Research ArticleBiochemistry

SUMOylation of the Transcriptional Co-Repressor KAP1 Is Regulated by the Serine and Threonine Phosphatase PP1

See allHide authors and affiliations

Science Signaling  27 Apr 2010:
Vol. 3, Issue 119, pp. ra32
DOI: 10.1126/scisignal.2000781

Abstract

Krüppel-associated box (KRAB) domain–associated protein 1 [KAP1, also known as transcription intermediary factor–1β (TIF1β)] is a ubiquitous transcriptional co-repressor that is susceptible to phosphorylation at Ser824 by ataxia-telangiectasia mutated (ATM) and to modification by small ubiquitin-like modifying (SUMO) proteins. Here, we found that, whereas the protein phosphatase 1α isoform (PP1α) directly interacted with KAP1 under basal conditions, PP1β interacted with KAP1 only in response to genotoxic stress. Changes in the abundance of PP1α or PP1β had differential effects on the phosphorylation and SUMOylation states of KAP1 under basal conditions and in response to DNA double-strand breaks (DSBs). Chromatin immunoprecipitation and re-immunoprecipitation experiments revealed that PP1α and PP1β were recruited to KAP1 with different kinetics before and after the induction of DNA DSBs, which provided a mechanistic basis for the switch in the phosphorylation and SUMOylation states of KAP1. PP1β-dependent SUMOylation of KAP1 occurred by mechanisms that were dependent and independent of the phosphorylation status of Ser824. We posit a mechanism whereby the combined actions of PP1α and PP1β cause dephosphorylation of KAP1 at Ser824 and assure its SUMOylation to counter the effect of ATM, thereby regulating the transcription of KAP1 target genes in unstressed and stressed cells.

Introduction

Krüppel-associated box (KRAB) domain–associated protein 1 [KAP1, also known as transcription intermediary factor–1β (TIFIβ) and tripartite motif–containing protein 28 (TRIM28)] is a well-characterized transcriptional co-repressor protein that is recruited to target genes by zinc finger and BRCA1-interacting protein with a KRAB domain 1 (ZBRK1), one of the KRAB zinc finger proteins (KRAB-ZFPs) (1). KAP1 connects ZBRK1, through protein-protein interactions, to the transcriptional repression machinery, such as the histone lysine N-methyltransferase SETDB1 [for Su(var), enhancer of zeste, trithorax (SET) domain bifurcated 1], which specifically methylates histone H3 at Lys9 (K9) (2). We previously reported that KAP1- and SETDB1-mediated transcriptional repression of the proarrest gene p21 is dependent on the dimethylation of histone H3 K9 at its proximal promoter in a small ubiquitin-like modifying (SUMO)–dependent manner (3). On the basis of its interactions with chromatin modification factors, such as histone deacetylase 1 (HDAC1), SETDB1, and heterochromatin protein 1 (HP1), KAP1 is proposed to regulate chromatin structure and heterochromatin formation so as to cause the epigenetic silencing of target genes (35).

DeSUMOylation of KAP1 is required to relieve its transcriptional co-repressor function at a subset of genes, such as p21, Bax, Puma, and Noxa, whose products are involved in cell cycle arrest and proapoptotic processes during the DNA damage response (DDR) (3, 6). In addition, KAP1 is phosphorylated at Ser824 by members of the phosphoinositide-3-kinase–related protein kinase (PIKK) family of kinases upon the induction of DNA double-strand breaks (DSBs) (3, 610). When KAP1 is phosphorylated at Ser824, it colocalizes with numerous DDR factors at DNA lesions (7, 9, 10). We further reported that the transcriptional co-repressor function of KAP1, which is mediated by SUMOylation, is inhibited by ataxia-telangiectasia mutated (ATM)–dependent phosphorylation of Ser824 in response to the induction of DNA DSBs (3, 6). In support of a role for ATM-mediated phosphorylation of Ser824 in mediating derepression of KAP1 target genes, Goodarzi and colleagues showed that depletion of KAP1 restored the process of DNA repair for compact chromatin in ATM-deficient cells (11); however, how the ATM-mediated effect on KAP1 function is inhibited remains ill-defined.

Protein serine and threonine phosphatases, including protein phosphatase 1 (PP1), PP2A, PP2B, PP4, PP5, PP6, and PP7, function by reversing the phosphorylation of key structural and regulatory proteins (12, 13). The PP1 catalytic subunit (PP1cs) exists in different isoforms (PP1α, PP1β, and PP1γ) that have distinct patterns of subcellular localization (1416). PP1 regulates a large number of cellular activities, including neurotransmission protein synthesis, muscle contraction, the DDR, and cell cycle progression (12, 13, 17). One critical question is how PP1 executes such pleiotropic effects at the right time and at the right location. Some studies have emphasized that the PP1cs is targeted to specific substrates by interacting with one of many regulatory subunits that also interact with the substrate. The consensus primary PP1-docking motif, [KR][X]0–1[VI]{P}[FW], is derived from most PP1-interacting proteins that associate with PP1 by binding to a hydrophobic groove at the opposite side of the catalytic center of PP1cs (18). Mutation or deletion of this motif often abolishes the binding between PP1cs and its interacting proteins. Several substrates of PP1 also contain this consensus PP1-docking motif (13, 19); however, to date, little is known about the role of PP1 in the regulation of KAP1.

Here, we identified PP1 as a specific stimulator of the SUMOylation of KAP1 to restore its co-repressor function. We showed that PP1α constitutively formed a complex with KAP1 through the PP1-binding site Lys366-Leu-Ile-Tyr-Phe370 (366KLIYF370), which is located in the coiled-coil region of KAP1, at the proximal promoter of p21 and that PP1α dephosphorylated Ser824 of KAP1, thereby augmenting the SUMOylation state of KAP1. Moreover, PP1β was recruited to KAP1 with distinct kinetics from that of PP1α and stimulated the SUMOylation of KAP1 independently of the phosphorylation status of Ser824. Our results help to unveil a previously uncharacterized role for PP1 in regulating a switch in the SUMOylation status of KAP1 and to redefine the contribution of PP1 to the ATM-to-KAP1 module in response to DNA damage.

Results

Depletion of KAP1 impairs cell cycle progression

To investigate the role of KAP1 in the DDR, we established a cell line (MCF-7/TR/sh-KAP1 cells) that expressed a doxycycline-inducible short hairpin RNA (shRNA) specific for the 3′ untranslated region (3′UTR) of KAP1 messenger RNA (mRNA) to deplete the cells of endogenous KAP1 protein (fig. S1A). As predicted, knockdown of KAP1 resulted in the MCF-7/TR/sh-KAP1 cells having a slower proliferation rate than that of the parental MCF-7 cells (fig. S1B). To evaluate the effect of knockdown of KAP1 on cell cycle progression, we arrested MCF-7/TR/sh-KAP1 cells at the G1-S stage by treatment with hydroxyurea (1 μM) for 16 hours and then released them from this arrested state by replenishment with fresh complete medium lacking hydroxyurea. Flow cytometry analysis revealed that ~24.9 and 42.6% of synchronized KAP1-depleted cells were in the G1 phase at 6 and 12 hours, respectively, after reentering the cell cycle (Fig. 1A). In contrast, less than 16.4 and 21% of KAP1-competent cells were in the G1 fraction at the same time points (Fig. 1A). The decrease in the number of doxycycline-treated MCF-7/TR/sh-KAP1 cells in the S phase was accompanied by an accumulation of MCF-7/TR/sh-KAP1 cells in the G2-M phase (Fig. 1A). Together, the increase in the percentage of KAP1 knockdown cells in G1 phase at 12 hours after reentering the next cell cycle may have resulted either from a cohort of cells from the G2-M phase that slowly entered the G1 phase or from a defect in the entry of cells into the S phase from the G1 phase.

Fig. 1

KAP1 is essential for genotoxicity-induced cell cycle arrest and apoptosis. (A) MCF-7/TR/sh-KAP1 cells, synchronized at G1-S phases by hydroxyurea (1 μM, 16 hours) with cotreatment of vehicle or doxycycline (2 μg/ml), were released from cell cycle arrest by replenishment with fresh complete medium lacking hydroxyurea for the indicated time periods and were analyzed by flow cytometry. (B) Unsynchronized MCF-7/TR/sh-KAP1 cells were treated with vehicle, doxorubicin (Dox, 1 μM), doxycycline (Doxy, 2 μg/ml) or a combination of the two for 24, 48, or 72 hours. Apoptosis was assessed by quantitation of the number of cells in the sub-G1 fraction. (C) MCF-7/TR/sh-KAP1 cells were pretreated with doxycycline overnight to knockdown endogenous KAP1 before treatment with doxorubicin for the indicated time periods. The steady-state abundance of p21, Bax, Puma, and Noxa mRNAs was quantified by real-time RT-PCR analysis with gene-specific primer pairs (table S1). Results represent the mean ± SD from three independent experiments. (D) Annexin V–containing cells in KAP1-depleted and KAP1-competent cells after treatment with doxorubicin for 9 hours were compared by flow cytometry. The data shown come from three independent experiments.

The effect of depletion of KAP1 on cell cycle progression was then tested in MCF-7 cells that were treated with the compound doxorubicin, which induces the formation of DNA DSBs. Whereas unsynchronized MCF-7 cells were increasingly arrested at the G2-M phase of the cell cycle at 24 and 48 hours after treatment with doxorubicin (fig. S1C), knockdown of KAP1 resulted in more MCF-7 cells becoming arrested at the G2-M phase at 48 hours after exposure to doxorubicin (fig. S1C), with increased apoptosis occurring during prolonged treatment (Fig. 1B). The effect of knockdown of KAP1 on the time-dependent expression profiles of Bax, Puma, Noxa, and p21 in response to doxorubicin was then assessed in MCF-7/TR/sh-KAP1 cells. Steady-state amounts of these four mRNAs were similarly increased starting at 6 hours and reached a maximum at 9 hours after treatment with doxorubicin in KAP1-depleted MCF-7/TR/sh-KAP1 cells (Fig. 1C). To confirm that knockdown of KAP1 led to increased apoptosis, we performed annexin V apoptosis assays on cells that had undergone treatment with doxorubicin for 9 hours. In agreement with the gene induction profile (Fig. 1C), there was a greater increase in the number of KAP1-depleted cells that stained with annexin V relative to that of KAP1-competent cells (Fig. 1D).

PP1α physically interacts with KAP1 through a PP1-binding site

Because ATM-mediated phosphorylation of Ser824 regulates KAP1 signaling in response to the induction of DNA DSBs (3, 610), we sought to identify the corresponding phosphatase(s) that regulated the phosphorylation status of KAP1 Ser824. We identified a putative PP1-binding site (366KLIYF370) in the coiled-coil region of KAP1. We performed reciprocal coimmunoprecipitation experiments followed by Western blotting analyses and showed that endogenous KAP1 interacted with endogenous PP1α and PP1γ rather than with PP1β in unstressed human embryonic kidney (HEK) 293 cells (Fig. 2A). Because myosin phosphatase targeting subunit 1 (MYPT1) reportedly targets PP1β to Polo-like kinase 1 (PLK1) (20) and HDAC7 (21), we sought to determine whether MYPT1 also formed a bridge between PP1β and KAP1. Coimmunoprecipitation experiments revealed that MYPT1 interacted with PP1β and PP1γ, but that it failed to form a complex with KAP1 (fig. S2A). Because PP2cA inactivates ATM (22), we also tested whether PP2cA interacted with KAP1. An increase in the abundance of PP1β and PP2cA, relative to that of PP1α, in transfected cells failed to force interactions between these proteins and KAP1 (Fig. 2B), confirming that KAP1 favored interacting with PP1α rather than PP1β or PP2cA in unstressed cells.

Fig. 2

A physical interaction between PP1α and KAP1 occurs through the PP1-binding motif. (A) Reciprocal coimmunoprecipitation experiments were performed to demonstrate the interaction of endogenous KAP1 with endogenous PP1α, PP1β, and PP1γ. Whole-cell extracts were subjected to immunoprecipitation (IP) with an antibody against KAP1, and coimmunoprecipitated PP1 was visualized with antibodies against PP1α, PP1β, or PP1γ. The reverse coimmunoprecipitation reactions were also performed by first immunoprecipitating with antibody against PP1 and then detecting with an antibody against KAP1. WB, Western blot. (B) Overexpression of PP1β fails to force an interaction with KAP1. HEK 293 cells were cotransfected with plasmids encoding FLAG-KAP1 and HA-PP1α, HA-PP1β, or HA-PP2cA, as indicated. Samples were immunoprecipitated with an antibody against HA and coimmunoprecipitated KAP1 proteins were visualized with antibody against FLAG. (C) Doxorubicin facilitates an interaction between PP1β and KAP1. Individually transfected HEK 293 cells, as indicated, were treated with vehicle (left panel) or doxorubicin (1 μM for 3 hours) (right panel) and subjected to coimmunoprecipitations as described for (B). (D) The I368G mutation of KAP1 favors an interaction with PP1α. Individually transfected HEK 293 cells were treated with vehicle or doxorubicin (1 μM, for 3 hours) and were analyzed as described for (B). The relative extent of binding (numbers in italics) of KAP1 proteins with PP1α, PP1β, or PP2cA before and after treatment of cells with doxorubicin, after normalization, is shown in (C) and (D). The binding of wild-type KAP1 to PP1α before treatment with doxorubicin was set as 1. s, short-term exposure; l, long-term exposure. The data shown come from three independent experiments.

HEK 293 cells that were cotransfected with plasmids encoding wild-type KAP1, a phosphorylation-defective mutant of KAP1 [KAP1(S824A)], or a phosphorylated mimetic mutant of KAP1 [KAP1(S824D)] together with plasmids encoding hemagluttinin (HA)-tagged PP1α (HA-PP1α), HA-PP1β, or HA-PP2cA were left untreated or were treated with doxorubicin in order to further characterize the interaction of PP1α with KAP1. Doxorubicin stimulated an interaction between KAP1 and PP1β or PP2cA (Fig. 2C); the extent of interaction between KAP1 and PP1β or PP2cA, relative to that with PP1α, increased from ~10 and ~0% to ~30 and ~53%, respectively. Although PP1α physically associated with KAP1 rather than with KAP1(S824A) or KAP1(S824D) before treatment with doxorubicin (Fig. 2C), induction of DNA DSBs facilitated an interaction between KAP1(S824A), but not KAP1(S824D), and either PP1α or PP2cA (Fig. 2C). The lack of a detectable interaction between PP1α and KAP1(S824D) was unexpected, and the underlying reason is still unclear. Although ATM phosphorylates two additional sites, Ser440 and Ser501, in KAP1 (23), the site-specific mutagenesis of either of these residues to alanine or aspartate did not affect the interaction between PP1 and KAP1 (fig. S3A).

To confirm the direct interaction between PP1α and KAP1, we engineered four mutant KAP1 proteins, each of which contained one of the substitutions K366G, I368G, F370G, and F370A in the putative PP1-binding site of KAP1. Coimmunoprecipitation experiments revealed that whereas the K366G substitution weakened the ability of KAP1 to interact with PP1α before and after treatment with doxorubicin, the KAP1(I368G) protein bound to PP1α with greater avidity than did wild-type KAP1 under both conditions (Fig. 2D). In contrast, the F370G and F370A mutations only modestly affected the interactions between the appropriate mutant KAP1 proteins and PP1α (fig. S2B). As expected, KAP1(K366G), a mutant protein with a decreased affinity for PP1α relative to that of wild-type KAP1, exhibited a higher basal p21-Luc activity in KAP1-depleted MCF-7/TR/sh-KAP1 cells (fig. S2C), confirming the relevance of the binding of KAP1 to PP1α to its transcriptional co-repressor activity. Although substitutions to the PP1-docking motif are thought to abrogate the binding of the respective mutant proteins to PP1 (18), results from our mutagenesis studies suggest otherwise. We postulated that the 366KLIYF370 motif at the coiled-coil region of KAP1 was not a conventional PP1-docking motif, but that instead, it might tether to the surface of PP1, close to its catalytic center. Structural analysis suggested that the PP1-docking motif of KAP1 assumed an α-helical conformation (fig. S3B), supporting its inability to dock to the hydrophobic channel in PP1, which is identified as the canonical PP1-docking motif (24). Given these findings, we concluded that KAP1 interacted with PP1α through an unconventional, albeit not an unprecedented (25, 26), PP1-binding site located at 366KLIYF370 in the coiled-coil region of KAP1.

PP1α dephosphorylates KAP1 at Ser824

Next, we assessed whether PP1cs affected the DNA DSB–induced phosphorylation of KAP1 Ser824. Because transfection of cells with a plasmid encoding PP1γ2 had no effect on doxorubicin-induced phosphorylation of KAP1 Ser824 in the absence or presence of exogenous MYPT1 (fig. S4A), we focused in the remaining studies on the effects of PP1α and PP1β. As expected, PP2cA attenuated doxorubicin-induced phosphorylation of ATM Ser1981 (22) (Fig. 3A), which led to a decrease in the extent of phosphorylation of KAP1 Ser824 (Fig. 3A). An increase in the abundance of PP1β in transfected MCF-7 cells attenuated doxorubicin-induced, but not basal, phosphorylation of Ser824 of endogenous KAP1 (Fig. 3A). In contrast, in cells transfected with a plasmid encoding PP1α, the basal phosphorylation of KAP1 Ser824 was decreased, and this effect of PP1α seemed transient and more effective at an earlier time after treatment with doxorubicin than at later times (Fig. 3A). These results, together with earlier data (Fig. 2, B and C), suggested that although PP1α associated with KAP1 during basal dephosphorylation of Ser824, PP1β was recruited to KAP1 to dephosphorylate Ser824 that had been phosphorylated in response to the induction of DNA DSBs. In vitro phosphatase assays further confirmed that purified rabbit muscle PP1 dephosphorylated KAP1 at Ser824 in a dose-dependent manner (Fig. 3B). It appeared that the extent of PP1-mediated dephosphorylation of Ser824 of various KAP1 proteins in vitro increased in proportion to their respective affinities for PP1α (Fig. 2D) [KAP1(I368G) > KAP1 > KAP1(K366G)] (Fig. 3B). The correlation between the strength of binding to PP1α of a KAP1 protein and the extent of PP1-mediated dephosphorylation of that protein in vitro was further supported by observations from experiments with the KAP1(F370G) and KAP1(F370A) mutants (figs. S2B and S4B).

Fig. 3

PP1α dephosphorylates KAP1 pSer824 that was phosphorylated in response to DNA DSBs. (A) MCF-7 cells, transfected with the indicated plasmids, were treated with vehicle or doxorubicin (5 μM) for 30 min. Protein extracts were analyzed by Western blotting with antibodies against HA, KAP1, KAP1 pSer824, ATM pSer1981, or tubulin. (B) In vitro phosphatase assays were performed with increasing amounts of purified rabbit skeletal muscle PP1 and pSer824-containing FLAG-KAP1(wt) (wild-type) or FLAG-KAP1(K366G). (C) Control MCF-7 cells and MCF-7 cells depleted of PP1α or PP1β, as indicated, were treated with vehicle or doxorubicin (5 μM) for 1 hour. Protein extracts were analyzed by Western blotting with antibodies against HA, KAP1, KAP1 pSer824, or tubulin. (D) Transfected MCF-7 cells were treated as indicated and the phosphorylation profile of KAP1 Ser824 was analyzed by Western blotting (n = 3 independent experiments). (E) Whole-cell extracts from transfected HEK 293 cells, as indicated, were immunoprecipitated with an antibody against HA, and coimmunoprecipitated I-2 was visualized with an antibody against Myc (left panel). MCF-7 cells were transiently transfected with plasmids encoding HA-PP1α or Myc-I-2 or with empty vector as a control and incubated for 24 hours. Cells were then treated for 30 min with vehicle or doxorubicin (5 μM) (right panel). Protein extracts were analyzed by Western blotting with antibodies against KAP1, KAP1 pSer824, or actin. Antibodies against PP1α and Myc were used to detect total PP1α or Myc-I-2 proteins, respectively, in MCF-7 cells (n = 3 experiments). Relative amounts (numbers in italics) of doxorubicin-induced KAP1 pSer824 protein, normalized to the amounts of total KAP1 protein, are shown in (A) to (E).

To further clarify the effect of PP1α or PP1β on the phosphorylation of KAP1 Ser824, we established stable MCF-7 cell lines with specific knockdown of PP1α (MCF-7/sh-PP1α cells) or PP1β (MCF-7/sh-PP1β cells) (fig. S4C). Although both basal and doxorubicin-induced phosphorylation of KAP1 Ser824 were greatly enhanced in MCF-7/sh-PP1α cells relative to that in MCF-7/sh-control cells, the lack of PP1β had, unexpectedly, a limited effect (Fig. 3C). These results indicated a role for PP1α in basal dephosphorylation of phosphorylated Ser824 (pSer824) of KAP1 that was not fully compensated for by either PP1β or PP1γ. As predicted, KAP1(I368G), a mutant with an enhanced affinity for PP1α, exhibited decreased basal and doxorubicin-induced phosphorylation at Ser824 after normalization for protein abundance (Fig. 3D), which presumably resulted from the increased association of KAP1(I368G) with PP1α compared to that of wild-type KAP1. Consistent with this, an increase in the abundance of PP1α in transfected cells markedly attenuated doxorubicin-induced phosphorylation of Ser824 of KAP1(I368G) relative to that of KAP1 (Fig. 3D). In addition, the increased abundance of PP1α did not enforce increased dephosphorylation of pSer824 of KAP1(K366G), a mutant protein with a reduced affinity for PP1α relative to that of wild-type KAP1 (Fig. 3D). The lack of a marked effect of the K366G mutation on the phosphorylation of KAP1 Ser824 in the absence of increased PP1α abundance under either unstressed or stressed conditions (Fig. 3D) was unexpected. Considering earlier results (Figs. 2D and 3C), we postulated that the 50 to 60% decrease in the association of KAP1(K366G) with PP1α alone was not sufficient to affect the phosphorylation status of Ser824 under either unstressed or stressed conditions. Consistent with their respective interactions with PP1α (fig. S2B) and with in vitro dephosphorylation results (fig. S4B), the phosphorylation status of Ser824 of KAP1(F370G) and KAP1(F370A) was regulated by PP1α (fig. S4D). Together, these results led us to postulate that PP1α dephosphorylated KAP1 Ser824, at least in part, by forming a physical association with KAP1 through its PP1-binding site.

To ascertain a role for PP1α in regulating the phosphorylation of KAP1 Ser824, we investigated whether other factors, such as the protein phosphatase inhibitor (I-2), which inhibits PP1 activity during various cellular events (27), could affect the extent of phosphorylation of KAP1 Ser824 under unstressed or stressed conditions. We examined whether I-2 could affect the doxorubicin-mediated modulation of KAP1 Ser824 phosphorylation. Coimmunoprecipitation experiments showed that I-2 formed complexes with PP1α of the three catalytic subunits examined (Fig. 3E). In addition, doxorubicin did not alter the interaction between I-2 and PP1α, PP1β, or PP2cA (Fig. 3E). Whereas PP1α attenuated the doxorubicin-induced phosphorylation of KAP1 Ser824 (Fig. 3E), ectopic expression of I-2 further increased the doxorubicin-dependent phosphorylation of endogenous KAP1 Ser824 (Fig. 3E). These data led us to suggest that I-2 might sequester PP1α from KAP1, thus leading to the enhanced phosphorylation of KAP1 Ser824, which would support our contention that PP1α is one of the key molecules that regulates the phosphorylation status of KAP1 Ser824.

PP1α and PP1β differentially regulate the SUMOylation of KAP1

Next, we performed experiments to differentiate the effects of PP1α or PP1β on the SUMOylation of KAP1. Consistent with our previous reports (3, 6), SUMOylation of KAP1 in MCF-7/sh-control cells decreased on exposure to doxorubicin (Fig. 4A). Surprisingly, knockdown of PP1β markedly decreased the abundance of SUMOylated KAP1 in vehicle-treated and doxorubicin-treated cells (Fig. 4A), and depletion of PP1α reproducibly had a relatively modest effect on the overall extent of SUMOylation of KAP1 (Fig. 4A). To reconcile the discrepancy between the Ser824 phosphorylation profile (Fig. 3C) and the corresponding SUMOylation profile (Fig. 4C) of KAP1 in MCF-7/sh-PP1α and MCF-7/sh-PP1β cells, we performed immunoprecipitation experiments to assess the interaction of KAP1 with PP1β in MCF-7 cells. Our results showed that the extent of the interaction between PP1β and KAP1, endogenously or ectopically expressed, increased, with diminished concentrations of available PP1α at the ground state, in the order of MCF-7/sh-PP1α cells > MCF-7/sh-control cells > MCF-7/sh-PP1-I-2 cells (Fig. 4B). Consistently, doxorubicin further enhanced the interaction between PP1β and either endogenous or exogenous KAP1 protein in the same order in these three cell lines (Fig. 4B). These data, together with earlier findings (Fig. 2, A and C), suggested that PP1α might compete with PP1β for binding to KAP1 and that doxorubicin stimulated an additional interaction between KAP1 and PP1β.

Fig. 4

PP1 regulates the SUMOylation of KAP1. (A) In vitro SUMOylation assays were performed by treating HEK 293 cells cotransfected with plasmids encoding KAP1 and enhanced green fluorescent protein (EGFP)-SUMO-1 with vehicle or doxorubicin, as indicated, and visualizing SUMOylated proteins by Western blotting. (B) Cells were untransfected (left panel) or were transfected with plasmids encoding FLAG-KAP1 and HA-PP1β (right panel) to assess the extent of the interaction between PP1β and KAP1. After treatment with vehicle or doxorubicin (1 μM) for 3 hours, whole-cell extracts from transfected MCF-7 cells that contained PP1α, PP1β, or I-2 were used in immunoprecipitations with an antibody against HA. Coimmunoprecipitated KAP1 was visualized with an antibody against the FLAG tag. Relative amounts (numbers in italics) of PP1β-interacting KAP1 normalized for the amount of input KAP1 protein are shown. (C) In vitro SUMOylation assays were performed with HEK 293 cells cotransfected with plasmids encoding FLAG-tagged wild-type KAP1 or its mutants and EGFP-SUMO-1, as indicated, and SUMOylated proteins were visualized by Western blotting analysis (n = 3 independent experiments). (D) In vitro SUMOylation assays were performed with HEK 293 cells cotransfected with plasmids encoding KAP1(S824A) and EGFP-SUMO-1 in the absence or presence of a combination of knockdown of PP1α and treatment with doxorubicin, as indicated, and SUMOylated proteins were visualized by Western blotting analysis. (E) In vitro SUMOylation assays were performed with KAP1(wt), KAP1(I368G), KAP1(K366G), and KAP1(824A) proteins in the absence or presence of a combination of overexpression of PP1α or PP1β and treatment with doxorubicin. SUMOylated proteins were visualized by Western blotting (n = 3 independent experiments).

To ascertain whether PP1α regulated the SUMOylation of KAP1 that had undergone ATM-induced phosphorylation at Ser824, we generated KAP1 mutants containing mutations in which Ser440, Ser501, or both were substituted with the phosphomimetic residue aspartate to examine whether ATM-induced phosphorylation at these residues (23) also played a critical role in inhibiting the SUMOylation of KAP1. Consistent with our premise, the SUMOylation profiles of KAP1(S440D), KAP1(S501D), and KAP1(S440/501D) were largely similar to those of wild-type KAP1 (Fig. 4C). In contrast, KAP1(S440/501/824D), similar to KAP1(S824D), exhibited decreased SUMOylation compared to that of wild-type KAP1 (Fig. 4C). After establishing that phosphorylation of Ser824 was a critical factor in modulating the SUMOylation of KAP1, we next performed assays to examine the effect of depletion of PP1α on the SUMOylation of KAP1(S824A). As predicted, SUMOylation of KAP1(S824A) in the absence and presence of doxorubicin remained nearly independent of the knockdown of PP1α (Fig. 4D).

To further explore the specific effects of PP1α and PP1β on the SUMOylation of KAP1, we examined the SUMOylation profiles of wild-type and mutant KAP1 proteins in the context of overexpression of PP1α or PP1β. Consistent with our previous results (6), attenuated phosphorylation of KAP1 Ser824 by ectopic expression of PP1α or PP1β stimulated the SUMOylation of KAP1 after treatment with doxorubicin (Fig. 4E). We further observed that overexpression of PP1α markedly stimulated the SUMOylation of KAP1(I368G) in untreated cells and in doxorubicin-treated cells (Fig. 4B), presumably by further suppressing the phosphorylation of Ser824. In contrast, overexpression of PP1β had a modest stimulatory effect on the mono- and tri-SUMOylation of KAP1(I368G) after treatment with doxorubicin (Fig. 4E). Overexpression of either PP1α or PP1β elicited a moderate effect in enhancing the SUMOylation of KAP1(K366G) (Fig. 4E), and doxorubicin failed to render a decrease in PP1α- or PP1β-dependent SUMOylation of KAP1(K366G) (Fig. 4E). Lastly, we further confirmed that KAP1(S824A) was resistant to PP1α-induced SUMOylation (Fig. 4E). Together, these results supported the notion that the phosphorylation of KAP1 Ser824 played a critical role in the PP1α-stimulated SUMOylation of KAP1; however, dephosphorylation of KAP1 Ser824 was not the sole factor that regulated the SUMOylation of KAP1.

To address whether PP1 stimulated the SUMOylation of proteins other than KAP1, we observed that PP1β, but not PP1α, stimulated the SUMOylation of endogenous Ran guanosine triphosphatase (GTPase)–activating protein 1 (RanGAP1), a cellular protein known to be SUMOylated (fig. S5B). Conversely, knockdown of PP1β, but not PP1α, abolished the SUMOylation of RanGAP1 (fig. S5B), which suggested that PP1β affected the SUMOylation state of proteins other than KAP1. Together, these results indicated that PP1α augmented the amount of SUMOylated KAP1 in response to doxorubicin, mainly through binding to the PP1-binding site of KAP1 to dephosphorylate pSer824, and that PP1β stimulated the SUMOylation of KAP1 by a more general means that was both dependent and independent of the dephosphorylation status of pSer824. The observed discrepancy in the inverse co-regulation of the extent of phosphorylation of KAP1 Ser824 (Fig. 2A) and the SUMOylation of KAP1 (Fig. 4A) in MCF-7/sh-PP1α cells relative to that of MCF-7/sh-control cells might have resulted from the effects of the increased amount of PP1β that was bound to KAP1 in the absence of PP1α (Fig. 4B).

Knockdown of PP1α promotes derepression of KAP1 target gene in MCF-7 cells with DNA damage

A real-time cell growth monitoring system and annexin V apoptosis assays were used to monitor the effects of shRNAs specific for PP1α or PP1β on cell proliferation. Both MCF-7/sh-PP1α and MCF-7/sh-PP1β cells showed relatively slower cell growth rates compared to that of MCF-7/sh-control cells (fig. S5C). Consistent with earlier data (Fig. 1B), knockdown of KAP1 resulted in an increase in the number of annexin V positive MCF-7 cells (Fig. 5A). Whereas depletion of PP1α rendered a phenotype similar to that which resulted from knockdown of KAP1 (Fig. 5A), knockdown of PP1β resulted in substantially more apoptosis in doxorubicin-treated MCF-7 cells than did knockdown of PP1α (Fig. 5A). We postulated that the S824A mutation of KAP1 might render KAP1 capable of escaping from ATM-mediated, sh-PP1β-associated deSUMOylation, thus protecting cells from apoptosis. As predicted, introduction of the KAP1(S824A) mutant protein decreased the percentage of annexin V positive signals in unstressed MCF-7/sh-PP1α cells (Fig. 5A) and doxorubicin-treated MCF-7/sh-PP1β cells (Fig. 5A), supporting the proposed roles of PP1α and PP1β in modulating the phosphorylation of KAP1 Ser824. Lastly, we examined the effect of knockdown of PP1α or PP1β on doxorubicin-induced expression of KAP1 target genes and compared it to the effect of depletion of KAP1. On the basis of earlier results (Fig. 1C), we compared the abundance of p21, Bax, Puma, and Noxa mRNAs in MCF-7, MCF-7/sh-KAP1, MCF-7/sh-PP1α, and MCF-7/sh-PP1β cells before and 9 hours after treatment with doxorubicin. Knockdown of PP1α substantially increased the doxorubicin-induced expression of p21, Bax, and Noxa, mimicking the effect of depletion of KAP1 (Fig. 5B). The induction of basal expression of Bax in MCF-7/sh-PP1α cells (Fig. 5B) was consistent with the observed increase in the number of annexin V positive MCF-7/sh-PP1α cells under unstressed conditions (Fig. 5A). The lack of an effect of depletion of PP1β on the induced expression of pro-arrest and pro-apoptotic KAP1 target genes could be due to PP1β functioning at a later time point after the induction of DNA DSBs.

Fig. 5

PP1α governs KAP1-mediated regulation of the cell cycle and expression of target genes. (A) MCF-7/sh-control, MCF-7/sh-PP1α, and MCF-7/sh-PP1β cells transfected with plasmid encoding KAP1(S824A) or with empty vector were treated with vehicle or doxorubicin (1 μM) for 72 hours and subjected to annexin V analysis by flow cytometry (left panel). Pooled data were also summarized (right panel). (B) MCF-7/sh-control, MCF-7/sh-PP1α, and MCF-7/sh-PP1β cells were treated with vehicle or doxorubicin (1 μM) for 9 hours; the steady-state amounts of p21, Bax, Puma, and Noxa mRNAs were analyzed as described for Fig. 1C. Results represent the mean ± SD of three independent experiments. P < 0.05 was considered statistically significant.

PP1α and PP1β form complexes with KAP1 at the proximal promoter of p21

We previously showed that the SUMOylation status of KAP1 affects the acetylation state of Lys9 (K9) and Lys14 (K14) and the dimethylation state of K9 of histone H3 at the proximal promoter of p21, thereby affecting its expression (3, 6). Because PP1α and PP1β interacted with KAP1 in a context-dependent manner (Fig. 2C), we investigated whether distinct complexes of KAP1 and PP1α or PP1β at the proximal promoter of p21 affected the histone H3-K9/K14 acetylation state or H3-K9 dimethylation state. First, we tested whether PP1α or PP1β was recruited to the KAP1 complex at the p21 promoter. PP1α, but not PP1β, was constitutively bound at the p21 proximal promoter region and doxorubicin increased the amount of both PP1α and PP1β that bound to KAP1 at the −20 region of the proximal promoter of p21 (Fig. 6A). In addition, no colinear chromatin regions of −713 or −3038 in p21 were immunoprecipitated by an antibody against the HA tag (which would bind to PP1α or PP1β) in the same chromatin immunoprecipitation (ChIP)–re-immunoprecipitation (ReIP) experiments (fig. S6A). These results, in conjunction with earlier findings (Figs. 2C and 4B), suggested that whereas PP1α constitutively formed a functional unit with KAP1 at the proximal promoter of p21, doxorubicin stimulated the recruitment of additional PP1α and PP1β proteins to KAP1 at the same region.

Fig. 6

PP1α and PP1β differentially regulate acetylation of K9 and K14 and dimethylation of K9 of histone H3 at the proximal promoter of p21. (A) ChIP-ReIP experiments were performed on vehicle- or doxorubicin-treated MCF-7 cells with beads conjugated with antibody against FLAG-M2 (to immunoprecipitate KAP1 proteins) followed by re-immunoprecipitations with an antibody against HA (to immunoprecipitate PP1α or PP1β) or with control IgG. Quantification was performed by real-time PCR with primer pairs against −20 amplicons of endogenous p21 (table S1). (B) ChIP-ReIP experiments were performed and analyzed as described for (A), where re-immunoprecipitations were performed with antibodies against acetylated K9 or K14 or dimethylated K9 of histone H3, or with control antibody. Results represent the mean ± SD of three independent experiments. P < 0.05 was considered statistically significant.

Lastly, to test the importance of PP1α or PP1β for KAP1-associated chromatin remodeling, ChIP-ReIP experiments performed in MCF-7 cell lines showed that knockdown of PP1α increased the accessibility of K9 and K14 of histone H3 to KAP1-associated acetylation before treatment with doxorubicin, and that overexpression of PP1β decreased the co-occupancy of acetylated K9 and K14 of histone H3 with KAP1 after treatment with doxorubicin at the proximal promoter of p21 (Fig. 6B). Conversely, overexpression of PP1α or PP1β stimulated an increase in the co-occupancy of dimethylated K9 of histone H3 with KAP1 at the proximal promoter of p21 after treatment with doxorubicin (Fig. 6B). In addition, overexpression of PP1β resulted in a substantial increase in the occupancy of KAP1-associated dimethylated K9 of histone 3 before treatment with doxorubicin (Fig. 6B). In contrast, the respective KAP1-associated profiles of acetylation of K9 and K14 and dimethylation of K9 of histone H3 at the distal 5′-flanking region of p21 in vehicle-treated or doxorubicin-treated MCF-7 cells with distinct PP1α or PP1β context were almost undistinguishable (fig. S6B).

Discussion

KAP1 provides a prosurvival advantage to cells by contributing to the transcriptional repression of the DDR genes p21, Bax, Noxa, and Puma (Fig. 1). Knowledge of the SUMOylation and phosphorylation states of KAP1 and of how these states are regulated in vivo is required to fully understand the biological effects of this key transcriptional co-repressor. Our results suggest a previously uncharacterized role for PP1 in regulating the dynamic function of KAP1 and indicate a molecular framework through which PP1 functions in unstressed and DNA-damaged cells (Fig. 7).

Fig. 7

Model depicting regulation of the SUMOylation of KAP1 and transcriptional repression. PP1α and PP1β use mechanisms that are dependent and independent of the phosphorylation state of KAP1 Ser824 to regulate the SUMOylation of KAP1 with distinct kinetics.

Studies have indicated that PP1 is involved in the DDR (17), including checkpoint activation (28), DNA repair (29, 30), and recovery from DNA damage checkpoint arrest (31). SUMOylation is one of the critical events in cellular responses to a wide range of DNA-damaging reagents (32); however, there is no previous experimental evidence for a role for PP1 in regulating SUMOylation in response to DNA damage. In this work, detailed experimental analyses were performed of the role of PP1α and PP1β in regulating the SUMOylation of KAP1 as a means to regulate the expression of key KAP1 target genes during the DDR. We showed that PP1α physically bound to KAP1 and that the interaction between PP1α and KAP1 was essential to establishing the minimal extent of phosphorylation of Ser824 that was required for the co-repressor function of KAP1 in unstressed cells. Site-specific mutagenesis studies further revealed that PP1α interacted with KAP1 proteins containing the mutations K366G and I368G with different affinities, which in turn displayed differential SUMOylation and pSer824 profiles. In contrast, PP1β was recruited to KAP1 after treatment of cells with doxorubicin (Figs. 2C and 6A), and its stimulatory effect on SUMOylation extended to proteins other than KAP1 (Fig. 4A), such as RanGAP1 (fig. S5B). Together with our observations that the co-repressor and chromatin remodeling functions of KAP1 are under the regulation of ATM after the induction of DNA DSBs (3, 68), we propose that PP1 serves a key role in unstressed cells and in cells with damaged DNA by coordinating the phosphorylation of KAP1 Ser824 and the switch in the SUMOylation state of KAP1.

A pertinent and intriguing question is how PP1 stimulates the SUMOylation of KAP1. Phosphorylation of SUMOylation targets may serve both as a positive and negative signal for respective SUMOylation in a context-dependent manner (33). For instance, phosphorylation at a serine residue adjacent to a SUMO consensus site augments the binding of the SUMO conjugase Ubc9 to myocyte-enhancement factor 2 (MEF2) and heat-shock transcription factor 1 (HSF1), which leads to increased SUMOylation (34). In contrast, phosphorylation at Thr188 attenuates the SUMOylation of SATB1 (special AT-rich sequence-binding protein-1) by inhibiting its binding to PIAS1, a SUMO E3 ligase (35). Because KAP1 is SUMO E3 ligase that acts on itself through the recruitment of Ubc9 (36), it is conceivable that PP1 might stimulate the self-SUMOylation of KAP1 by enhancing the binding of Ubc9 to KAP1. Alternatively, PP1 may enhance the SUMOylation of KAP1 through the recruitment of other SUMO E3 ligases or by precluding deSUMOylation enzymes from KAP1. In either scenario, dephosphorylation of KAP1 at Ser824 by PP1α may lead to structural changes that either recruit or exclude proteins involved in the SUMOylation of KAP1. We also observed that, in addition to KAP1, PP1β regulated the SUMOylation state of RanGAP1, which suggests that the effect of PP1β on the regulation of SUMOylation is not limited to KAP1 per se. In addition, modification of Ubc9 is also involved in discriminating between its targets Sp100 and RanGAP1 (37). Collectively, it is possible that PP1α forms complexes with KAP1 to maintain the unphosphorylated state of Ser824, thereby facilitating an environment that favors the SUMOylation of KAP1, whereas PP1β may stimulate the SUMOylation machinery in addition to attenuating ATM-activated phosphorylation of Ser824.

On the basis of our data, we propose the following model for the role of PP1 in regulating the co-repressor activity of KAP1 (Fig. 7). KAP1 exists in a dynamic balance between a pSer824 and unSUMOylated state (an inactive co-repressor) and a dephosphorylated but SUMOylated state (an active co-repressor) in unstressed cells. PP1α constitutively binds to KAP1 and sets a basal transcription rate for a subset of KAP1 target genes whose products are involved in pro-arrest and pro-apoptotic processes. After the induction of DNA DSBs, resident PP1α dephosphorylates pSer824, which was phosphorylated by ATM, and cooperates with newly recruited PP1α and PP1β to timely restore SUMOylation of KAP1, which in turn resumes its role as a co-repressor. In our model, we further predict that the amount of PP1α bound to KAP1 sets the threshold for the extent of activation of ATM that is required to overcome PP1α-mediated dephosphorylation of pSer824. In contrast, PP1β interacts with KAP1 in a spatiotemporal manner (in the absence of PP1α and after the induction of DNA DSBs) to promote the SUMOylation of KAP1. Because both PP2A and PP4C directly dephosphorylate γ-H2AX—to differentially regulate the abundance of γ-H2AX that originates from different stressors, different degrees of DNA damage, or both (38, 39)—it is not surprising that both PP1α and PP1β are involved in the temporal regulation of the dephosphorylation of KAP1 Ser824 and the SUMOylation of KAP1. With this caveat, PP1-dependent stimulation of the SUMOylation of KAP1 may represent an example of the interplay between posttranslational modifications and gene regulation networks.

Yamashiro et al. reported that PP1 antagonizes the function of PLK1 by interacting with MYPT1 during mitosis (20). Phosphorylation of PLK1 at Thr210 activates PLK1 during mitosis, and the PP1-MYPT1-PLK1 complex restrains the activation of PLK1 to retard untimely mitosis. In addition, PP1 and MYPT1 dephosphorylate HDAC7 and promote its transcriptional repression of the orphan nuclear receptor Nur77, which is involved in the induction of negative selection or apoptosis after the activation of the T cell receptor, thereby inhibiting apoptosis in thymocytes (21). We report here that MYPT1 is involved in the interaction of KAP1 with PP1β, but not PP1α, thus presumably regulating the phosphorylation of KAP1 Ser824. In addition to functioning as a transcriptional co-repressor, KAP1 phosphorylated at Ser824 as a result of an ATM-dependent event upon the induction of DNA DSBs, also promotes the relaxation of chromatin (8). Furthermore, knockdown of KAP1 results in increased chromatin accessibility in heterochromatin and in the restoration of DNA repair competence in ATM-deficient cells (22). This study further showed that phosphorylation of KAP1 at Ser824 perturbed its interactions with silencing factors, such as HDAC1, without attenuating the binding of KAP1 to its target. In light of the multiple functions of KAP1 in different contexts (2, 5, 7, 22, 34, 4048), it is tempting to speculate that PP1 might serve a critical role in regulating diverse biological processes, including the DDR.

In summary, our findings provide previously uncharacterized mechanistic insights into how PP1 affects the DDR, influencing not only the dephosphorylation of KAP1 pSer824, but also the SUMOylation of KAP1 and the expression of its target genes. Our findings highlight that PP1 confers its effects through crosstalk between different posttranslational modifications. Considering the number of transcription factors and associated activator and repressor complexes that are SUMOylated, it is possible that the regulation of the SUMOylation state of each may be essential to enable appropriate cell responses to occur. Dephosphorylation-dependent SUMOylation of targets integrates two important signal transduction pathways. The work presented here on the PP1-dependent conjugation of SUMO proteins to KAP1 expands the concept of how SUMOylation is regulated and how it participates in cellular responses to DNA damage.

Materials and Methods

Constructs and establishment of MCF-7 cell lines with stable expression of shRNA

The various KAP1 expression constructs used in this study were generated as previously described (3, 6) or as follows. KAP1(S440D), KAP1(S501D), KAP1(K366G), KAP1(I368G), KAP1(F370G), and KAP1(F370A) constructs were generated from the FLAG-KAP1 template (3) with a site-directed mutagenesis kit (Clontech). The mutagenesis oligonucleotides used were as follows (underlined sequences denote mutations): KAP1(S440D), 5′-GGGCTCTGGCAGCGACCAGCCCATGGAG-3′; KAP1(S501D), 5′-CCTCACAGCTGACGACCAGCCACCCGTC-3′; KAP1(K366G), 5′-CTTTTGCTTTCTAAGGGGTTGATCTACTTCCAG-3′; KAP1(I368G), 5′-CTTTCTAAGAAGTTGGGCTACTTCCAGCTGCAC-3′; KAP1(F370G), GCTTTCTAAGAAGTTGATCTACGGCCAGCTGCACCGGGCC; KAP1(K370A), GCTTTCTAAGAAGTTGATCTACGCCCAGCTGCACCGGGCC. MCF-7 cells were transduced with the empty lentiviral vector pLKO.1 [obtained from the RNA interference (RNAi) consortium at Academia Sinica] or with pLKO.1 vectors containing shRNAs specific for human KAP1, PP1α, or PP1β or a random sequence, as a control. The packaging of self-inactivating lentiviruses containing sh-KAP1, sh-PP1α, sh-PP1β, sh-I-2, or sh-control sequences and the transduction of MCF-7 cells with these viruses were performed as previously described (49). The transduced cells were selected and maintained in MCF-7 medium containing puromycin (2 μg/ml) and pools of stable cells were used in our studies. Protein phosphatase inhibitor-2 (I-2) was cloned into pCMV-Tag3A (Stratagene) by standard reverse transcription polymerase chain reaction (RT-PCR) assay with complementary DNA (cDNA) reverse transcribed from human total RNA, and amplified with the I-2–specific primer pair, 5′-GGATCCAATGGGCGCCTCGACGGC-3′ (forward) and 5′-GAATTCAACAAATCTCGTGCTATGAACTTCG-3′ (reverse), encompassing Bam HI and Eco RI sites, respectively. Expression constructs encoding human PP1α, PP1β, and PP1γ2 were cloned into pcDNA3X(+)HA by standard PCR with cDNA clones with GenBank accession numbers BE001888, BC002697, BC014073, and BC019275, respectively, as templates, and amplified with the following PP1 isoform–specific primer pairs: PP1α: 5′-GTGTGTGGATCCATGTCCGACAGCGAGAAGCTC-3′ (forward) and 5′-GAATTCCTCGAGCTATTTCTTGGCTTTGGCGGAATTG-3′ (reverse); PP1β: 5′-GTGTGTGGATCCATGGCGGACGGGGAGCTGAAC-3′ (forward) and 5′-GAATTCCTCGAGTCACCTTTTCTTCGGCGGATTAGC-3′ (reverse); PP1γ2: 5′-GTGTGTGGATCCATGGCGGATTTAGATAAACTCAAC-3′ (forward) and 5′-GAATTCCTCGAGCTATTTCTTTGCTTGCTTTGTGATC-3′ (reverse). The PP1 isoform-specific primer pairs all encompassed Bam HI and Xho I restriction sites to facilitate cloning. PP2cA was cloned into pcDNA3xHA with the PP2cA-specific primer pairs, 5′-TACGCTGAATTCATGGACGAGAAGGTGTTCACCAAG-3′ (forward) and 5′-GATTAAGGGCCCTTACAGGAAGTAGTCTGGGGTACG-3′ (reverse), which encompassed Eco RI and Apa I restriction sites, respectively, to facilitate cloning. DNA sequencing analysis further verified the identity of the positive clones.

Inducible sh-KAP1

Tet-inducible sh-KAP1 specific for the 3′UTR of KAP1 was constructed according to van de Wetering et al. (50). The oligonucleotides used were as follows: KAP1, 5′-GATCCCCCAGCCAACCAGGGGAAATC-3′ and 5′-AGCTTTTCCAAAAACCACCCAACCAG-3′; control, 5′-GATCCCTCCTCTTTCTTATCCTCGTAT-3′ and 5′-AGCTTTTCCAAAAATCCTCTTTCTTAT-3′. pTER+ was digested with Bgl II and Hind III and the pTER backbone was purified by gel electrophoresis. Sense and antisense oligonucleotides (100 pmol of each) were annealed in 50 μl of annealing buffer [10 mM potassium acetate, 3 mM Hepes-KOH (pH 7.4), and 0.2 mM magnesium acetate]. This mixture of oligonucleotides (1 μl) and 100 ng of purified pTER+ backbone were used in each ligation reaction. Clones expressing pTER+-sh-KAP1 were selected by ampicillin resistance. Subsequently, MCF-7/TR cell clones were established from the stable integration of pCDNA6/TR, which express the Tet repressor. Stable MCF-7/TR/sh-KAP1 cells were established by transfecting MCF-7/TR cells with pTER+-sh-KAP1 followed by selection with blasticidin (5 μg/ml) and zeocin (100 μg/ml). To test the effect of sh-KAP1–mediated knockdown of KAP1, we cotransfected MCF-7/TR/sh-KAP1 cells with the plasmids p21-Luc and pRL-TK. Twenty-four hours after transfection, cells were treated with doxycycline (1 or 2 μg/ml) for 4 to 48 hours to induce the expression of the shRNA. Luciferase activity was measured and normalized against pRL-TK; pCDNA4-TO-Luc (Invitrogen) was used as a positive control for doxycycline induction.

In vitro phosphatase assay

Purified rabbit muscle PP1 was purchased from Millipore (catalog # 14-110). HEK 293 cells that were transfected with plasmids encoding wild-type or mutant KAP1 proteins were treated with doxorubicin (5 μM) for 30 min and lysed with radioimmunoprecipitation assay (RIPA) buffer (see Supplementary Materials for details), and 1 mg equivalent of whole-cell lysates was used in immunoprecipitation assays with agarose beads conjugated to an antibody against FLAG, according to the manufacturer’s instructions. Bound KAP1 proteins were washed three times with 1× phosphate-buffered saline (PBS) and incubated with 0.25 or 1 U of PP1 in a total volume of 50 μl of incubation dilution buffer (Millipore, catalog # 20-169) at 37°C for 30 min. KAP1 proteins were subsequently eluted with 50 μl of 2× SDS sample buffer, and the extent of phosphorylation of KAP1 Ser824 was evaluated by Western blotting analysis with antibodies specific for the FLAG tag and KAP1 pSer824, respectively.

ChIP-ReIP assays

MCF-7 cells were treated with 1% formaldehyde in growth medium for 10 min to cross-link chromatin-DNA complexes. Cross-linking was terminated by the addition of glycine to a final concentration of 125 mM and incubation at 25°C for 10 min. Cells were rinsed with cold PBS, harvested, allowed to swell in 200 μl of SDS lysis buffer supplemented with protease inhibitor mixture (Roche Applied Science) and 1 mM phenylmethylsulfonylfluoride (PMSF) per 1 × 106 cells, and sonicated to shear the DNA to yield fragments of 500 base pairs (bp) or less. Samples were then diluted with 9 volumes of ChIP dilution buffer before being precleared for 1 hour with 40 μl of a mixture of protein A agarose and salmon sperm DNA. Agarose beads conjugated to an antibody against FLAG (60 μl) were added to the supernatant and incubated overnight. Bound proteins were eluted with 10 mM dithiothreitol (DTT) (30 μl), and eluates were diluted with 50× volumes (~1.5 ml) of ChIP dilution buffer before the second immunoprecipitation. Approximately 3 to 5 μg of antibodies specific for HA, acetylated K9 and K14 of histone H3, or dimethylated K9 of histone H3 or of mouse normal immunoglobulin G (IgG) were incubated overnight at 4°C with postclearance supernatant. Protein A agarose conjugated to salmon sperm DNA was added the next day and was incubated for an additional 4 hours. Washes were sequentially performed with low-salt buffer, high-salt buffer, LiCl buffer, and twice with tris-EDTA (TE) buffer [10 mM tris-HCl (pH 8.0), 1 mM EDTA]. Immunoprecipitates were eluted in 500 μl of elution buffer (1% SDS, 0.1 M NaHCO3) followed by the addition of 20 μl of 5 M NaCl. Cross-linking was reversed by overnight incubation of samples at 65°C and treatment with protease K (Sigma) for 2 hours at 45°C, and the recovered DNA was extracted with phenol-chloroform and precipitated with ethanol. Quantification was performed by real-time PCR with primer pairs against the −3038, −713, and −20 amplicons of p21, respectively, as previously described (table S1) (3) with the My IQ real-time PCR detection system and an IQ SYBR Green Supermix (Bio-Rad). Control IgG and input DNA values were used to normalize values from ChIP-ReIP samples. Three independent experiments were performed to generate the data shown in each figure panel, unless specified otherwise.

Acknowledgments

Acknowledgments: We are sincerely grateful to Y. Yen, M.-L. Kuo, R. Natarajan, and Y. Chen for their critical reading and helpful suggestions. We also thank Y.-C. Yuan and H. Li of Bioinformatics Core Laboratory at City of Hope for modeling analyses; the RNAi consortium at Academia Sinica for providing the lentivirus vectors against human KAP1, PP1α, PP1β, and I-2; S. Liu and A. Lee for technical assistance; members of the Ann laboratory for helpful discussions; and S. R. da Costa for editing. Funding: This work was supported in part by NIH Research Grants R01DE10742 and R01DE14183 (to D.K.A.), Natural Science Foundation of China Grants 30570371, 90608014, and 30711120570 (to X.X.), and National Science Council Grants 97-2321-B-001-107 and 98-2321-B-001-012 (to H.-M.S.). Author contributions: X.L., H.H.L., H.C., X.X., H.-M.S., and D.K.A. participated in the experimental design; X.L., H.H.L., and D.K.A. participated in data acquisition and analyses; and X.L. and D.K.A. wrote the manuscript. Competing interests: The authors declare no conflicts of interest.

Supplementary Materials

www.sciencesignaling.org/cgi/content/full/3/119/ra32/DC1

Methods

Fig. S1. Tetracycline-inducible knockdown of KAP1 decreases the proliferation of MCF-7 cells.

Fig. S2. The F370A and F370G mutations have opposing effects on the interactions of KAP1 with PP1α.

Fig. S3. The S440/501A and S440/501D double mutations of KAP1 do not affect its interaction with PP1α or PP1β or the predicted secondary structure of the PP1-docking motif.

Fig. S4. The F370G mutation of KAP1 attenuates the PP1α-dependent dephosphorylation of pSer824.

Fig. S5. PP1α and PP1β and the regulation of KAP1 Ser824 phosphorylation, RanGAP1 SUMOylation, and MCF-7 cell proliferation.

Fig. S6. PP1α and PP1β do not occupy the −713 and −3038 distal regions of p21 and they fail to affect the modification of K9 in histone H3 at the −3038 distal region.

Table S1. Primer pairs used in real-time PCR assays.

References

References and Notes

View Abstract

Navigate This Article