The G(2) DNA damage checkpoint delays expression of genes encoding mitotic regulators.

Transcriptional control of gene expression contributes to the regulation of diverse cellular processes including cell cycle progression and the cellular response to DNA damage. Global gene expression profiling was performed using p53-deficient human cells to identify genes with G(2)/M-specific and DNA damage-responsive expression. Numerous cell cycle-regulated genes were identified, but surprisingly the analysis failed to identify genes activated by ionizing radiation. Instead, significant delays in expression of G(2)/M-specific genes, including known mitotic regulators, were observed following DNA damage. Thus, in the absence of p53, gene induction does not contribute to the G(2) arrest following DNA damage. Rather, the DNA damage checkpoint elicits a G(2) cell cycle arrest, in part, by delaying accumulation of proteins required in mitosis.

Transcriptional control of gene expression contributes to the regulation of diverse cellular processes including cell cycle progression and the cellular response to DNA damage. Global gene expression profiling was performed using p53-deficient human cells to identify genes with G 2 /M-specific and DNA damage-responsive expression. Numerous cell cycle-regulated genes were identified, but surprisingly the analysis failed to identify genes activated by ionizing radiation. Instead, significant delays in expression of G 2 /M-specific genes, including known mitotic regulators, were observed following DNA damage. Thus, in the absence of p53, gene induction does not contribute to the G 2 arrest following DNA damage. Rather, the DNA damage checkpoint elicits a G 2 cell cycle arrest, in part, by delaying accumulation of proteins required in mitosis.
Eukaryotic cells coordinate DNA replication and cellular division through a series of integrated pathways that are cumulatively known as the cell division cycle. Progression through the cell division cycle is regulated, in part, by the activities of protein kinase complexes composed of two subunits, a cyclin and a cyclin-dependent protein kinase. Increased transcription of cyclin genes and reduced destruction of cyclin proteins account for the periodic accumulation of cyclin subunits throughout the cell cycle (for reviews, see Refs. [1][2][3]. In addition to the cyclins, cell cycle-specific expression of several other genes has been reported (for reviews, see Refs. 2, 4, and 5). Genes with G 1 /S-specific expression include the transcription factor E2F-1 and genes whose protein products are involved in nucleotide synthesis such as dihydrofolate reductase and thymidylate synthase (6,7). S phase-specific genes include those encoding the histones (reviewed in Ref. 8) and accessory components of DNA polymerase such as PCNA (9,10). Genes exhibiting G 2 /M-specific expression patterns include those encoding cyclin B1, CDC25C, CKSH1, and CKSH2 (regulators of cyclin-dependent protein kinases) as well as PLK1, a regulator of mitotic entry and exit (11)(12)(13)(14)(15).
Transcriptional activation is also a major component of the cellular response to DNA damage. DNA damage activates checkpoint pathways that delay cell cycle progression to provide time for repair of damaged DNA (16). The G 1 DNA damage checkpoint arrests cells in the G 1 phase of the cell cycle prior to DNA synthesis, and the G 2 DNA damage checkpoint arrests cells in the G 2 phase of the cell cycle prior to mitosis. Paramount to the G 1 DNA damage checkpoint is the p53 transcription factor (17). In response to DNA-damaging agents such as ionizing radiation, p53 becomes stabilized and activates the transcription of many genes including the cyclin-dependent protein kinase inhibitor p21 (18,19). In turn, p21 blocks the activity of cyclin⅐cyclin-dependent protein kinase complexes required for S phase entry. The G 1 cell cycle arrest observed after DNA damage requires both p53 and p21 and is dependent on p53-mediated transcriptional regulation (20,21). p53-independent mechanisms also contribute to the G 1 cell cycle arrest observed after DNA damage (22).
In contrast to the G 1 cell cycle arrest observed after DNA damage, the G 2 cell cycle arrest is independent of p53, since cells are able to initiate a G 2 cell cycle arrest in its absence. The G 2 DNA damage checkpoint operates, at least in part, by preventing the CDC25C protein phosphatase from activating the CDC2 protein kinase. This is accomplished by maintaining CDC25C in a phosphorylated form that binds 14-3-3 proteins (23)(24)(25)(26). The binding of 14-3-3 proteins prevents CDC25C from accumulating in the nucleus (25,(27)(28)(29)(30)(31)(32). This presumably prevents activation of CDC2 in the nucleus as it shuttles between the cytoplasm and the nucleus during the G 2 checkpoint response. This component of the G 2 DNA damage checkpoint is independent of p53 and does not require new gene transcription. Although it is not known whether transcriptional activation contributes in any way to the initiation of the G 2 DNA damage checkpoint, maintenance of the G 2 arrest does involve p53-mediated transcriptional activation (33). p53-induced genes whose protein products contribute to maintenance of the G 2 cell cycle arrest include GADD45, p21, and 14-3-3 (34 -36). 14-3-3 has been proposed to contribute to the maintenance of the G 2 cell cycle arrest by preventing the nuclear accumulation of cyclin B1-CDC2 complexes (33). G 1 and G 2 cell cycle arrest following DNA damage promotes survival by allowing time for cells to repair mutations. Many cancer cells lack a functional p53 protein and, as a result, are incapable of arresting in the G 1 phase of the cell cycle following DNA damage. However, these cells maintain their ability to arrest in G 2 . A complete understanding of p53-independent pathways involved in G 2 checkpoint control is desirable because p53-deficient cancer cells are more vulnerable to G 2 checkpoint abrogation than are cells containing functional p53 (37,38). Thus, drugs that abrogate G 2 checkpoint function might improve the killing of p53-defective tumors and thereby enhance chemotherapy and radiotherapy regimens for many human malignancies (37).
In this study, global gene expression profiling was performed both to identify genes whose expression is altered as cells traverse the S and G 2 phases of the cell cycle and to identify genes whose expression is altered during a G 2 DNA damage checkpoint response. These studies were carried out in human cells with compromised p53 function to determine whether there is a p53-independent transcriptional component to the G 2 DNA damage checkpoint.

EXPERIMENTAL PROCEDURES
Cell Culture and Synchronization-HeLa cells were obtained from the ATCC and were grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum, 100 units/ml penicillin and streptomycin, and 1 mM glutamine (culture medium). Synchronization was performed using a double thymidine block or a sequential thymidine/mimosine protocol (39). Cells were first treated with 2 mM thymidine for 18 h. Cells were then washed with phosphate-buffered saline (PBS) 1 and released from the first block by trypsinization and replating at ϳ50% confluency in release medium (culture medium supplemented with 25 M deoxycytidine and 25 M thymidine). Eight hours later, the cells were treated with either 2 mM thymidine (double thymidine block) or 400 M mimosine for an additional 18 h. The cells were released from the second block by washing with PBS and incubating in release medium (double thymidine block) or fresh culture medium (thymidine/mimosine protocol).
Irradiation and Drug Treatment-For ␥-irradiation, cells were transported in a 37°C insulated container in sealed flasks and exposed to 10 Gy from a 60 Co source. For UV irradiation, the culture medium was removed, and cells were irradiated in uncovered tissue culture dishes with 254-nm UV light at a dose of 20 J/m 2 using a Stratalinker (Stratagene). In some cases, cells were incubated with 0.5 g/ml (0.9 M) doxorubicin (Sigma) for 1 h. Heat shock was performed by incubating cells at 42°C for 1 h.
Affymetrix DNA Microarray Analysis-GeneChip microarray analysis was performed according to methods recommended by the manufacturer (Affymetrix, Santa Clara, CA). Adherent cells were removed by trypsinization and pelleted after the addition of an equal volume of culture medium. Pellets were washed with PBS and collected by centrifugation. An aliquot of cells was removed for fluorescence-activated cell sorting (FACS). Poly(A) ϩ RNA was prepared from the remaining cells using Oligotex mRNA preparation kit (Qiagen) according to the manufacturer's instruction. RNA was converted to double-stranded cDNA using the SuperScript Choice system (Life Technologies) following the supplier's protocol except that an high pressure liquid chromatography-purified T7-dT24 primer (Geneset) was utilized for first strand synthesis. After synthesis, the double-stranded cDNA was purified by phenol/chloroform extraction and ethanol precipitation. Synthesis of cRNA and GeneChip hybridization, washing, and scanning were performed at the University of Texas Southwestern Medical School. Briefly, the purified cDNA was used to generate biotinylated cRNA target using the Bioarray High Yield RNA transcript labeling kit (Enzo) and following the supplier's protocol. The labeled cRNA was then purified using the RNeasy total RNA clean up protocol (Qiagen) and quantitated by 260-nm absorbance.
Twenty micrograms of cRNA was fragmented by heating at 94°C for 35 min in fragmentation buffer (40 mM Tris acetate, pH 8.1, 125 mM KOAc, 30 mM MgOAc). A hybridization "mixture" was constructed containing 20 g of cRNA, 50 pM control oligonucleotide B2 (Geneset), 1ϫ control cRNA mix, 0.1 mg/ml herring sperm DNA, and 1ϫ MES hybridization buffer in a volume of 300 l. One hundred microliters of the hybridization mixture was heated to 95°C and hybridized at 45°C for 16 h sequentially to Test, Hu6800, Hu35KSubA, Hu35kSubB, Hu35kSubC, and Hu35kSubD GeneChip arrays. This analysis compared expression of ϳ42,000 probe sets.
All arrays were washed in the Affymetrix GeneChip Fluidics Station 400 using preprogrammed steps. The wash program involved a low stringency wash, high stringency wash, steptavidin/phycoerythrin stain, and a final low stringency wash. After washing and staining, each GeneChip was placed in the Affymetrix GeneChip array scanner, and image data were captured and converted to numerical output using the GeneChip Analysis Suite version 3.1. Arrays from base-line and experimental conditions were directly compared using normalization across all probe pair sets. Affymetrix comparison files were exported as flat text files and imported into Microsoft Excel for further analysis. Data were initially filtered to exclude all genes that were scored "A" (not detected) in both the base-line and experimental arrays. Genes whose calculated "-fold change" value was less than 2.0 or whose qualitative change call was "NC" (no change) between base-line and experimental samples were also excluded from further analysis. Using these exclusion criteria, analysis of cell cycle progression on gene expression (mock-irradiated cells at 7 h after release compared with mock-irradiated cells at 1 h after release) yielded 974 and 1247 remaining genes in the first and second independent analyses over all five arrays (Hu6800 and Hu35kSubA-D). A similar analysis of the effect of ionizing radiation on gene expression (␥-irradiated cells at 7 h after release compared with mock-irradiated cells at the 7 h after release) yielded 489 and 437 genes in the first and second independent analyses. Of the remaining genes, those that were induced or repressed at least 2.5-fold in each of two independent analyses and at least 3-fold in the average of the two analyses were evaluated further. An exception to these criteria was used in the case of genes that were induced by ionizing radiation. Using the criteria outlined above, no genes were induced in response to radiation. Therefore, the data were reanalyzed to identify all genes that were at least 2.2-fold induced following irradiation in both of the two independent experiments.
RNA Preparation for Northern Blotting-RNA was prepared for Northern blotting using standard methodologies (40) with minor modifications. Cells were removed from culture dishes by trypsinization and added to tubes containing cold culture medium in order to inactivate the trypsin. Cells were pelleted at 1000 ϫ g and then resuspended in PBS. An aliquot was removed for FACS analysis, and the remaining cells were once again subjected to centrifugation. Cell pellets were resuspended in PBS and then lysed by the addition of PBS supplemented with 1% Nonidet P-40 and 100 units/ml ammonium heparin. After centrifugation at 14,000 ϫ g for 1 min, supernatants were transferred to tubes containing phenol and an equal volume of TE/SDS (10 mM Tris, 1 mM EDTA, 0.5% SDS, pH 8.0). After extraction, the aqueous phase was sequentially extracted with phenol-chloroform and twice with chloroform prior to precipitation with isopropyl alcohol. RNA was collected by centrifugation and washed with 70% ethanol. RNA was resuspended in TE/SDS and denatured using 10% glyoxal at 65°C prior to electrophoresis through a 1.2% agarose gel (41). RNA was transferred to a nylon membrane and cross-linked using UV light with a Stratalinker (Stratagene). Membranes were prehybridized in ExpressHyb (CLONTECH) and probed with radiolabeled cDNA probes according to the recommendations of the manufacturer.
Immunoblotting-HeLa cells were synchronized using a double thymidine block. One hour (p53 immunoblotting) or 3 h (immunoblotting for other proteins) after release, the cells were subjected to 10 Gy of ␥-irradiation. For the analysis of p53 expression, asynchronously growing U2-OS cells were irradiated with the same dose. Cells from individual flasks of cells were harvested by trypsinization at time points thereafter. Cells were added to tubes containing cold culture medium to inactivate the trypsin. Cells were collected by centrifugation at 1000 ϫ g, washed with cold PBS, and again collected by centrifugation. Cell pellets were lysed by the addition of mammalian cell lysis buffer (50 mM Tris, pH 8.0, 0.5% Nonidet P-40, 100 mM NaCl, 2 mM dithiothreitol, 5 mM EDTA, 1 mM Na 3 VO 4 , 1 M microcystin, 50 mM NaF) and 1ϫ protease inhibitor solution (Sigma) and rocking for 30 min at 4°C. Lysates were cleared by centrifugation. Protein concentration was determined using a Bradford assay (Bio-Rad) according to the manufacturer's instructions. An equal amount of total protein (200 g) from each lysate was loaded on an SDS-polyacrylamide gel. After electrophoresis, separated proteins were transferred to nitrocellulose. Ponceau S staining was performed to confirm equal loading and transfer of lysates. The nitrocellulose membrane was blocked with 5% milk and probed for the indicated proteins using specific antiserum and the manufacturer's instructions. Antibodies used included a mouse monoclonal antibody to p53 (p53 Ab-2; Oncogene Research Products), a mouse monoclonal antibody to cyclin B1 (SC-245; Santa Cruz Biotechnology, Inc., Santa Cruz, CA), a rabbit polyclonal antibody to PLK1 (33-1700; Zymed Laboratories Inc.), a goat polyclonal antibody to CDC20 (SC-1906; Santa Cruz Biotechnology), a rabbit polyclonal antibody to TTK (SC-540; Santa Cruz Biotechnology), and a rabbit polyclonal antibody to C-TAK1 (42). Horseradish peroxidase-conjugated secondary antibodies included a goat anti-mouse antibody (ICN), a rabbit anti-goat antibody (Zymed Laboratories Inc.), and a goat antirabbit antibody (Zymed Laboratories Inc.). Antibody recognition of proteins was determined using enhanced chemiluminescence reagents (Amersham Pharmacia Biotech).
cDNA Probes-In some cases, probes used for Northern blotting were obtained commercially as ESTs (Genome Systems) or isolated by PCR amplification from a HeLa cDNA library. ESTs were obtained from Genome Systems for many of the genes identified in this study, including TTK, PLK1, CDC20, KIAA1333, E2-EPF, DDA3, EST AA417350, tastin, PIF-1, topoisomerase II␣, CAB92444, Rab6-interacting kinesin, EST AA286911, EST R70506, EST AA001296, EST AA398563, KIAA0069, KIAA0008, FLJ10491, RAN GTPase-activating protein 1, KIAA0110, and melanoma ubiquitously mutated protein. In each case, the insert was excised using restriction enzymes described as cloning sites in the GenBank TM record for the clone in question. A 1.5-kilobase pair cyclin B1 cDNA fragment and a 0.5-kilobase pair fragment of the p21 cDNA were also used as probes for Northern blotting.
Several cDNA fragments were obtained by reverse transcriptase-PCR from a HeLa cell cDNA library. The HeLa cell cDNA library was purchased from Stratagene, or in some cases, RNA was prepared from HeLa cells synchronized in the G 2 phase of the cell cycle. In the latter case, synchronization and RNA preparation were performed as described above. A cDNA library was made using the RACE cDNA library preparation kit (Stratagene) according to the manufacturer's recommendations. cDNA fragments obtained by PCR amplification were KAP/Cdi1, KIAA0159, CENP-A, KIAA0220, and STK-15. Primers for amplification of KAP/Cdi1 were 5Ј-AGCGATGAAGCCGCCCA and 5Ј-GGAATATCTGCACATGTACATTTACATTCAAG. For amplification of KIAA0159, primers were 5Ј-GAGTTCCATCTGCCATTATCCCCAGAG and 5Ј-GCCCATAGACTCACGGCTGCAAC. For amplification of CENP-A, primers were 5Ј-GGCCCTACAAGAGGCAGCAGAAGC and 5Ј-CCACTAATGGTGAGGCTACTCTGAAGGG. For amplification of KIAA0220, primers were 5Ј-CCAGCCTATCTCAGGGGAATGAATTG and 5Ј-CAGCTGCAAGGTACTTAGAACTGGCAAAG. For amplification of STK-15, primers were 5Ј-GAGGCGCCCTGTAGGATACTGCTTG and 5Ј-GCATGTTCCTGTCAGGTTATATGGCAGC. Primers are shown in 5Ј 3 3Ј orientation in all cases. PCR products of the correct size were subcloned into pCRII using the TA cloning kit (Invitrogen) according to the manufacturer's instructions and sequenced to confirm isolation of the correct cDNA fragment. For probe preparation, the insert fragment was excised using EcoRI, and the insert was gel-purified prior to use as a template in the probe preparation reaction.
Probes for Northern blotting were prepared using the Megaprime kit (Amersham Pharmacia Biotech) and [ 32 P]dCTP (PerkinElmer Life Sciences) according to the manufacturer's recommendation. In each case, ϳ25 ng of the cDNA fragment was used as a template in the reaction. Unincorporated [ 32 P]dCTP was removed using spin column filtration through P-10 beads (Amersham Pharmacia Biotech) prior to heat denaturation of the probe and the addition to the hybridization solution.
FACS-Adherent HeLa cells were removed by trypsinization and collected by centrifugation. After removal of the supernatant by aspiration, cells were resuspended in PBS. Cells were fixed by mixing with 95% ethanol (10 times the aqueous volume). Cells were again collected by centrifugation and then resuspended in PBS supplemented with 1% fetal bovine serum. After centrifugation, cells were resuspended in PBS supplemented with 1% fetal bovine serum, 100 ng/ml RNase A, and 30 g/ml propidium iodide. Cells were filtered to remove cell aggregates (Falcon filter top tubes) and analyzed for DNA content by FACS using a FACScalibur cytometer (Becton-Dickinson Instruments). Data were analyzed using Cell Quest analysis software (Becton-Dickinson).

RESULTS
The major goals of this study were to analyze changes in gene expression that occur as cells progress through the S and G 2 phases of the cell cycle and to analyze changes in gene expression that occur during a G 2 DNA damage checkpoint response. In particular, we examined p53-deficient cells because the ability of cells to arrest in G 2 following DNA damage is independent of p53. In addition, we wished to determine whether the G 2 DNA damage response pathway contains a p53-independent transcriptional component. HeLa cells were chosen for these analyses because they are easily synchronized, have an intact G 2 DNA damage checkpoint, and have compromised p53 function. HeLa cells lack functional p53 due to the expression of the papillomavirus E6 protein, which targets p53 for destruction (43). However, the p53 gene is intact in these cells, and under some circumstances a functional p53 protein can be induced (44). Therefore, we examined p53 protein levels by immunoblotting and p21 gene induction by Northern blotting following treatment of HeLa cells with ionizing radiation (Fig. 1). U2-OS cells were used as a positive control because they respond to DNA damage with p53 stabilization and p21 induction. p53 expression was minimal and remained constant following irradiation of synchronized HeLa cells, whereas irradiation of U2-OS cells led to stabilization and accumulation of the p53 protein (Fig. 1A). p21 mRNA levels did not change upon irradiation of HeLa cells, whereas p21 mRNA levels were observed to rise after irradiation of U2-OS cells (Fig. 1B). These data, along with the lack of identification of known p53 transcriptional targets in our microarray analysis (see below), suggest that p53 does not contribute to the transcriptional responses observed in this study.
Next, transcriptional profiling on a genome-wide scale was performed on synchronized HeLa cells in order to identify genes with cell cycle-regulated expression and DNA damage responsive expression. HeLa cells were synchronized at G 1 /S using a double thymidine block protocol. One hour after release from the second block, cells were mock-irradiated or treated with 10 Gy of ␥-irradiation and then returned to normal culture conditions. Duplicate samples were harvested 1 h after release (prior to ␥-irradiation) as well as 2 and 6 h after mock or ␥-irradiation (3 and 7 h after release, respectively). FACS analysis demonstrated that in the 7 h following release from the double thymidine block, cells had progressed from early S to late G 2 (data not shown). cDNA synthesized from poly(A) ϩ RNA was used to generate biotinylated cRNA probe, which was then hybridized to Affymetrix microarray chips representing ϳ42,000 GenBank TM entries.
Comparisons were made between mock-irradiated samples to identify genes whose expression increased or decreased as cells progress through the S and G 2 phases of the cell cycle (Tables I and II). In addition, comparisons were made between mock-and ␥-irradiated samples at each time point to assess the effects of ionizing radiation on gene expression (Table I). In general, a change in expression of 2.5-fold or greater on independent analyses was used for assessment of up-or down-FIG. 1. Ionizing radiation fails to induce p53 protein and p21 mRNA in HeLa cells. HeLa cells were synchronized using a double thymidine block protocol. One hour after release from the second block, cells were mock-irradiated or subjected to 10 Gy of ␥-irradiation. Separately, asynchronously growing U2-OS cells were irradiated with 10 Gy of ␥-irradiation. Cells were harvested at the time of irradiation (IR) (0 h) and at various times after irradiation. A subset of cells were lysed, and the proteins were resolved by SDS-polyacrylamide gel electrophoresis and examined for p53 protein levels by immunoblotting (A). Alternatively, RNA was prepared from the remainder of the cells and monitored for RNAs encoding p21 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) by Northern blotting (B).
regulation. In addition, only changes that were noted in each of two independent analyses were concluded to represent real changes, and only those changes that averaged at least 3-fold were analyzed further. In certain circumstances, genes that showed at least a 2.2-fold change in expression in two independent experiments were analyzed further. Extensive overlap between genes that had a G 2 /M-specific expression pattern and those that were down-regulated in response to ␥-irradiation allowed the presentation of these two data sets in a single table (Table I).
Changes in Gene Expression That Accompany S/G 2 Cell Cycle Progression-To analyze changes in gene expression that occur as cells move through the S and G 2 phases of the cell cycle, samples prepared at 2 and 6 h after mock irradiation (3 and 7 h after release from the block, respectively) were compared with mock-irradiated samples prepared 1 h after release from the block. Only one gene was found to be up-regulated by 2 h after mock irradiation. The protein encoded by this gene (EST AA046674/Unigene ID Hs.108350) has significant homology with three predicted membrane proteins of unknown function in budding yeast, Cdc50, YNRO48w, and YNK323w (45,46). Although this gene with homology to Cdc50 was the only gene observed to be up-regulated by 3 h after release from the G 1 /S block, many more genes (53) were found to be induced by 7 h when cells were in the late G 2 phase of cell cycle ( Table I).
Several of the genes included in Table I (CKS2, TTK, STK15, PLK1) did not fully meet the criteria for inclusion on this list, since the average -fold induction for these genes was less than 3-fold. However, based on the proposed function of the encoded proteins as cell cycle regulatory molecules, these genes were analyzed, and G 2 /M-specific expression was confirmed by Northern blotting (Figs. 2-4 and data not shown). Whenever possible, genes were grouped based on their proposed or speculated functions. Several of the gene products, including cyclin B1, the protein kinases PLK1 and TTK, and the APC regulator CDC20, modulate mitotic entry and/or progression. Several genes included on this list were independently identified as having G 2 or M phase-specific expression in primary human fibroblasts (47) and in Saccharomyces cerevisiae (48,49).
There were also many genes (45 genes) whose expression was reduced as cells exited S phase and progressed through G 2 (Table II). Nine out of these 45 genes were also down-regulated at 3 h after release from the block (denoted by the asterisk in Table II). The down-regulation of some of the genes shown in Table II (i.e. CDC21 and histones) is probably a result of cells completing S phase and progressing through the G 2 phase of the cell cycle. Several of the genes were independently identified to have cell cycle-regulated or serum-responsive expression (47,50).
Changes in Gene Expression during a G 2 DNA Damage Checkpoint Response-In addition to monitoring changes in gene expression that accompany progression through the S and G 2 phases of the cell cycle, we also analyzed changes in gene expression that accompany the cellular response to DNA damage. The experiments were designed to identify genes with increased or decreased expression during G 2 in response to ␥-irradiation in a p53-independent manner. Since there are many previous reports of DNA damage-inducible genes in a variety of cell lines, we predicted that this analysis would identify numerous genes that were induced by ␥-irradiation. Surprisingly, there were no genes that were reproducibly induced either at 2 or 6 h after irradiation under these conditions. When the data were reexamined using less stringent criteria (2.2-fold induction in both analyses), two genes were identified whose expression was induced 6 h after ␥-irradiation. However, Northern blotting failed to confirm the induction of either of these genes after ionizing radiation (data not shown). These findings suggest that gene activation does not contribute significantly to the initiation of the G 2 cell cycle arrest following DNA damage in HeLa cells.
While there were very few, if any, genes induced by ␥-irradiation, there were many genes (24 genes) whose expression was delayed or repressed following DNA damage (Table I). Northern blotting demonstrated that expression of tastin, KIAA0220, EST AA001296, and CKS2 was similarly delayed by DNA damage despite the microarray analysis indicating no change (Figs. 2-4 and data not shown). This suggests that the pattern of DNA damage-induced down-regulation of G 2 /M-specific genes may be more common than predicted by the microarray data. For many of the genes listed in Table I, an expression pattern similar to that of cyclin B1 was observed, i.e. expression of the gene increased as cells progressed through the S and G 2 phases of the cell cycle, but this increase in expression was delayed following DNA damage. The G 2 /M expression pattern of these genes predicts a role in mitotic regulation, and the delay in expression of these genes by genotoxic stress might contribute to delays in mitotic entry and/or exit. In fact, the DNA damage-mediated repression of cyclin B1 expression has been shown to contribute to the G 2 cell cycle delay (51).
A detailed evaluation by Northern blotting was performed to better characterize the expression pattern for eight of the G 2 / M-specific genes found to be down-regulated by ␥-irradiation (Fig. 2). Cells were synchronized at G 1 /S using a double thymidine block protocol. One hour following release from the block, cells were mock-or ␥-irradiated and then returned to normal growth conditions. Samples were harvested at the time of release (prior to ␥-irradiation) and at 3, 7, 11, and 17 h following mock or ␥ irradiation (4, 8, 12, and 18 h after release, respectively). FACS analysis demonstrated that mock-irradiated cells progressed through S phase and were in late G 2 by 8 h after release, and by 12 h most had completed mitosis and were in the G 1 phase of the cell cycle ( Fig. 2A). Irradiated cells were delayed in S phase by ϳ4 h (Fig. 2A). These cells completed S phase between 4 and 8 h and were in late G 2 by 12 h, where they remained at 18 h after release ( Fig. 2A). RNA was isolated from each sample and prepared for Northern blotting. As seen in Fig. 2B, expression of most genes in mock-irradiated samples peaked at 8 -12 h after release from the block and decreased thereafter. A similar G 2 /M-specific expression pattern for each of these genes was also noted when cells were synchronized using mimosine (data not shown). In irradiated cells, accumulation of the mRNAs was delayed relative to mockirradiated samples until 18 h after release, at which time the mRNAs began to accumulate as cells remained arrested in G 2 (Fig. 2B). Irradiated cells completed S phase and reached late G 2 by 12 h after the release in contrast to mock-irradiated cells, which completed S phase and reached late G 2 by 8 h. Since the mRNAs analyzed in these experiments are expressed in G 2 /M, the S phase delay might contribute to the delay in expression of these genes merely by delaying progression into G 2 . However, at 12 h after irradiation, the expression of many of these mRNAs was still markedly repressed (Fig. 2B), despite maximal accumulation of the cells in late G 2 ( Fig. 2A), demonstrating that the effect was not exclusively a result of S phase delay.
To specifically monitor effects of DNA damage on gene expression as a function of cell cycle position, cells were irradiated at 1, 4, and 6 h after release from the double thymidine block (Fig. 3). FACS analysis revealed that the majority of cells were in early S phase 1 h after release, in late S phase 4 h after release, and in G 2 by 6 h after release. Irradiation at each time point led to a delay in expression of each of the genes that were analyzed (Fig. 3B). Interestingly, mRNA levels were elevated TABLE I Genes with G 2 /M-specific and DNA damage-responsive expression In the first column, genes are grouped according to predicted functions. Genes shown in boldface type were independently demonstrated to have a G 2 -or M-phase specific expression pattern (47). An asterisk after the gene name in column one indicates that a homologous protein in yeast is also expressed in a cell cycle-dependent manner (48,49). GenBank ™ accession numbers for the gene or EST are shown in the second column. In some cases, the gene in question was represented more than once on the microarray (examples include cyclin B1, STK15, and KIAA1333). The relative increase in expression during G 2 (as compared with S phase, i.e. at 7 h after release when compared with 1 h after release) is indicated in the third and fourth columns, and the average of these two independent analyses are in boldface type in the fifth column. In the sixth and seventh columns are the relative decrease in expression during G 2 in cells that were irradiated, as compared with those that were mock-irradiated. The average of these two independent analyses of the effect of irradiation is shown in boldface type in the eighth column. Positive numerical values indicate the -fold increase in expression, whereas negative numbers indicate -fold decrease in expression. NC indicates that there was less than 2-fold difference in expression. ϩ/Ϫ indicates that there was an increase or decrease in one of the two analyses but not in the other analysis. by 8 h after release when cells were irradiated in G 2 (6 h after release) but then fell by 12 h after release. Thus, while an S phase delay may contribute to the reduced expression of G 2 /Mspecific genes, the effect is reproducible even when the S phase delay was reduced or avoided by irradiation in late S or G 2 . G 2 /M-specific expression and DNA damage-induced delay in expression was also observed for CKS2, TTK, PLK1, CDC20, KIAA1333, topoisomerase II␣, CENP-A, CAB92444, EST AA417350, tastin, AA001296, KIAA0159, KIAA0069, KIAA0008, melanoma ubiquitous mutated protein, and PIF-1 (data not shown). We were unable to detect mRNAs for EST R70506, poly-(rC)-binding protein, STAT utron, EST AA398563, and EST AA286911 by Northern blotting. Two genes, RAN-GTPase-1 and KIAA0110, did not have the expression pattern indicated by the microarray analysis and were deleted from the list of genes in Table I. These results suggest that there was a low false-positive rate in our data set.
Since our analyses employed synchronized cells, it was for-mally possible that effects of irradiation were specific to irradiation during S or G 2 . Therefore, an analysis of gene expression was performed following ␥-irradiation of asynchronously growing HeLa cells. The majority of cells in this population were in the G 1 phase of the cell cycle at the time of irradiation, and irradiation caused a G 2 accumulation of cells, as expected (Fig. 4A). In each case, irradiation caused a transient decrease in expression of each of the genes with a G 2 /M-specific expression pattern (Fig. 4B). Some of the genes observed to have delayed expression following ionizing radiation in this study have previously been reported to be induced by various forms of DNA damage. For example, genes encoding cyclin B1, tastin, KIAA0069, and KIAA0159 were reported to be induced following treatment with doxorubicin, and murine DDA3 gene expression was reported to be induced following ionizing radiation (52,53). These studies used asynchronously growing populations of cells for their analysis and assayed for changes in gene expression at relatively late times after irradiation. The   Genes with expression that is decreased in G 2 phase relative to S phase are listed according to proposed function in column one. Genes in boldface type were independently identified as cell cycle-regulated genes by Cho et al. (47). Expression of several of the genes is also relatively decreased at 3 h after release from the double thymidine block as compared with expression at 1 h after release and are identified by an asterisk after the gene name. The gene or EST represented on the microarray is listed in the second column. The relative decreases in expression at 7 h after release (when compared with 1 h after release) in each of the two microarray analyses are listed in the third and fourth columns, and the average of these is shown in bold in the fifth column. The change in expression that was noted following serum stimulation of primary fibroblasts (50), if any, is listed in the seventh column. NC indicates that an analysis was performed, and no change with serum was detected. If the gene was not analyzed for serum responsiveness, the column is blank for the gene in question. "I" indicates an increase in expression following serum stimulation at the indicated times, while "D" represents a decrease in expression at the indicated times.  Fig. 4 reveal a transient decrease in RNA levels at early time points after irradiation followed by an increase in RNA levels as cells accumulate in the G 2 phase of the cell cycle. Thus, conclusions concerning whether a gene is up-or down-regulated following DNA damage is influenced by the time point chosen for analysis. Genes that are specifically up-regulated in G 2 /M can appear to be induced by DNA damage at late time points due to the accumulation of cells in the G 2 phase of the cell cycle following DNA damage. We next determined whether similar changes in gene expression would be observed when cells were treated with DNAdamaging agents other than ionizing radiation (Fig. 5). For these experiments, HeLa cells were synchronized using a double thymidine block protocol as before. One hour after release, the cells were untreated (mock) or were treated with agents that induce various forms of DNA damage including UV light, ionizing radiation, and doxorubicin. In addition, cells were subjected to heat shock, a stimulus that can delay cell cycle progression (54, 55) but does not cause a significant accumulation of HeLa cells in G 2 (56). Cells were harvested 7 and 21 h after treatment (8 and 22 h after release, respectively). As expected, mock-irradiated cells and cells exposed to heat progressed through the cell cycle without delay, while cells treated with the DNA-damaging agents arrested in G 2 (Fig. 5A). In both control and heat shock-treated cells, mRNA levels of each of the eight genes were elevated at 8 h after release and were relatively decreased by 22 h after release (Fig. 5B). In contrast, cells treated with the various DNA-damaging agents showed a delay in expression of each of the genes, resulting in lower levels of the mRNA at the 8-h time point. At 22 h after release, RNA levels increased for some of the mRNAs as the cells remained arrested in G 2 , while for others it remained decreased. These data demonstrate that the delay in gene expression of G 2 /M-specific genes noted with ␥-irradiation is also seen with other chemotherapeutic and radiation treatments that exert their effects through diverse mechanisms. Thus, agents that induce the G 2 DNA damage checkpoint alter the timing of expression of many genes that probably play important roles in regulating mitotic entry and/or progression.
Ionizing Radiation Delays Accumulation of Proteins Encoded by G 2 /M-specific Genes-Since mRNA levels may not always correlate with accumulation of the encoded protein, we analyzed effects of cell cycle phase and DNA damage on expression of a subset of proteins encoded by the G 2 /M-specific genes identified in this study. For this experiment, HeLa cells were synchronized with a double thymidine block. Cells were either mock-or ␥-irradiated at 3 h after release and harvested at the Immunoblotting demonstrated the G 2 /M-specific accumulation of cyclin B1, PLK1, CDC20, and TTK proteins. As seen in Fig.  6, protein levels rose as cells progressed from S through the G 2 phases of the cell cycle (0 -8 h in mock-irradiated samples). Cyclin B1 and CDC20 levels decreased precipitously thereafter as the cells progressed from M phase through G 1 (12 and 16 h in mock-irradiated samples), whereas PLK1 and TTK levels decreased more gradually. Levels of C-TAK1, a protein kinase that is not regulated in a cell cycle-dependent manner, was monitored as a control for protein loading (42). Ionizing radiation delayed the accumulation of cyclin B1, PLK1, CDC20, and TTK (compare 8-h time points) but did not affect levels of C-TAK1, as expected. As irradiated cells accumulated in G 2 , the G 2 /M-specific proteins continued to accumulate (16-h time point). Therefore, both the G 2 /M-specific expression pattern and the delay in expression in response to ionizing radiation that we observed for these mRNAs encoding cyclin B1, PLK1, CDC20, and TTK are also seen at the level of protein accumulation.

DISCUSSION
In this study, genome-wide expression profiling was performed to identify genes whose expression is regulated as cells progress from S phase through the G 2 phase of the cell cycle. In addition, this analysis was used to identify genes whose expression is altered when the G 2 DNA damage checkpoint is activated. In particular, we focused on p53-independent transcriptional responses because initiation of the G 2 DNA damage checkpoint occurs normally in the absence of p53. Many genes with a G 2 /M-specific expression pattern were identified in this study (Table I). Several of the genes, including cyclin B1, PLK1, FIG. 5. Various DNA-damaging agents delay expression of genes with a G 2 /M expression pattern. HeLa cells were synchronized using a double thymidine block protocol. One hour after the second release, cells were mock-irradiated (mock), or were subjected to heat shock for 1 h at 42°C (heat), irradiated with 20 J/m 2 UV light (UV), irradiated with 10 Gy of ␥-irradiation (IR), or incubated with 0.5 g/ml doxorubicin (Dox) for 1 h. Cells were harvested at the time of release (0 h) and at 8 and 22 h after release following each treatment condition. Cells were analyzed for DNA by flow cytometry (A) or were processed for RNA isolation, and specific RNAs were detected by Northern blotting (B).
FIG. 6. G 2 /M-specific protein expression is delayed following ionizing radiation. HeLa cells were synchronized using a double thymidine block protocol. Three hours after the second release, the cells were either mock-irradiated (mock) or treated with 10 Gy of ␥-irradiation (IR). Cells were harvested at the time of release (0 h) and 8, 12, and 16 h after release. Cells were analyzed for DNA content by flow cytometry (A) or were processed for analysis by immunoblotting (B) with antibodies to the indicated proteins.
TTK, CDC20, CKSH2, and topoisomerase II␣, have known roles in regulating mitotic entry and/or progression. In addition, many ESTs encoding proteins of unknown function were identified. Several of these ESTs encode proteins with interesting domains including a cyclin box (EST AA608668), HECT domain (ESTs AA281251 and AA233231), and RhoGAP domains (ESTs AA251010, AA213817, and N30185). In addition, we identified the human homolog of the mouse DDA3 protein (EST AA047880), which is implicated in growth suppression (53), and a novel protein encoded, in part, by EST AA417350, which is related to GAS2 and GAR22 (57,58). Characterization of the novel protein products encoded by these ESTs is expected to contribute to our understanding of mitotic and G 2 checkpoint control.
Eight of the genes with G 2 /M-specific expression patterns identified in our analysis are homologous to cell cycle-regulated genes in S. cerevisiae (48,49). These genes are indicated with an asterisk in Table I, and several encode proteins with known mitotic functions. Two recent studies have examined global changes in gene expression as a function of the cell cycle in human fibroblasts with normal p53 function. In the first of these two studies (50), cells were treated by serum starvation followed by serum repletion. While this study was designed to analyze serum-responsive gene expression, cell synchronization is one result of this protocol. In the second study, cells were synchronized with a double thymidine block protocol (47). Although unique genes were identified in each study, these studies, like ours, identified cyclin B1, CKS2, STK-15, PLK1, topoisomerase II␣, and KIAA0069 as having G 2 /M-specific expression patterns or serum-responsive expression patterns (and by inference cell cycle phase-specific expression). Interest-ingly, several of the radiation responsive G 2 /M-specific genes identified in our study (discussed below) were also noted to be down-regulated with cellular senescence (59,60). These findings suggest that cellular responses to DNA damage and aging may share certain features, and it will be interesting to examine mechanisms of gene regulation under both conditions.
We also identified 45 genes whose expression decreased as cells completed S phase and progressed through G 2 (Table II). The down-regulation of some of these (i.e.: CDC21 and histones) is probably a result of cells completing S phase and progressing through the G 2 phase of the cell cycle. Interestingly, nine of the genes in Table II were determined to be cell cycle-regulated in primary human fibroblasts (47). In addition, 10 of the genes shown in Table II were also found to have serum-responsive expression (50), and most of these were found to be induced in the first several hours after serum repletion. Several genes shown in Table II including CTGF, CYR-61, and c-fos have been shown to be transiently induced in response to serum stimulation in other studies (61)(62)(63). Therefore, a portion of the genes listed in Table II may represent growth-responsive genes rather than genes with S phase-specific expression. Further experimentation will be required to fully characterize the expression pattern of these genes.
The paucity of DNA damage-induced genes was one of the most striking findings in this study. The failure of ionizing radiation to induce gene expression in our analysis is in marked contrast to previous studies that identified numerous genes whose expression is induced by a variety of DNA-damaging agents (47,64,65). The lack of p53 induction in our experimental system is one likely explanation for the failure to identify up-regulated genes in this study. However, additional FIG. 7. Redundant mechanisms ensure G 2 arrest after DNA damage. DNA damage activates both p53-dependent and p53-independent pathways that lead to a G 2 cell cycle arrest. p53-dependent transcriptional activation of several genes, including p21, GADD45, and 14-3-3 enforces a persistent G 2 arrest. p53-independent mechanisms include both post-transcriptional mechanisms leading to inactivation of cyclin B1-CDC2 complexes and transcriptional repression of many genes. p53-independent pathways induce a transient G 2 cell cycle arrest. Genes repressed by DNA damage include those encoding known mitotic regulators as well as those encoding proteins predicted to function as mitotic regulators (see Table I).
effects of the viral oncoproteins E6 and E7, the specific DNAdamaging agent used, and the cell synchronization protocol employed cannot be excluded as possible explanations for some of the differences. While there were no genes induced by DNA damage in our analysis, 24 genes whose expression was delayed in response to ionizing radiation were identified (Table I). The delayed expression of these genes, many of which have a G 2 /M-specific expression pattern, might be expected to delay mitotic entry and/or progression. In fact, a delay in cyclin B1 expression following ionizing radiation has previously been shown to play a role in delaying mitotic entry (51). Similarly, delayed activation of PLK1 after DNA damage in late G 2 has been shown to delay exit out of mitosis (66). The mechanism(s) leading to down-regulation of mitotic regulatory genes after genotoxic stress is not yet known. It seems unlikely that a single transcriptional mechanism is employed for all of these genes, since the extent and timing of down-regulation varies. Transcriptional regulatory elements, known as CDE and CHR, repress expression of some of these genes during G 1 (67) and might lead to DNA damage-mediated repression as well. CONCLUSION Genome-wide expression profiling identified numerous human genes with G 2 /M-specific expression patterns. Within this set were genes encoding proteins with known mitotic functions and several genes encoding proteins of unknown function. Global gene expression profiling failed to identify genes induced by ionizing radiation. This suggests that gene induction does not contribute significantly to initiation of G 2 cell cycle arrest following DNA damage in the absence of p53. However, a delay in expression of numerous G 2 /M-specific genes was observed following DNA damage. Thus, human cells have evolved multiple mechanisms to delay the onset and completion of mitosis in the presence of DNA damage, as shown schematically in Fig. 7. p53-independent mechanisms include functional inhibition of the CDC25C protein phosphatase (23) as well as delays in expression of genes encoding proteins required in mitosis (the data presented here). In addition, p53-dependent transcriptional activation of genes such as 14-3-3, p21, and GADD45 contribute to maintenance of the G 2 cell cycle arrest (34 -36), but this pathway is absent in the majority of human cancers that lack p53. These results highlight the redundant nature of the G 2 DNA damage checkpoint response. It is not clear why multiple mechanisms are necessary for prevention of mitosis after genotoxic stress. However, we speculate that the redundancy is present to ensure that the G 2 checkpoint functions flawlessly and may allow p53-independent mechanisms to initiate the arrest prior to the time that p53 accumulates. Our identification of numerous mitotic regulatory genes that are modulated by DNA damage may ultimately provide new molecular targets for the treatment of p53-deficient malignancies.