CDC25A Phosphatase: a Rate-Limiting Oncogene That Determines Genomic Stability Free

Deregulated cell cycle progression is a hallmark of cancer (1). Regulatory components of the cell cycle machinery often become targets of genetic alterations during carcinogenesis. Cyclin-dependent kinases (CDK) play central roles in promoting cell cycle progression, and uncontrolled activation of CDKs is the driving force of cancer cell proliferation. Overexpression of CDKs and cyclins, as well as down-regulation of CDK inhibitors, is found in a majority of human cancers. High levels of functional redundancy seem to exist among the components of the CDK system, according to remarkably mild developmental phenotypes of mice deficient for each component, such as cyclin D1, D2, D3, E1, E2, CDK2, CDK4, or CDK6 null mice (2). However, recent studies showed critical rate-limiting roles for each of cyclin D1, cyclin E, and CDK4 in transformation and tumorigenesis mediated by other oncogenes (e.g., RAS; refs. 3–6). These observations imply that oncogenic cell cycle progression during tumor initiation requires nonredundant regulation of specific CDK complexes or higher capacity of total CDK activity. CDK activation depends not only on the association with a cyclin subunit but also on proper phosphorylation of the catalytic subunit. CDK-activating kinases mediates phosphorylation of a threonine residue located in the T-loop domain to activate CDKs. In contrast, phosphorylation of the tyrosine and threonine residues located in the ATP binding domain, i.e., Tyr¹⁵ and Thr¹⁴ of CDK1 and CDK2, inhibits the CDK activity. This inhibitory phosphorylation is mediated by the WEE1 and MYT1 kinases, and subsequently eliminated (dephosphorylated) by the CDC25 dual specificity phosphatases (7). Whereas the levels of cyclin B/CDK1 complexes accumulate progressively from early G₂ phase, CDK1 activation at the G₂-M transition depends on the activity of CDC25 phosphatases. Thus, CDC25 phosphatases are critical for timely CDK activation in cell cycle progression. Mammalian cells have three CDC25 proteins, CDC25A, B, and C. The expression patterns of these proteins are mostly overlapping, and the knowledge on the function of each CDC25 protein has been limited. Studies using cultured cells suggest that CDC25A regulates both G₁-S and G₂-M transitions, whereas CDC25B and CDC25C are involved only in G₂-M regulation.

Recent studies provided convincing evidence that the regulation of CDC25 proteins is important for cell cycle checkpoint in response to DNA damage, acting as effective targets of the checkpoint-signaling pathways. Damage caused by ionizing irradiation, UV light, oxidative stress, and other DNA-damaging agents leads to activation of ataxia-telangiectasia–mutated (ATM) and ATR (ATM- and Rad3-related) kinases, which then activate the checkpoint transducer kinases, Chk1 and Chk2. Chk1 phosphorylates CDC25A on multiple sites including Ser⁷⁶, Thr¹²⁴, and Ser¹⁷⁸. Whereas Ser¹⁷⁸ phosphorylation creates a docking site for 14-3-3 proteins, Ser⁷⁶ phosphorylation promotes subsequent phosphorylation of Ser⁸² and Ser⁸⁸ by undefined kinase(s) (refs. 8, 9). Phosphorylation of Ser⁸² and Ser⁸⁸ facilitates the association with the SCF^β−TrCP ubiquitin ligase complex. Thus, the Chk1-mediated priming phosphorylation of CDC25A initiates the multistep process of ubiquitin-mediated proteasomal degradation. The Chk1-dependent degradation of CDC25A is now thought to be a major p53-independent mechanism of cell cycle arrest or delay in G₁, S, and G₂, given the role for CDC25A in activating CDKs throughout the cell cycle. CDC25B and CDC25C are also substrates of CHK1 and/or CHK2. Whereas the significance of CDC25B phosphorylation remains unclear, CHK2-mediated phosphorylation leads CDC25C to association with 14-3-3 and cytoplasmic sequestration.

The cell cycle–promoting action of CDC25A and its role as a p53-independent checkpoint target suggest that the expression level of this phosphatase must be tightly controlled in normal cells. Misregulation of CDC25A levels results in impaired checkpoint response to DNA damage with aberrantly high CDK activity. Impaired checkpoint could lead to genomic instability, a hallmark of cancer. Indeed, CDC25A overexpression has been reported in a variety of human cancers, including breast, liver, esophageal, endometrial, and colorectal cancers, and non–Hodgkin lymphomas (7, 10). Furthermore, CDC25A overexpression often correlates with more aggressive diseases and poor prognosis. Recent studies on in vivo roles of the checkpoint pathways suggest that genomic instability caused by checkpoint defects could be a major mechanism of tumor initiation and progression (11), although it is still controversial. CDC25A is an E2F target gene, and its high expression in cancer tissues may simply reflect enhanced proliferation of cancer cells. Therefore, the central questions that awaited experimental evaluation were whether deregulated overexpression of CDC25A is tumorigenic in vivo and how altered regulation of CDC25A levels affects cell cycle control and genomic stability during tumor initiation. We addressed these questions by generating and characterizing mouse mammary tumor virus (MMTV)-CDC25A transgenic mice and Cdc25A-knockout mice, as reported in recent two articles (12, 13).

MMTV-CDC25A transgenic mice exhibit modest alveolar hyperplasia but no spontaneous tumorigenesis in mammary glands, suggesting that CDC25A overexpression alone is insufficient for tumor initiation (12). However, MMTV-CDC25A;MMTV-H-ras double transgenic mice develop mammary tumors with much shorter latency than MMTV-H-ras mice (12 versus 20 weeks). This is consistent with a previous study on transformation of cultured fibroblasts (14). The tumor latency in MMTV-CDC25A;MMTV-neu double transgenic mice is comparable with that in MMTV-neu mice; however, mammary tumors grow faster with more invasive characteristics in MMTV-CDC25A;MMTV-neu double transgenic mice. Importantly, tumor cells with CDC25A overexpression display miscoordination of S phase and mitosis, and have severe genomic instability indicated by aneuploidy, deletion of fragile chromosomal regions (e.g., a telomeric region of chromosome 4 including the Trp73 and Casp9 loci), and other chromosomal aberrations. The chromosome 4 locus is homologous to human chromosome 1p31-36, a hotspot for cancer-associated alterations. Although it is yet to be determined how these chromosomal aberrations are involved in the aggressive phenotype of CDC25A-overexpressing tumors, this study clearly indicates that CDC25A cooperates with the neu-ras oncogenic pathway in mammary tumor development and presents the first in vivo evidence of oncogenic function of CDC25A. Mammary tissues in Chk1 heterozygous (+/−) mice exhibit similar cell cycle miscordination with increased expression of CDC25A (15), and Chk1 (+/−) mice are more susceptible to MMTV-wnt–induced tumorigenesis (16). These observations support the notion that CDC25A down-regulation is a major mechanism of CHK1-mediated checkpoint response and critical for tumor suppression in vivo.

To obtain further insight into the roles of CDC25A in tumorigenesis, we recently generated a Cdc25A-null mouse strain (13). Cdc25A-homozygous (−/−) mice die in utero by embryonic day 7, indicating that Cdc25A is an essential gene for early embryogenesis. This lethal phenotype is in sharp contrast with viable Cdc25B (−/−) and Cdc25C (−/−) mice with limited developmental defects (17, 18). Cdc25A-heterozygous (+/−) mice do not show appreciable developmental defects. The expression levels of CDC25A protein in Cdc25A (+/−) mouse embryonic fibroblasts (MEF) were significantly reduced and correlated with increased Tyr¹⁵ phosphorylation of CDK1 and CDK2. Although Cdc25A (+/−) MEFs display normal kinetics of cell cycle progression, they have enhanced G₂ checkpoint, arresting more efficiently in response to ionizing irradiation. Importantly, Cdc25A (+/−) MEFs show markedly reduced efficacy in undergoing transformation upon expression of activated RAS and a dominant-negative p53 mutant. These data suggest that CDC25A plays a crucial role in RAS-mediated oncogenic transformation. Consistently, Cdc25A (+/−) mice exhibit significant resistance to tumorigenesis involving RAS activation. The latency of mammary tumorigenesis induced by MMTV-neu- or MMTV-ras transgene is markedly prolonged in the Cdc25A (+/−) background. Decreased proliferation and increased Tyr¹⁵ phosphorylation of CDK1/2 are observed in premalignant mammary tissues of Cdc25A (+/−);MMTV-neu mice, compared with those in Cdc25A (+/+);MMTV-neu mice. Taken together, we concluded that the expression level of CDC25A is rate limiting for in vivo tumorigenesis induced by activation of the HER2/neu-RAS oncogenic pathway.

These studies suggest that cells must control CDC25A levels within an appropriate range so that the coordination of cell cycle progression and checkpoint response is maintained (Fig. 1). Aberrantly high levels of CDC25A cause deregulated CDK activation with decreased Tyr¹⁵ phosphorylation and impair checkpoint response, which cooperate with the oncogenic action of RAS. Lower levels of CDC25A result in increased Tyr¹⁵ phosphorylation of CDKs, enhanced checkpoint response, and relative resistance to RAS-mediated tumor initiation. It is well-known that activation of RAS triggers not only proliferation-stimulatory signals (e.g., extracellular signal-regulated kinase activation, p27^Kip1 down-regulation, and Cyclin D1 up-regulation) but also proliferation-inhibitory or senescence-inducing signals (e.g., p53 activation and p21^Cip1 up-regulation; ref. 19). The senescence-like response of primary cells to RAS activation is likely to result from oxidative stress with accumulated reactive oxygen species (ROS). RAS activation in p53-deficient MEFs can cause transformation, with the p53-dependent senescence checkpoint abrogated. We hypothesize that RAS activation affects CDC25A levels with a balance of conflicting signals. RAS may activate CDK4/6 and CDK2 by Cyclin D1 up-regulation and p27^Kip1 down-regulation, respectively. As a consequence, E2F transcription factors are activated, which may transactivate CDC25A expression. On the other hand, RAS-induced oxidative stress may activate not only the p53-dependent pathway but also the CHK1-dependent (p53-independent) pathway, which leads to CDC25A degradation. CHKI is also an E2F target gene (20) and seems to down-regulate CDC25A protein levels during S and G₂ even during unperturbed proliferation. If the CHK1-dependent signaling pathway or the CDC25A ubiquitination process is somehow perturbed during RAS-mediated tumorigenesis, CDC25A should become robustly overexpressed at both transcriptional and posttranscriptional levels, and genomic stability will be severely compromised. This hypothesis is consistent with the clinical correlations between CDC25A expression and aggressive forms of cancers. Thus, it is critical to fully define the mode of CDC25A activation during RAS-mediated tumor initiation and the mechanisms of CDC25A overexpression in human cancers.

The rate-limiting role for CDC25A in tumor initiation and progression provides an encouraging scientific support to the ongoing efforts in developing therapeutic strategies targeted on CDC25 phosphatases (7). Multiple CDC25 inhibitory compounds have been developed, such as para-quinonoid compounds derived from vitamin K. The potent para-quinoid compounds NSC663284, N82685, and IRC083864 can inhibit proliferation of cancer cells and in vivo growth of xenografted human tumors. More characterizations about selectivity and toxicity are needed before clinical trials of such compounds are validated. It is plausible that pharmacologic inhibition of CDC25 phosphatases in vivo would induce senescence-like irreversible cell cycle arrest, especially in cancer cells that accumulate ROS due to aberrant oncogene activation. Also, partial inhibition of CDC25A might be beneficial for genomic stability and tumor suppression of normal tissues. Another implication from our studies is regarding recent clinical trials of CHK1 inhibitors such as UCN-01 and XL-844. Although these compounds could be a new class of drugs to sensitize cancer cells in patients to cytotoxic drugs or radiation, the inhibition of CHK1 is likely to increase the CDC25A level not only in cancerous tissues but also in normal tissues. This may lead to decreased genomic stability and potential secondary carcinogenesis, together with DNA damage induced by chemotherapy or radiation therapy. Further investigations are needed to fully understand the pathophysiologic roles of the CHK1-CDC25A pathway and better design targeted intervention.

Grant support: NIH (R01-CA100204, CA112282, and HD38085), the Department of Defense (DAMD 17-02-1-0413; H. Kiyokawa), the Searle Leadership Fund, the Zell Scholar Fund, and the Robert H. Lurie Comprehensive Cancer Center.

We thank our collaborators, especially Sophia Y. Tsai and Eleftherios T. Papoutsakis, for their invaluable contributions.