A-type lamins bind both hetero- and euchromatin, the latter being regulated by lamina-associated polypeptide 2 alpha

1The confined space of metazoan cell nuclei requires chromatin to be tightly packed while maintaining a high degree of organization to provide genome stability and coordinated operation of the transcriptional machinery. Conceptually, gene-rich and transcriptionally active chromatin localizes to the nuclear interior, while gene-poor and transcriptionally repressed chromatin preferentially resides at the nuclear periphery.

2The nuclear lamina, a scaffold structure at the periphery of metazoan nuclei, has a major role in chromatin organization by anchoring heterochromatin. Lamins, type V intermediate filaments, are the major constituents of the lamina. They are categorized into A-type lamins, comprising the major isoforms lamin A and C in mammals, and B-type lamins, including lamin B1 and B2. Whereas B-type lamins are ubiquitously expressed throughout development, A-type lamins are expressed in a differentiation-dependent manner. Lamins interact with a plethora of proteins in the inner nuclear membrane, which constitute important additional components of the nuclear lamina.

3At the molecular level, genome-wide interactions of chromatin with the nuclear lamina have first been mapped by the DamID technique identifying genomic regions attached to the nuclear periphery, so-called lamina-associated domains (LADs). LADs cover nearly 40% of the human genome and are up to 10 Mb long, gene-poor, and enriched in repressive histone marks H3K9me3 and H3K27me3. These studies led to the concept that the lamina anchors heterochromatin at the nuclear periphery, thereby contributing to gene repression. Solovei et al. showed that two protein complexes of the lamina redundantly link heterochromatin to the nuclear periphery, the inner nuclear membrane (INM) protein lamin B receptor (LBR), most likely in a complex with B-type lamins, and a complex of A-type lamins with LAP-Emerin-MAN1 (LEM) domain proteins of the INM. LBR binds to H3K9me3 via the chromobox 5 (CBX5) protein and to H4K20me2 directly via its Tudor domain. LEM proteins contain a bihelical structural motif, the LEM domain that mediates association with chromatin via Barrier-to-Autointegration Factor (BANF1). Most LEM proteins are integral components of the INM and interact with lamins or require A-type lamins for their proper localization. A well-studied group among mammalian LEM proteins comprises isoforms of lamina-associated polypeptide (LAP) 2, encoded by thymopoietin (Tmpo, also known as LAP2). The major LAP2 isoform in the INM, LAP2beta binds chromatin-associated BANF1 and DNA via its LEM and LEM-like domains, respectively, and contributes to gene repression by binding to transcriptional repressors germ cell-less and ZBTB7B together with histone deacetylase (HDAC) 3.

4Unlike B-type lamins, A-type lamins are also found in a mobile pool throughout the nuclear interior. B-type lamins are tightly associated with the INM through post-translational addition of a hydrophobic farnesyl group to their C terminus. In contrast, lamin C is not farnesylated at all, and pre-lamin A is only transiently farnesylated, as a C-terminal peptide, including the farnesyl group, is removed in a final processing step. Thus, A-type lamins can dissociate from the membrane into the nucleoplasm, where they interact with the non-membrane-bound LAP2 isoform, LAP2alpha. Deletion of exon 4 of the Tmpo gene generated mice specifically lacking LAP2alpha and leads to the selective loss of nucleoplasmic lamin A/C. Loss of LAP2alpha causes tissue-specific phenotypes, including increased proliferation of tissue progenitor cells in epidermis, colon, and the hematopoietic system, delayed skeletal muscle differentiation, and impaired heart function. However, the molecular mechanisms remain elusive.

5In view of the recently reported role of A-type lamins and LEM proteins in anchoring heterochromatin at the nuclear periphery, we hypothesized that complexes of A-type lamins and LAP2alpha may also associate with chromatin in the nuclear interior. LAP2alpha has several means to interact with chromatin and DNA, including a LEM and LEM-like motif, and was recently reported to associate with chromatin in vivo on a genome-wide level.

6Using various protocols for chromatin immunoprecipitation (ChIP) in immortalized wild-type (Tmpo WT) and LAP2alpha-deficient (Tmpo KO) murine dermal fibroblasts (imMDFs), we found that A-type lamins not only associate with heterochromatic LADs as reported but also show a genome-wide interaction with euchromatic regions outside of LADs. Interestingly, euchromatic regions bound by lamin A/C largely overlap with those bound by LAP2alpha and were significantly rearranged and shifted toward heterochromatic regions in Tmpo KO cells resulting in changes in gene expression and epigenetic histone marks.

7Lamins associate with heterochromatic LADs at the nuclear periphery. To test whether nucleoplasmic lamin A/C and LAP2alpha also interact with chromatin, we performed ChIP for LAP2alpha and for two distinct domains of lamin A/C, using antibodies N18, directed against lamin A/C's N terminus, and 3A6, detecting the C-terminal Ig fold. As a control for peripheral lamina-chromatin interactions, we performed lamin B1 ChIP. Our ChIP protocol applied moderate sonication (12 cycles in a Bioruptor) yielding DNA fragments of 100–500 bp and allowing efficient immunoprecipitation of lamins and LAP2alpha (Supplemental Fig. 1). ChIP and respective input samples were analyzed by deep sequencing (ChIP-seq). Integrative Genomics Viewer (IGV) tracks, representing the loge (ChIP/input) signal, display wide-range chromatin interactions of all proteins (Fig. 1A). The enriched domain detector (EDD) peak-calling algorithm identified around 250 EDD peaks in lamin A/C, LAP2alpha, and lamin B1 ChIP samples with an average length of 1–2 Mb, in total covering 15%–20% of the genome (Table 1). The SICER algorithm identified significantly more (7000–9000) and smaller (average size of 25 kb) regions to be associated with LAP2alpha, lamin A/C, and lamin B1 (Table 1). Regions with a high SICER peak density overlapped with EDD peaks (Supplemental Fig. 2).

8Next, we tested whether the lamin A/C-, LAP2alpha-, and lamin B1-enriched chromatin domains match published lamin B1 LADs in mouse embryonic fibroblasts (MEF) identified by DamID. Whereas 95% of the lamin B1-associated genomic regions overlapped with lamin B1-LADs, only 27%–41% of lamin A/C- and LAP2alpha-associated regions overlapped (Fig. 1A,C). Accordingly, in contrast to LADs, lamin A/C- and LAP2alpha- EDD peaks were gene-rich (18–24 genes/Mb versus 5 genes/Mb in LADs) (Fig. 1A,D) and enriched in active H3K4me3 and H3K9ac histone marks, as well as in H3K27me3 (Fig. 1E). Lamin B1 predominantly associated with gene-poor regions (Fig. 1D), devoid of active histone marks and maintaining the highest levels of the heterochromatic H3K9me3 mark (Fig. 1E). Thus, whereas lamin B1 interacted mainly with heterochromatic LADs, lamin A/C and LAP2alpha surprisingly occupied active, euchromatic regions of the genome.

9Elucidating the disparity between our data and the reported lamin A/C binding to LADs, we noted that lamin A/C ChIP samples obtained after moderate sonication (12 cycles) contained, in addition to the main pool of DNA fragments of ∼350 bp, a significant fraction of larger fragments (∼1.5 kb) (Supplemental Fig. 1). This fraction likely represents heterochromatic genomic regions, which were excluded from Illumina sequencing due to size selection. In order to make these regions accessible to analysis, we extended the sonication to 30 cycles and reperfomed the ChIP. This resulted in a clear reduction of the larger DNA fragment pool, while still allowing for efficient immunoprecipitation (Supplemental Fig. 1). Under these conditions, lamin A/C-3A6-associated chromatin regions are shifted toward gene-poor heterochromatin and overlap with LADs by ∼96%, resembling lamin B1-associated regions (Fig. 1B–D). Concomitantly, active histone marks are lost and heterochromatic H3K9me3 is increased in lamin A-3A6-recovered regions (Fig. 1E). Moreover, lamin A/C-3A6 EDD peaks of the 30-cycle sample show only minimal overlap (11%) with peaks obtained from the 12-cycle sample, while the lamin B1 peaks overlapped nearly completely, although the recovery of lamin B1 LADs was significantly improved in the 30- versus 12-cycle preparation (Fig. 1F). Similar effects were observed when comparing lamin A/C-N18-associated chromatin regions in 30- versus 12-cycle ChIP preparations (Fig. 1C–F). Taken together, these results suggest that A-type lamins occupy both heterochromatic LADs and euchromatic regions outside of LADs, while lamin B1 is enriched in heterochromatic LADs only.

10Our data showed that sonication conditions are a critical parameter for the nature of recovered lamin A/C-associated chromatin regions: moderate sonication (12 cycles) enriching for euchromatic regions, and extended sonication (30 cycles) for heterochromatic LADs. Intriguingly, this effect was more pronounced in the lamin A/C-3A6 compared to lamin A/C-N18 ChIP samples. Whereas in the 12-cycle ChIP, the lamin A/C-associated chromatin regions retrieved by the two lamin A/C antibodies coincided by nearly 80%, they considerably differed in the 30-cycle ChIP (Fig. 2A, lower panel), indicating that the 3A6 and N18 antibodies target different pools of A-type lamins. However, the overall overlap of lamin A/C-associated genomic regions with lamin B1-associated LADs strongly increased in the 30- versus 12-cycle preparation for both antibodies, N18 and 3A6 (Fig. 2A, bar graph).

11The two lamin A/C ChIP samples also differ in their epigenetic marks. Lamin A/C-N18-associated domains contained more active histone marks (H3K4me3, H3K9ac) and less repressive H3K9me3 than the lamin A/C-3A6-bound regions in euchromatin-enriched samples, and they had more H3K27me3 marks in the heterochromatin-enriched samples (Fig. 1E). These data suggest that antibody 3A6 targets lamin A/C associated with heterochromatin, as opposed to antibody N18 that may preferentially detect lamin A/C complexes in an open chromatin context. This hypothesis was further supported by immunofluorescence analyses of mixed cultures of Tmpo WT and Tmpo KO fibroblasts. While lamin A/C-N18 detects both peripheral and nucleoplasmic lamin A/C pools in Tmpo WT (but not in Tmpo KO) cells, lamin A/C-3A6 and lamin B1 detect primarily the peripheral lamina in both cell types (Fig. 2B). Furthermore, coimmunoprecipitation of lamin A/C and LAP2alpha revealed that lamin A/C-N18-antibody coprecipitated considerably more LAP2alpha than lamin A/C-3A6 antibody (Fig. 2C). We concluded that antibody 3A6 mainly detects lamin A/C complexes not bound to LAP2alpha, whereas antibody N18 targets nucleoplasmic lamin A/C in complexes with LAP2alpha.

12Having shown that lamin A/C binds in a heterochromatic and a euchromatic environment, we next tested to what extent the lamin A/C binding sites correlate with those of LAP2alpha. By overlapping of lamin A/C- and LAP2alpha-associated regions, we found that, in euchromatin-enriched samples (12 cycles), the majority of lamin A/C EDD peaks (85% for 3A6 and 90% for N18) concurred with LAP2alpha EDD peaks (Fig. 2D). After extended sonication (30 cycles), the congruency of LAP2alpha with lamin A/C-3A6 and -N18 is diminished to 21% and 20%, respectively. Thus, the overlap of lamin A/C- and LAP2alpha-associated chromatin regions decreased markedly in 30- versus 12-cycle preparations (Fig. 2D, bar graph). However, still more than half (56%) of lamin A/C-N18-associated chromatin (but only 24% of 3A6-associated regions) is also bound by LAP2alpha under these conditions (Fig. 2D). Similar results were obtained by comparing the number of SICER peaks per Mb bin between LAP2alpha versus lamin A/C-3A6 and LAP2alpha versus lamin A/C-N18 ChIP samples (Supplemental Fig. 2B). Furthermore, sequential ChIP of lamin A/C and LAP2alpha followed by qPCR revealed that lamin A/C and LAP2alpha co-occupy various genomic regions only in the euchromatin-enriched sample (Supplemental Fig. 3). Overall, we concluded that LAP2alpha and lamin A/C occupy similar genomic regions under conditions enriching for euchromatin, whereas in heterochromatin-enriched regions, lamin A/C-chromatin association seems to be mostly independent of LAP2alpha.

13Given that LAP2alpha and lamin A/C predominantly co-occur in euchromatic genomic regions, we next aimed to investigate the role of LAP2alpha in lamin A/C chromatin interaction by extending our ChIP-seq experiments to Tmpo KO fibroblasts (Supplemental Fig. 4). We assessed changes in lamin A/C EDD peaks between Tmpo WT and Tmpo KO samples on a genome-wide scale in euchromatin-enriched regions. This analysis revealed lamin A/C-associated chromatin regions exclusively in Tmpo WT cells (“lost in KO”) (Fig. 3A, black arrowheads) or exclusively in Tmpo KO cells (“gained in KO”) (Fig. 3A, white arrowheads). At the genome-wide level, only 61% of the lamin A/C-N18- and 47% of the lamin A/C-3A6-bound regions were detected in both Tmpo WT and KO cells (“WT^KO”); the remaining regions were either lost (39% of N18 and 53% of 3A6 peaks) or gained (41% of N18 and 51% of 3A6 peaks) in Tmpo KO cells (Fig. 3B). Interestingly, regions that lost or retained lamin A/C interaction in the Tmpo KO cells overlap with LAP2alpha-associated regions by 81%–96%, while the newly gained lamin A/C associated regions showed only 24% (N18) and 7% (3A6) overlap (Fig. 3B). This indicates a reorganization and partial loss of lamin A/C within the LAP2alpha-associated regions in Tmpo WT versus KO cells and a gain of lamin A/C binding primarily outside these regions. Within the LAP2alpha-associated regions, 45%–59% of lamin A/C-associated regions were retained in the Tmpo KO cells, while 31%–51% were lost and only a marginal fraction was gained (Fig. 3C). This rearrangement of lamin A/C-chromatin association is also reflected by the low correlation of SICER peak densities in Tmpo WT versus KO cells (R = 0.539 for N18 and R = 0.438 for 3A6) (Fig. 3D).

14Interestingly, partial reorganization of lamin A/C-N18- and lamin A/C-3A6-associated domains in Tmpo KO versus WT cells was also detectable in the 30-cycle ChIP analyses (Supplemental Fig. 5), suggesting that, in the absence of LAP2alpha, the association of lamin A/C with chromatin is reorganized at a genome-wide level.

15In view of our findings that lamin A/C and LAP2alpha co-occupy particularly euchromatic genomic regions, while lamin A/C association with heterochromatic LADs occurs largely independent of LAP2alpha, we hypothesized that loss of LAP2alpha might affect the equilibrium between euchromatic and hetereochromatic lamin A/C pools. Indeed, we found that the overlap of lamin A/C-3A6 EDD peaks with LADs increased from 41% in wild-type to 74% in Tmpo KO cells and that of lamin A/C-N18-identified EDD peaks from 27% to 41% (Fig. 4A), indicating that lamin A/C tends to relocalize from euchromatic to heterochromatic regions in the absence of LAP2alpha.

16To investigate whether the loss of LAP2alpha-chromatin interaction and the rearrangement of lamin A/C on chromatin affects epigenetic marks and gene expression, we analyzed active and repressive histone mark abundance (see histone mark tracks in Fig. 3A) in genomic regions that gained or lost lamin A/C binding in Tmpo KO versus WT cells.

17Repressive H3K9me3 and H3K27me3 marks accumulated in regions that gained lamin A/C binding (“gain”) and diminished in regions that lost lamin A/C binding (“loss”) in Tmpo KO, whereas the abundance of these marks did not change in regions of unaltered lamin A/C-association (“WT^KO”) (Fig. 4B). Intriguingly, also the level of active H3K4me3 and H3K9ac marks decreased in regions that lost lamin A/C-binding in Tmpo KO cells and increased in regions that gained lamin A/C binding. Thus, changes in lamin A/C binding to chromatin in Tmpo KO versus WT cells affect repressive and active marks similarly. Loss of LAP2alpha binding had a similar effect on the epigenetic profile as the loss of lamin A/C binding (Fig. 4B, upper panel). Changes in epigenetic marks were most prominent in ChIP samples enriched in euchromatin (12 cycles); in heterochromatin-enriched ChIP samples (30 cycles) these marks changed only marginally (Supplemental Fig. 6).

18Having determined that loss or gain of lamin A/C binding affects both active and repressive histone marks in a similar manner, we wondered whether these changes are reflected by changes in gene expression. Assessing global gene expression changes in Tmpo KO versus WT in imMDFs by RNA-seq, we found 616 genes to be differentially expressed (false discovery rate [FDR] < 0.05), of which 341 were up-regulated and 275 were down-regulated (Supplemental Table 1). Of the 616 de-regulated genes, 239 were located in genomic regions of altered lamin A/C association (Supplemental Table 2). As expected, promoters of up-regulated genes showed an increase in active and a decrease in repressive histone marks and vice versa for down-regulated genes (Fig. 4B, lower panels, 4D; Supplemental Fig. 6). However, in the euchromatin-enriched samples, gain and loss of lamin A/C-gene association in Tmpo KO cells did not strictly correlate with overall down- and up-regulation of gene expression, respectively (Fig. 4B,C), although expression of several genes is clearly increased upon loss of lamin A/C- and decreased upon gain of lamin A/C-binding in Tmpo KO cells (Fig. 4D). In the heterochromatin-enriched samples (30 cycles), loss of lamin A/C binding was linked to a general increase in gene expression, and gain of lamin A/C binding slightly decreased gene expression (Fig. 4C,D; Supplemental Fig. 6). Loss of LAP2alpha-association correlated with neither up- nor down-regulation of gene expression in both samples.

19Most studies performed to date report predominant binding of lamin A/C to heterochromatic regions. We show here that lamin A/C, but not lamin B1, binds to both hetero- and euchromatin.

20We found this discrepancy to arise from the extent of chromatin shearing, where moderate conditions preserve and enrich for open euchromatin, while heterochromatin yields larger fragments usually excluded from deep sequencing. Conversely, prolonged sonication enriches heterochromatin, while euchromatin may be lost due to hyperfragmentation. Our findings are in line with recent data showing that sonication versus micrococcal nuclease digestion in lamin A/C ChIP identified only partially overlapping lamin A/C-bound genomic regions.

21Furthermore, the location and accessibility of the lamin A/C epitope recognized by the antibodies used for ChIP can ostensibly affect ChIP results. A lamin A/C antibody against the N terminus (N18) favored lamin A/C complexes containing LAP2alpha, while an antibody to lamin A/C's C-terminal Ig fold (3A6) preferentially precipitated heterochromatin-associated lamin A/C not bound by LAP2alpha. As LAP2alpha interacts with the Ig fold of lamin A/C, it may compete with binding of the C-terminal antibody 3A6.

22Mature lamin A/C in the nuclear interior colocalizes with LAP2alpha. As euchromatic regions associated with lamin A/C were largely congruent with those bound by LAP2alpha, it is conceivable that euchromatin-associated lamin A/C is predominantly localized in the nuclear interior, whereas heterochromatin-bound lamin A/C is part of the peripheral lamina. Overall, our data are consistent with a model in which lamin A/C, together with lamin B1, provide a stable anchor for heterochromatin at the nuclear lamina. Lamin A/C binding to euchromatin in the nuclear interior may be more dynamic, providing a flexible chromatin environment allowing efficient response to epigenetic and transcriptional regulators in a context-dependent manner.

23In a LAP2alpha-deficient background, lamin A/C relocalizes toward heterochromatic regions. How can LAP2alpha affect lamin A/C-chromatin interactions? Lamin A/C and LAP2alpha complexes in the nuclear interior are highly mobile and may thus bind chromatin in a dynamic manner, while immobile lamin A/C structures in the absence of LAP2alpha may bind chromatin more stably. Alternatively, LAP2alpha may compete with lamin A/C for binding to the same chromatin regions, ensuring flexibility of chromatin organization.

24What are the consequences of altered lamin A/C chromatin interactions in Tmpo KO versus WT cells? Interestingly, in euchromatin both repressive and active histone marks were decreased in regions that lost lamin A/C binding and increased in regions that gained lamin A/C, without affecting overall gene expression. Thus, lamin A/C-LAP2alpha complexes may primarily be involved in establishing a chromatin environment permissive for epigenetic regulation, rather than regulating gene expression directly by binding to promoters as suggested for lamin A/C. In contrast, in heterochromatin epigenetic marks do not change in regions that gain or lose lamin A/C binding in Tmpo KO versus WT cells, but loss of lamin A/C binding correlated with increased gene expression, indicating that lamin A/C may serve different functions in euchromatin versus heterochromatin. The open and transcriptionally active environment in euchromatin allows efficient access of chromatin regulators such as epigenetic modifiers, nucleosomal remodeler complexes, and transcriptional regulators. Lamin A/C-LAP2alpha complexes may contribute to this open chromatin state and thereby favor dynamic chromatin regulation. In such a scenario, gene expression changes occur in a context-dependent manner, for instance, during differentiation due to the presence or absence of specific transcription factors. In contrast, heterochromatin is tightly packed and less accessible for regulators. Lamin A/C in the lamina is known to contribute to heterochromatin formation and stable gene repression. Therefore, loss of lamin A/C binding in heterochromatic regions may have profound effects on gene expression.

25It is also tempting to speculate that an impairment of lamin's chromatin-regulating functions in the diseases linked to LMNA mutations (laminopathies) may contribute to the disease phenotype. In line with this hypothesis, expression of a muscle disease-causing lamin mutant in Caenorhabditis elegans impairs cell-type–specific reorganization of repetitive DNA arrays during muscle differentiation. In addition, expression of progerin, a lamin A mutant causing the premature aging disease, Hutchinson Gilford Progeria Syndrome (HGPS), affects heterochromatin organization. Interestingly, while mature lamin A is not farnesylated, progerin retains its farnesyl moiety and is tightly bound to the INM at the nuclear periphery similar to lamin B1. We recently showed that expression of progerin causes loss of nucleoplasmic lamin A/C and LAP2alpha, leading to down-regulation of the expression of extracellular matrix components.

26Immortalized murine dermal Tmpo WT and KO fibroblasts were cultured at 37°C, 5% CO2 in Dulbecco's modified Eagle's medium (DMEM), supplemented with 10% FCS, 2 mM L-glutamine, 100 U/mL penicillin, 100 µg/mL Streptomycin, 0.1 mM nonessential amino acids.

27ChIP was performed as previously described (see also Supplemental Material). Briefly, cells were cross-linked for 10 min with 1% formaldehyde and chromatin prepared by lysis (2× 10 min on ice). Nuclei-containing pellets were resuspended in lysis buffer (50 mM Tris-HCl [pH 8.1], 10 mM EDTA, 1% sodium dodecyl sulfate [SDS], 1 mM PMSF, 1× complete protease inhibitor) and sonicated with a Bioruptor Plus (Diagenode) for 12 or 30 cycles. For histone mark ChIPs, 12 cycles of sonication were applied. Fifty micrograms of chromatin were used for ChIP. Input chromatin was diluted 1:10 in ChIP dilution buffer (16.7 mM Tris-HCl [pH 8.1], 167 mM NaCl, 1.2 mM EDTA, 1.1 % Triton X-100, 0.01% SDS) and incubated with antibodies overnight. Chromatin-antibody complexes were captured with magnetic protein A/G beads and crosslinks were reversed. Eluted DNA was purified with the ChIP DNA Clean & Concentrator kit (Zymo Research).

28Sequencing libraries were prepared from DNA fragments ranging from 100 to 800 bp by the Vienna Biocenter CSF NGS unit (www.csf.ac.at) and sequenced either on an Illumina HiSeq 2000 or HiSeq 2500 at the CSF NGS unit. After de-multiplexing, adapters were clipped with cutadapt. Reads were mapped to the mouse genome (NCBI37/mm9 annotation from July 2007) with Bowtie 2 version 2.1.0 using default parameters. Reads with a mapping quality below 20 were discarded. As suggested by, reads stemming from PCR duplicates were removed and all ChIP read files were down-sampled to match the read count of their respective input sample using Picard (http://broadinstitute.github.io/picard/). Supplemental Table 3 lists the number of sequenced reads and mapped reads after filtering and de-duplication for all samples.

29Regions of enrichment (peaks) were identified using EDD version 1.0.2 and SICER version 1.1. Since EDD's auto parameter estimation was unsuited for our data sets, we used the parameter set “-b 11 –g 5 –fdr 0.1” (bin size 11 kb, gap penalty 5) suggested for lamin A by. Based on visual inspection, these settings yielded the best results for our data. For SICER, we used a window size of 1000 bp, a gap size of 3000 bp, and a false discovery rate of 0.01, to account for the characteristics of lamin and LAP2alpha binding. See Supplemental Material for more details and Table 1 for information on peak number and genome coverage for all ChIP samples.

30Ten micrograms of purified, rRNA-depleted RNA (>200 bp) were fragmented by hydrolysis (40 mM TrisOAc at pH 8.2, 100 mM KOAc, 150 mM MgOAc) at 94°C for 3 min. First-and second-strand cDNA synthesis was performed as described in detail in the Supplemental Material. Sequencing libraries were prepared by the Vienna Biocenter CSF NGS unit using the NEBNext Library Prep Reagent Set for Illumina (NEB), multiplexed (2 samples/lane), and sequenced on HiSeq 2000 (Illumina) at the CSF NGS unit. RNA-seq data analysis was done as described by and in the Supplemental Material. All significantly up- or down-regulated genes are listed in Supplemental Table 1. Significant changes in expression in Lamin A/C gain, loss, and WT^KO regions were detected by comparing the actual log2-fold change to fold changes obtained from 10,000 random permutations of differentially expressed genes (P < 0.05).

31ChIP-seq data (peaks and log ratios) were visualized with the Integrative Genomics Viewer. Log ratios (base e) between Input and ChIP samples were computed for nonoverlapping 10-kb bins using EDD-tools (https://github.com/CollasLab/edd). Genes shown in the browser view were obtained from the RefSeq data track provided by the UCSC Genome Browser. cLAD, ciLAD, and MEF LAD annotations were downloaded from the GEO database (accession number GSE17051 and GSE36132).

32Fifty to one hundred micrograms of cross-linked chromatin, diluted 1:10 in ChIP dilution buffer, were incubated with antibodies overnight at 4°C. Antibody-protein complexes were captured with magnetic protein A/G beads, 1/10 of the IP supernatant was put aside for later analysis, beads were washed (see ChIP-seq, Supplemental Material), and antibody-protein complexes were eluted in 1× Laemmli Buffer.

33Two times 107 Tmpo WT imMDFs were harvested in 3 mL IP buffer (20 mM Tris-HCl [pH 7.5], 150 mM NaCl, 2 mM EGTA, 2 mM MgCl2, 1 mM DTT, 0.5% NP-40, 1 U/mL Benzonase [Novagen], 1 mM PMSF, 1× complete protease inhibitor) and incubated on ice for 10 min. One-twentieth of the lysate was kept as “Input.” One-third of the soluble supernatant fraction was incubated with antibodies overnight at 4°C. Antibody-protein complexes were captured with magnetic protein A/G beads, 1/10 of the IP supernatant was put aside, beads were washed twice with IP buffer (w/o Benzonase) and IP samples eluted in 1× Laemmli buffer.

34Polyacrylamide gel electrophoresis was performed according to. After blotting onto nitrocellulose (Amersham), membranes were blocked with 5% BSA or Odyssey Blocking Buffer (Licor), incubated overnight with primary antibody and for 1 h with secondary antibodies.

35One to two times 105 cells on coverslips were washed once with PBS++ (1 mM CaCl2 and 0.5 mM MgCl2), fixed for 10 min with 4% formaldehyde/PBS, permeabilized with 50 mM NH4Cl and 0.5% Triton X-100/PBS for 10 min, and blocked with 0.5% gelatine/PBS for 15 min. Primary and secondary antibody dilutions were prepared in 0.5% gelatine/PBS and applied for 1 h at room temperature.

36Samples were imaged using a Zeiss LSM 700 with a Zeiss Plan-Apochromat 63×/1.4 oil DIC objective and processed with ImageJ and Photoshop CS4.

37Lamin A/C (3A6-4C11, Active Motif 39287, hybridoma supernatant, 100 µL for ChIP and Co-IP, 1:500 for WB, 1:100 for IF); lamin A/C (N18, Santa Cruz sc-6215, 10 µg for ChIP and Co-IP, 1:100 for WB, 1:50 for IF); lamin B1 (Proteintech 12987-1-AP, 3 µg for ChIP, 1:1000 for WB, 1:1000 for IF); mouse LAP2alpha (MFPL Monoclonal Antibody Facility, 8C10-1H11, hybridoma supernatant, 50 µL for ChIP and Co-IP, 1:500 for WB, 1:50 for IF; [specificity validated in Supplemental Fig. 3]); LAP2alpha (245.2, Abcam ab5162, 1:1000 for IF); H3K9me3 (Abcam ab8898, 3 µg for ChIP); H3K4me3 (Millipore 07-473, 3 µg for ChIP); H3K9ac (Millipore 06-942, 3 µg for ChIP); H3K27me3 (Millipore 07-449, 3 µg for ChIP); mouse normal IgG (Millipore 12-371, 10 µg for ChIP); rabbit normal IgG (Abcam ab46540, 10 µg for Co-IP).

38Secondary antibodies for WB: Licor IRDye 680RD donkey anti-mouse, 680RD donkey anti-rabbit, 800CW donkey anti-rabbit, 800CW donkey anti-goat; all 1:15000.

39Secondary antibodies for IF: DyLight donkey anti-mouse 594, donkey anti-rabbit 594, goat anti-mouse 488, goat anti-rabbit 488, donkey anti-goat 549; all 1:500.

40The data from this study have been submitted to the NCBI Gene Expression Omnibus (GEO; http://www.ncbi.nlm.nih.gov/geo/) under accession number GSE70149.