The Co-Repressor SMRT Delays DNA Damage-Induced Caspase Activation by Repressing Pro-Apoptotic Genes and Modulating the Dynamics of Checkpoint Kinase 2 Activation

Checkpoint kinase 2 (Chk2) is a major regulator of DNA damage response and can induce alternative cellular responses: cell cycle arrest and DNA repair or programmed cell death. Here, we report the identification of a new role of Chk2 in transcriptional regulation that also contributes to modulating the balance between survival and apoptosis following DNA damage. We found that Chk2 interacts with members of the NCoR/SMRT transcriptional co-regulator complexes and serves as a functional component of the repressor complex, being required for recruitment of SMRT on the promoter of pro-apoptotic genes upon DNA damage. Thus, the co-repressor SMRT exerts a critical protective action against genotoxic stress-induced caspase activation, repressing a functionally important cohort of pro-apoptotic genes. Amongst them, SMRT is responsible for basal repression of Wip1, a phosphatase that de-phosphorylates and inactivates Chk2, thus affecting a feedback loop responsible for licensing the correct timing of Chk2 activation and the proper execution of the DNA repair process.


Introduction
The induction of DNA damage by environmental carcinogens induces a very complex response aimed at repairing the alterations in the DNA structure and sequence [1,2]. The detection of base modifications or strand breaks by sensor proteins induces the activation of a checkpoint response that allows the cell to block proliferation and to repair DNA [3][4][5]. If the damage is too extensive to be reversed, checkpoint proteins induce a switch from repair to programmed cell death, thereby preventing the transmission of mutations to the next cellular generations [6]. Caspase activation is a central event in the induction of apoptosis after DNA damage as well as other stimuli [6]. The mechanisms by which a cell discriminates between reparable damage and lethal genotoxic stress leading to caspase activation and apoptosis are not well understood.
Checkpoint kinase 2 (Chk2) is a central DNA damage checkpoint regulator [7,8], responding mostly to stimuli that cause double strand breaks, such as ionizing radiation or topoisomerase II inhibitors such as doxorubicin. The kinase ATM is involved in sensing the damage and initiating the checkpoint response [9]. Among other substrates, ATM phosphorylates Chk2 on threonine 68 [10,11], inducing dimerization of Chk2 and auto-phosphory-lation of multiple sites, leading to full activation [12][13][14][15]. Chk2 activation results in cell cycle arrest [16] and activation of the DNA repair process [17,18]. Finally, if the damage is too extensive to be repaired, Chk2 is able to induce apoptosis, through phosphorylation of the pro-apoptotic transcription factors p53 [19,20] and E2F1 [21].
It has been shown that the ATM-Chk2-p53 system is activated with an oscillatory pattern, as a consequence of the equilibrium between activating stimulus (DNA damage) and a double feedback system activated by p53 and including Mdm2, which induces p53 degradation [22], and Wip1, a protein phosphatase that dephosphorylates both ATM and Chk2 [23]. The oscillations are considered a mechanism to check on the progress of the DNA damage and shut down the checkpoint response if the repair has been successful [24]. It has also been proposed that the oscillatory activation of p53 provides a means to fine-tune the cellular response and produce different outcomes according to the severity of the DNA damage [25].
Besides p53, other transcription factors play different roles in modulating the apoptotic response to DNA damage. Among these, NFkB and AP1 have a fundamental function. NFkB shows preeminent pro-survival actions in non-lymphoid tissues [26][27][28][29]; however, it has been shown that DNA damaging stimuli such as UV-C and doxorubicin/daunorubicin induce a switch of NFkB action from activator to repressor on the anti-apoptotic genes X-IAP, Bcl-X(L) and survivin, thus contributing to the induction of programmed cell death [30].
AP1 has been associated with regulation of apoptosis in response to a wide variety of stimuli, such as growth factor deprivation and environmental stresses, including DNA damage [31,32]. The pro-apoptotic action is mediated by the transcriptional activation of FasL, that triggers the extrinsic apoptotic pathway [33], as well as Bim [34]. AP1-mediated repression also plays an important role in regulating the apoptotic switch: proteins of the AP-1 family are able to repress pro-apoptotic genes such as p53 [35,36] and Fas [37]; therefore, the final outcome depends on the balance between gene activation and repression, and is highly influenced by several considerations including cell type, conditions of growth, and the presence of growth factors [38]. It is generally accepted that DNA damage activates initially both pro-apoptotic and anti-apoptotic pathways, each one subjected to subtle and complex regulation by different stimuli and transcription factors, and that the overall balance between genes promoting survival and genes inducing death signals dictates the cellular outcome [31].
Transcriptional co-repressors, recruited by transcription factors bound on DNA regulatory elements, act as a platform for further recruitment of repressive proteins including histone deacetylases or methyltransferases, which locally modify the structure of chromatin in a way that restricts access to activator complexes and RNA polymerases [39]. The core NCoR/SMRT co-repressor complexes contain the histone deacetylase HDAC3 [40], the exchange factors TBL1 and TBLR1 [41][42][43], and the G-protein pathway suppressor, GPS2 [44]. The NCoR/SMRT complexes are required for basal repression of genes by unliganded nuclear receptors, such as thyroid hormone receptor (TR) and retinoic acid receptor (RAR) [43], or for active repression mediated by ligand-bound receptors, such as tamoxifen-bound estrogen receptor [45]. NCoR/SMRT have also been associated with repression by a number of other transcription factors, including the regulators of apoptosis NFkB [46] and AP1 [47].
Although sharing significant sequence and structure homology and the interaction with members of the respective co-repressor complexes [43,48,49], NCoR and SMRT have been shown to play distinct and specific roles in differentiation and development [50][51][52], and to be subjected to different regulatory pathways [51,53,54].
Here we report the finding that, upon DNA damage, Chk2 is required to specifically recruit the co-repressor SMRT to repress pro-apoptotic genes, including AP1 and NFkB targets, thus keeping the apoptotic response on hold, while the cells repair the damage. In particular, SMRT-dependent repression of the protein phosphatase Wip1, a major regulator of checkpoint response that de-phosphorylates Chk2, ATM, p53 and p38 [55], is fundamental for caspase activation after DNA damage by modulating the oscillatory frequency of Chk2 activation.

Chk2 interacts with SMRT and regulates different transcription factors, acting as a repressor
To gain insight into the role of the checkpoint kinase Chk2 in the regulation of the apoptotic switch after DNA damage [19,56,57], we first sought to identify the protein complex associated with Chk2. To this end the protein interaction domain (FHA domain) of human Chk2 was expressed in bacteria and the purified protein incubated with HeLa cell extract. This approach identified, amongst others, TBLR1, a component of the NCoR/ SMRT co-repressor complexes, as a Chk2 interacting protein ( Figure S1A-B). To confirm the interaction, flag-tagged full-length human Chk2 was expressed both in HeLa (data not shown) and U2OS cells and anti-Flag antibody was used to immunoprecipitate associated proteins, followed by Western blot analysis with antibodies specific for members of the NCoR/SMRT complexes (NCoR, SMRT, TBL1, TBLR1). This approach confirmed the interaction of Chk2 with SMRT, TBL1 and TBLR1, but not with NCoR ( Figure S1C). When the same experiment was performed using flag-tagged Chk1 for comparison, no interaction was detected.
To initially explore if Chk2 had an effect on transcription, an array of luciferase expression vectors containing binding sites for different transcription factors (AP1, NFkB and RAR), which use NCoR/SMRT as co-repressors, were used to perform standard reporter assays. Knock-down of Chk2 in U2OS cells potentiated the transcriptional response to AP1 and RAR-driven reporters ( Figure S1D), suggesting a repressive function for Chk2.
SMRT, but not NCoR, significantly affects DNA damageinduced transcription Because Chk2 is a major regulator of DNA damage checkpoint response, we sought to determine potential roles of the NCoR/ SMRT co-repressor complex in the transcriptional response to DNA damage. Osteosarcoma cells U2OS were chosen because they have a functional p53 and represent a good model for studying DNA damage checkpoint. A time-course experiment in U2OS cells showed that the genotoxic drug cisplatin (CDDP) at the concentration of 100 mM caused phosphorylation of Chk2 on T68 starting at 4 h and reaching a peak at 6 h ( Figure S2). The peak of Chk2 activation was followed by caspase activation reaching a plateau at 8 h ( Figure S2) Caspase activation was evaluated through an antibody that specifically recognizes the PARP fragment after the cleavage by caspase 3 on Asp214, while not interacting with the full-length protein or with other PARP degradation fragments ( Figure S2).
To study the role of NCoR and SMRT in CDDP-induced gene transcription, U2OS cells were transfected with siRNAs against SMRT or NCoR, and 48 hours later treated with 100 mM CDDP for 6 h to induce damage and activate Chk2. RNA was extracted, labeled and hybridized on Illumina BeadChip microarrays. To control the efficiency of knock-down, RT-qPCR was performed with primers specific for NCoR or SMRT. As shown in Figure 1A, the siRNAs against NCoR and SMRT caused a reduction in the respective mRNA levels of 71% and 65%. The microarray profiling showed that treatment with CDDP induced a massive wave of gene repression, with negatively-regulated genes (1074) being twice as frequent as positively-regulated (507) ones ( Figure 1B and Table S1), suggesting that transcriptional repression is an important aspect of the response to CDDP-induced DNA damage. Gene Ontology analysis of CDDP-regulated genes showed enrichment, among the most significant biological processes, of negative regulation of transcription, cell cycle, apoptosis, and cell death ( Table 1). The reduction of NCoR levels did not significantly affect the CDDP-induced transcriptional program, with only 36 genes being significantly regulated by CDDP differently in the NCoR knock-down compared to the scramble siRNA-transfected cells (FDR $0.2, Table S2). Conversely, knockdown of SMRT resulted in significant changes in the transcriptional program triggered by CDDP ( Figure 1C and Table S3). Among the CDDP-repressed genes, 186 (16%) were no longer repressed or were significantly less repressed with SMRT knockeddown ( Figure 1C, upper panel and Class 1 in Table S3), while 99 genes (9%) where repressed more intensely in the absence of SMRT. Among the CDDP-activated genes, 37 (7.8%) were activated more intensely in the SMRT knock-down cells ( Figure 1C, lower panel), indicating that this co-repressor limits the activation of these genes after CDDP treatment, while 94 (19.9%) were not activated or activated less intensely when SMRT was knocked down. Moreover, 17% of the genes activated by treatment with CDDP were also activated by SMRT knock-down in the absence of treatment (Class 2 in Table S3), suggesting a basal repression by SMRT, which is removed by treatment with CDDP. Figure 1D reports a heat map of a selected group of genes regulated by CDDP differently in the SMRT knock-down compared to the NCoR knock-down, showing how the profile in the NCoR siRNA resembled the profile in the scramble siRNA-transfected cells, while the SMRT siRNA stood out for both activation and repression of genes. Interestingly, in the group of genes that were differentially regulated by CDDP in the cells where SMRT was knocked down compared to scramble siRNA-transfected cells, some of the most enriched GO terms were cell death and apoptosis, along with protein amino acid phosphorylation ( Table 2).

SMRT protects against apoptosis through repression of pro-apoptotic genes
Because apoptosis was among the most enriched Gene Ontology (GO) terms in SMRT-dependent genes, we selected a group of pro-apoptotic genes regulated by CDDP (FOS, PPM1D, SMAD7, SRPK2, BCL2L11, MAX), and performed RT-qPCR after transfection of the cells with siRNA against SMRT and subsequent treatment with CDDP. As shown in Figure 2A, SMRT limited CDDP-dependent activation of FOS and PPM1D and exhibited a repressive function on SMAD7, SRPK2, BCL2L11 and MAX, as their inhibition by CDDP was lost in the cells transfected with SMRT siRNA. Because AP1 is a transcription factor implicated in induction of apoptosis, itself utilizing the NCoR/SMRT complex for repression of target genes, the regulation of c-FOS, a major component of the AP1 family, was selected for further investigation. Chromatin immunoprecipitation of SMRT following CDDP treatment revealed increased recruitment of SMRT on the c-FOS promoter after DNA damage ( Figure 2B), suggesting direct regulation of c-FOS by SMRT. Knock-down of Chk2 was able to abrogate the occupancy of SMRT on the c-FOS promoter, supporting the requirement for Chk2 in DNA damage-dependent co-repressor recruitment ( Figure 2C).
In order to investigate the biological consequences of the Chk2-SMRT regulatory events, U2OS and 293 cells were transfected with siRNAs against Chk2 or SMRT, followed by treatment with toxic doses of CDDP and Western blot analysis of the caspase 3dependent cleavage of PARP1. As shown in Figure 3A, while the siRNAs against Chk2 had little effect on the activation of caspase 3, knock-down of SMRT increased PARP cleavage in CDDP-treated cells. The knock-down of SMRT caused a very slight increase in PARP cleavage also in non-treated cells, only visible after very long exposure ( Figure 3B). To confirm activation of caspase 3 by SMRT siRNA, a Western blot was performed on U2OS protein extracts with an antibody which specifically recognized the 17-kDa and 19-kDa cleavage products of caspase 3, showing increased caspase 3 cleavage when SMRT was knocked-down ( Figure 3C). This effect was not detected in cells transfected with the same siRNAs, but treated with TNF-a in combination with cycloheximide, which has been shown to activate the extrinsic apoptotic pathway by the TNF receptor [58] ( Figure 3D-E). Interestingly, SMRT was required for CDDP-induced repression of the Wip1 phosphatase (PPM1D gene, Figure 2A), a major down-regulator of the Chk2 pathway, as well as of other phosphorylated proteins such as p53 and p38 [55]. To confirm the repressive action of SMRT on PPM1D, U2OS cells were transfected with siRNA against SMRT and treated with another genotoxic drug, doxorubicin, which inhibits topoisomerase II and causes double strand breaks. As expected, Wip1 protein levels were augmented by doxorubicin treatment and increased by siRNA against SMRT, even in the absence of DNA damage, indicating basal repression of PPM1D by SMRT ( Figure 4A).
We next investigated the role of PPM1D repression in the initial caspase activation induced by SMRT knock-down in the presence of DNA damage. We first treated the cells with doxorubicin for different time points, and detected cleavage of PARP starting between 5 h and 6 h of treatment ( Figure 4B). While knock-down of SMRT caused an increase in doxorubicin-induced caspase 3 activation, co-treatment with a chemical inhibitor of Wip1 attenuated this phenotype ( Figure 4C). Consistently, co-transfection of siRNA against PPM1D reduced caspase activation when compared to actions of SMRT siRNA alone ( Figure 4D, compare lane 4 and 2).
Wip1 has many substrates, all involved in the regulation of apoptosis in response to DNA damage or cellular stresses. To determine whether Chk2 is required for the activation of caspase that occurs when SMRT is down-regulated, we co-transfected U2OS cells with siRNAs for SMRT and Chk2. Indeed, cotransfection with Chk2 siRNA attenuated the caspase activation caused by SMRT knock-down, after treatment with doxorubicin ( Figure 4D, compare lanes 6 and 2).
are either activated or repressed by CDDP, using a log 2 fold change cutoff of 0.585. C) The two histograms report the number of genes whose regulation by CDDP is affected by SMRT knock-down, grouped by CDDP repression (upper panel) or activation (lower panel). D) Genes regulated by CDDP differently in the SMRT knock-down compared to the NCoR knock-down. The criteria for selection of SMRT-specific genes were the following: false discovery rate $0.2; log 2 fold change (CDDP/vehicle) in scramble siRNA cells $|0.32|; difference between log 2 fold changes in the SMRT knock-down and the scramble siRNA [(siSMRT+CDDP/ siSMRT+vehicle) -(siCtl+CDDP/siCtl+vehicle)] $0.32; difference between log 2 fold changes in the SMRT knock-down and the NCoR knockdown [(siSMRT+CDDP/siSMRT+vehicle) -(siNCoR+CDDP/siNCoR+ vehicle)] $0.32. Genes were grouped with the ''Cluster'' software (using the ''Average linkage clustering'' function) and the heat maps were created with the ''TreeView'' software. doi:10.1371/journal.pone.0059986.g001  Because the Wip1-dependent feedback has been reported to cause oscillatory behavior in Chk2 activation [24,59], and because in genetic oscillators the intensity of the feedback loop dictates the oscillatory behavior of the system [60][61][62], we sought to examine the possibility that the SMRT-dependent repression of Wip1 might be responsible for influencing the dynamics of Chk2 activation after DNA damage. U2OS cells were treated with doxorubicin at 109 intervals for 2 hours, comparing cells transfected with scramble siRNA or siRNA against SMRT. In control conditions, an oscillatory pattern of Chk2 phosphorylation was observed, with two discrete peaks at 409 and 809. SiRNA against SMRT altered this pattern, resulting in only one, more persistent phospho-Chk2 peak ( Figure 4E). Interestingly, when the cells transfected with the   SMRT siRNA were also treated with a Wip1 inhibitor, the single peak of Chk2 activation ( Figure 4F, upper panel) was replaced by two peaks of activation after doxorubicin treatment, one at 409-509 and one at 709 ( Figure 4F, lower panel). Even if the peaks are not perfectly overlapping with the original oscillation observed in the wild type cells, this result suggests that the lack of inhibition of Wip1 might play a role in the altered dynamics observed in the SMRT knock-down.

SMRT is required for proper repair of DNA damage
It has been argued that the cycles of oscillation in checkpoint proteins allow cells to monitor DNA repair and re-activate the checkpoint response if the damage is still present at the end of each cycle [24,63]. To assess the applicability of this model to Chk2, we treated U2OS cells with doxorubicin with or without co-treatment with the antioxidant N-Acetyl-Cysteine (NAC), which reduces the amount of DNA damage by scavenging free radicals. As expected, treatment with the antioxidant reduced caspase activation in response to doxorubicin. However, if the cells were transfected with siRNA specific for SMRT, the ability of the antioxidant to rescue the DNA damage-induced apoptosis was lost ( Figure 5A), suggesting that SMRT is involved in monitoring repair and shutting down the response if the DNA damage is repaired. Interestingly, co-treatment with the Wip1 inhibitor partially restored the ability of the SMRT-deficient cells to block apoptosis in the presence of NAC ( Figure 5A, lane 5), suggesting a role of Wip1 de-repression in the observed phenomenon.
Finally, we argued that if SMRT is required for properly sensing and responding to DNA damage, then SMRT knock-down should affect the ability of the cells to repair DNA damage. As expected, when SMRT was down-regulated by siRNA, the number of cH2AX foci, a marker of active repair of DNA double strand breaks, was doubled ( Figure 5B-C).

Discussion
The decision between life and death of a cell is a very important one, with relevant implications for cancer therapy. Here we present a role of the transcriptional co-repressor SMRT in delaying the induction of apoptosis after DNA damage. We suggest that Checkpoint kinase 2, activated by DNA damage, induces recruitment of SMRT on the promoters of some proapoptotic genes, thus delaying activation of caspase and allowing DNA repair ( Figure 5D, left panel). When SMRT is removed from the cells, pro-apoptotic genes are de-repressed and cells activate the caspase pathway ( Figure 5D, right panel). Our data showed that SMRT is down-regulated both at the RNA level ( Figure 1A) and at the protein level (data not shown) after treatment with apoptotic doses of CDDP.
The RNA profiling results suggest that SMRT plays a significant role in gene repression after DNA damage. However, SMRT actually exhibited an activating action on some CDDPregulated genes, possibly due to secondary effects. The genes that we confirmed by RT-qPCR to be inhibited by SMRT are all important regulators of apoptosis, belonging to different pathways. BCL2L11 (or Bim) is a member of the ''BH3-only'' family of apoptosis facilitators; members of this family bind and inactivate pro-survival Bcl2-like proteins, thus allowing permeabilization of mitochondrial membrane by Bax/Bak channels [64]. SRPK2 is a protein Serine-Arginine kinase that, after DNA damage, migrates to the nucleus and induces apoptosis through only partially defined mechanism [65]. MAX forms a functional dimer with MYC and binds the promoter of target genes thus inducing cell proliferation as well as apoptosis [66,67]. Of particular interest, among the proapoptotic genes repressed by SMRT there are some components or regulators of two major pathways involved in the regulation of apoptosis: AP1 and NFkB. FOS is a major transcription factor of the AP1 family [68], and has been associated with induction of apoptosis after different cellular stresses, including DNA damage [31,32]. SMAD7 is an inhibitory transcription factor activated by TGF-b signaling, required for repression of anti-apoptotic NFkB target genes [69]. The SMRT complex is already known to be required for gene repression by AP1 [47] and NFkB [46]. In our experiments, SMRT exerts a basal repressive function on NFkB genes, as its removal from the cells by siRNA induces upregulation of NFkB targets such as PPM1D (Figure 2A) as well as IL-8 and TNF-a (data not shown). In contrast, active recruitment of SMRT on the promoter of AP1 targets such as c-FOS occurs only after DNA damage.
As expected, the SMRT knock-down sensitizes cells to DNA damaging agents such as CDDP and doxorubicin, increasing drug-induced caspase activation. An apparently inconsistent finding is that the Chk2 knock-down does not have the same effect. Although we were not able to demonstrate direct phosphorylation of SMRT by Chk2, our data suggest that phosphorylation of SMRT occurs after treatment with CDDP, which is reduced, but not entirely abolished, by Chk2 siRNA (data not shown). Therefore, we are tempted to suggest that SMRT might be phosphorylated by different kinases after DNA damage, such as Chk1, p38, and JNK, which can complement the absence of Chk2.
Of particular interest to us is the SMRT-dependent repression of the protein phosphatase Wip1 or PPM1D, a down-regulator of the Chk2 pathway. De-repression of Wip1 by SMRT knock-down is associated with increased caspase activation. It is tempting to speculate about potential explanations of these observations. One hypothesis is that Wip1 de-repression changes the kinetics of Chk2 activation. Chk2 has been reported to be activated in cycles [24]. It has been shown that the frequency of oscillations of a negative feedback system is dictated by the balance between the activating stimulus and the negative feedback [60,62]. Therefore, Chk2 oscillations might change in frequency as the system evolves: initially, when Wip1 levels are low, one observes high-frequency oscillations (409 period); however, as Wip1 levels increase due to p53-dependent transcriptional activation (mimicked by SMRT knock-down), one might expect lower frequency oscillations and, therefore, more persistent phosphorylation of Chk2 and activation of caspases. While more detailed kinetic analysis of Chk2 activation would be required to fully characterize the system's evolution during time, the idea of frequency modulation as a strategy to induce different cellular outcomes is a possible explanation.
Indeed, the idea of alteration in activation dynamics as a regulatory strategy has been previously proposed. The ATM-Chk2-p53 pathway has been shown to be activated in cycles after IR treatment [24]; it has been suggested that the cycles of p53 activation are required to sense the intensity of the DNA damage [59] and to modulate the response according to the severity of the damage [70]. Moreover, the kinetics of activation of JNK-1 has been proposed to dictate the cellular outcome, with transient activation being pro-inflammatory and pro-survival, while sustained activation inducing cell death [71].
The mechanism of the inhibition of PPM1D gene by SMRT is not clear. PPM1D has been reported to be regulated by NFkB [72], which uses SMRT as a co-repressor. However, we were not able to detect any SMRT protein on the NFkB binding site in the PPM1D promoter (data not shown). It is possible that the effect on Wip1 is indirect, mediated by increased autocrine secretion of TNF-a by the SMRT knock-down cells. Indeed, treatment of the cells with exogenous TNF-a has an effect on caspase activation similar to that of siRNA for SMRT: increased doxorubicin-induced apoptosis, which can be rescued by siRNA for Wip1 (data not shown).
Finally, Wip1 regulates many different kinases [55] and, therefore, its regulation has most likely wider consequences than only those dependent on Chk2 oscillation. We speculate that oscillatory behavior occurs simultaneously in many different systems and for many different regulatory proteins, wherever a negative feedback loop is present, and the final ''decision'' on cell fate depends on the balance between the signals sent by each of the regulatory checkpoint proteins.
SiRNAs specific for Chk2, NCoR, SMRT and PPM1D were purchased from Invitrogen, and a mix of three independent siRNAs was used for each gene.

Cell culture
All cells were grown in Dulbecco Modified Eagle Medium (GIBCO) with low glucose, supplemented with 10% FBS and 50 mg/ml Gentamycin. For transfection, Lipofectamine 2000based liposomes were incubated in serum-free medium (5 ml/ plate) and Optimem medium (3 ml/plate) for 6 hours, followed by incubation in complete medium for 2 days before the appropriate treatments. Cells were incubated for 2 days in serum-free DMEM medium before treatments with 9cisRA, TNF-a or TPA.

Reverse Transcription-qPCR
Total RNA was extracted with RNEasy kit (QIAGEN) according to the manufacturer's recommendations, and 1 mg of RNA was used for RT reaction by SuperScript III Reverse Transcriptase kit (Invitrogen); cDNA samples were diluted 1:20 for test genes and 1:200,000 for the reference gene (18S), and 4 ml per reaction were used for qPCR, using Brilliant III SYBR Green master mix (Agilent). Results from three technical replicates were analyzed by the DDCt method, using 18S as a normalizer. Briefly, statistical significance was calculated by normalizing the values of three independent biological replicates for the respective 18S values, eliminating the outliers and averaging the DCts; the pvalue was calculated by Student's T-test.

Microarrays
U2OS cells were transfected with siRNAs for NCoR or SMRT, incubated in complete medium for 2 days, and then treated with 100 mM CDDP for 6 hrs, followed by RNA extraction, labeling with Illumina TotalPrep RNA Amplification Kit (Ambion) and hybridization on Illumina Expression Beadchips (HumanRef 8.0), following the manufacturer's recommendations. Data was analyzed as described in Methods S1. The data was uploaded on the GEO database (ID number: GSE34226).

Cell extracts and fractionation
For whole cell extracts, Cells were washed in cold PBS, harvested and resuspended in IPH buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 5 mM EDTA, 0.5% NP-40, 1 mM PMSF, 50 mM NaF, 2 mM Na 2 VO 3 , 1X Complete protease inhibitor mix [Roche] and 1 mM glycerol phosphate), incubated for 209 on ice and subsequently centrifuged for 109 at 13,000 rpm. The supernatant (whole cell extract was used for Western blot analysis. Immunofluorescence U2OS cells were plated at 80% confluence in 8-well plates, transfected as indicated with Lipofectamine 2000 (Invitrogen) and then treated with 5 mM doxorubicin for 1 h. Cells were fixed in 2% paraformaldehyde for 109 at room temperature and then washed in PBS 4 times (209 each), followed by permeabilization in 0.1% NP-40 in PBS (309 at room temperature) and incubation in PGBA-Super blocking solution (0.1% gelatin, 10% BSA, 0.01% sodium azide, 1% normal goat serum in PBS) for 309 at room temperature. Primary antibody anti-cH2AX was diluted 1:200 in PGBA solution (0.1% gelatin, 1% BSA, 0.01% sodium azide in PBS) and added to the cells over-night. Cells were washed 4 times in PBS and then incubated for 1 h with secondary antibody antimouse (1:300 in PGBA) conjugated with Alexa-488 (Invitrogen). Cells were washed 3 times in PBS and then analyzed at the fluorescence microscope at 20X magnification. Five pictures were taken for each experimental point. cH2AX foci were counted with ImageJ, using the FociPicker3D application, followed by counting the number of cells for each picture with the 'Analyze Particles' function, and the number was used to calculate the average number of foci per cell. Data from three independent experiments were used to calculate statistical significance, with Student's T-test. Figure S1 Purification of proteins interacting with the FHA of Chk2. A) A PATH-tagged FHA domain of Chk2 was purified from bacteria and incubated with cellular extracts from HeLa cells. As a control, a mutant lacking the ability to bind target phosphopeptides was used. Purified proteins were run on polyacrylamide gels and visualized with silver stain. B) Proteins identified to specifically interact with the FHA domain of CHK2. C) U2OS cells were transfected with Flag-tagged Chk1 or Chk2 expression vectors, and protein extracts were immunoprecipitated with anti-Flag antibody, followed by SDS-PAGE and Western Blot analysis. D) U2OS cells were transfected with the indicated reporters with or without siRNA against Chk2, incubated for 2 days in serum-free medium and then treated with 10 ng/ml TPA, 20 ng/ml TNF-a or 5610 28 M 9cisRA, as indicated. Statistical significance was calculated on the ratio between luciferase luminescence and Renilla luciferase from three independent experiments, each one including four technical replicates, by Student's T-test. (One star: p-value#0.05, two stars: p-val-ue#0.01, three stars: p-value#0.001). (TIF) Figure S2 Time-course of cisplatin (CDDP) treatment. U2OS cells were treated with 100 mM CDDP for the indicated time points and the whole cell extracts were used for Western blot with specific antibodies against phospho-Chk2 (T68), cleaved PARP (Asp214), or b-tubulin.

(TIF)
Table S1 Genes regulated by cisplatin. The table reports the total number of genes resulting regulated by CDDP in the microarrays experiment, ranked by significance as described in Methods S1. (XLS) Table S2 Genes whose regulation by cisplatin is affected by NCoR. The table reports the total list of genes whose regulation by CDDP in the microarray experiments was different in the cells transfected with NCoR siRNA compared to the cells transfected with scramble control siRNA.

(XLS)
Table S3 Genes whose regulation by cisplatin is affected by SMRT. The table reports the total list of genes whose regulation by CDDP in the microarray experiments was different in the cells transfected with SMRT siRNA compared to the cells transfected with scramble control siRNA. (XLS) Methods S1 Methods for the data included in Supporting Figures and detailed descriptions of microarray data analysis, plasmids and qPCR primers are included. (DOC)