The Chk2 tumor suppressor is not required for p53 responses in human cancer cells.

Ionizing radiation damages chromosomal DNA and activates p53-dependent transcription in mammalian cells. The Chk2 protein kinase has been hypothesized to be the primary mediator of this response. We have rigorously tested this hypothesis in human cells by disrupting the CHK2 gene through homologous recombination. We found that the p53 response was unexpectedly robust in such cells. Phosphorylation of p53 at serine 20, accumulation of p53 protein, transcriptional activation of p53 target genes, and cell cycle arrest and apoptotic death phenotypes were completely intact regardless of CHK2 status. Our results indicate that Chk2 kinase is not required for p53 activation in human cells and explain why CHK2 and TP53 mutations can jointly occur in human tumors.

Ionizing radiation damages chromosomal DNA and activates p53-dependent transcription in mammalian cells. The Chk2 protein kinase has been hypothesized to be the primary mediator of this response. We have rigorously tested this hypothesis in human cells by disrupting the CHK2 gene through homologous recombination. We found that the p53 response was unexpectedly robust in such cells. Phosphorylation of p53 at serine 20, accumulation of p53 protein, transcriptional activation of p53 target genes, and cell cycle arrest and apoptotic death phenotypes were completely intact regardless of CHK2 status. Our results indicate that Chk2 kinase is not required for p53 activation in human cells and explain why CHK2 and TP53 mutations can jointly occur in human tumors.
The tumor suppressor protein p53 is a central mediator of the cellular response to DNA damage and other forms of stress and is inactivated in the majority of human cancers (1). Upon exposure of cells to ionizing radiation (IR), 1 the p53 protein is stabilized and activates transcription of downstream target genes that implement cell cycle arrest or apoptosis (2,3). This activation is tightly coupled to phosphorylation of a number of key residues, particularly at the amino terminus (4,5). Phosphorylation of these residues is thought to block the interaction of p53 with MDM2 (6), a ubiquitin-protein ligase that normally targets p53 for rapid turnover and thus suppresses its accumulation in undamaged cells (7).
Although the relative importance of individual residues to p53 activation remains under active study, phosphorylation of serine 20 appears to be essential for p53 stabilization (8,9). Upstream serine/threonine kinases that perform this phosphorylation are believed to be essential components of the signaling pathway that leads to the post-transcriptional activation of p53. Among the cellular kinases that can phosphorylate p53 at serine 20 is the product of the CHK2 gene (6,10), which is mutated in a subset of kindreds with the Li-Fraumeni-like syndrome (LFLS) (11,12). LFLS overlaps clinically with Li-Fraumeni syndrome (LFS), a cancer-prone disease caused by inherited mutations in the TP53 gene. The DNA damage-dependent activation of Chk2 kinase, in turn, has been found to require the kinase activity of ATM (13)(14)(15), encoded by the gene that is defective in patients with another inherited cancer syndrome called ataxia telangiectasia (16). Taken together, the cancer-prone phenotypes of kindreds with mutations in the ATM, CHK2, or TP53 genes and the biochemical data have given rise to the idea that these genes function along a linear pathway in mammalian cells (17). In support of this model, IR-dependent activation of Chk2 kinase and up-regulation of p53 are both defective in fibroblasts derived from ataxia telangiectasia patients and in murine ATM Ϫ/Ϫ cells (16). Moreover, it has been reported that IR fails to activate p53-dependent transcription in mouse embryonic fibroblasts (MEFs) deficient in Chk2 (18). However, re-examination of these MEFs (19) and independently derived MEFs (20) has led to somewhat different conclusions. A recent study of the relationship between the homologs of CHK2 and TP53 in Drosophila, DmCHK2, and DmP53, respectively, has shown that DmChk2 activity is required for Dmp53-mediated apoptosis (21). Evolutionary conservation of the regulation of p53 function by Chk2 would support the idea that Chk2 is required for p53 responses in human cells.
Although the model of a linear ATM-Chk2-p53 pathway is attractive, recent studies of CHK2 and TP53 mutations in human cancers prompted us to evaluate this model in human cells. In particular, the most frequent CHK2 variant described in LFLS kindreds, the 1110delC mutation, is also found in ϳ 1% of the total population and thus must be a very low penetrance allele (22). By contrast, germline TP53 mutations are invariably high penetrance alleles that result in the severe cancer predisposition phenotype typical of LFS. In addition, the linear pathway model predicts an epistatic relationship between CHK2 and TP53, i.e. mutation of both genes in a single tumor should be extremely rare, as rare as concurrent alterations in APC and CTNNB1 (␤-catenin) (23), RB1 and CDKN2A (p16) (24), MDM2 and TP53 (25), or KRAS (c-Ki-ras) and BRAF genes (26). However, the coexistence of mutations in CHK2 and TP53 has been documented in multiple human cancers (27,28), including a well characterized colorectal cancer cell line (HCT-15; also called DLD-1) (11,29).
Alternatively, if CHK2 is not required for, but only partially contributes to, p53 stabilization, the coincidence of mutations in TP53 and CHK2 might not be unexpected. The finding of a partial requirement of Chk2 in a non-linear p53 stabilization pathway would make it difficult to understand how CHK2 mutation alone might cause the marked phenotype of LFLS. To evaluate the contribution of human Chk2 kinase to p53 responses in an unambiguous manner, we chose to examine human colorectal cancer cells in which CHK2 had been disrupted through gene targeting, as described below.

EXPERIMENTAL PROCEDURES
Cell Culture-The human colon cancer cell lines HCT116, DLD-1, and their derivatives were cultured in McCoy's 5A medium supplemented with 10% fetal calf serum and penicillin/streptomycin (Invitrogen). Stably transfected clones were initially selected with either Geneticin or hygromycin as appropriate, then routinely propagated in the absence of selective drugs.
Targeted Deletion of the Human CHK2 Locus-The targeting constructs pCHK2ex6a and pCHK2ex6b.1 were made by PCR, using bacterial artificial chromosome clone RP11-444G7 (Research Genetics) as the template. Homology regions were cloned into a bipartite genetargeting system that has an additional recombination site designed into the neomycin resistance gene (30). Transfection, selection, and PCR screening were performed as described (30) using Chk2-43 (denoted as a in Fig. 1B) and NEOreverse primers. Following isolation of heterozygous targeted clones the neomycin-resistance gene was excised by infection with an adenovirus expressing cre recombinase. A second round of gene targeting was then performed, yielding three independently isolated clones in which both alleles of CHK2 had been disrupted. All clones gave identical results in the experiments reported here. Details of the vector design and sequences of all PCR primers are available from the authors upon request.
Irradiation and Drug Treatment-Cells were grown as subconfluent monolayers in T-25 flasks and exposed to a 137 Cs source that delivered a measured dose of 0.8 Gy/min. Where indicated, nocodazole was added to the cell culture medium at a final concentration of 0.2 g/ml immediately following irradiation. 5-Fluorouracil (5-FU) was used at a final concentration of 50 g/ml.
Immunoblotting and Northern Blotting-In preparation for immunoblotting, cells were harvested and lysed in HB2 buffer as described (30). Protein extracts were resolved via SDS-PAGE and transferred to polyvinylidene difluoride membranes. Antibodies specific for Chk2 and Chk1 were obtained from Santa Cruz Biotechnology. Phosphoepitopespecific antibodies were obtained from Cell Signaling Technology. Electrophoresis and hybridization of total RNA was performed as described (31).
Cell Cycle Analysis-Cells were collected by incubation in trypsin/ EDTA, pelleted by centrifugation, and fixed in a solution containing 3.7% formaldehyde, 0.5% Nonidet P-40, and 10 g/ml Hoechst 33258 in phosphate-buffered saline. 10,000 cells were analyzed per sample on a flow cytometer (BD Biosciences). Mitotic indices were determined by visual scoring of at least 300 nuclei.

RESULTS
To develop a system in which Chk2 could be systematically studied in human cells, we deleted both alleles of the CHK2 gene in HCT116 cells through homologous recombination. The HCT116 colon cancer cell line is a useful system to study the effects of Chk2 upon p53 activation. It has been demonstrated that endogenous Chk2 kinase is activated in HCT116 cells following IR treatment (32). Moreover, the p53-dependent phenotypes of HCT116 cells have been examined in detail. These cells exhibit intact G 1 /S and G 2 /M checkpoints following DNA damage (33). Targeted deletion of p53 itself (34) or its target genes p21 and 14 -3-3 abrogates one or both of these checkpoints (35), confirming the integrity of the DNA damage response pathway and its dependence on p53 in the parent HCT116 cells.
Examination of the CHK2 genomic locus revealed a total of 15 exons, including a previously unreported exon that lies upstream of the first coding exon. The targeting constructs were designed so that 160 bp of genomic sequence within the encoded kinase homology domain was deleted upon homologous recombination (Fig. 1A). Because a number of CHK2related pseudogenes containing significant homology to exons 12-15 have been identified (36), we intentionally restricted our targeting strategy to the proximal exons (Fig. 1B). Even if the resulting allele were expressed, it would encode a truncated protein that would be predicted to be catalytically inactive. Two sequential rounds of targeting were carried out as described under "Experimental Procedures." Successful targeting of both CHK2 alleles was initially assessed by PCR and confirmed in multiple independent clones by genomic Southern blotting (Fig. 1C). In these clones, no Chk2 protein, either full-length or truncated, was detected using three different antibodies directed against the amino terminus of Chk2, including one that recognizes phosphorylation of threonine 68 and thus would be expected to detect the presence of a truncated phosphoprotein if it had been present ( Fig. 2A).
Following exposure to DNA damaging agents such as IR, parental HCT116 cells undergo growth arrest in the G 1 and G 2 phases of the cell cycle (33), as is typical of many cell types that harbor wild type p53. Within one hour after IR treatment, activated upstream kinases phosphorylate p53 and Chk2 in a fashion similar to that of other cell types that have been described (Fig. 2, A and B). This phosphorylation is accompanied  (12)(13)(14)(15) have homology to CHK2 pseudogenes found in multiple copies throughout the human genome (36) and were therefore excluded from the targeting vectors. B, targeting strategy. A bipartite vector system (30) was used to delete exon 6, which is located within the kinase domain of Chk2. The genomic positions of BamHI (B) and EcoRI (E) restriction sites and PCR primers used for screening (a, neo) have been marked for reference. The predicted sizes of restriction fragments corresponding to wild type alleles (WT), correctly targeted alleles (KO), and correctly targeted alleles after cre-mediated excision of the neomycin-resistance cassette (KO⌬neo) are also shown. C, confirmation of CHK2 targeting by Southern blotting. Genomic DNAs were digested with BamHI and EcoRI and hybridized with an appropriate probe (see panel B). The asterisk denotes a heterozygote harboring a tandem insertion of the neomycin-resistance cassette within the CHK2 locus. All neomycin-resistant heterozygotes (Ϯ lanes) were fully resolved to simple 160-bp deletions by cre-mediated excision (lanes marked Ϯ c ). by a sustained and dramatic accumulation of p53 protein over a period of several hours (Fig. 2B). Pretreatment of cells with caffeine prevents both serine 20 phosphorylation and p53 stabilization in response to IR (37), supporting the hypothesis that these events are functionally linked. Notably, in CHK2 Ϫ/Ϫ cells, p53 accumulated to high levels with time (Fig. 2B). In addition, this accumulation of p53 was accompanied by robust phosphorylation of p53 on serine 20 (Fig. 2C). The kinetics and amplitude of these responses were essentially identical in CHK2 Ϫ/Ϫ cells and their isogenic wild type controls.
Robust induction of p53 was evident within 3 h of exposure to 12 Gy of IR (Fig. 2B). We took advantage of this observation to determine whether CHK2 Ϫ/Ϫ cells could up-regulate p53 levels in response to lower doses of IR. Throughout the broad range of doses, p53 was consistently induced to a similar extent, regardless of CHK2 status (Fig. 2D).
IR-mediated activation of p53 results in transcriptional activation of downstream target genes. The most well known of these p53 targets is the p21 cyclin-dependent kinase inhibitor, which participates in the G 1 and G 2 arrest phenotypes associated with wild type p53 function (34). We found that p21 was induced to equivalent levels in wild type and Chk2-deficient cells (Fig. 2B), indicating that p53 was not only stabilized by IR but also functioned as a transcriptional activator.
One of the most important results of p53 deficiency is the failure to implement cell cycle checkpoints. Following IR treatment, we observed a clear diminution of cells in S phase in both CHK2 ϩ/ϩ and CHK2 Ϫ/Ϫ cells. Cells arrested into stable populations with either 2N or 4N DNA content, regardless of genotype. These results indicated that the G 1 /S checkpoint was intact despite the lack of Chk2 protein (Fig. 3A).
We performed several experiments to evaluate the ability of CHK2 Ϫ/Ϫ cells to initiate and maintain a G 2 arrest following IR. First, we measured the ability of irradiated cells to block entry into mitosis immediately following IR treatment. Initiation of this pre-mitotic block is thought to be mediated by phosphorylation of the Cdc25C phosphatase on serine 216, which sequesters Cdc25C in the cytoplasm and prevents activation of mitotic cyclin B/Cdc2 kinase in irradiated cells (38). In vitro, Cdc25C can be phosphorylated on this site by either Chk2 or Chk1 kinases (14,39). Within 1 h of IR exposure, both CHK2 ϩ/ϩ and CHK2 Ϫ/Ϫ cells had phosphorylated Cdc25C on serine 216 ( Fig. 2A) and blocked entry into mitosis (Fig. 3B), demonstrating that the ability to initiate G 2 arrest is independent of CHK2 status.
To assess the long-term maintenance of this G 2 arrest, we incubated IR-treated cells in the presence of nocodazole. We previously showed that p53-deficient cells fail to maintain the G 2 checkpoint after IR treatment and eventually progress into mitosis (34). In contrast, however, very few Chk2-deficient cells entered mitosis, similar to wild type cells (Fig. 3C). We conclude that the maintenance of the G 2 cell cycle checkpoint was ting. C, induction of p53 in Chk2-deficient cells is accompanied by robust phosphorylation at serine 20. Cells treated as in B were subjected to immunoblot analysis using a phosphoepitope-specific antibody that detects p53 serine 20 phosphorylation. D, Chk2-deficient cells exhibit normal p53 induction in response to a wide range of IR exposures. Cells of the indicated genotypes were exposed to 0, 2, 4, 8, or 12 Gy of IR and harvested three hours later. Immunoblotting was performed as in B. E, IR-dependent phosphorylation of p53 serine 20 is preserved in a naturally occurring hCHK2 mutant cell line. DLD-1 cells stably transfected with either vector alone (left lanes) or wild type Chk2 kinase (right lanes) were exposed to 12 Gy of IR and harvested one or two hours later. Immunoblot analysis of Chk2 and p53 was performed as described above. The extent of p53 serine 20 phosphorylation was slightly lower in DLD-1 cells expressing wild type Chk2, suggesting that elevated Chk2 levels may actually interfere with the physiologic p53 serine 20 kinase in these cells.

FIG. 2. IR-dependent signaling to p53 is intact in human colorectal cancer cells lacking Chk2.
A, the acute response to ionizing radiation (IR) is intact in human cells lacking Chk2. Exponentially growing cells were exposed to 12 Gy of IR and harvested one hour later. The abundance of Chk2, Chk1, and Cdc25C proteins and the extent of their IR-dependent phosphorylation were determined by immunoblotting with specific antibodies. B, human cells lacking Chk2 display a robust p53 response to IR. Cells were exposed to 12 Gy of IR and cultured for a 24-hour period. Samples were collected at the indicated time points. Total p53 and p21 levels were evaluated by immunoblot-unaffected by targeting CHK2, even though this checkpoint was markedly altered in HCT116 cells in which the TP53 gene had been analogously targeted (34).
The activation of p53 under certain conditions results in apoptosis rather than growth arrest (40). In HCT116 cells, the apoptotic response is strongly elicited by the chemotherapeutic agent 5-FU (41), the mainstay of colorectal cancer chemotherapy. This apoptotic response is absolutely dependent on functional p53 (41) and its downstream transcriptional target, ferredoxin reductase (FDXR) (31). In both CHK2 ϩ/ϩ and CHK2 Ϫ/Ϫ cells, 5-FU exposure triggered the phosphorylation of serine 20 and a dramatic accumulation of p53 protein (Fig. 4B). The end result was strong up-regulation of FDXR transcripts and high levels of apoptotic cell death (Fig. 4A) in both wild type and Chk2-deficient cells.
Finally, we wished to determine whether these results were applicable to other colorectal cancer cell lines. Though generation of other knock-out lines would not have been feasible in a reasonable time frame, some information could be gained through the examination of DLD-1 cells. This human colorectal cell line contains one CHK2 allele with an inactivating point mutation (R145W), whereas the other allele is transcriptionally silenced, rendering the cells Chk2 deficient (11,29). Serine 20 phosphorylation of the endogenous mutant p53 protein in this line was easily detectable after IR exposure (Fig. 2E). Moreover, we introduced a Chk2 expression vector into the line and established subclones of DLD-1 that produced high levels of Chk2, comparable with that observed in HCT116 cells. The phosphorylation of p53 serine 20 was not increased in these subclones (Fig. 2E), arguing that Chk2 is not the physiologically limiting kinase responsible for phosphorylation at this site.

DISCUSSION
The results described above are quite different from what we expected based on the phenotype of Chk2-deficient mouse cells (18). In particular, Chk2 disruption in human cells did not lead to defects in p53 stabilization, phosphorylation, or transcriptional activation of downstream targets, or to defects in G 1 /S or G 2 checkpoints or apoptotic induction at the physiologic level. Why are the results we have obtained so different from those previously described in mice? We presume these differences are due to the different cell types and species examined. For example, we studied neoplastic epithelial cells in humans, whereas previous studies (18,19) focused on normal thymocytes or embryonic cells of murine origin. Further studies will be necessary to distinguish which of these differences are the most critical. Regardless, our results rigorously show that at least some human cells (and perhaps all) do not require Chk2 to activate p53 pathways for growth control and that no simple linear model connecting ATM to Chk2 to p53 can be generally applicable.
It has recently been suggested that Chk2 activation may require a functional mismatch repair system (42). However, independent assessments performed in multiple laboratories (10,29,32,43) have demonstrated that Chk2 is robustly activated in human cancer cells that have been exposed to DNA damage, irrespective of mismatch repair status. Furthermore, mismatch repair-deficient cancer cell lines clearly respond to DNA damage exposure with a sustained G 2 checkpoint arrest that requires both p53 and its downstream target, the p21 CDK inhibitor (34). Taken together, these data indicate that mismatch repair-deficient cells are competent for the biochemical activation of both Chk2 kinase and the transcriptional upregulation of p53-regulated target genes.
Our genetic analysis indicates that Chk2 is not required for p53 responses in human cells. One might initially interpret our results as suggesting that another kinase substitutes for Chk2 and activates p53 in human cells, but not in mouse or insect cells (18,21), that lack Chk2. However, it has recently been discovered that Chk2 kinase purified from IR-treated human cells phosphorylates p53 very poorly in vitro (44). Taken together, these genetic and biochemical data support an alternative explanation, namely that Chk2 does not participate significantly in the p53 responses of human cells. One exciting implication of this interpretation is that another, as yet unknown, kinase functions to communicate the DNA damage signal to p53 in human cells. Chk1 would appear to be an obvious candidate for such a kinase, because it can phosphorylate serine 20 of p53 in vitro (10). However, Ahn et al. (44) have found that down-regulation of either or both Chk1 or Chk2 by RNA interference does not prevent human breast and colorectal cells from inducing p53 in response to DNA damage.

FIG. 3. G 1 /S and G 2 /M checkpoint responses in human colorectal cancer cells do not require Chk2. A, CHK2
Ϫ/Ϫ cells have an intact G 1 /S checkpoint. Cells of the indicated genotypes were irradiated and cultured for 24 h. The fraction of cells in G 1 , S, and G 2 /M phases was determined by flow cytometry. B, normal establishment of G 2 arrest in human cells lacking Chk2. Wild type and Chk2-deficient cells were either irradiated (ϩIR) or mock-treated (ϪIR) and then harvested one hour later. The percentage of cells in mitosis was determined by microscopic examination. C, long-term maintenance of G 2 arrest does not require Chk2 in human cells. Cells of the indicated genotypes were either irradiated (ϩIR) or mock-treated (ϪIR) and then immediately incubated in the presence of nocodazole to trap cells traversing the G 2 /M boundary (34). The fraction of cells in mitosis at various times after irradiation was determined by microscopic examination.
Thus, the identity of the authentic, physiologically relevant kinase upstream of p53 in human cells remains an important open question.
Finally, our results readily explain why some tumors arise with alterations of both Chk2 and p53, as our data show that mutations of these genes do not lead to similar defects. And our data raise an intriguing question: if an essential function of Chk2 does not involve p53 or the checkpoints studied here, what is the role of this highly conserved gene in human cells, and why is it mutated in cancers and cancer-prone individuals?
The lines we have generated should be instrumental for answering such questions in the future.