Altered modulation of lamin A/C‐HDAC2 interaction and p21 expression during oxidative stress response in HGPS

Abstract Defects in stress response are main determinants of cellular senescence and organism aging. In fibroblasts from patients affected by Hutchinson–Gilford progeria, a severe LMNA‐linked syndrome associated with bone resorption, cardiovascular disorders, and premature aging, we found altered modulation of CDKN1A, encoding p21, upon oxidative stress induction, and accumulation of senescence markers during stress recovery. In this context, we unraveled a dynamic interaction of lamin A/C with HDAC2, an histone deacetylase that regulates CDKN1A expression. In control skin fibroblasts, lamin A/C is part of a protein complex including HDAC2 and its histone substrates; protein interaction is reduced at the onset of DNA damage response and recovered after completion of DNA repair. This interplay parallels modulation of p21 expression and global histone acetylation, and it is disrupted by LMNAmutations leading to progeroid phenotypes. In fact, HGPS cells show impaired lamin A/C‐HDAC2 interplay and accumulation of p21 upon stress recovery. Collectively, these results link altered physical interaction between lamin A/C and HDAC2 to cellular and organism aging. The lamin A/C‐HDAC2 complex may be a novel therapeutic target to slow down progression of progeria symptoms.


| INTRODUCTION
Several evidences link lamin A/C to stress response. Prelamin A, the precursor of lamin A, is transiently accumulated during oxidative or replicative stress Liu, Drozdov, Shroff, Beltran, & Shanahan, 2013). Moreover, proteins involved in repair of stressinduced DNA damage are recruited by lamins to damaged sites or inside the nuclear compartment (Gibbs-Seymour, Markiewicz, Bekker-Jensen, Mailand, & Hutchison, 2015;Gonzalez-Suarez et al., 2011;Lattanzi et al., 2014). Consistent with these functions, lamin A/C has been implicated in mechanisms related to physiological  and pathological aging (Evangelisti, Cenni, & Lattanzi, 2016), above all in progeroid laminopathies (Camozzi et al., 2014). Here, we analyzed cells from patients affected by HGPS, a premature aging syndrome linked to LMNA mutations, and observed an altered modulation of CDKN1A, encoding p21, in HGPS under oxidative stress. p21, alternatively p21 WAF1/Cip1 , is a cyclin-dependent kinase inhibitor that targets CDK2 and CDK1 complexes and regulates cell cycle progression at G 1 /S border (Cazzalini, Scovassi, Savio, Stivala, & Prosperi, 2010). Moreover, p21 has been shown to play a role in the maintenance of G 2 -phase arrest and to be the principal mediator of cell cycle blockade in response to DNA damage (Bell & Sharpless, 2007;Prives & Gottifredi, 2008). In fact, persistent upregulation of p21 is associated with geroconversion (Bell & Sharpless, 2007;Karimian, Ahmadi, & Yousefi, 2016;Leontieva, Demidenko, & Blagosklonny, 2015). Previous studies had shown that lamin A/C depletion is a trigger of p21 expression (Moiseeva, Bourdeau, Vernier, Dabauvalle, & Ferbeyre, 2011). Moreover, accumulation of toxic levels of prelamin A or progerin, the mutated prelamin A form found in HGPS, was associated with upregulation of p53 target genes, including CDKN1A (Kudlow, Stanfel, Burtner, Johnston, & Kennedy, 2008;Varela et al., 2005). However, the molecular link between lamin A/C and p21 modulation remained elusive. Thus, having observed altered p21 regulation upon oxidative stress in HGPS, we set out to identify which molecule could mediate lamin A/C effects on p21 expression. It has been demonstrated that histone deacetylase 2 (HDAC2) is involved in the regulation of CDKN1A gene. It was demonstrated (Peng et al., 2015) that HDAC2 is recruited to CDKN1A promoter by FOXO3a and regulates p21 expression in cerebellar granule neuron. Furthermore, HDAC2 has been shown to suppress p21 expression in human hepatocellular carcinoma via its binding to an Sp1-binding site (Noh et al., 2011).
Finally, a clear link has been established between stress-induced chromatin remodeling, including acetylation or methylation of HDAC2 substrates H3 histone lysine 9 (H3K9) and H4 histone lysine 16 (H4K16), and lamin A/C posttranslational modifications (Ghosh et al., 2015;Lattanzi et al., 2007Lattanzi et al., , 2014Liu et al., 2013;Mattioli et al., 2008). Based on the whole evaluation of those reported data, we wondered if HDAC2 could mediate lamin A/C-dependent effects on p21 expression during DDR. Our data show that lamin A/C, which binds CDKN1A promoter, interacts with HDAC2 to promote deacetylase activity, and the interaction is reduced at the onset of DDR and recovered after completion of DNA repair. This interplay occurring during oxidative stress response parallels modulation of p21 expression and global histone acetylation, all mechanisms disrupted by LMNA mutations leading to progeroid phenotypes.

| Altered regulation of p21 expression during oxidative stress response in HGPS
It has been demonstrated that HGPS fibroblasts start acquiring a senescent phenotype at late passages (Columbaro et al., 2005;Goldman et al., 2004;Meaburn et al., 2007). We hypothesized that an altered response to stress stimuli could be a major determinant of cellular aging in those cells. To test this hypothesis, we induced oxidative stress in HGPS fibroblasts and age-and passage-matched controls ( Information Figure S1a,b), while prelamin A was significantly increased and its levels were decreased after stress recovery (Supporting Information Figure S1a-c), as previously reported . Of note, also progerin, the truncated prelamin A form accumulated in HGPS cells tended to increase during DDR (Supporting Information Figure S1a-d). In this context, we evaluated the expression pattern of CDKN1A, whose modulation during DDR is a key event to avoid shift into a senescence program (Cazzalini et al., 2010 Figure 1b) and high transcript levels persisted after oxidative stress recovery (Figure 1b). Modulation of p21 protein level followed the same pattern ( Figure 1c). These results suggested deregulation of CDKN1A expression. However, as proteasome-mediated degradation of p21 is known to contribute to modulation of protein levels during DDR (Cazzalini et al., 2010), we wanted to test the possibility that proteasomal degradation of p21 could be impaired in HGPS. The same extent of proteasome-mediated proteolysis was observed during stress recovery in control and HGPS cells, as determined by measuring protein accumulation upon MG132 treatment (Supporting Information Figure S2a). On the other hand, we did not observe any autophagic degradation of p21 during oxidative stress recovery neither in controls nor in HGPS cells, as determined by chloroquine treatment (Supporting Information Figure S2b). We concluded that p21 accumulation in HGPS cells during recovery from oxidative stress was mainly due to increase in CDKN1A transcripts.
This could also involve a p53-dependent mechanism, as p53 is a major player in stress response and regulator of p21 expression.
However, by analyzing the stress-response transcriptome, we found that BRCA1, TGF beta 1, 2, and 3, SMAD1, and interferon beta 1, all genes affecting p53 activity, were not dysregulated in HGPS cells (Supporting Information Figure S3). Moreover, while levels of phospho-p53 (Serine 15, pp53) were elevated under basal conditions in HGPS fibroblasts, both p53 and pp53 dynamics during stress were comparable to age-and passage-matched controls ( Figure 1c). Thus, p21 increase in HGPS cells subjected to oxidative stress appeared also due to a p53-independent dysregulation of CDKN1A expression.

| Regulation of stress response and p21 by lamin A/C
However, within the small group of genes showing an altered transcriptional response to oxidative stress in HGPS cells (Table 2), the majority were interconnected in the p53-p21 pathway ( Figure 2a).
In support of a role of lamin A/C in p21 regulation, we were able to detect lamin A/C binding to CDKN1A promoter ( Figure 2b). Specificity of lamin A/C binding was demonstrated, as an unrelated antibody, anti-NF-YA, did not bind the same region. As negative control, we used CXCR4 promoter that is bound neither by lamin A/C nor by NF-YA ( Figure 2b). Lamin A/C binding was detected in a region spanning 1,450 bp upstream the TSS, which includes an HDAC2 binding site ( Figure 2b) (Peng et al., 2015). Consistent with a major involvement of lamin A/C and HDAC2 in p21 regulation, we observed increase in p21 protein levels in HDAC2-depleted as well as in lamin A/C-depleted control fibroblasts ( Figure 2c). Interestingly, increased lamin A/C binding to CDKN1A promoter was observed in HGPS cells, but the interaction between the promoter and HDAC2 was significantly reduced (Figure 2d). The latter result suggested that loss of lamin A/C-HDAC2 interaction in HGPS could affect HDAC2 recruitment to the p21 promoter.

| Lamin A/C interacts with HDAC2, and binding is decreased in progeroid cells
Thus, we decided to investigate the interplay between HDAC2

| Lamin A/C activates HDAC2 more strongly than progerin
Then, we examined lamin A/C and HDAC2 interplay with the HDAC2 substrates H4 histone acetylated on lysine 16 (acH4K16) and H3 histone acetylated on lysine 9 (acH3K9) in control and HGPS cells. As expected, HDAC2 bound acH4K16 ( Figure 4a). However, the interaction was significantly reduced in HGPS, although acH4K16 levels were increased ( Figure 4a). As a control, we used the HDAC2 inhibitor MS275 (Panella et al., 2016) that significantly reduced HDAC2-acH4K16 binding, while increasing H4K16 acetylation (Figure 4a). Importantly, the HDAC2-lamin A/C-containing platform also included HDAC2 substrates. In fact, an interaction of lamin A/C with acH4K16 ( Figure 4b) and acH3K9 ( Figure 4c) was observed by PLA. In T A B L E 2 Genes differently regulated upon stress in HGPS cells. Changes in gene expression in control and HGPS cells subjected to 4-hr H 2 O 2 treatment are reported. A ratio >1.8 or a ratio <0.55 in control or HGPS was used in the study   Figure S5).
In support of the hypothesis that lamin A/C could affect HDAC2 activity, we were able to determine an interaction between lamin A/C and the phosphorylated form of HDAC2 (serine 394, pHDAC2), which was enriched at the nuclear periphery in 33% of examined nuclei (Figure 4d). Lamin A/C interaction with pHDAC2 was significantly F I G U R E 2 Regulation of stress response and p21 by lamin A/C. (a) String map (https://string-db.org/) indicating interconnections among genes analyzed in the microarray reported in Table 2. Genes upregulated in HGPS after H 2 O 2 treatment with respect to H 2 O 2 -treated human normal fibroblasts are indicated by an arrow, and genes whose regulation after H 2 O 2 treatment is hampered in HGPS are indicated by the symbol ┤. (b) (i) Chromatin immunoprecipitation (ChIP) of lamin A/C on the CDKN1A promoter in normal fibroblasts. The promoter of CDKN1A and CXCR4 (unrelated promoter) was detected by qPCR with specific primers listed in Experimental procedures. Protein binding is expressed as the percentage of the total DNA input. HDAC2 phosphorylation did not influence binding affinity between the lamin A/C and HDAC2 (Figure 4f). In fact, phosphomimetic or nonphosphorylable forms of HDAC2 were equally recovered in lamin A/ C-containing immunocomplexes (Figure 4f). Importantly, data obtained in an experimental model, HEK293 cells overexpressing LMNA mutants, suggested that wild-type lamin A promotes HDAC2 activity toward both acH4K16 and acH3K9, while progerin fails to properly regulate histone acetylation (Figure 4g).

| Lamin A/C-dependent oxidative stress response and recovery is impaired in HGPS
Then, we set out to investigate the fate of lamin A/C-HDAC2 com- These results demonstrated that defects in modulation of lamin A/C-HDAC2 interaction alter heterochromatic H3K9 and H4K16 histone acetylation pattern and oxidative stress recovery in HGPS cells.

| DISCUSSION
The main achievement of this study is the characterization of a functional interplay between lamin A/C and HDAC2 in human fibroblasts, which is modulated during DDR and contributes to regulation of HDAC2 activity and p21 expression. Importantly, we show that the protein platform, which, also includes HDAC2 substrates acetylated H3K9 and H4K16, is disrupted in HGPS, where CDKN1A downregulation upon stress recovery is affected leading to cellular senescence.
The observation that p21 modulation during oxidative stress response is altered in HGPS cells, both at the mRNA and protein level, suggested that functional lamin A/C was required for CDKN1A regulation. Our data show that in fact lamin A/C binds the CDKN1A promoter and the interaction between lamin A/C and HDAC2 favors HDAC2 recruitment (Noh et al., 2011), as suggested by reduced deacetylase binding in HGPS, despite increased protein levels.  expressing Y259D mutated LMNA (Mattioli et al., 2011) or control fibroblasts were included in this study (Table 1).

| Plasmids and siRNA
Plasmids used for transfections: GFP-lamin A, FLAG-lamin A, FLAGprogerin or wild-type FLAG-HDAC2, nonphosphorylable FLAG-HDAC2 S394A, and phosphomimetic FLAG-HDAC2 S394D (Peng et al., 2015). Transfection of HEK293 cells was performed using To define a statistically significant variation in expression level, we considered a F I G U R E 3 Lamin A/C interacts with HDAC2 in human normal fibroblasts more strongly than progeroid mutants. (a) Co-IP of GFP-lamin A and FLAG-HDAC2 (IP FLAG) in HEK293 cells. IP control IgG, negative control. Molecular weight markers are indicated. (b) (i) Immunofluorescence staining of lamin A/C and HDAC2 and PLA of lamin A/C and HDAC2 (PLA) in human normal fibroblasts. PLA of lamin A/ C and HDAC2 and lamin A/C staining are merged in the right picture (merge); (ii) percentage of nuclei showing less than 50% of signals at the periphery (PLA lamin A/C-HDAC2 homogenous distribution) or more than 50% of signals at the periphery (PLA lamin A/C-HDAC2 enrichment at the periphery); (iii) fluorescence intensity profile of lamin A/C and HDAC2 in a representative nucleus. (c) (i) PLA of lamin A/C and HDAC2 in the presence (left picture) or absence of lamin A/C antibody (no anti-lamin A/C antibody) and (ii) quantitative analysis; (iii) PLA of lamin A/C and HDAC2 after HDAC2 knockdown (HDAC2 siRNA); (iv) PLA of lamin A/C and HDAC2 after lamin A/C knockdown (lamin A/C siRNA) and (v) quantitative analysis of PLA signals in the indicated samples; (vi) lamin A/C and MEF2c staining and PLA of lamin A/C and MEF2c (no signals were detected). (d) (i) PLA of lamin A/C and HDAC2 in fibroblasts from healthy subjects, HGPS, APS, MADA, or EDMD2 patients (see Table 1 for details) and (

| Antibodies and drugs
Antibodies used in this study are listed in Table 3. Anti-lamin A/C and anti-prelamin A antibodies used in this study have been previously characterized in control and HGPS cells Columbaro et al., 2005;Lattanzi et al., 2014

| Statistical analysis
Graphs in each panel represent mean values from at least three independent experiments ± standard error for PLA assay and IF or ± standard deviation for WB and PCR. Statistically significant differences (p < 0.05) are calculated by Student's t test.

ACKNOWLEDG MENTS
The

CONFLI CTS OF INTEREST
None declared.