Introduction
DNA damage induces activation of intricate signalling networks that control growth arrest, DNA repair, cellular senescence and cell death1, 2. These networks are governed by DNA damage-activated protein kinases, including ATM (ataxia telangiectasia-mutated) and ATR (ATM and Rad3-related), which coordinate the cellular response to DNA damage and oppose genomic instability and cancer development3, 4, 5.
The serine/threonine kinase homeodomain-interacting protein kinase 2 (HIPK2; ref. 6) is conserved in evolution and is a key regulator of the apoptotic programme induced by DNA damage7, 8, 9. HIPK2 is a tumour suppressor and has been shown to be inactivated in tumour cells through multiple mechanisms, including downregulation, mutation and mislocalization10, 11, 12, 13. DNA damage causes activation of HIPK2 through an unclear mechanism. HIPK2 forms a complex with the tumour suppressor p53, mediates phosphorylation of p53 at Ser 46 and stimulates CBP-mediated p53 acetylation at Lys 382 (refs 7, 8). Phosphorylation of p53 at Ser 46 is associated with lethal DNA damage and initiates the apoptotic programme through upregulation of pro-apoptotic p53 target genes, including Bax and p53AIP1 (refs 7, 8, 14). HIPK2 localizes to nuclear domains and when DNA is damaged, a fraction of HIPK2 is recruited to promyelocytic leukaemia nuclear bodies (PML-NBs) where HIPK2 associates with p53. This suggests that p53 Ser 46 phosphorylation takes place in PML-NBs7, 8, 15. Accordingly, tumour suppressor protein PML is an essential cofactor for HIPK2-mediated p53 Ser 46 phosphorylation16. Furthermore, Axin may also serve as an adaptor protein for HIPK2-dependent p53 phosphorylation17. During the initiation phase of apoptosis induced by DNA damage, the carboxy-terminal inhibitory domain of HIPK2 is removed by caspase-dependent cleavage, producing a hyperactive kinase that promotes p53 activation and apoptosis induction18. DNA damage can also drive p53-deficient cells into apoptosis, a process facilitated, at least in part, by HIPK2-mediated phosphorylation and subsequent proteasomal degradation of the anti-apoptotic co-repressor CtBP9. CtBP represses growth inhibitory and apoptotic target genes, including Noxa and Bax19. Given the apoptotic functions of HIPK2, it is important for unstressed and sublethally damaged cells to keep HIPK2 activity in check. However, the underlying mechanisms by which cells restrain or activate HIPK2 remain unclear.
In this study, we have investigated HIPK2 regulation in unstressed and DNA-damaged cells. We show that HIPK2 stability and function is controlled by the ubiquitin ligase seven in absentia homologue-1 (Siah-1) and the DNA-damage checkpoint kinases ATM and ATR.
Results
HIPK2 is unstable in unstressed cells and is stabilized upon DNA damage
To assess HIPK2 regulation, we UV-irradiated cells and analysed HIPK2 levels by immunoblotting. Damage caused a marked increase in HIPK2 levels (Fig. 1a). Similar upregulation of HIPK2 caused by cisplatin treatment, UV damage or ionizing radiation has been reported previously8, 20, 21; however, the underlying mechanism remained unclear. RT–PCR analysis revealed no upregulation of HIPK2 mRNA levels in damaged cells (Fig. 1b), indicating that HIPK2 accumulation originates from a post-transcriptional mechanism. Proteasome inhibition resulted in increased HIPK2 steady-state protein levels (Fig. 1c), which was accompanied by accumulation of polyubiquitin-reactive signals in the HIPK2 immune-complexes (Fig. 1d). This suggests that HIPK2 is regulated by the ubiquitin–proteasome system.
Figure 1: HIPK2, an unstable protein, is stabilized when DNA is damaged and its stability is regulated by Siah-1.
(a) HIPK2 levels increase after DNA damage. Cells were irradiated as indicated and analysed 24 h after treatment by immunoblotting (upper panels) and colony formation (lower panels). (b) RT–PCR analysis of HepG2 cells treated as indicated. (c, d) Proteasome inhibition increases HIPK2 levels (c) and HIPK2 ubiquitination (d) in HepG2 cells. (c) Cells were treated for 8 h with MG-132 or solvent and cell lysates were analysed by immunoblotting. (d) Immunoblot analysis of HIPK2 immunoprecipitated from MG-132 or solvent treated cells. (e) DNA damage causes stabilization of HIPK2. HT1080 cells were treated as indicated. After 24 h, protein synthesis was inhibited with 50 
g ml-1 cycloheximide (CHX). Cell lysates were analysed by immunoblotting. (d, e) Signals were quantified using the ImageQuant software. (f) Siah proteins induce HIPK2 degradation. H1299 cells were transfected as indicated along with pEGFP vector to normalize transfection. Cell lysates were analysed by immunoblotting. (g) Proteasome inhibition prevents Siah-1-mediated HIPK2 degradation. H1299 cells were transfected with the expression plasmids indicated. After 24 h, cells were treated for 12 h with solvent or MG-132 and analysed by immunoblotting. (h) Siah-1-mediated HIPK2 degradation is independent of Mdm2. mdm2;p53 double-knockout MEFs were transfected as indicated and cell lysates were analysed by immunoblotting. (i) Siah-1 controls HIPK2 steady-state levels. Endogenous Siah-1 and Siah-2 expression was downregulated in HepG2 cells using specific siRNAs. HIPK2 levels were analysed by immunoblotting. Knockdown efficiency was determined by RT–PCR and quantified by densiometry (lower panels). (j) Siah-1 regulates HIPK2 stability. Siah-1 was downregulated in HepG2 cells by RNAi and HIPK2 stability was determined using cycloheximide treatment and immunoblotting. Uncropped images of the scans in c, i and j are shown in Supplementary Information, Fig. S7.
Next, we determined HIPK2 stability in unstressed and DNA-damaged cells by blocking de novo protein synthesis with cycloheximide and analysing the remaining HIPK2 protein amounts by immunoblotting. In unstressed cells, HIPK2 had a short half-life of approximately 30 min (Fig. 1e). This was increased markedly after UV damage, as protein amounts remained virtually unchanged 120 min after cycloheximide treatment (Fig. 1e). Similar HIPK2 stabilization was also observed with ionizing radiation (Fig. 1e). These data indicate that HIPK2 is unstable in unstressed cells and is stabilized by DNA damage.
Ubiquitin ligase Siah-1 regulates HIPK2 steady-state levels
As HIPK2 is unstable in unstressed cells, we hypothesized that an unidentified ubiquitin ligase regulates HIPK2 turnover. To identify the ligase, we performed yeast two-hybrid screens using HIPK2 as bait22. Among several HIPK2-interacting clones we identified one full-length clone encoding Siah-1, a RING domain ubiquitin ligase implicated in fundamental cellular functions, including cell-cycle control and apoptosis23, 24, 25, 26, 27, 28, 29.
Expression of Siah-1 resulted in HIPK2 depletion, whereas ligase-deficient Siah-1 (Siah-1C44S) failed to do so (Fig. 1f). This indicates that ligase activity is required for Siah-1-mediated HIPK2 degradation. Expression of Siah-2, a Siah-1-related protein, also triggered HIPK2 degradation (Fig. 1f). Siah-1-dependent HIPK2 depletion was inhibited by the proteasome inhibitor MG-132, indicating that Siah-1 facilitates proteasomal degradation of HIPK2 (Fig. 1g).
Recently, HIPK2 was reported to be degraded by the ubiquitin ligase MDM2/HDM2 (ref. 30). Therefore, we used mdm2;p53 double-knockout cells31 to examine the role of MDM2 in Siah-1-mediated HIPK2 degradation. Siah-1 efficiently facilitated HIPK2 degradation in these cells, demonstrating that Siah-1 acts independently of MDM2 (Fig. 1h).
To determine whether Siah-1 regulates HIPK2 steady-state levels, we used RNA interference (RNAi) to downregulate Siah-1. Immunoblotting showed HIPK2 accumulation when Siah-1 or Siah-2 were depleted (Fig. 1i). Consistently, HIPK2 protein stability was markedly increased in response to Siah-1 knockdown (Fig. 1j). Taken together, these findings indicate that Siah-1 regulates HIPK2 steady-state levels.
Siah-1 mediates HIPK2 ubiquitination
Proteasome-dependent degradation is often induced by polyubiquitination. To determine whether Siah-1 facilitates HIPK2 ubiquitination, we performed in vitro ubiquitination assays using recombinant E1 and E2 enzymes. To clarify whether Siah-1 mediates HIPK2 ubiquitination directly, we expressed recombinant His–HIPK2, GST–Siah-1 and GST–Siah-1
RING in Escherichia coli and purified the proteins by affinity chromatography. Siah-1 efficiently facilitated HIPK2 ubiquitination in this assay system, in contrast to Siah-1RING, which failed to catalyse HIPK2 ubiquitination (Fig. 2a, b). We also purified His–HIPK2 under denaturing conditions from in vitro ubiquitination reactions by pulldown using Ni-NTA beads to ensure that the polyubiquitin-specific signals in our assay system are indeed caused by HIPK2 polyubiquitination and not by autoubiquitinated Siah-1 (Fig. 2b). Consistent with our finding that Siah-1 facilitates HIPK2 degradation independently of MDM2 (Fig. 1h), HIPK2 was polyubiquitinated by Siah-1 when the proteins were immunopurified from cell lysates of mdm2;p53 double-knockout cells (Supplementary Information, Fig. S1). Thus, Siah-1 directly mediates HIPK2 polyubiquitination.
Figure 2: Siah-1 mediates HIPK2 polyubiquitination and degradation.
Siah-1 mediates HIPK2 ubiquitination in vitro. (a, b) In vitro ubiquitination of HIPK2 using purified recombinant proteins. Proteins were expressed in E.coli and purified by affinity chromatography. For the assay shown in b, His–HIPK2 was purified under denaturing conditions using Ni-NTA beads from the in vitro ubiquitination reactions. Reactions were analysed by immunoblotting with a polyubiquitin and HIPK2-specific antibody. Input controls are shown. (c) Siah-1 mediates HIPK2 ubiquitination in vivo. 293T cells were transfected with expression vectors encoding HA–ubiquitin, His–HIPK2 and HA–Siah-1 or HA–Siah-1C44S. After 20 h cell lysates were prepared and His–HIPK2 was pulled down under denaturing conditions with Ni-NTA beads. Eluted proteins were boiled in SDS loading buffer and analysed by immunoblotting. Aliquots of the extracts were analysed for protein expression by immunoblotting (lower panels). Uncropped images of the scans are shown in Supplementary Information, Fig. S7. (d, e) Mapping of the HIPK2 region required for degradation by Siah-1. H1299 cells were transfected with expression vectors encoding Flag–HIPK2 wild-type (WT) or Flag–HIPK2 deletion mutants and HA–Siah-1 as indicated. An EGFP expression vector was included for normalization of the transfections. After 24 h, total cell lysates were prepared and analysed by immunoblotting. (e) A graphical overview, summarizes the results shown in d. The p53, axin (Ax) and Siah-1 binding domain of HIPK2 are indicated.
Full size image (141 KB)To assess whether Siah-1 also catalyses HIPK2 ubiquitination in vivo, we co-expressed His-tagged HIPK2 with wild-type Siah-1 or ligase-deficient Siah-1C44S in the presence of a proteasome inhibitor and pulled down His–HIPK2 from cell lysates using Ni-NTA beads under denaturing conditions. Immunoblot analysis showed clear HIPK2 polyubiquitination in the presence of wild-type Siah-1, but not by ligase-deficient Siah-1 (Fig. 2c). Together, these results show that Siah-1 facilitates HIPK2 polyubiquitination.
Mapping of the HIPK2 region required for Siah-1-dependent degradation
We used various HIPK2 deletion mutants to map the region required for Siah-1-mediated HIPK2 degradation (Fig. 2d, e). Deletion of amino acids 600–800 (HIPK2600–800) rendered HIPK2 resistant to Siah-1-mediated degradation (Fig. 2d), demonstrating a crucial role of this region in Siah-1-mediated HIPK2 proteolysis. Within this region, HIPK2 harbours a potential Siah-binding motif32. Unexpectedly, deletion of this motif (724–734) did not abolish Siah-1-dependent proteolysis or HIPK2–Siah-1 binding (Supplementary Information, Fig. S2), demonstrating that it is dispensable for HIPK2–Siah-1 interaction and proteolysis.
To identify Lys residues essential for Siah-1-dependent HIPK2 proteolysis, we examined Lys 25 and Lys 1183, two residues previously implicated in HIPK2 stability and post-translational modification22, 30, 33. Mutation of these residues failed to protect HIPK2 from Siah-1-dependent degradation (Supplementary Information, Fig. S3). Next, we used a truncated HIPK2 mini-protein (amino acids 551–1191), which binds Siah-1 (Fig. 3d) and is degraded in a Siah-1-dependent manner (Fig. 2e). Exchange of Lys 545, 558, 664, 796 and 798 for Arg (HIPK2 545–1191K5R) inhibited Siah-1-dependent degradation of the mini-protein (Supplementary Information, Fig. S4). However, mutation of these Lys residues in the HIPK2 full-length context (HIPK2K5R) failed to prevent Siah-1-mediated HIPK2 proteolysis (Supplementary Information, Fig. S4), suggesting that Siah-1 can switch to other Lys residues to target HIPK2 for proteolysis. Thus, multiple Lys residues seem to be used to target HIPK2 for Siah-1-dependent degradation.
Figure 3: Colocalization and in vitro and in vivo interaction of Siah-1 and HIPK2.
(a–c) HIPK2 and Siah-1 colocalize. U2OS cells expressing HA–Siah-1 were fixed and Siah-1 (red) was detected by indirect immunofluorescence microscopy. Nuclear DNA (blue) was stained with Draq5. (b) Confocal image showing colocalization (yellow) of GFP–HIPK2 (green) and HA–Siah-1C44S (red) in U2OS cells. (c) Confocal image showing colocalization (yellow) of endogenous HIPK2 (green) and Siah-1 (red) in WI-38 fibroblast cells. (a–c) Scale bars, 10 m. (d, e) Siah-1 and HIPK2 interact in vitro. GST and GST–Siah-1 proteins were incubated with the indicated in vitro translated 35S-labelled HIPK2 proteins. GST pulldowns were analysed by SDS–PAGE and Coomassie brilliant blue staining (lower panels). Gels were dried and exposed to X-ray films. Autoradiograms are shown (upper panels). A schematic drawing indicating the Siah-1 binding domain of HIPK2 is shown (e, bottom panel). (f–h) HIPK2 and Siah-1 interact in vivo. (f) Lysates from H1299 cells expressing HA–Siah-1C44S and the indicated HIPK2 proteins were immunoprecipitated (IP) with Flag antibodies. Immune complexes were analysed by immunoblotting (upper panels). Aliquots of the lysates were analysed for protein expression by immunoblotting (lower panels). (g) Lysates of 293T cells expressing HA–Siah-1C44S were immunoprecipitated with IgG or HIPK2 antibodies. Immune complexes were analysed by immunoblotting. 5% of the cell lysates used for immunoprecipitation was loaded as control. (h) Interaction of endogenous HIPK2 and Siah-1. WI-38 cells were crosslinked and lysates were subjected to immunoprecipitation with control and HIPK2 antibodies (left panels) or Erk2 antibody (right panels). Immune complexes were analysed by immunoblotting. Lysates of H1299 cells expressing inducible p53 to drive endogenous Siah-1 expression (see Fig. 5c) were used as positive controls for Siah-1 detection (control). Uncropped images of the scans in f and g are shown in Supplementary Information, Fig. S7.
Siah-1 colocalizes and interacts with HIPK2
Immunofluorescence staining and confocal microscopy showed that ectopically expressed Siah-1 localized predominantly to the nucleus (Fig. 3a). As overexpression of Siah-1 triggers HIPK2 degradation, we co-expressed HIPK2 along with ligase-deficient Siah-1C44S. HIPK2 and Siah-1 colocalized in nuclear domains (Fig. 3b). Similar colocalization was found for endogenous HIPK2 and Siah-1 in WI-38 fibroblasts (Fig. 3c).
To confirm the interaction between Siah-1 and HIPK2 suggested by our yeast two-hybrid screen, we performed GST-pulldown assays using recombinant GST-fusion proteins and in vitro translated 35S-labelled proteins. Specific interaction of 35S-labelled HIPK2 with GST–Siah-1 was observed (Fig. 3d) and the HIPK2 region spanning amino acids 551–1191 was sufficient for Siah-1 binding (Fig. 3d).
Interaction-mapping demonstrated that HIPK2545–800 and HIPK2600–800, which were resistant to Siah-1-dependent degradation (Fig. 2d, e), are deficient in Siah-1 binding in vitro (Fig. 3e) and in intact cells, as demonstrated by co-immunoprecipitation assays (Fig. 3f). These results indicate that lack of Siah-1 binding accounts for the resistance of these HIPK2 mutants to Siah-1-dependent degradation. Additional in vitro mapping assays suggest that Siah-1 binds HIPK2 through a region which harbours its substrate-binding domain34 (Supplementary Information, Fig. S5).
To assess whether Siah-1 interacts with endogenous HIPK2, we expressed HA–Siah-1C44S in 293T cells and performed co-immunoprecipitation assays. We found that Siah-1 co-precipitated efficiently with endogenous HIPK2 (Fig. 3g). Endogenous Siah-1 also co-precipitated with endogenous HIPK2, but not with the abundantly expressed mitogen-activated protein kinase (MAPK) Erk2 from lysates of WI-38 cells, which were in vivo crosslinked with DTBP (Fig. 3h). These results demonstrate that HIPK2 and Siah-1 interact in vitro and in vivo.
HIPK2 is degraded by the proteasome during recovery from DNA damage
By comparing lysates from sublethally and lethally UV-irradiated cells by immunoblotting, we found that although HIPK2 levels increased in response to both lethal and sublethal damage, HIPK2 was depleted during recovery from damage in numerous cell lines, confirming that the effect was not cell-line specific (Fig. 4a–e). Similar depletion of HIPK2 was also found with sublethal adriamycin treatment (Supplementary Information, Fig. S6). Consistent with a previous report14, p53 Ser 46 phosphorylation and apoptosis induction were specifically associated with lethal damage (Fig. 4a, b). Sublethally damaged cells recovered and were able to proliferate, as revealed by colony formation assays (Fig. 4a–e).
Figure 4: Siah-1 controls HIPK2 degradation during recovery from damage and regulates DNA damage-induced apoptosis.
(a–e) HIPK2 is degraded on sublethal DNA damage. Cells were treated with sublethal or lethal UV irradiation and were analysed by immunoblotting or colony formation. Data shown in b are means 
s.d. of three independent experiments. (f) HepG2 cells were treated as indicated and analysed by immunoblotting. (g) Proteasome-dependent HIPK2 degradation during recovery from damage. HepG2 cells were UV-irradiated and treated 24 h later with MG-132, zVAD-fmk or DMSO as a solvent control. Cell lysates were analysed by immunoblotting. (h–k) Siah-1 is important for recovery from DNA damage and HIPK2 degradation. (h) HepG2 cells were transfected with the siRNAs indicated and 24 h later RT–PCR analysis was performed. (i, j) Siah-1 is important for recovery from UV damage. Cell-cycle analysis of HepG2 cells transfected with the siRNAs indicated and treated with 20 J m-2 UV. Sub-G1 DNA content (apoptotic DNA fragmentation) was analysed by propidium iodide staining and FACS analysis. Results shown in j are means of three independent experiments. (k) Siah-1 is essential for HIPK2 degradation on recovery from damage. HepG2 cells transfected with the siRNAs indicated were irradiated with UV and lysates were analysed by immunoblotting. (l) HIPK2 degradation during recovery from damage is independent of HDM2 in HepG2 cells. (m) Siah-1 silencing potentiates apoptosis induced by lethal DNA damage. Cell-cycle (FACS, upper panels) and immunoblot analysis (lower panels) of HepG2 cells transfected with the indicated siRNAs and treated with lethal UV irradiation. Uncropped images of the scans in a, c, k and m are shown in Supplementary Information, Fig. S7.
Time course analysis revealed that HIPK2 levels declined sharply 32 h after damage (Fig. 4f), indicating that HIPK2 is rapidly degraded during the recovery phase. Proteasome inhibition prevented HIPK2 degradation during recovery from damage, whereas a pan-caspase inhibitor (zVAD) had no effect (Fig. 4g). Thus, HIPK2 is degraded by the proteasome during recovery from DNA damage.
Siah-1 controls HIPK2 degradation during recovery from damage and regulates DNA damage-induced apoptosis
To address the relevance of Siah-1 in HIPK2 degradation during recovery from DNA damage, we knocked down Siah-1 by RNAi (Fig. 4h–k). This resulted in HIPK2 accumulation and apoptosis induction after otherwise sublethal DNA damage (Fig. 4i–k), demonstrating that Siah-1 is essential for HIPK2 degradation and recovery from DNA damage.
Elevation of HIPK2 levels caused by Siah-1 depletion did not result in p53 Ser 46 phosphorylation after mild DNA damage (Fig. 4k), confirming that p53 Ser 46 phosphorylation is associated with lethal damage14, 35. As HIPK2 can also induce apoptosis by facilitating degradation of the anti-apoptotic co-repressor CtBP9, we analysed CtBP levels. Indeed, elevated HIPK2 levels after Siah-1 knockdown coincided with CtBP degradation (Fig. 4k), suggesting that the apoptotic response is activated under these conditions through the HIPK2–CtBP pathway.
It has been reported previously that HIPK2 degradation in 293 cells is regulated by MDM2/HDM2 (ref 30). To analyse a potential role of HDM2 in HIPK2 degradation in HepG2 cells we downregulated HDM2. Consistent with previous reports, HDM2 silencing resulted in upregulation of p53 (Fig. 4l), which is an established substrate for HDM2 (ref. 36). HDM2 depletion did not prevent HIPK2 degradation on recovery from damage in HepG2 cells (Fig. 4l), suggesting that the action of HDM2 may be cell-type specific.
We also investigated whether Siah-1 depletion affects apoptosis in response to lethal DNA damage. Siah-1 silencing resulted in increased HIPK2 steady-state levels and potentiated apoptosis induction (Fig. 4m), which was accompanied by an increased reduction in HIPK2 levels, probably caused by amplified caspase-dependent HIPK2 processing under these apoptosis-potentiating conditions18. Interestingly, Siah-1 depletion did not additionally increase HIPK2 levels after lethal DNA damage when compared with control cells, indicating that DNA damage uncouples HIPK2 from Siah-1-dependent negative control.
p53 regulates Siah-1-dependent HIPK2 degradation
As Siah-1 is a p53 target gene27, 28, 29, 37, 38, we assessed the effect of p53 on HIPK2 degradation in cells recovering from damage. Notably, HIPK2 steady-state levels were elevated in p53-deficient cells, which, in contrast to p53-proficient cells, failed to downregulate HIPK2 during recovery from damage (Fig. 5a). Comparable results were obtained using a p53-proficient and p53-deficient isogenic cell system (Fig. 5b), confirming the role of p53 in HIPK2 degradation.
Figure 5: p53 regulates Siah-1-dependent HIPK2 degradation.
(a) HepG2 (p53+/+) or Hep3B (p53-/-) cells were left untreated or UV-irradiated. Cell lysates were prepared and analysed by immunoblotting. (b) HCT116 p53+/+ and p53-/- cells were sublethally UV-irradiated and analysed by immunoblotting. (c) H1299–tet-p53 and (d) H1299–tet-p53H175D cells were treated with doxycycline (doxy) to induce expression of the p53 proteins. Cell lysates were analysed by immunoblotting. (e) H1299–tet-p53 cells were left untreated or treated with doxycycline for 18 h. 12 h before collection, cells were treated with solvent (DMSO), MG-132 or zVAD-fmk. Cell lysates were analysed by immunoblotting. (f) p53-driven HIPK2 degradation is independent of Mdm2. mdm2;p53 double-knockout cells were transfected and treated as indicated with MG-132 or the pan-caspase inhibitor zVAD-fmk. Lysates were analysed by immunoblotting. (g, h) Siah-1 is crucial, whereas HDM2 is dispensable, for p53-induced HIPK2 degradation. H1299–tet-p53 cells were transfected with control or Siah-1 siRNA (g) or control or HDM2 siRNAs (h). Cells were treated with doxycycline for 14 h or left untreated and lysates were analysed by immunoblotting. Uncropped images of the scans in g and h are shown in Supplementary Information, Fig. S7.
Full size image (87 KB)We used an inducible p53 expression system to further investigate the role of p53 in HIPK2 depletion. Induction of wild-type p53 efficiently triggered degradation of endogenous HIPK2 (Fig. 5c), indicating that p53 (in absence of other stress-activated pathways) is sufficient to drive HIPK2 degradation. Notably, HIPK2 depletion correlated with p53-induced Siah-1 upregulation, whereas mutant p53H175D failed to induce Siah-1 expression and HIPK2 degradation (Fig. 5d).
To analyse whether p53-driven HIPK2 depletion in this system is facilitated through proteasomal degradation or caspase-dependent HIPK2 cleavage18, we inhibited proteasome and caspase activity. In contrast to HIPK2 degradation during recovery from DNA damage (Fig. 4g), both proteasome and caspase inhibition partially prevented HIPK2 depletion, indicating that p53 engages HIPK2 degradation in this system through both mechanisms (Fig. 5e).
MDM2 is a p53-responsive gene36, 39 and was thought to mediate HIPK2 degradation30. To determine whether p53-mediated HIPK2 proteolysis requires MDM2 we used mdm2;p53 double-knockout cells. Caspase-dependent HIPK2 proteolysis by p53 (ref. 18) was excluded in our experimental setting by treatment with a pan-caspase inhibitor. p53 expression efficiently triggered HIPK2 degradation in mdm2;p53-deficient cells (Fig. 5f), demonstrating that p53-driven HIPK2 degradation does not require MDM2.
We knocked down Siah-1 expression to analyse the relevance of Siah-1 in p53-driven HIPK2 degradation. Strikingly, Siah-1 silencing rescued HIPK2 from p53-driven degradation (Fig. 5g). In contrast to a previous report30, HDM2 downregulation failed to inhibit p53-driven HIPK2 degradation (Fig. 5h). These results indicate that Siah-1 is essential for p53-mediated HIPK2 depletion.
ATM and ATR are essential for HIPK2 stabilization induced by DNA damage
Remarkably, HIPK2 is stabilized in response to DNA damage (Fig. 1), although p53, which drives Siah-1-dependent HIPK2 degradation (Fig. 5), is activated under these conditions. Therefore, we hypothesized that DNA damage activates a pathway that protects HIPK2 from degradation. We have previously shown that irradiation damage causes HIPK2 activation through an ATM-dependent pathway21.
To study the role of the checkpoint kinases ATM and ATR in HIPK2 stabilization we inhibited ATM/ATR with caffeine and analysed HIPK2 levels following DNA damage. Caffeine blocked HIPK2 stabilization induced by DNA damage, whereas a MAPK inhibitor (PD98059) had no effect (Fig. 6a). Accordingly, ATR downregulation by RNAi comparably inhibited HIPK2 stabilization, confirming an important role of ATR in UV-induced HIPK2 stabilization (Fig. 6b).
Figure 6: ATM and ATR regulate HIPK2 stability and HIPK2–Siah-1 interaction.
![Figure 6 : ATM and ATR regulate HIPK2 stability and HIPK2|[ndash]|Siah-1 interaction.](/ncb/journal/v10/n7/thumbs/ncb1743-f6.jpg)
(a) ATR/ATM blockade inhibits HIPK2 stabilization. HepG2 cells pre-treated with the indicated compounds (5 mM caffeine, 50 M PD98059) were UV-irradiated and 24 h later, cell lysates were analysed by immunoblotting. (b) ATR is required for UV-induced HIPK2 stabilization. HepG2 cells transfected with control or ATR-specific siRNAs were UV-irradiated as indicated and 24 h later, total lysates were analysed by immunoblotting. (c) ATM/ATR checkpoint activity is important for HIPK2 stability. UV-damaged HepG2 cells were left untreated or treated with 5 mM caffeine. Total cell lysates were analysed by immunoblotting. (d, e) ATR and ATM protect HIPK2 against Siah-1-mediated degradation. H1299 cells were transfected as indicated and analysed by immunoblotting. For ATR and ATM detection aliquots of the lysates were subjected to immunoprecipitation with anti-Flag antibodies and immunoprecipitates were analysed by immunoblotting. (f) UV damage attenuates HIPK2/Siah-1 binding. H1299 cells transfected with the indicated expression constructs were left untreated or UV-irradiated. HIPK2–Siah-1 interaction was assessed by immunoprecipitation analysis (upper panels). Protein expression was confirmed by immunoblotting of cell lysates (lower panels). (g, h) ATM and ATR disrupt the HIPK2–Siah-1 complex. H1299 cells were transfected as indicated and treated with MG-132. HIPK2–Siah-1 interaction was assessed by immunoprecipitations (upper panels). Aliquots of the lysates were analysed by immunoblotting (lower panels). (i) ATM kinase activity is essential for HIPK2–Siah-1 complex disruption. H1299 cells expressing the indicated proteins were treated with MG-132 and 10 M Ku55933 or solvent. Immunoprecipitates and cell lysates were analysed by immunoblotting. Uncropped images of the scans in g, h and i are shown in Supplementary Information, Fig. S7.
To determine whether continuous ATM/ATR checkpoint signalling is required to maintain HIPK2 stability after DNA damage, we inhibited ATM/ATR with caffeine and measured HIPK2 stability. ATM/ATR inhibition markedly impaired HIPK2 stability in damaged cells (Fig. 6c), indicating that continuous ATM/ATR activity is essential for maintaining DNA damage-induced stabilization of HIPK2.
We also examined whether ATR and ATM expression per se are sufficient to shield HIPK2 from Siah-1-dependent degradation. Remarkably, ATM and ATR rescued HIPK2 from Siah-1-mediated proteolysis (Fig. 6d, e). Collectively, these results indicate that DNA damage uncouples HIPK2 from Siah-1-dependent negative regulation through an ATM/ATR-dependent mechanism.
Regulation of the HIPK2/Siah-1 complex by ATM and ATR
To gain insight into the mechanism by which ATM/ATR controls HIPK2 stability, we studied the effect of DNA damage on the interaction between HIPK2 and Siah-1. Co-immunoprecipitation assays showed that HIPK2–Siah-1 complex formation was attenuated after DNA damage (Fig. 6f). As UV damage preferentially triggers ATR activation, which in turn may activate ATM40, we examined the role of ATM and ATR in the regulation of the HIPK2–Siah-1 complex. Ectopic expression of ATR and ATM, which results in their activation as determined by phosphorylation of the substrate p53 (see Fig. 7a and data not shown), was sufficient to trigger disruption of the HIPK2–Siah-1 complex (Fig. 6g, h). ATM-mediated HIPK2–Siah-1 complex disruption was prevented by the ATM-specific inhibitor Ku55933 (ref. 41; Fig. 6i), indicating that the kinase function of ATM regulates HIPK2–Siah-1 complex dissociation.
Figure 7: ATM and ATR phosphorylate Siah-1 at Ser 19 and regulate HIPK2–Siah-1 interaction.
![Figure 7 : ATM and ATR phosphorylate Siah-1 at Ser 19 and regulate HIPK2|[ndash]|Siah-1 interaction.](/ncb/journal/v10/n7/thumbs/ncb1743-f7.jpg)
(a, b) ATM phosphorylates Siah-1 in vitro. Autoradiogram showing in vitro phosphorylation assays for ATM wild-type (WT) or kinase-dead (KD) ATM. (a) Phosphorylated p53, Siah-1 and autophosphorylated ATM are marked by arrows (left panel). Aliquots of the immunoprecipitated substrates and ATM proteins used for the phosphorylation assays were analysed by immunoblotting (input controls, right panel). (b) Autoradiogram showing an ATM kinase assay using GST–Siah-1 1–80 and GST–Siah-1 80–282 (upper panel). GST protein loading was analysed by Coomassie staining (input control). (c) ATM phosphorylates Siah-1 at Ser 19 in vitro. ATM kinase assay with GST–Siah-1 1–80 and GST–Siah-1 1–80S19A proteins as substrates is shown (upper panel). Protein input (lower panel) was analysed by Coomassie staining and immunoblotting. (d) Phospho-specific Siah-1 Ser 19 antibodies recognize phosphomimetic Siah-1S19D. Phospho-deficient HA–Siah-1S19A and phosphomimetic Siah-1S19D were precipitated from 293T cell lysates and analysed by immunoblotting. (e, f) ATM (e) and ATR (f) phosphorylate Siah-1 Ser 19 in vivo. HA–Siah-1 was precipitated from 293T cell lysates and analysed by immunoblotting. ATM and ATR expression was confirmed by immunoprecipitation and immunoblotting. (g) Reduced HIPK2-degrading capacity of phosphomimetic Siah-1. H1299 cells transfected with the expression constructs indicated were analysed by immunoblotting. (h) Siah-1 Ser 19 is essential for ATM-mediated HIPK2–Siah-1 complex disruption. H1299 cells were transfected as indicated and treated with MG-132. HIPK2–Siah-1 interaction was assessed by immunoprecipitations (left panels). Aliquots of the cell lysates were analysed by immunoblotting (right panels). (i) Decreased interaction of phosphomimetic Siah-1 and HIPK2. H1299 cells were transfected as indicated and treated with MG-132. HIPK2–Siah-1 interaction was assessed by immunoprecipitations (upper panels). Uncropped images of the scans in e, f and h are shown in Supplementary Information, Fig. S7.
Full size image (117 KB)ATM and ATR phosphorylate Siah-1 at Ser 19 to modulate HIPK2–Siah-1 interaction
We speculated that ATM may regulate the HIPK2–Siah-1 module through phosphorylation. Thus, we examined whether Siah-1 is an ATM substrate. In fact, in vitro phosphorylation assays demonstrated ATM-mediated phosphorylation of Siah-1 (Fig. 7a). ATM preferentially phosphorylated the N-terminal 80 amino acids, which comprise the RING domain of Siah-1 (Fig. 7b). As ATM phosphorylates its substrates preferentially at SQ/TQ motifs3, 4, 5, we scanned this region for potential phosphorylation motifs. A single SQ motif at Ser 19 was identified and substitution of Ser 19 to Ala (S19A) inhibited ATM-mediated Siah-1 phosphorylation in vitro (Fig. 7c).
To study the in vivo phosphorylation of Siah-1 we raised an antibody against a Siah-1 peptide harbouring phosphorylated Ser 19. The affinity-purified antibodies showed selective reactivity against a phosphorylation-mimetic Siah-1S19D point mutant (Fig. 7d), indicating that it recognizes phosphorylated Ser 19. Consistent with our in vitro findings, ATM mediated Siah-1 Ser 19 phosphorylation in vivo (Fig. 7e). Furthermore, ATR also induced Ser 19 phosphorylation, and this was abolished by caffeine treatment (Fig. 7f). Thus, both ATM and ATR mediate Siah-1 phosphorylation in vivo. Due to lack of antibodies capable of precipitating endogenous Siah-1 we were not able to address the phosphorylation of endogenous Siah-1.
Next we assessed whether Ser 19 phosphorylation has an effect on Siah-1-dependent HIPK2 degradation. Phosphomimetic Siah-1S19D showed a reduced HIPK2-degrading capacity when compared with wild-type Siah-1 (Fig. 7g), supporting a regulatory function of Ser 19. Phospho-deficient Siah-1S19A was as efficient as wild-type Siah-1 in mediating HIPK2 degradation (Fig. 7g), excluding the possibility that Ser 19 is essential for Siah-1 function. Therefore, the effect of phosphomimetic Siah-1S19D is specific and not due to a general blockade of Siah-1 function.
We also analysed whether Siah-1 Ser 19 phosphorylation modulates HIPK2–Siah-1 interaction. In contrast to the HIPK2–Siah-1 wild-type complex, the HIPK2–Siah-1S19A complex was partially protected from ATM-mediated dissociation (Fig. 7h). Conversely, phosphomimetic Siah-1S19D showed reduced interaction with HIPK2, compared with wild-type Siah-1 (Fig. 7i). Thus, the decreased HIPK2-degrading capacity of Siah-1S19D, at least in part, stems from reduced HIPK2 binding. These findings indicate that ATM modulates HIPK2–Siah-1 interaction through site-specific phosphorylation of Siah-1.
Discussion
HIPK2 is activated by DNA damage and is involved in apoptosis induced by DNA damage7, 8, 9, 21. However, the mechanisms underlying HIPK2 regulation have been unclear. On the basis of our results, we propose a molecular framework for HIPK2 regulation (Fig. 8).
Figure 8: A model for HIPK2 regulation in unstressed and DNA-damaged cells.
(a) In unstressed cells HIPK2 forms a complex with Siah-1, resulting in HIPK2 polyubiquitination and proteasomal degradation. (b) In response to DNA damage, checkpoint kinase ATM and/or ATR is activated and mediates HIPK2 stabilization through disruption of the HIPK2–Siah-1 complex by phosphorylating Siah-1 at Ser 19. In response to lethal damage, HIPK2 activates the apoptotic machinery, whereas with sublethal damage, it is degraded by a p53-controlled, Siah-1-dependent mechanism to facilitate cellular recovery from DNA damage.
Full size image (74 KB)In unstressed cells HIPK2 is a short-lived protein and its steady-state levels are balanced through proteasome-dependent degradation, which is facilitated by ubiquitin ligase Siah-1. Siah-1 is also crucial for HIPK2 degradation during recovery from DNA damage, and interference with HIPK2 degradation through Siah-1 depletion leads to HIPK2 accumulation and sensitization for DNA damage-induced apoptosis.
HIPK2 degradation during recovery from DNA damage has been previously reported to be regulated by p53-induced MDM2/HDM2 (ref. 30). Our data here demonstrate that Siah-1 facilitates HIPK2 polyubiquitination and proteasomal degradation independently of MDM2/HDM2. Consistently, we found that p53-driven HIPK2 degradation is independent of MDM2, but is sensitive to downregulation of Siah-1, which is an established p53 target gene27, 28, 29, 37.
Caspase-mediated HIPK2 processing induced by p53 was previously reported to amplify HIPK2–p53 signalling during the initiation phase of adriamycin-induced apoptosis18. Accordingly, we found that HIPK2 downregulation by induced expression of p53 required caspase activity as well as proteasome activity. HIPK2 depletion during recovery from UV damage was not blocked by caspase inhibition, but relied on proteasome activity, suggesting no direct relationship between proteasome-dependent HIPK2 degradation by Siah-1 and caspase-dependent HIPK2 processing.
Our findings demonstrate that HIPK2 is stabilized when DNA is damaged. As p53 is also activated by DNA damage36, 39 and is capable of driving Siah-1-dependent HIPK2 degradation, this argues for a mechanism that protects HIPK2 from p53-driven proteolysis to facilitate HIPK2 stabilization. Consistent with this hypothesis and our previous findings21, HIPK2 stabilization induced by DNA damage depends on ATM and ATR function, which maintains HIPK2 stability in damaged cells and protects it from Siah-1-dependent proteolysis. ATM and ATR phosphorylate Siah-1 at Ser 19 to disrupt the HIPK2–Siah-1 complex and thereby protect HIPK2 from Siah-1-dependent proteolysis. Cells that recover from damage downregulate HIPK2 through Siah-1-dependent proteolysis to prevent unscheduled activation of apoptosis. During recovery from damage, the DNA damage checkpoints are resolved and ATM is inactivated42, 43. This may facilitate reformation of the Siah-1-HIPK2 complex and HIPK2 degradation.
Recently, Siah proteins have been linked to Ras-mediated transformation and tumorigenesis44. It will be interesting to see whether the DNA damage response can modulate these Siah functions. Finally, as HIPK2 is a tumour suppressor and mediator of DNA damage-induced apoptosis, our results propose a new strategy to improve the efficiency of DNA-damage-causing cancer therapies by targeting the HIPK2–Siah-1 complex.
Methods
Cell culture and transfection.
293T, H1299, HT1080, HepG2, Hep3B, MCF7, U2OS (all obtained from ATCC), mdm2;p53 double knockout mouse embryonic fibroblasts31 (a gift from G. Lozano, Anderson Cancer Center, Houston, TX) were maintained in DMEM supplemented with 10% FCS, 1% (w/v) penicillin/streptomycin and 20 mM Hepes buffer at 37 °C at 5% CO2. WI-38 (ATCC) cells were maintained in DMEM supplemented with 15% FCS. Transient transfections were performed using Lipofectamine 2000 (Invitrogen) or by standard calcium phosphate precipitation. The doxycycline-inducible H1299–tet-p53 and H1299–tet-p53H175D expressing cells were provided by G. Rohaly (Heinrich-Pette-Institut, Hamburg, Germany). Transgene expression was induced in H1299–tet-p53 and H1299–tet-p53H175D cells by adding doxycycline (5 ng ml-1 and 1 g ml-1, respectively; Sigma) for the time periods indicated.
Antibodies.
The following antibodies were used: p53 (DO-1 and FL-393), p21 (F–5), poly-ubiquitin (P4D1), HDM2 (SMP14), CtBP (E–12) and Erk2 (D–2) from Santa Cruz Biotechnologies, Flag (M2) and 
-tubulin (DM1A) from Sigma, actin (C4) from MP Biomedicals, ATR (Ab-2) from Calbiochem, ATM (2C1) from Abcam, HA (clones 12CA5 and 3F10) and GFP (clones 7.1 and 13.1) from Roche. All antibodies were used at dilution of 1:1000. The affinity-purified HIPK2 antibody has been described previously7. The p53 phospho Ser 46 antibody was from Cell Signaling Technologies. Chicken Siah-1 antibodies were raised against the following KLH-coupled Siah-1 peptide H2N-MSRQTATALPTGTSKC-COOH. Siah-1 antibodies were purified from egg yolks and affinity-purified against the peptide before use. For endogenous interaction of HIPK2 and Siah-1, immunoprecipitations were performed with rabbit HIPK2 antibodies raised against the following KLH-coupled HIPK2 peptide H2N-QKCGLKRKSEEIENTSC-CONH2. HIPK2 antibodies were affinity-purified against the peptide before use. Rabbit phospho Ser 19 antibodies were generated by Eurogentec and were raised against the following KLH-linked phosphopeptide H2N-TSKCPPS(PO3H2)QRVPALC-CONH2. Phospho-specific antibodies were affinity-purified using the phosphorylated- and the non-phosphorylated-peptide. ELISA tests showed strong phospho-specificity of the phospho-specific sera for the phosphopeptide, in contrast to the unphosphorylated peptide. All antibodies were generated with Eurogentec.
Expression constructs.
Human Flag–HIPK2 constructs and GST–HIPK2 fusion constructs were generated using standard PCR techniques or have been described previously7, 22. All PCR-generated constructs were verified by DNA sequencing. The human HA–Siah-1a, HA–Siah-1aC44S and GST–Siah-1 expression vectors were gifts from P. Matthias, R. Thiedt (Friedrich Miescher Institut, Basel, Switzerland) and T. Wirth (University of Ulm, Germany). Human Siah-2 was provided by R. Marschalek (Institute of Pharmaceutical Biology, Frankfurt, Germany). HA-tagged ubiquitin was a gift from K. Harbers (Heinrich-Pette-Institut), ATR constructs were from R.T. Abraham (The Burnham Institute, La Jolla, CA), ATM expression constructs from M.B. Kastan (St Jude Children's Research Hospital, Memphis, TN).
RT–PCR analysis.
Total RNA was isolated using the RNAeasy kit (Qiagen). 2 g total RNA was reverse transcribed using the cDNA Cycle Kit (Invitrogen) according to the manufacturer's instruction. 20% of the RT reaction was used as a template for PCR using the following primer pairs: HIPK2 (sense 5´-GGCCTCACATGTGCAAGTTTTC, antisense 5´-TTGGTAGGTATCAAGGAGGCTC), Siah-1 (sense 5´-CCTGTAAATGGCAAGGCTCTC, antisense 5´-CTAGTCTTTGACACCAGCATTG), ATR (sense 5´-GATGTTAGAAGGAATTGCTGTTGTCTTAC, antisense 5´-GTCAGATAAAACATGCCGTGAAGAGTAC), Siah-2 (sense 5´-GCAGCCCCAGCACACTCCGTC, antisense 5´-GTTACACACCAGGTGCCCGGC), 
-actin (sense 5´-CCTCGCCTTTGCCGATCC; antisense 5´-GGATCTTCATGAGGTAGTCAGTC). PCR reactions were performed using the following conditions: 1 min 95 °C, 1 min 58 °C, 1 min 72 °C, 25 cycles. Reactions were analysed on 1.2% agarose gels.
GST-pulldown assays.
GST fusion protein expression, protein purification and GST pulldown assays were performed as described previously7.
Immunoprecipitations, crosslinking and immunoblotting.
Immunoprecipitation analysis and immunoblotting were performed as described previously, and proteins were detected by enhanced chemiluminescence (western blot Dura and Femto, Pierce)7. In vivo crosslinking of WI-38 cells with 5 mM Dimethyl 3,3´-dithiobisproprionamidate-2-HCl (DTBP, Pierce) was performed as described previously45.
RNA interference.
siRNA duplexes were produced by Dharmacon Research and Qiagen. siRNA was directed against the following target site of Siah-1 which was selected from a custom 'Smart Pool' against human Siah-1 (Dharmacon): 5´-GATAGGAACACGCAAGCAA. The other Siah-1 specific siRNAs included in the Smart Pool were inefficient in depleting endogenous Siah-1 message (data not shown). For Siah-2 knockdown the custom 'Smart Pool' from Dharmacon was used. The siRNA specifically targeting HDM2 (5´-CCACCUCACAGAUUCCAGCTT) and ATR (5´-AAGCGCCTGATTCGAGATCCT) have been published30, 46. For control experiments the non-specific control duplex IX siRNA (Dharmacon; D-001206-09-20) and a GL-2 luciferase duplex were used. siRNA duplexes (final concentration 75–100 nM) were transfected using Dharmafect 4 (Dharmacon) or HiPerFect (Qiagen) as specified by the manufacturer. Twenty-four hours after transfection cells, were treated as indicated.
Immunofluorescence staining.
Immunofluorescence staining and confocal microscopy were performed as described previously47. Primary antibodies used were chicken Siah-1, rabbit HIPK2 and mouse anti-HA (12CA5, Roche). Secondary antibodies were Alexa-488-coupled goat anti-mouse and Alexa-594-coupled goat anti-rabbit (Molecular Probes). DNA was observed with Draq5 (Apotech), which was false-coloured in blue. Cells were examined using a confocal laser scanning microscope (LSM510 META, Zeiss) with a 
63 oil objective. Images were collected and processed using the AxioVision software (Zeiss) and sized in Adope Photoshop 7.0.
In vitro and in vivo ubiquitination.
In vitro ubiquitination assays were carried out in ubiquitination buffer (50mM Tris, pH 7.4, 5mM MgCl2, 2mM DTT) using human recombinant E1 (100 ng, Boston Biochem), human recombinant E2 UbcH5b (200 ng, Boston Biochem), His-tagged ubiquitin (10 g, Boston Biochem). E3 ligase Siah-1 and HIPK2 proteins were expressed in E.coli BL21 (HIPK2–6His; GST–Siah1-, GST–Siah-1RING) and purified by standard protocols or immunopurified from cells lysates of cells transfected with respective expression vectors as mentioned in the results section. 250 ng GST–Siah-1 or GST–Siah-1RING and 200 ng HIPK2–6His were used for ubiquitination reactions. Reactions (30 l volume) were incubated at 30 °C for 2 h. Where indicated, His–HIPK2 was subsequently purified from the reactions by Ni-NTA pulldown. Reactions were stopped by adding 5 SDS loading buffer, heated to 95 °C for 5 min, separated on 6% SDS–PAGE gels and transferred by wet blotting to a PVDF membrane. Proteins were detected by immunoblotting.
In vivo ubiquitination was performed by transfecting 293T cells in 10-cm dishes with 3 g HA–ubiquitin, 5 g HA–Siah-1 or Siah-1C44S, 15 g HIPK2–6His constructs and pcDNA3 vector to keep equal amounts of plasmid DNA in each transfection. Twenty hours later the cells were lysed in buffer A (6 M guanidinium-HCl, 0.1 M Na2HPO4/NaH2PO4, 0.01 M Tris–HCl pH 8.0, 5 mM imidazole, 10 mM -mercaptoethanol) and incubated with Ni2+-NTA beads (Qiagen) for 4 h at 20 °C. Beads were washed with buffer A, B (8 M urea, 0.1 M Na2HPO4/NaH2PO4, 0.01 M Tris–HCl pH 8.0, 10 mM -mercaptoethanol) and C (8 M urea, 0.1 M Na2HPO4/NaH2PO4, 0.01 M Tris–HCl pH 6.3, 10 mM -mercaptoethanol), and bound proteins were eluted with buffer D (200 mM imidazole, 0.15 M Tris–HCl pH 6.7, 30% glycerol, 0.72 M -mercaptoethanol, 5% SDS) and analysed by immunoblotting.
DNA-damaging treatments and apoptosis determination.
Culture medium was removed and cells were exposed to the indicated doses of UV-C irradiation, as described previously7 using a Stratalinker 1800 (Stratagene). Subsequently, fresh medium was added, cells were incubated and harvested at the indicated time points and further processed as specified. For adriamycin treatment, cells were incubated for 12 h in culture medium supplemented with adriamycin at the indicated concentrations. Subsequently, medium was changed and cells were collected after further incubation at the time points indicated. Ionizing radiation treatment was performed as described previously21. Apoptosis was determined by FACS-based subdiploid DNA content measurement as published48.
In vitro phosphorylation assays.
For ATM and ATR in vitro phosphorylation assays Flag-tagged ATR and ATM proteins expressed in 293T cells were immunoprecipitated using Flag antibodies and incubated with Flag–HIPK2K221A, Flag–p53, HA-tagged Siah-1 or bacterially expressed GST–p53 or GST–Siah-1 proteins. Kinase reactions were performed as described49, 50 using 30 l kinase buffer containing 40 M cold ATP and 5 Ci [
-32P] ATP. After incubation for 20 min at 30 °C, the reaction was stopped by adding 5 SDS loading buffer. After separation by SDS–PAGE, gels were fixed, dried and exposed to X-ray films.
Note: Supplementary Information is available on the Nature Cell Biology website.
Authour contributions
M.W., D.S., I.D., J.M., J.C., K.S and T.G.H. performed the experiments, analysed and discussed the data. T.G.H. planned and supervised the entire project, designed the experiments and wrote the manuscript.
