Identification and Characterization of HIPK2 Interacting with p73 and Modulating Functions of the p53 Familyin Vivo *

To study the biological role of p73α, a member of the p53 tumor suppressor family, we performed a yeast two-hybrid screen of a human cDNA library. Using a p73α fragment consisting of amino acids 49–636 as bait, we found that p73α is functionally associated with the human homologue of mouse and hamster homeodomain-interacting protein kinase 2 (HIPK2). The hamster homologue, also known as haHIPK2 or PKM, was used for further characterization of interactions between HIPK2 and members of the p53 protein family. Systematic yeast two-hybrid assays indicated a physical interaction between the oligomerization domains of p73α and p53 (amino acid regions 345–380 and 319–360, respectively) and amino acid region 812–907 of haHIPK2. This region of haHIPK2 includes a PEST sequence, an Ubc9-binding domain, and a partial speckle retention sequence and is identical to amino acid residues 846–941 of human HIPK2 (hHIPK2). The interaction was confirmed by glutathioneS-transferase pull-down assays in vitro and immunoprecipitation assays in vivo. HIPK2 colocalized with p73 and p53 in nuclear bodies, as shown by confocal microscopy. Overexpression of HIPK2 stabilized the p53 protein and greatly increased the p73- and p53-induced transcriptional repression of multidrug-resistant and collagenase promoters in Saos2 cells but had little effect on the p73- or p53-mediated transcriptional activation of synthetic p53-responsive and p21WAF1 promoters. Stable expression of HIPK2 in U2OS cells enhanced the cisplatin response of sub-G1 and G2/M populations, and it also increased the apoptotic response to cisplatin and adriamycin as demonstrated by fluorescence-activated cell sorter and 4′,6-diamidino-2-phenylindole-staining analyses. HIPK2 potentiated the inhibition of colony formation by p73 and p53. These results suggest that physical interactions between HIPK2 and members of the p53 family may determine the roles of these proteins in cell cycle regulation and apoptosis.

The tumor suppressor protein p53 is one of the most important regulators of cellular growth functions, such as cell cycle arrest, DNA repair, and apoptosis, and is mutated in about 50% of all human tumors (1). Some human tumors containing a normal p53 gene nonetheless exhibit functional inactivation of p53 at the protein level by cellular MDM2 or by viral oncoproteins, such as human papilloma virus E6, SV40 T antigen, and adenovirus E1B. As a transcriptional regulator, p53 modulates expression of various genes in response to cellular genotoxic stresses including DNA damage, oncogene activation, and hypoxia (2).
The effect of p53 on transcription is dependent upon the promoter context, type of stimulus, and cellular environment, so that it can function both as an activator and as a repressor (3). One gene activated by p53 is p21WAF1, which encodes an inhibitor of cyclin-dependent kinases and thus can initiate cell cycle arrest (4). Other positively regulated targets include the GADD45, MDM2, cyclin G, KILLER/DR5, IGF-BP3, and Bax genes, whose products function as regulators of several aspects of cell growth (5). Overexpression of p53 represses transcription of the MDR1 gene, which encodes a transmembrane glycophosphoprotein that mediates resistance to chemotherapeutic agents (6). Certain serum-inducible genes (7), cell cycle regulators (8 -10), viral enhancers (11,12), apoptotic regulators (13)(14)(15)(16)(17), and tumor progression genes (18 -22) are also negatively regulated by p53. The mechanism of transcriptional repression by p53 may involve direct binding to response elements, indirect interactions with other transcriptional regulators, or both. The C-terminal oligomerization domain (OD) 1 of p53 has been reported to be necessary for p53-mediated repression (23)(24)(25), and histone deacetylases (HDACs) and mSin3a also appear to be involved (26,27).
Recently, several groups identified p73 and p63 (also known as p51, KET, p40, p73L, and p53CP) as members of the p53 tumor suppressor family (28 -33). Like p53, both p73 and p63 can form homo-oligomers, bind to canonical p53 DNA binding sites, modulate transcription of p53-responsive genes, and suppress growth or induce apoptosis when overexpressed in certain human tumors (34 -37). Furthermore, their polypeptide sequences appear to contain the three principal domains of p53: (a) an N-terminal transcriptional activation domain (AD), (b) a sequence-specific DNA-binding domain (DBD), and (c) an OD that mediates tetramerization.
The p73 and p63 proteins are unlike p53 in certain notable ways, however. First, p73 and p63 undergo alternative splicing, giving rise to a family of isoforms of unknown physiological significance (38,39). Second, the C-terminal regions of the p73␣ and p63␣ isoforms harbor a sterile ␣ motif (SAM) domain that is not found in p53 and may be involved in protein-protein interactions or developmental regulation (30). Differences in the SAM domains of the different family members may reflect significant divergence in signaling and function (40,41). Third, distinct developmental abnormalities were observed in mice lacking either p73 or p63 (42,43) and in humans with germline p63 mutations (44). Finally, p73 and p63 are rarely mutated in tumors and are thus unlikely to be classical tumor suppressors (45). Thus, although certain characteristics of p73 and p63 are similar to those of p53, they cause some different physiological responses to extracellular signals and developmental cues. Here, we report our investigation into the molecular bases for the differences in function among these three proteins of the p53 family.
To study the biological role of p73, we first performed yeast two-hybrid screening of a human cDNA library with a p73 and found that p73 is functionally associated with the human homologue of mouse and hamster homeodomain-interacting protein kinase 2 (HIPK2). This protein has been identified previously as a nuclear serine/threonine kinase that interacts with the NK homeodomain transcription factor (46), acts as a corepressor for the NK homeodomain, and cooperates with Groucho and HDAC-1 in enhancing transcriptional repression (47). Furthermore, SUMO-1 modification of HIPK2 correlates with its localization to nuclear speckles (dots) or nuclear bodies (48). Yeast two-hybrid screens have shown that HIPK2 also interacts with various other proteins including interferon type I-induced MxA (49), Fas/CD95 (50), and HMGI(Y) (51). The biological significance of these interactions has not been determined. The human homologue of HIPK2 (hHIPK2) was recently cloned and mapped to chromosome 7q32-q34 (52).
HIPK2 has very recently been reported to activate p53-mediated transcription through down-regulation of MDM2, which may result in enhanced p53 protein levels (53). The two-hybrid results of our two-hybrid described here provide strong evidence for a functional association between other p53 family proteins and HIPK2. We have examined the sites of interaction on these proteins, the biological significance of these interactions, and whether association with HIPK2 stabilizes p53 family proteins. Using the hamster HIPK2 homologue (haHIPK2), also known as PKM, for further characterization both in vivo and in vitro, we have found evidence for a physical interaction between the OD of p53 family proteins and a region of ha-HIPK2 near a PEST sequence. These proteins colocalize in nuclear bodies, and the repressor functions of p73 and p53 are enhanced by HIPK2 in a dose-dependent manner. The corepressor activity of HIPK2 is abrogated in the presence of trichostatin A (TSA), an inhibitor of HDAC-1. Finally, modulation of transcription of p53 family proteins by HIPK2 correlates with inhibition of colony formation, suppression of the cell cycle, and induction of apoptosis.

MATERIALS AND METHODS
Cell Lines and Cell Culture-All cells used in our experiments were routinely maintained in RPMI 1640 medium supplemented with 10% heat-inactivated fetal bovine serum and penicillin/streptomycin (all from Invitrogen). Stock solutions of G418 (neomycin) (Invitrogen), adriamycin (Sigma), and cisplatin (Amersham Biosciences) were 2 mg/ml in H 2 O.
cDNA Constructions-All cDNAs were made according to standard methods (54) and verified by sequencing. The multicopy yeast expression plasmids used in the two-hybrid assays were pBTM116 (a Trp1-LexA DBD vector) and pASV3 (a Leu2-VP16 acidic transactivation domain vector). In these plasmids the target fusion genes are expressed under control of the ADH1 and phosphoglycerate kinase promoters, respectively (55). Deletion and point mutants of the desired genes were created by PCR amplification and subcloned into the pBTM116 or pASV3. Hemagglutinin (HA)-tagged p73␣, p73␤, and p53 expression plasmids were described previously (56) and modified by subcloning into yeast vectors for two-hybrid assays and pEGFP-N1 vector (CLON-TECH, Palo Alto, CA) for colocalization assays, respectively. Expression plasmids for p53 point and deletion mutants were kindly provided by Drs. K. Roemer (Hamburg, Germany) and C. C. Harris (Bethesda, MD). HA-tagged p63␣ (or p51B) and p63␥ (or p51A) expression plasmids were kindly provided by Dr. Y. Ikawa (Tokyo, Japan) and modified by subcloning into yeast vectors for two-hybrid assays. haHIPK2 cDNA (a kind gift from Otto Haller, Freiburg, Germany) was PCR-amplified to construct HA-tagged haHIPK2 in pCDNA-HA, a derivative of vector pCDNA3 encoding HA tag, and GFP-tagged haHIPK2 in pEGFP-N1 vector (CLONTECH) for colocalization assays. MDR1-CAT reporter plasmid was kindly provided by Dr. S. L. Doong (Taiwan, Taipei Republic of China). An expression plasmid for a glutathione S-transferase (GST)-haHIPK2 fusion protein was created by the subcloning of PCRamplified haHIPK2 into vector pGEX2T (Amersham Biosciences). Details of plasmid constructions are available upon request.
Yeast Two-hybrid Assays-A human liver cDNA library (Matchmaker, CLONTECH) in the prey plasmid pGAD10 (CLONTECH) was screened for proteins that interact with p73 using yeast reporter strain L40 (MATa, his3⌬200, trp1-901, leu2-3, 112, ade2, LYS:: (lexAop) 4 -HIS3, URA3::(lexAop) 8 -lacZ) (55). The transactivation domain of p73 was deleted so that amino acids 49 -636 of p73 were encoded as a fusion with LexA on the bait plasmid pBTM116. The prey and bait plasmids were cotransformed into L40 using lithium-acetate. Transformed cells were spread directly on minimal medium lacking histidine, leucine, and tryptophan and supplemented with 5 mM 3-aminotriazole. Positive clones were isolated and then retested for ␤-galactosidase activity on permeabilized cells. Library plasmids from positive isolates were transformed into and recovered from Escherichia coli strain HB101 (leu2 Ϫ ) and were then analyzed by restriction digests. Unique inserts were sequenced and analyzed by comparison to the GenBank TM sequence data bank. The longest insert was systematically tested for interactions with a protein containing the LexA DBD fused to p73␣- (49 -636) or to other p53 family proteins by ␤-galactosidase assays (56). To precisely map the p73 interaction domain on the haHIPK2 insert, full-length and deletion derivatives of haHIPK2 were fused with a VP16 AD by subcloning into vector pASV3. The resulting VP16 AD-haHIPK2 fusion vectors were cotransformed with a pBTM116 derivative encoding the LexA DBD-p73␣-(49 -636) fusion protein.
Immunoprecipitation-After transfection with plasmid DNA, U20S osteosarcoma cells were washed in phosphate-buffered saline (PBS), and cell lysates were prepared by adding 1 ml of ice-cold RIPA buffer (50 mM HEPES, pH 7.0, 250 mM NaCl, 0.1% Nonidet P-40, 5 mM EDTA) supplemented with protease inhibitors. Lysates were precleared by pre-incubation with protein A-Sepharose beads for 1 h and then incubated overnight at 4°C with protein A-Sepharose beads and a 1:200 dilution of either a mouse anti-HA monoclonal antibody (mAb) 12CA5 (Roche Molecular Biochemicals) or rabbit anti-GFP polyclonal antibody sc-8334 (Santa Cruz Biotechnology, Santa Cruz, CA). The beads were washed five times in RIPA buffer and twice with PBS, and the immune complexes were released from the beads by boiling in sample buffer for 5 min. Following electrophoresis on 8% SDS-polyacrylamide gels, immunoprecipitation products were analyzed by Western blotting using either rabbit polyclonal anti-GFP antibody sc-8334 (1:200), mouse anti-p53 mAb NCL-p53P (Novocastra, Burlingame, CA) (1:100), or mouse anti-p73 mAb Ab-2 (Neomarkers, Fremont, CA) (1:100).
Confocal Immunofluorescence Microscopy-Saos2 cells seeded on glass coverslips in 6-cm plates were transfected with pCDNA3 containing a gene for HA-p73␣ or HA-p53, pEGFP-N1 containing the gene for a GFP-haHIPK2 fusion protein, or both plasmids together. Two days later, cells were washed with PBS, fixed with 4% paraformaldehyde in PBS for 10 min at room temperature, and permeabilized for 4 min at 4°C in PBS containing 0.3% Triton X-100 and 10% goat serum. Subsequent incubations were performed at room temperature. After washing, cells were blocked for 60 min in PBS containing 3% bovine serum albumin and then incubated for 2 h with mouse anti-HA mAb 12CA5 diluted 1:200 in blocking buffer. Cells were then incubated with Texas Red-conjugated anti-mouse IgG N2031 (Amersham Biosciences) (1:50) for 1 h and mounted with 50 l of VectaShield (Vector Laboratories, Burlingame, CA). Microscopy was performed on a Bio-Rad Radiance 2000 MP confocal microscope.
Transient Transfection and CAT Enzyme-linked Immunosorbent Assays-Saos2 cells were transiently transfected in 6-well plates using LipofectAMINE (Invitrogen) according to the manufacturer's instructions. The total amount of transfected DNA including an SV40-driven ␤-galactosidase internal control plasmid, was maintained at 2 g by including appropriate control plasmids lacking inserts as indicated. After overnight transfection, cells were washed, fed with complete medium, and incubated for an additional 24 h. Cells were then washed twice with ice-cold PBS, collected, resuspended in 100 l of 0.25 M Tris-HCl, pH 7.6, and subjected to three freeze-thaw cycles. Cell lysates were cleared by centrifugation (12,000 rpm, 10 min, 4°C), and protein concentrations were determined using a Bio-Rad protein assay dye reagent. ␤-Galactosidase activity was determined in 96-well plates. Aliquots (30 -70 l) of the cleared lysates were assayed for CAT by a CAT enzyme-linked immunosorbent assay according to the manufacturer's instructions (Roche Molecular Biochemicals). The CAT concentration of each sample was normalized with respect to ␤-galactosidase activity.
Construction and Growth of Cells Stably Expressing haHIPK2-An expression vector for haHIPK2 was constructed by subcloning PCRamplified cDNA into the G418-resistant plasmid pCDNA3. U2OS cells were transfected with either this vector (pCDNA3-haHIPK2) or pCDNA3 in the presence of LipofectAMINE (Invitrogen) for 48 h and treated with 500 g/ml G418. After 5-7 days, cells were incubated with fresh culture medium containing G418, and G418-resistant colonies were selected for 2 months. Cell growth was followed by direct cell counting with a hemocytometer. Cells were grown in 6-well culture plates at an initial density of 10 5 cells in 2 ml of medium per well. Results were expressed as mean values of cell numbers for at least four duplicate wells.
DAPI Staining and FACS Analysis-U2OS stable cells were treated with 0.2 g/ml adriamycin (or doxorubicin) or 25 mM cisplatin for 72 h, washed with 1ϫ PBS, fixed with 70% ethanol for 20 min at room temperature, and washed again with 1ϫ PBS. Cells were then treated with DAPI (1 g/ml) (Sigma) for 12 min, washed with 1ϫ PBS for 5 min, and treated with 50 l of VectaShield (Vector Laboratories). DAPI staining of the cells was observed by fluorescence microscopy using a Provis AX70 microscope (Olympus Optical Co.).
After adriamycin or cisplatin treatment, adherent and detached cells were combined and fixed overnight in 70% ethanol in wash buffer (PBS containing 5 mM EDTA) at 4°C. After centrifugation at 3000 rpm for 1 min, cell pellets were incubated for 30 min with 500 l of wash buffer and 50 l of RNase A (10 mg/ml). Cells were then stained in 500 l of PBS containing 100 g/ml propidium iodide. A total of 1 ϫ 10 6 cells were analyzed by flow cytometry (FACS-vantage; Becton Dickinson Biosciences, San Jose, CA).
Colony Formation Assays-Saos2 or U2OS cells (5 ϫ 10 5 cells/dish) were transfected with 2 g of a pCDNA3 derivative carrying wild-type p53 or p73␣ and/or haHIPK2 using Effectene (Qiagen, Hilden, Germany) according to the manufacturer's protocol. G418-resistant colonies were selected by growth in 400 g/ml G418 for 2 weeks. The G418-resistant colonies were then stained with Coomassie Blue and counted. The experiments were repeated at least three times.

RESULTS
HIPK2 Interacts with Members of the p53 Family-The yeast two-hybrid system was used to identify proteins that interact with p73␣-(49 -636), which lacks the transactivation domain of p73␣ (Fig. 1A). We isolated 285 His gene-positive clones from a screen of a human liver cDNA library fused to the GAL4 AD. When tested for activation of a GAL1-lacZ reporter gene, which contains eight LexA-binding sites in the promoter region, 73 of the clones exhibited high-level expression of lacZ. We isolated and amplified the plasmids from these clones and subjected them to agarose gel electrophoresis. The insert sequences from a total of 42 unique library plasmids were used in a GenBank search, and five of these clones were found to be human homologues of mouse HIPK2 (mHIPK2) (46) and haHIPK2 (49). The longest insert corresponded to amino acids 552-945 of mHIPK2 and 552-918 of haHIPK2 (Fig. 1B). This clone showed a strong interaction with p73␣-(49 -636) but not with the unrelated IRF-2 protein in a yeast two-hybrid assay ( Fig. 2A). Subsequent yeast two-hybrid assays indicated that hHIPK2 interacts with other members of the p53 family including p73␤, p53, p63␣, and p63␥, all of which were truncated to remove their transactivation domains for these assays (Fig. 2B).
To map the HIPK2-interaction domain on p53 family proteins, genes for LexA DBD fusions of these proteins and their deletion derivatives were placed on pBTM116. The full-length LexA DBD-p73␣ clone showed transcriptional activity when cotransformed into yeast along with an expression plasmid encoding a GAL4 AD-hHIPK2 fusion protein. Assays with a series of C-terminal deletion mutants of p73␣ indicated that a region near the OD is required for interaction with HIPK2 (Fig.  3A). Further analysis with p73-(49 -380) and p73␣-(345-400) revealed that the minimal interaction domain is the region bounded by amino acids 345-380. Interestingly, another, inde- Like p73, p53 strongly associated with HIPK2 in two-hybrid assays. A region between amino acid residues 319 and 360 of p53, which harbors the OD, was found to be sufficient for mediating an interaction with HIPK2 (Fig. 3B). Likewise, the interaction domain in p63 was the amino acid segment 352-388, which includes the OD (Fig. 3C). A second interaction domain was mapped to near the C terminus of p63␣ that includes the SAM domain. Thus far, no functional domains have been identified near the C termini of p73␣ and p63␣.
Mapping of the p73-interacting Domain of haHIPK2-To map the p73-interaction domain on haHIPK2 proteins, genes for VP16 AD-haHIPK2 fusions of haHIPK2 and its deletion derivatives were placed on pASV3. These plasmids were cotransformed into yeast along with a LexA DBD-p73␣-(49 -636) expression plasmid, and levels of interaction were determined by ␤-galactosidase assays in permeabilized yeast (Fig. 4). In control assays, the LexA DBD-p73␣-(49 -636) fusion and the LexA DBD alone were assayed for activation in the presence of the VP16 AD. No increase in reporter activity was detected when LexA DBD-p73␣-(49 -636) was coexpressed with the VP16 AD. In contrast, a strong interaction was observed between LexA DBD-p73␣-(49 -636) and VP16 AD-haHIPK2. Analysis of a series of N-and C-terminal truncation mutants of HIPK2 revealed that a C-terminal amino acid segment 812-907 of haHIPK2 is required for interaction with p73␣. Further   FIG. 3. Specific mapping of the domains of p73 (A), p53 (B), and p63 (C) responsible for association with hHIPK2. Yeast two-hybrid and ␤-galactosidase assays were performed using genes for LexA DBD-fused truncations of the p53-family proteins carried on pBTM116 and the gene for GAL4 AD-fused hHIPK2 carried on pGAD10 as described under "Materials and Methods. "   FIG. 2. Human HIPK2 interacts with p53 family proteins. A, identification of hHIPK2 by yeast two-hybrid screening of genes encoding the GAL4 AD fused to a human liver cDNA library. B, interaction of p53-family proteins with hHIPK2 as analyzed by yeast two-hybrid and ␤-galactosidase assays. The genes for LexA DBD-fused p53-family proteins were carried on pBTM116, and the GAL4 AD-fused hHIPK2 gene was carried on pGAD10, as described under "Materials and Methods." deletions in that region abolished interaction with p73␣, and assays with three other mutants indicated that haHIPK2-(812-907) is sufficient for the interaction. Thus, this haHIPK2 region, which includes a PEST sequence, appears to be required and sufficient for the interaction with p73␣. Notably, the corresponding region of hHIPK2 has been shown to interact with several proteins including Ubc9 (48), Fas/CD95 (50), and HMGI(Y) (51).
Members of the p53 Family Associate with haHIPK2 in Vitro and in Mammalian Cells-To further examine whether a direct physical interaction between haHIPK2 and p53 family proteins is present, we performed in vitro binding assays. GST-p53 and GST-HIPK2 fusion proteins were expressed in E. coli, purified, and mixed separately with in vitro-translated [ 35 S]methionine-labeled haHIPK2, p53, p73␣, p73␤, p63␣, or p63␥ in a rabbit reticulocyte lysate. In agreement with the results of our yeast two-hybrid experiments, 35 S-labeled HIPK2 was retained by GST-p53 immobilized on glutathione-Sepharose beads (Fig. 5A). In a reciprocal experiment, 35 Slabeled p53 was retained by immobilized GST-HIPK2. Further assays with labeled p73 and p63 showed that HIPK2 associates with other members of the p53 family, although these interactions are somewhat weaker (Fig. 5, B and C). Overall, these data confirm that HIPK2 directly interacts with members of the p53 family in vitro.
To demonstrate that the interaction between HIPK2 and the p53 family also occurs in vivo, p53-null Saos2 mammalian cells were cotransfected with HA-tagged p73␣ or p53 and GFPtagged haHIPK2 constructs. Immunoprecipitation with an anti-HA antibody and subsequent Western blotting with an anti-GFP antibody revealed that both p73␣ and p53 interact with HIPK2 in vivo (Fig. 6A). In other experiments in which untagged p73␣ or p53 was cotransfected with GFP-tagged ha-HIPK2, both p73␣ and p53 were detected in precipitates obtained with an anti-GFP antibody. The interaction of p73␣ with haHIPK2 was weaker than that of p53 (Fig. 6, B and C). The kinase-defective HIPK2 mutant K221W (49) was also able to interact with p73␣ and p53 as expected from two-hybrid mapping experiments (data not shown).
HIPK2 and p53 Family Proteins Colocalize in Nuclear Bodylike Structures-We next sought to determine the subcellular localization of p73␣, p53, and HIPK2. For this purpose, p53null Saos2 cells were transiently cotransfected with plasmids encoding HA-tagged p73␣ or HA-tagged p53 and GFP-tagged HIPK2. Immunofluorescence and confocal laser microscopy showed that the majority of HIPK2 localized to nuclear speckles in agreement with previous reports (49). Texas Red staining of cells transfected singly with a p73␣ or p53 plasmid yielded a diffuse nuclear red fluorescence (data not shown). However, when p53 or p73␣ and HIPK2 plasmids were cotransfected, p53 and p73␣ were localized to distinct spots within the nucleus (Fig. 7, A and B). These findings are reminiscent of previous reports that PML targets p53 into PML speckles or nuclear bodies (57,58). Taken together, our in vitro and in vivo inter- FIG. 6. Immunoprecipitation assay for in vivo interactions between haHIPK2 and p73␣ or p53. U2OS cells were cotransfected with genes for fusion proteins GFP-haHIPK2 and HA-tagged p73␣ or p53. Cells were lysed in RIPA buffer supplemented with protease inhibitors, and lysates were precleared by incubation with protein A-Sepharose for 1 h. Immunoprecipitation was carried out with anti-HA or anti-GFP antibodies. Precipitated proteins were eluted with SDS sample buffer, resolved by electrophoresis on 10% SDS-polyacrylamide gels, and visualized by Western blotting using anti-GFP (A), anti-p73␣ (B), or anti-p53 (C) antibodies.
action data indicate that p53 and p73␣ are physiologically associated with HIPK2.
HIPK2 Modulates Transcriptional Regulation by Members of the p53 Family-To investigate the biological significance of the physical interaction between haHIPK2 and p53 family proteins, we measured the effect of HIPK2 expression on transcriptional regulation by p53 family proteins in vivo. When plasmids containing genes for HIPK2 and G5p53-CAT (a CAT reporter gene under the control of the p53 response element (a gift from Dr. Robbins, Pittsburgh, PA)) were cotransfected into mammalian cells, HIPK2 enhanced transcriptional activation by p53 (Fig. 8A). Activation of the reporter gene was also enhanced by p73␣ and p63␣, and HIPK2 potentiated this enhancement in a dose-dependent manner. Transient coexpression of genes for HIPK2 and p53 or p73␣ in the presence of a p21WAF1 promoter-CAT reporter also led to a weak enhancement of p21WAF1 expression (data not shown).
Several studies have implicated a role for transcriptional repression in p53-dependent cellular functions. To determine the effect of HIPK2 on the transrepression function of p53, we used the known target promoters for MDR1 and collagenase (Coll) fused to the CAT reporter gene. When MDR-CAT or Coll-CAT was cotransfected with the p53 expression vector, CAT expression was down-regulated. Coexpression of HIPK2 further increased the p53-mediated repression of the MDR-CAT gene (Fig. 8B) and the p53 family protein-mediated repression of the Coll-CAT gene (Fig. 8C) in a dose-dependent manner.
To determine whether HDACs are involved in p53/HIPK2mediated transcriptional repression, the HDAC inhibitor TSA was used to treat cells cotransfected with p53 and HIPK2 expression vectors and the Coll-CAT reporter plasmid. As shown in Fig. 8D, the down-regulation of the Coll-CAT gene by p53 in the absence and presence of HIPK2 was mitigated by TSA in a dose-dependent manner. We also observed that expression of a GAL4 DBD-HIPK2 fusion protein inhibited tran-scription of the 5ϫ GAL-TATA-CAT reporter gene, and that this inhibition was relieved by TSA (data not shown). Overall, our transfection data indicate that haHIPK2 modulates the transcriptional regulatory activity of p53 family proteins.
Stable Expression of HIPK2 Retards the Growth of U2OS Cells-To correlate transcriptional modulation by HIPK2 with its in vivo function, we first transiently transfected p53-positive U2OS cells with a haHIPK2 expression vector. Western blotting showed that HIPK2 increased the amount of p53 protein produced in these cells (data not shown). Next, we generated U2OS cells stably overexpressing HIPK2 (HIPK2-U2OS cells) and examined the effect of HIPK2 on the growth of these cells and on the levels of p53 and p21WAF1 proteins. As shown in Fig. 9A, stable overexpression of HIPK2 retarded U2OS cell growth during 5 days in a modest level (Fig. 9B). As in our transient transfection experiments, stable overexpression of HIPK2 also increased the amount of p53 in the cell. These observations raise the possibility that transiently or stably expressed HIPK2 stabilizes p53 in U2OS cells through direct protein-protein interactions, leading to enhanced p21WAF1 protein production and subsequent growth arrest. We did not detect any differences in levels of p73 protein in the presence or absence of stably expressed HIPK2, however (data not shown).
To further examine the role of HIPK2 in p73-and p53mediated apoptosis, we carried out FACS and DAPI analyses of HIPK2-U2OS and U2OS cells treated with DNA-damaging agents such as cisplatin or adriamycin. After cisplatin treatment, the DNA content of propidium iodide-stained cells was determined by FACS. In FACS analysis, a distinct quantifiable region below the G 1 phase (the sub-G 1 peak; designated as M1) is indicative of apoptotic cells. As shown in Fig. 10A, cisplatin treatment increased the sub-G 1 fraction (hypodiploid) by 2-fold compared with control cells (from 7.87 to 13.83%). Cisplatin treatment caused a further 2-fold increase (from 13.83 to 24.13%) in HIPK2-U2OS cells.
Morphological markers of apoptosis, such as cell shrinkage, nuclear segmentation, and chromatin condensation, were investigated by fluorescence microscopy following DAPI staining. These morphological changes were more evident in DAPIstained nuclei of HIPK2-U2OS cells after treatment with cisplatin or adriamycin (Fig. 10B). Like our FACS data, these studies showed that HIPK2 increases cell death induced by cisplatin or adriamycin. These overall data suggest that HIPK2 may be a critical mediator for the cellular functions of p73 and p53 such as cell cycle arrest and apoptosis.
HIPK2 Potentiates the Inhibition of Colony Formation by p53 Family Proteins-The roles of HIPK2 in mediating cellular functions of p53 family proteins were further investigated by colony formation assays. Saos2 and U2OS cells were transfected with expression vectors for HIPK2 and/or p53-or p73␣ or with a negative control vector (pCDNA3) and selected for resistance to G418. As shown in Fig. 11A, overexpression of p53 or HIPK2 alone in p53-null Saos2 cells dramatically reduced the number of G418-resistant colonies to only 17 or 15%, respectively, of those observed for the negative control. Overexpression of p73␣ only mildly suppressed the growth of Saos2 cells (65% of control), whereas p73␣ was undetectable (data not shown). When HIPK2 was cotransfected with p53 or p73␣, the number of colonies formed decreased to 1.5% (HIPK2 and p53) or 2.2% (HIPK2 and p73␣) of the control value. As was also observed in p53-positive U2OS cells, overexpression of p73␣ or HIPK2 resulted in a significant reduction in the number of colonies to 35 and 8% of the control value, respectively. Coexpression of p73␣ and HIPK2 severely inhibited colony formation in U2OS cells as well (Fig. 11B). Together, these observations suggest that HIPK2 significantly suppresses tumor cell growth independent of p53 or 73␣ and that the suppressing ability of HIPK2 increases in the presence of members of p53 or 73␣. However, it remains to be conclusively shown whether this synergistic effect is due to a direct interaction between HIPK2 and p53 family proteins. DISCUSSION The tumor suppressor protein p53 is one of the most important regulators of cellular growth functions such as cell cycle arrest, DNA repair, and apoptosis. Recently, two p53 family proteins, p73 and p63, were discovered that appeared to have functions similar to those of p53. To begin further investigations of the differential roles of these proteins, we performed yeast two-hybrid screens of a human liver cDNA library using p73␣ as the bait protein. These screens identified the human homologue of mouse and hamster HIPK2 as a p73␣-interacting protein. Fine mapping of the interaction domains indicated that the OD of the p53 family proteins and the amino acid segment 812-907 of hHIPK2, which includes PEST and SRS sequences and Ubc9 binding regions, are required for the interaction between p53 family proteins and HIPK2. The p53 OD is essential for the tetramerization that is required for the tumor-suppressive activity of p53 (59) and is also required for p53-mediated transcriptional repression (23)(24)(25). Interestingly, an additional interacting region was found in the Cterminal region of p73␣ and p63␣, which includes a SAM domain of undetermined function.
The amino acid sequence of the haHIPK2 interaction domain is identical to the sequences of amino acids 846 -941 and 839 -934 of human and mouse HIPK2, respectively. This region may be an interaction hot spot, as was described for the Mx interaction domain (49), the Ubc9-binding domain (48), and the CD95 binding site (50) that are all located near PEST/SRS sequences. However, this region is different from the NK homeodomain binding region (46). Because we have shown that the SRS of HIPK2 is involved in the interaction with p53 family proteins, it is not surprising that p73␣ and p53 colocalize to nuclear speckles.
The above findings are reminiscent of previous reports that PML targets p53 to PML speckles or nuclear bodies (57,58). It has been suggested that SUMO-1 modification of PML increases the formation of nuclear bodies and thus augments recruitment of p53 into these structures. Although the sumoylation of p53 family proteins has been reported (60 -62), it is unlikely that sumoylation is required for nuclear body localization (63). Because the sumoylation of HIPK2 directs its localization to the nuclear speckles, we expected that a physical association with HIPK2 might target p53 family proteins to the nuclear speckles as well. However, whether SUMO-1 modifica-tion of HIPK2 is required for localization of p53 family proteins to the nuclear speckles has yet to be determined. Other modifications of p53, such as phosphorylation by HIPK2 and/or acetylation by p300/CBP, may promote its colocalization into nuclear speckles with HIPK2. Because p300/CBP has also been demonstrated to accumulate in PML bodies (64), the network of interactions between p73/p53, PML, HIPK2, and p300/CBP may be important for various modifications of p73 and p53 that result in differential regulation of transcription in PML bodies. Clearly, more work is required to establish the impact of modifications of p73/p53 by HIPK2 and p300/CBP in these subnuclear structures.
Our in vitro and in vivo binding assays revealed that HIPK2 directly associates with p53 family proteins, suggesting that this association modulates the transcriptional regulatory activity of p73␣ and p53. Expression of HIPK2 in p53-null Saos2 cells increased the transactivation function of p53 and p53 family proteins as shown by the response of a CAT reporter gene. This response was dependent upon the kinase function of HIPK2 because it was abrogated when the kinase-defective HIPK2 mutant K221W was used (data not shown). Other transfection experiments indicated that HIPK2 enhances the transrepression activity of p73␣ and p53 with negatively p53responsive MDR and Coll-CAT reporter genes. Because this transrepression was prevented by treatment with TSA, this result supports the conjecture that an HDAC is critical for the corepressor function of HIPK2 (47). When the amount of mutant K221W in the cells was increased, the corepressor activity of HIPK2 diminished (data not shown), suggesting that K221W may function as a dominant-negative mutant.
Because HIPK2 kinase activity is required for modulation of the transactivation and transrepression functions of p73␣ and p53, the phosphorylation states of p73␣ and p53 and the mechanism for switching HIPK function between coactivation and corepression are of considerable interest. We have found that the interaction between HIPK2 and CBP involves amino acid residues 1-520 of HIPK2 and 662-1095 of CBP (data not shown). This region of HIPK2 includes the kinase domain. We speculate that interactions with CBP and HDAC are required for the coactivation and corepression functions of HIPK2, respectively. Two recent works provided some clues to how HIPK2 associates and cooperates with p53 and CBP in the p53-dependent transactivation by demonstrating that HIPK2 phosphorylates p53 at Ser-46 and that the phosphorylation promotes the CBP-mediated acetylation of p53 at Lys-382 in the nuclear bodies (65,66). However, it was not determined whether HIPK2 affects the functions of p73 and p63 and how HIPK2 mediates the p53-dependent transrepression. Therefore, precise determination of the acetylation/deacetylation levels of p53, p73, and p63 in the presence of CBP-or HDACassociated HIPK2 will provide further insight into the regulation of p53-family proteins in nuclear bodies.
The biological significance of the interactions between HIPK2 and p73␣ or p53 was examined using HIPK2-U2OS cells, which stably express HIPK2. In agreement with other studies (53), p53 levels in these cells were greater than in the parental U2OS cells. Levels of p53 were also greater in Saos2 cells transiently expressing HIPK2 than in parental Saos2 cells. In HIPK2-U2OS cells, the p53-responsive gene product p21WAF1 was up-regulated, and cell growth was slowed compared with U2OS cells. These cells also exhibited increased apoptosis after cisplatin or adriamycin treatment. Up-regulation of p53 family protein levels via protein stabilization is not likely to be a common mechanism for the effect of HIPK2, however, because p73 levels appeared to be unaffected by stable or transient expression of HIPK2. In other experiments, FIG. 11. Colony formation by Saos2 (A) or U2OS (B) cells transfected with a negative control vector or expression vectors for wild-type p53, p73␣, and/or haHIPK2. After selection of stably transfected cells for 2 weeks using G418, G418-resistant colonies were stained with Coomassie Blue, photographed, and counted. Numerical results represent the mean of three independent experiments. A photograph of one representative result is shown.
tumor cell growth was significantly inhibited by expression of either HIPK2 or p53 family proteins.
The p53-independent function of HIPK2 may reflect the presence of various cellular phosphorylation targets of HIPK2 that are involved in growth regulation independent of p53. This supposition is supported by the existence of other targets of HIPK2 including NK homeodomain transcription factor (46), interferon type I-induced MxA (49), Fas/CD95 (50), STAT3 (67), and HMGI(Y) (51). Alternatively, HIPK2 may suppress tumor cell growth through p53 family proteins other than p53 itself. Quantitative studies of the effects of HIPK2 on p73 and p63 expression in Saos2 and U2OS cells have not yet been reported. However, we found that the growth-suppressing activity of HIPK2 was synergistically increased in the presence of p53 or p73␣. Collectively, our results reveal a novel function for HIPK2 and support the speculation that a functional association between HIPK2 and members of the p53 family may be an important determinant of p53 family protein functions in regulation of the cell cycle and apoptosis in subnuclear structures.