K4 methyltransferase complex) in the IP-EHMT1 or IP-LMNB1 confirmed the specificity of IP reaction (Supplementary Fig 1A).
Mapping experiments (in vivo and in vitro) to identify the LMNB1 interacting domain of EHMT1, revealed that the SET domain of EHMT1 is sufficient to bind to LMNB1 ( Supplementary Fig 1B and C; Fig 1C). These results confirmed that the EHMT1/2 directly associates with LMNB1 via its SET domain.
Both EHMT and LMNB1 are known to interact with chromatin independently or via mediator proteins [17,22,23]. To identify EHMT1-LMNB1 co-bound regions in the genome, we performed ChIP-Seq analysis. Individually, EHMT1 and LMNB1 occupied 36807 and 32688 number of peaks respectively and 8584 peaks were cobound by EHMT1 and LMNB1 (Fig 1D). A majority of EHMT1 and LMNB1 reads were distributed on non-TSS regions such as introns and gene-poor regions (represented as "others") ( Fig 1E and F), whereas only 1.5% of reads were found on the upstream region of genes ( Fig 1F). Functional category analysis of the genes  Fig 1G). These data suggested that EHMT1-LMNB1 associates on gene-poor areas that are the critical determinants of higher-order chromosome structure at the NP.

EHMTs methylate LMNB1
Next, we tested if LMNB1 is a substrate for methylation by the EHMT enzymes.
Using an in vitro fluorometric methyltransferase assay we demonstrate an increase in fluorescence upon incubation of EHMT1-SET domain with GST-LMNB1 in presence of S-adenosyl methionine (SAM) ( Supplementary Fig 2A). To confirm that EHMT proteins indeed methylate LMNB1, we used lysine methyl-specific (Methyl-K) antibody to probe for methylated LMNB1. Purified LMNB1 C terminus protein containing the rod domain and tail domains (LMNB1-CT) ( Supplementary Fig 2B) was used in this assay. Towards this, we performed in vitro methyltransferase assay using different concentrations of LMNB1 and incubated with an equimolar ratio of 7 the EHMT1/2-SET domain in presence or absence of SAM. When products of these reactions were immunoblotted using the Methyl-K antibody, specific methylation signal was observed upon incubation of LMNB1 with EHMT1/2-SET in presence of SAM (Fig 2A and B).
Further, to examine if LMNB1 is methylated in vivo, we performed anti-Methyl-K or anti-LMNB1 IPs from HEK nuclear lysates. Products of IPs were split into two halves and probed with either anti-LMNB1 or anti-Methyl-K specific antibodies.
Several lysine residues are methylated on the H3 tail, detecting the histone signal in IP-Methyl-K confirmed the specificity of IP reaction and served as a positive control ( Fig 2C). The presence of a LMNB1 band in the same Methyl-K IP samples indicated the presence of endogenous methylated LMNB1 ( Fig 2C). In reciprocal IP reaction in which LMNB1 was IP'ed and probed with Methyl-K antibody, identification of Methyl-K signal in the IP-LMNB1 confirmed LMNB1 methylation in vivo (Fig 2D).
It has been reported that EHMT2 is capable of methylating lysine on dipeptide Arg-Lys (RK) sequence of non-histone proteins [24]. We synthesized peptides for such motifs present at the C-terminus of LMNB1 and identified K417 as the methylation site targeted by EHMT1 and EHMT2 ( Fig 2E). K417A peptide mutation abolished methylation of LMNB1 ( Fig 2E). To investigate the function of methylated LMNB1 in vivo, we mutated the 417K residue to alanine (K417A) in the wild-type (Wt.) LMNB1 construct. As opposed to Wt.LMNB1, which was localized at the NP, much of K417A-LMNB1 was accumulated in the nucleoplasm (Fig 2F and Supplementary   Fig 2C-E). We also observed aggregates of mutant LMNB1 transported into the cytoplasm, and was accompanied by abnormal nuclear morphology (Fig 2F and Supplementary Fig 2C, 2F). Co-staining with LMNB1 antibody showed localization of endogenous LMNB1 and overexpressed Wt.LMNB1 at the NP in Wt.LMNB1 8 expressing cells (Fig 2F). However, in mutant-LMNB1 expressing HDFs, endogenous LMNB1 was localized in K417A-LMNB1 aggregates indicating a dominant negative function of the mutant protein ( Fig 2F). Further mislocalization of LMNA/C in the aggregates of mutant-LMNB1 ( Supplementary Fig 2G) indicated LMNB1 methylation is critical for maintaining the NL meshwork composition at the periphery. Based on these data we speculated that the methylation modification prevents the degradation of LMNB1 and confers protein stability.

EHMTs regulate LMNB1 levels
To test the consequence of the loss of EHMTs on LMNB1 levels we used shRNAs and achieved approximately 60% and 70% depletion of EHMT1 and EHMT2, Recent studies demonstrate that the B-type lamins are long-lived proteins [25].
Therefore, loss of such proteins is a combination of transcriptional inhibition and protein degradation. Thus, we tested if EHMT proteins regulated LMNB1 expression transcriptionally. Our data demonstrate approximately 60% loss of LMNB1 transcript upon depletion of EHMT proteins (Fig 3D). Overall our results demonstrate that EHMT1 and 2 regulate LMNB1 expression transcriptionally and directly via posttranslational modification. 9 EHMT mediated histone and non-histone methylation influences peripheral heterochromatin organization H3K9 methylation and LMNB1 are the critical determinants for the formation of the LADs at the NP [7,[18][19][20]. To test the requirement of EHMT mediated H3K9 dimethylation in tethering peripheral heterochromatin, we depleted EHMTs and monitored the co-localization of H3K9me2 with LMNB1. Global H3K9me2 was decreased by 50% in shEHMT1 and 80% in shEHMT2 cells ( Supplementary Fig 4A   and Fig 4A).
Next, we performed EM to investigate the status of heterochromatin in EHMT depleted cells. shCnt transduced HDFs exhibited a layer of electron dense peripheral heterochromatin just beneath the nuclear envelope (NE) (Fig 4B). Knockdown of EHMT2 led to the partial disruption of heterochromatin from the periphery to the interior of the nucleus (Fig 4C). This result was correlated with redistribution of H3K9me2 marks towards the interior of the nucleus. In shEHMT1 HDFs, we noticed near complete detachment of peripheral heterochromatin and a distorted NE (Fig 4D).
We also detected floating islands of heterochromatin in the nuclei. Interestingly, the severity of heterochromatin detachment and compromised NE integrity were unique to shEHMT1 knockdown wherein H3K9me2 activity was modestly affected.
We also looked at the effects of pharmacological inhibition of H3K9me2 on overall heterochromatin positioning and nuclear distortion. HDFs treated with BIX 01294 showed 40% less H3K9me2 staining compared to controls ( Supplementary Fig 4F-H) 1 0 with no changes in the LMNB1 methylation levels ( Supplementary Fig 4H). Unlike EHMT depleted cells, we did not notice any significant changes in the nuclear morphology of BIX treated cells ( Supplementary Fig 4F and I). EM imaging indicated modest changes in heterochromatin anchorage ( Supplementary Fig 4I). Interestingly addition of BIX did not influence EHMT mediated methylation of LMNB1 in fluorometric methyltransferase assay using recombinant LMNB1 protein (Supplementary Fig 4J).
To obtain a clearer picture of the role of LMNB1 methylation, HDFs were    We next examined whether progressive loss of EHMTs and altered distribution of H3K9me2 impacts heterochromatin organization in aging cells. Compared to fetal cells, in 18Y and 31Y HDFs, heterochromatin was still retained at the periphery where there was a substantial loss of EHMT2 (Fig 5G and A). We also observed a redistribution of peripheral heterochromatin, which correlated with redistribution of 1 3 heterochromatin in 40Y old cells with complete depletion observed in 65Y aged nuclei ( Fig 5G). Interestingly complete loss of peripheral heterochromatin organization was correlated with loss of EHMT1 and EHMT2 in aged cells as shown in Fig 5A- Fig 7F), indicating that the blockade of UPS activity can restore the EHMT2 degradation only in early age group and such mechanisms do not operate in aged cells wherein EHMT2 is already drastically low.
Taken together our data on the loss of EHMT1 either as a consequence of physiological aging or by forced depletion in fetal HDFs implicates EHMTs in heterochromatin organization at the NP. 1 4 Low levels of LMNB1 have been observed in senescent cells and fibroblasts derived from Progeria patients [27]. In this study, expression analysis of nuclear lamins in intrinsically aged cells showed decline in LMNA/C from fetal to 18Y age and then the protein levels remained constant in further age groups ( Fig 5H). Contrary there was a dramatic reduction of LMNB1 starting in the 18Y age group with a significant loss at 65Y (Fig 5H and I). Diminishing levels of LMNB1 were correlated with the reduction in EHMT2 protein with drastic loss upon depletion of EHMT1 protein in 65Y cells (Fig 5A-D). This is consistent with the data in Figure 3 where we found that EHMT proteins directly regulate levels of the LMNB1 protein.
Next, we investigated whether diminishing perinuclear heterochromatin organization in aging nuclei is a result of the loss of EHMT1, EHMT2 and LMNB1 interaction.
Towards this, we performed IP experiments using Fetal, 18Y and 65Y HDF nuclear extracts. Our results revealed an association between EHMT2 and LMNB1 occurred in fetal cells and was highly diminished in adult and aged cells (Supplementary Fig   2H). On the contrary, EHMT1 associated with LMNB1 in all the age groups and the interaction was reduced gradually in age-dependent manner (Supplementary Fig 7I).
The complete absence of perinuclear heterochromatin in 65Y-aged nuclei corresponded to over 80% reduction in the interaction between EHMT1 and LMNB1.
These data indicated that EHMT1 and LMNB1 association is involved in maintaining peripheral heterochromatin in aging fibroblasts.
Further, we tested the status of LMNB1 methylation during the physiological aging and found the reduced intensity of a methylated LMNB1 signal from Fetal to 31Y with virtually no methylation in 65Y-aged cells (Fig 5J-L). Taken together our results indicate that the loss of peripheral heterochromatin in EHMT depleted cells or aged cells occur due to loss of H3K9 activity coupled with the loss of LMNB1, which are 1 5 critical determinants of peripheral heterochromatin anchorage.

Overexpression of EHMT proteins rescues peripheral heterochromatin defect in aged cells
To test if depletion of EHMT proteins is indeed responsible for the loss of peripheral heterochromatin in aged cells we transfected full-length V5 tagged EHMT1 or Flag-EHMT2 (set domain) plasmids in 65Y HDFs. Immunostaining using V5 or Flag antibodies confirmed the overexpression of EHMT1 and EHMT2 proteins ( Fig 6A).
EHMT1 levels enhanced the expression of LMNB1 but the Flag-SET of EHMT2 had no effect on LMNB1 levels ( Fig  We further examined H3K9me2 localization and organization of heterochromatin in EHMT1 and EHMT2 overexpressing cells. In both the cases H3K9me2 was colocalized with LMNB1 at the NP ( Fig 6B). Consistent with this, EM imaging revealed peripherally organized heterochromatin upon EHMT1 and EHMT2 overexpression, which was completely absent in untransfected aged cells (Fig 6C-E). These results demonstrate that the reduction of EHMT proteins contributes to the loss of peripheral heterochromatin organization during aging and this defect can be reversed upon reexpression of EHMT proteins.
To investigate the contribution of LMNB1 methylation in reversing peripheral heterochromatin tethering in aged cells we co-expressed wild-type or mutant-LMNB1 with V5-EHMT1 or Flag-EHMT2. As expected Wt.LMNB1 localized with H3K9me2 at the NP in EHMT1 overexpressing cells ( Supplementary Fig 8C, upper panel).
Instead, H3K9me2 showed aggregated staining in the nucleoplasm and did not 1 6 localize with mutant-LMNB1 ( Supplementary Fig 8D and E) thereby implicating a role for methylated LMNB1 in organizing heterochromatin to the NP.

Proliferation rate of HDFs correlate with EHMT expression
Previous reports have demonstrated that the reduction in LMNB1 expression results in reduced proliferation and induction of premature senescence [28,29]. Therefore, we  Table 2  The telomerase enzyme prevents the replicative senescence in primary fibroblast.
Examining the levels of telomerase activity in shEHMT and aged cells identified a correlation between the loss of EHMT proteins with reduced telomerase activity ( Fig   7F and G). Overall our results demonstrated that the depletion of EHMT1 and EHMT2 was correlated with reduced proliferation.

Discussion
The NL is a meshwork of lamins that constitutes the nucleoskeleton required for nuclear structure and function [8][9][10]31]. The NL undergoes extensive 1 7 posttranslational modifications (PTMs) that are crucial for their localization to regulate a variety of biological processes [32]. While uncovering the mechanism by which EHMTs organize heterochromatin, we have identified lysine methylation as a novel PTM on the nuclear localization signal (NLS) of LMNB1 that is critical for its retention at the NP and maintaining NL stability. Interestingly, this NLS motif is conserved in LMNA/C and is required for lamin-chromatin interactions [33]. High resolution imaging of endogenous LMNA and LMNB1 demonstrated that individual homopolymers exist in close contact with each other [28]. Our results showing that aggregates of mutant-LMNB1 contain LMNA/C indicate the potential crosstalk between the two proteins via PTMs, thereby opening new avenues to explore the role of methylated LMNB1 towards the assembly of NL and its integrity. LMNA is extensively studied with several known binding partners and disease causing mutations [34]. However, patients with laminopathies rarely exhibit a defect in the NLS sequence likely due to its lethality. In this regard, our study offers a new perspective on the less studied LMNB1 in the context of normal physiology and perhaps in laminopathies/disease.
Methylation of lysine residues facilitates a variety of functions including protein stability [35][36][37][38]. Altogether our data demonstrated EHMT1 and EHMT2 as upstream regulators of LMNB1 that influences its protein levels via PTMs. While EHMT2 is known to methylate a variety of non-histone proteins [24], our study for the first time demonstrates the competency of EHMT1 to methylate non-histone proteins and utilizing it as a mechanism to attach heterochromatin to the NP during aging.
Structurally similar EHMT1/2 proteins form a heteromeric complex in mammalian cells [23] and are known to fulfill both overlapping and unique physiological roles in developing and adult animals [39]. In the quest to understand the individual 1 8 contributions of EHMT proteins in regulating peripheral heterochromatin, we identified that both EHMTs regulate LMNB1. However, unique molecular changes seen upon EHMT1 loss such as disruption of NE integrity coupled with the loss of architectural proteins like HP1 and HMG that influence heterochromatin organization requires further investigation. Nonetheless, our studies provide a broader role for EHMTs by which it impacts the spatial distribution of the genome.
There are a number of studies demonstrating redistribution or loss of chromatin modifiers and their implications in aging [40,41]. These studies mainly focused on the consequence of the global loss of chromatin structure but none addressed the mechanisms underlying the alteration of genome architecture. Our study not only demonstrates the correlation between the expressions of EHMTs with peripheral heterochromatin organization during aging but also provides a mechanism by which EHMT regulates higher-order chromatin structure via stabilization of the NL and architectural proteins. These results are supported by previous observations wherein defects in the sophisticated assembly of nuclear lamins along with architectural proteins results in disease or aging [41][42][43][44][45]. Reorganization of heterochromatin at the NP by restoration of EHMT1 or EHMT2 in aged cells further reinforces the fact that EHMT proteins are key determinants of higher-order chromatin organization.
Aging associated defects in chromatin organization exhibit a variety of functional consequences such as misregulation of gene expression via alteration of the epigenome, activation of repeat elements and susceptibility to DNA damage [46,47].
Additionally, loss of lamins leads to altered mechano-signaling [48]. Together, these processes make aged cells stressed and also influence the stress response contributing towards reduced proliferation and enhanced senescence. Our studies revealed a direct correlation between loss of peripheral heterochromatin and reduced proliferation in 1 9 shEHMT and 65Y HDFs. Taken together, we conclude that the steady loss of EHMT proteins drives normal aging in fibroblasts. It remains a mystery as to how EHMT proteins are regulated. We propose that EHMT2 degradation leads to gradual destabilization of EHMT1 in response to aging via currently unknown mechanisms. It is important to note that these results are in cultured HDFs and yet to be established if this extends in the context of tissue aging.

Antibodies & Inhibitors:
The following antibodies were used in the current study: EHMT1 (

Cell growth curve:
Fetal fibroblasts transduced with shCnt or shEHMT1 and shEHMT2, were independently seeded per well of a 6 well plate, with one well each for different time points. After each time point, cells were harvested and cell count was determined.
Cell count was plotted against the time points to determine the growth curve.

SA-β-Galactosidase Assay:
SA-β-Galactosidase staining was performed using the Senescence Cells Histochemical Staining Kit (CS0030, Sigma). Cells were rinsed with PBS followed by fixing with 1X Fixation buffer provided with the kit for 8 min at RT. After rinsing thrice with PBS, 0.5ml of the staining mixture was added and incubated at 37°C without CO 2 for 18 h. The percentage of β-gal positive cells were quantified from the images taken at 10 randomly selected microscopic fields.

Cloning:
The Ankyrin and SET domains of EHMT1 were amplified from cDNA prepared using the Superscript III cDNA synthesis kit (11752-050, Thermo Fisher Scientific) from HDFs with the help of Ankyrin (737-1004 AA) and SET (1013-1265 AA) domain primers (Appendix Table S1). The PCR amplified EHMT1-Ankyrin and SET products were then cloned into pEGFPC1 vector (6084-1, Clontech) to generate the plasmid constructs, pEGFP-ANK and pEGFP-SET. For cloning the C-Term of LMNB1, the cDNA from HDFs was PCR amplified using primers LMNB12F and LMNB1-1R (Appendix Table S1) and cloned into pET28a+ vector between BamHI and HindIII restriction digestion sites. The identity of all plasmids was confirmed by sequencing.  The purified protein was concentrated, buffer exchanged and protein dialysis was performed using Amicon Ultra-4 centrifugal concentrators (UFC801008, Millipore,) with a molecular weight cut off of 10 kDa and the final concentration was estimated using the Bradford protein assay (5000006, Bio-Rad). The protein was also subjected to mass spectrometry to assess its purity and molecular weight. In-gel digestion for 2 6 mass-spectrometry analysis revealed a Mascot Score of 2354.46 for 6X His EHMT1-SET.

Protein-protein interaction assays:
For interaction assays, Ni-NTA beads pre-washed with IP100 buffer (

Immunoprecipitation (IP):
For IP experiments, cell or nuclear lysates (400 ug) prepared from HEK293 or HDFs to Dynabeads Protein A (10001D, Thermo Fisher Scientific) that were prebound with 2-3µg of indicated antibody and incubated overnight at 4°C. Beads were then washed and eluted in 2X loading dye. Eluted proteins were subjected to western blotting with indicated antibodies.

Sequential IP:
For sequential IP, nuclear extract was added to Dynabeads Protein A (10001D,

ChIP-seq library preparation:
ChIP DNA was subjected to library preparation using TruSeq ChIP sample preparation kit from Illumina (IP-202-1012). Briefly, ChIP samples were processed for end repair to generate blunt ends using end repair mix. A single 'A' nucleotide was added to the 3' ends of the blunt fragments to prevent them from ligating to one another during the adapter ligation reaction. In the next step, indexing adapters were ligated to the ends of the DNA fragments. The ligated products were purified on a 2% agarose gel and narrow 250-300bp size range of DNA fragments were selected for ChIP library construction appropriate for cluster generation. In the last step, DNA fragments with adapter molecules on both ends were enriched using PCR. To verify the size and quality of library, QC was done on high sensitivity bioanalyzer chips from Agilent and the concentration was measured using Qubit dsDNA HS assay kit (Q32851, Thermo Fisher Scientific). After passing QC, samples were sequenced 75 2 9 paired end (PE) on NextSeq Illumina platform. Genotypic Technology Pvt. LTD.

ChIP-seq Analyses:
Alignment of ChIP-seq derived short reads to the human reference genome (UCSC hg19) was done using Bowtie2 short read aligner [50] with default parameters.
Subsequently, aligned ChIP-seq reads from two replicates were merged. Peak calling was done for each sample with their respective control using MACS 1.4 algorithm [51]. The following parameters deviated from their default value: -effective genome size = 2.70e+09, bandwidth = 300, model fold = 10, 30, p-value cutoff = 1.00e-03.
In order to identify regions enriched for EHMT1 and LMNB1, we employed a twostep approach, first total peak counts for EHMT1 and LMNB1 was calculated in 1MB window for all the chromosomes. Next, ratio of total peak count over expected peak count (total peaks from a chromosome divided by total 1 MB window for the same chromosome) was calculated for each 1MB window. Raw data has been deposited in NCBI {SRP110335 (PRJNA391761)}.

RNA-seq and data analysis
Fetal HDFs transduced with shEHMT1, shEHMT2 or shCnt were harvested and RNA was extracted by Trizol method. RNA concentrations were estimated using Qubit fluorometer and quality was assessed using Bioanalyzer. After passing the QC, samples were subjected for library preparation and QC for the same. Samples were sequenced at Genotypic Technology Pvt. LTD. Bengaluru, India.
From the sequencing reads, adapters were trimmed using Trimmomatic program [52].

Immunostaining:
Briefly, cells were fixed with 4% Paraformaldehyde (PFA, P6148, Sigma) for 10 min at room temperature (RT) and permeabilized with 0.5% triton X-100. The blocking was done with 5% BSA for 1 h. Antibodies mentioned in the antibodies and inhibitors section were used at desired dilution and imaging was carried out on FV3000 Confocal Microscope (Olympus). Image analysis and extraction of raw files was done with the Fiji software [62].

Transmission Electron Microscopy (TEM):
For TEM sample preparation cells of different age groups or treated with different conditions as mentioned in results sections were trypsinized and the pellet was fixed

Telomerase assay:
Telomerase activity was detected using the PCR-based Telomeric Repeat Amplification Protocol (TRAP) assay kit (Millipore, S7700). Briefly, different age groups HDFs or cells transduced with shEHMT1 and shEHMT2 were harvested and resuspended in 1X CHAP Buffer provided with the kit. The protocol was followed as per the manufacturer's instructions.
Quantitation for mean fluorescence intensity: 3 2 For quantitation of fluorescence intensity in Fig 4A, Supplementary Figure 2E

Statistics:
The detailed statistical analysis and methods have been described in the figure legends along with the p-values for respective data sets. The data is represented as mean ± SD. For statistical analysis, GraphPad Prism version 7 software was used.

Acknowledgments
We thank Drs. Colin Jamora, Arjun Guha and Pavan Kumar P for the helpful suggestions and critical reading of the manuscript. We thank members of Jamora lab and Sara Ripamonti for experimental help. We thank Ashish Dhayani for help with

Conflicts of Interest
The authors declare no competing financial interests.       peripheral heterochromatin in old cells compared with control cells (Scale bar: 1µm).
Arrows indicate the area zoomed and presented in the inset format.     A Heat map demonstrating differential expression of age related genes in EHMT1 and EHMT2 depleted fibroblasts. Representative genes that were altered similarly or distinctly in EHMT1 vs EHMT2 are indicated from few clusters.
B. Validation for differential expression of candidate genes obtained from RNA-Seq analysis by semi-quantitative PCR. H-I. Cell lysates prepared from human fibroblasts of indicated age groups were subjected for IP using EHMT1/EHMT2 antibody. IPed material was analyzed by immunoblotting using LMNB1 and LMNA/C antibodies. 30 µg of fetal cell lysate 4 6 was used as input control. Dotted lines indicate that different exposures were used for Input and IP of the same western blot.