G2 arrest of cells suffering DNA damage in S phase is crucial to avoid their entry into mitosis, with the concomitant risks of oncogenic transformation. According to the current model, signals elicited by DNA damage prevent mitosis by inhibiting both activation and nuclear import of cyclin B1-Cdk1, a master mitotic regulator. We now show that normal human fibroblasts use additional mechanisms to block activation of cyclin B1-Cdk1. In these cells, exposure to nonrepairable DNA damage leads to nuclear accumulation of inactive cyclin B1-Cdk1 complexes. This nuclear retention, which strictly depends on association with endogenous p21, prevents activation of cyclin B1-Cdk1 by Cdc25 and Cdk-activating kinase as well as its recruitment to the centrosome. In p21-deficient normal human fibroblasts and immortal cell lines, cyclin B1 fails to accumulate in the nucleus and could be readily detected at the centrosome in response to DNA damage. Therefore, in normal cells, p21 exerts a dual role in mediating DNA damage-induced cell cycle arrest and exit before mitosis. In addition to blocking pRb phosphorylation, p21 directly prevents mitosis by inactivating and maintaining the inactive state of mitotic cyclin-Cdk complexes. This, with subsequent degradation of mitotic cyclins, further contributes to the establishment of a permanent G2 arrest.
INTRODUCTIONCyclin B1-Cdk1 is a master mitotic regulator and is therefore an ultimate target toward which most, if not all, antiproliferative signals generated during S and G2 phase converge. In cycling cells, cyclin B1 synthesis increases during late S-early G2 phase (Pines and Hunter, 1989), but the newly formed cyclin B1-Cdk1 complexes are kept inactive by inhibitory phosphorylations on Thr 14 and Tyr 15 (Norbury et al., 1991) and by active and constant nuclear exclusion via Crm1-mediated nuclear export (Hagting et al., 1998;Yang et al., 1998). The rapid nuclear accumulation of cyclin B1-Cdk1 that occurs in prophase coincides with phosphorylation of cyclin B1 at the cytoplasmic retention sequence, which also contains its nuclear export sequence (Hagting et al., 1999; for review, see Porter and Donoghue, 2003). Nuclear import of cyclin B1-Cdk1 coincides with Cdc25 phosphatase-mediated dephosphorylation of Thr 14 and Tyr 15 on Cdk1 (Berry and Gould, 1996). Based on these observations, it has been assumed that cytoplasmic cyclin B1-Cdk1 is not active. However, both earlier and most recent work suggest that cyclin B1-Cdk1 might already be activated in the cytoplasm and most probably at the centrosome (Heald et al., 1993;De Souza et al., 2000;Hirota et al., 2003;Jackman et al., 2003).According to the current model, signals elicited by DNA damage in S and G2 phases prevent mitotic entry by inhibiting both activation (via inactivation of Cdc25) and nuclear import of cyclin B1-Cdk1 (Jin et al., 1998;Yang et al., 1998Yang et al., , 2001Deming et al., 2001). G2 arrest is initiated via phosphorylation of Cdc25 by Chk1/2 kinases, a process that does not require active p53. On the other hand, the...