A mammalian homolog of fission yeast Cdc5 regulates G2 progression and mitotic entry.

Progression through G2/M of the mammalian cell division cycle requires the coordinated expression of many gene products, but little is known of the transcriptional regulators involved. Schizosaccharomyces pombe Cdc5 is a putative transcription factor implicated in G2/M transit. We recently identified a cDNA encoding a putative human transcription factor, now designated human Cdc5 (hCdc5), with homology to S. pombe Cdc5. Widespread expression of hCdc5 in human tissues and homology with expressed sequences in other eukaryotes suggested an evolutionarily conserved general function. Nuclear import of hCdc5 upon serum stimulation of mammalian cells suggested a possible role in cell proliferation. We now report that overexpression of hCdc5 in mammalian cells shortened G2 and reduced cell size. A dominant negative mutant of hCdc5 lacking the carboxyl-terminal activation domain slowed G2 progression and delayed entry into mitosis. Thus, hCdc5 is the first transcriptional regulator shown to affect G2 progression and mitotic entry in mammalian cells.

Orchestration of the cell division cycle includes a series of checkpoints which ensure that some events are completed before others begin (1). One set of controls determines whether the cell replicates its genome in preparation for division (G1/S), whereas another checks that DNA replication is complete and that the cell has grown sufficiently for division to take place (G2/M) (2). Cyclins and cyclin-dependent kinases mediate these events and are themselves regulated through several mechanisms (3). In fission yeast, regulation of the Cdc2⅐Cdc13 cyclin complex by the Wee1 kinase and Cdc25 phosphatase is thought to be the primary mechanism controlling G2/M (4,5). The Cdc2⅐Cdc13 complex accumulates during S phase, but Cdc2 is phosphorylated and thereby maintained in an inactive state by Wee1 (6,7). As cells complete DNA replication, Cdc25 dephosphorylates Cdc2 (8,9), heralding progression through G2 and entry into mitosis. Upstream regulators of Wee1 and Cdc25 also have been described. Nim1 kinase inactivates Wee1 through phosphorylation (10 -12), whereas protein phosphatase 2A has been shown to negatively regulate mitotic entry by its interaction with Wee1 and/or Cdc25 (13).
The biochemical events controlling G2/M transit in mammalian cells are remarkably similar to those in Schizosaccharo-myces pombe. Mammalian Cdc2 kinase accumulates in S phase (14) and is regulated by a Wee1 kinase (15) and Cdc25 phosphatase (16,17). Less is known, however, about their upstream regulation. More generally, whereas G2/M progression clearly requires the coordinated expression of many gene products, little is known about how these are regulated at the level of transcription. Characterization of transcription factors regulating G2 progression and mitotic entry would significantly advance our understanding of the mechanisms controlling this portion of the cell cycle.
The cdc5 gene product in fission yeast is a putative transcription factor implicated in G2/M transit (18,19). Haploid yeast bearing a temperature-sensitive mutation in the cdc5 gene arrest with a diploid complement of DNA, single nucleus, and decondensed chromosomes without evidence of mitotic arrest or defective DNA replication (18,19). We recently identified a cDNA encoding a putative human transcription factor (20) with homology to S. pombe Cdc5 (19). The amino acid sequence of this new protein was 75% identical to that of Cdc5 over 223amino-terminal residues containing the DNA binding domain, but this identity decreased to 17% over the subsequent 535 amino acids (20). Its widespread expression in human tissues and homology with expressed sequences in other eukaryotes suggested an evolutionarily conserved general function (20). Nuclear import upon serum stimulation of mammalian cells suggested a possible role in cell proliferation (20). Without clear evidence that this was a functional Cdc5 homolog, however, the new protein initially was designated Pombe Cdc5related protein.
We now describe the functional analysis of Pombe Cdc5related protein in mammalian cells. These studies strongly suggest that it is a human homolog of S. pombe Cdc5, and so we henceforth refer to it as hCdc5. 1 Overexpression of hCdc5 shortened G2 and reduced cell size. A dominant negative mutant of hCdc5 lacking the carboxyl-terminal activation domain slowed G2 progression and delayed entry into mitosis. These results demonstrate that hCdc5 regulates G2 progression and mitotic entry and is the first transcriptional regulator shown to effect G2/M transit in mammalian cells.

EXPERIMENTAL PROCEDURES
Cell Transfections and Generation of Stable Lines-All cell lines were obtained from ATCC and maintained according to their recommendations. COS-7 and BALBc/3T3 cells were transfected using Lipo-fectAMINE (Life Technologies, Inc.) according to the manufacturer's instructions. Jurkat cells were transfected by electroporation at 0.25 V, 960 microfarads. Synchronization of BALBc/3T3-derived cultures in G0 was accomplished by washing cells with serum-free media and maintaining in media containing 0.5% bovine calf serum for 48 h. Release into G1 was accomplished by the addition of prewarmed bovine calf * This work was supported in part by National Institutes of Health Grants HL44907 and HL43821. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ Supported by National Institutes of Health Clinical Investigator Development Award HL03228. To whom correspondence should be addressed: University of California, San Francisco, 513 Parnassus Ave., Box 0632, San Francisco, CA 94143-0632. E-mail: HS_Bernstein @pedcardgateway.ucsf.edu. serum to 10%. The plasmid encoding FLAG-tagged hCdc5 in pcDNA3 (Invitrogen) was constructed as described previously (20). Cell line 3T15.8 was generated by transfecting BALBc/3T3 cells with pUHD 15-1 (Display Systems Biotechnology), which directs the expression of a tetracycline repressor-VP16 fusion protein, then selecting in the pres-ence of 400 g/ml G418. Cell line 3T15.8.22 was made by co-transfecting 3T15.8 cells with plasmids encoding FLAG-tagged hCdc5 downstream from a minimal CMV promoter and seven tandem repeats of the tet operator (pUHD 10 -1; Display Systems Biotechnology) and a marker for hygromycin resistance. Transfectants were selected with 200 g/ml hygromycin in the presence of G418 and 10 g/ml tetracycline. To assess the effects of constitutive hCdc5 overexpression, cultures were rinsed with tetracycline-free media and re-fed with the same media every 2 h for 6 h. Assays were performed after 36 -48 h in tetracycline-free media. pNGFP-N1, encoding GFP with F64L and S65T mutations (21), was obtained from CLONTECH.
Antibodies and Immunoblot Analysis-Monoclonal antibody B1 was generated against a peptide representing the carboxyl terminus of hCdc5 (residues 769 -788) (20) according to standard methods (22). Monoclonal antibody M2 directed against the FLAG epitope was ob-tained from Eastman Kodak Co. Polyclonal antibodies recognizing murine cyclins were obtained from Santa Cruz Biotechnology. Immunoblot analysis was performed as described previously (20). Monoclonal antibody B1, which recognized both recombinant and endogenous hCdc5, was used to compare expression levels of both proteins. Densitometric analysis was accomplished by scanning representative ECL results into Photoshop (Adobe), then measuring signal densities using NIH Image (NIH).
Transcription Reporter Assay-Portions of the hCdc5 coding region were cloned in-frame into pM (CLONTECH), downstream of nucleotides 1-441 of GAL4. These fusion constructs were co-transfected with a reporter plasmid, pFR-Luc (Stratagene), containing the entire coding region of Photinus pyralis (firefly) luciferase downstream of five tandem repeats of the GAL4 binding element and a basic promoter (TATATA). COS-7 cells were transfected as described above, and luciferase activity was assayed in whole cell lysates using the dual-luciferase reporter assay system (Promega) according to the manufacturer's instructions in a Monolight 3010 luminometer (Analytical Luminescence Laboratory). Plasmid pRL-SV40 (Promega), encoding Renilla reniformis (sea pansy) luciferase downstream of an SV40 promoter, was included in each transfection for normalization of transfection efficiency. Plasmid pM3-VP16 (CLONTECH), encoding a GAL4-VP16 fusion, was used as a positive control.
Cell Cycle Analysis-Asynchronously dividing or synchronized cultures were analyzed after 72 h of active division or at indicated timepoints after serum stimulation, respectively. 10 6 cells were stained with propidium iodide and Hoechst 33342 to allow measurement of DNA content in living cells. Flow cytometry was performed using a Becton Dickinson FACStar Plus with dual argon ion lasers at 488 and 363 nm light output. Propidium iodide, Hoechst, and GFP signals were acquired using 630/22-, 470/10-, and 530/30-mm bandpass filters, respectively. All analyses of DNA content were performed on 20,000 collected events using CellQuest software (Becton Dickinson). Prophase index assay was accomplished by the addition of 0.5 g/ml nocodazole 48 h after transfection. Cells then were stained with Hoechst 33342 at various timepoints. Slides containing 5 ϫ 10 3 cells were prepared in triplicate from each transfection at each timepoint, and 200 GFP-positive cells per slide were scored for the presence of chromosomal condensation. For all experiments, analysis of variance was used to identify significant differences among all groups. Multiple comparisons were then performed using the Bonferroni t test to isolate specific differences.

RESULTS AND DISCUSSION
To study the effects of hCdc5 on the mammalian cell division cycle, we attempted its constitutive overexpression in a variety of cell types but were unable to generate stable lines. As this suggested that constitutive expression at high levels might be incompatible with cell survival, we generated a BALBc/3T3based cell line, designated 3T15.8.22, in which hCdc5 was expressed under control of a tetracycline-regulatable promoter (23, 24). Recombinant hCdc5 was not detectable in the presence of tetracycline; however, in the absence of tetracycline, recombinant hCdc5 was expressed at levels ϳ2-fold above that of the endogenous protein (Fig. 1A), as determined by densitometry.  hCdc5(432-675)), present in mutant ⌬675, failed to activate transcription in the absence of the carboxyl terminus.
As has been shown for transiently expressed (20) and endogenous hCdc5, 2 stably expressed recombinant hCdc5 translocated to the nucleus upon serum stimulation (data not shown).
To assess the effect of increased hCdc5 expression on cell cycle progression, 3T15.8.22 cells were synchronized in G0 with low serum then stimulated with high serum and measured for DNA content at various times (Fig. 1B). Induction of hCdc5 by removal of tetracycline accelerated progression through the cell cycle; cells returned to G1 by 18 h after release from quiescence in the absence of tetracycline compared with 24 h in the presence of tetracycline. By contrast, the parent cell line 3T15.8, which expressed the tetracycline repressor-VP16 fusion protein but not exogenous hCdc5, returned to G1 in 24 h in the presence or absence of tetracycline (data not shown). This cell cycle acceleration was maintained over time, as manifest by an increase in cell number over several cycles (Fig. 1C). Over 72 h of continuous growth, cells overexpressing hCdc5 exhibited an abbreviated cell cycle length of ϳ19 h, whereas cells in which recombinant hCdc5 expression was repressed by tetracycline grew at a slower rate, with a cycle length of ϳ28 h. The cell cycle length of 3T15.8 cultures was ϳ26 h and unaffected by tetracycline.
To determine which phase of the cell cycle was shortened by overexpression of hCdc5, we analyzed asynchronously dividing 3T15.8.22 and 3T15.8 cultures grown in the presence and absence of tetracycline (Fig. 1D). In 3T15.8.22 cultures, a highly significant tetracycline-dependent decrease in the calculated length of G2/M was noted (1.9 h versus 7.4 h). An effect in S phase also was seen (2.2 h versus 5 h). Control 3T15.8 cultures showed no significant tetracycline-dependent effect. Thus, hCdc5 overexpression significantly accelerated G2/M.
In fission yeast, cells must reach a size threshold to enter mitosis (5,25). In mammalian cells, size requirements for mitotic entry are presumed but have not been demonstrated. Since mammalian cells overexpressing hCdc5 displayed a remarkable acceleration of G2/M, we asked whether overexpression might affect cell size. Dividing 3T15.8.22 cultures were synchronized in G0 with low serum in the presence or absence of tetracycline, then stimulated with serum and analyzed over time for DNA content and cell size by flow cytometry (Fig. 2). In the absence of tetracycline, 3T15.8.22 cultures showed a decrease in average cell size as they completed a division cycle and entered G0. Upon stimulation with serum, this smaller average cell size was maintained over at least two cell cycles. Moreover, unlike the parent cell line or 3T15.8.22 cultures in the presence of tetracycline, cells overexpressing hCdc5 exhibited little increase in size as they approached mitosis.
The basis for the decrease in cell size with overexpression of hCdc5 is unknown. In fission yeast, wee1 ts mutants and Cdc25 overexpressors display shortened G2 and decreased cell size but a normal cell cycle length (5), supporting the notion that a size set point has been altered. In mammalian cells overexpressing hCdc5, decreased cell size occurred in the setting of a shortened cell cycle. A similar phenomenon has been observed 2 H. S. Bernstein, unpublished observation. in mammalian cells overexpressing cyclin D1. In these cells, decreased cell size was noted in the context of an uncompensated decrease in G1/S transit time and growth acceleration (26). The size reduction seen in hCdc5 overexpressing cultures may thus be due either to a readjustment of the size requirements for mitotic entry or, more simply, to a shortened cell cycle length with less time for cell growth.
Constitutive expression of recombinant hCdc5 at levels approximately 2-fold that of the endogenous protein in dividing cultures resulted in a surprisingly dramatic phenotype. If endogenous hCdc5 levels varied throughout the cell cycle, constitutive overexpression might result in relative changes in hCdc5 levels much greater than 2-fold during certain phases of the cell cycle. To investigate this possibility, we performed immunoblot analysis of synchronized BALBc/3T3 and 3T15.8.22 cultures with antibodies against endogenous and recombinant hCdc5. This demonstrated that endogenous hCdc5 was expressed early in the cell cycle (Fig. 3), suggesting that its effects in G2/M are not regulated at the level of expression. As little else is known about the regulation of this recently described protein and its effectors, the mechanism by which hCdc5 overexpression influences cell growth parameters remains to be elucidated.
To further examine the role of hCdc5 in G2 progression, we generated a mutant, designated ⌬675, that contained the DNA binding and nuclear localization domains of hCdc5 but lacked its carboxyl terminus (Fig. 4A). Studies of the carboxyl terminus of hCdc5 fused to a heterologous DNA binding domain suggested that this region contained a domain capable of activating transcription in vitro (Fig. 4B). Therefore, mutant ⌬675 should compete with wild-type hCdc5 for binding to as yet unidentified target promoters but should fail to activate them. Jurkat human T cell leukemia cultures were transiently transfected with a vector directing expression of wild-type hCdc5 or ⌬675 together with a plasmid directing expression of GFP as a marker for transfection (21). Cells transfected with the ⌬675 cDNA expressed a protein of the expected molecular weight by immunoblot, and immunostaining confirmed that the mutant protein localized to the nucleus (data not shown). GFP-positive cells in asynchronously dividing cultures were analyzed for cell cycle distribution by flow cytometry (Fig. 5A). Cells expressing the ⌬675 mutant accumulated in G2/M, whereas cells transfected with the empty vector had a phase distribution similar to untransfected cells. Cultures transfected with wild-type hCdc5 distributed toward G1, consistent with the experiments with stable lines described above. Co-expression of wild-type hCdc5 with ⌬675 titrated the effect of the mutant on G2/M accumulation (Fig. 5B), consistent with the mutant acting specifically in the hCdc5 pathway.
We next used mutant ⌬675 to ask whether inhibition of hCdc5 function delayed G2 versus mitosis itself (Fig. 5C). Jurkat cultures transfected as above were examined for chromosomal condensation at various times after exposure to nocodazole, which arrests these cells in prophase (27,28). GFPexpressing cells transfected with ⌬675 exhibited a significantly slower rate of accumulation of cells arrested in prophase compared with empty vector transfectants. Cells expressing wildtype hCdc5 displayed a slightly faster rise in prophase index compared with empty vector and a significantly accelerated increase compared with cells expressing the mutant. This result suggests that hCdc5 acts at least in part during G2 before the onset of prophase in mitosis. Taken together with the flow cytometry data (Fig. 5B), these findings strongly support a role for hCdc5 in the transcriptional regulation of G2 progression and mitotic entry. hCdc5, S. pombe Cdc5, and other Cdc5-like proteins are putative transcription factors by virtue of their primary structures (19, 20, 29 -31). In addition, we have shown that the carboxyl terminus of hCdc5 is capable of activating transcription in vitro (Fig. 4B). Genes regulated by Cdc5 family members, however, remain unknown. The Cdc2⅐cyclin B complex drives G2 progression in mammalian cells (3,32). Inhibitory phosphorylation of Cdc2 by Wee1 kinase, and stimulatory dephosphorylation by Cdc25C phosphatase constitutes the major regulatory mechanism governing Cdc2-cyclin B activity (15,16). S. pombe wee1 ts mutants and Cdc25 overexpressors display shortened G2 and reduced size (5). Since overexpression of hCdc5 in mammalian cells results in a similar phenotype, hCdc5 might be a transcriptional activator for Cdc25C or an inhibitor of Wee1 such as Nim1 (11,12). Another formal possibility is that Cdc5 family members are regulated by and act as effectors of Cdc2 kinase. Although this is supported by the recent finding that a Xenopus laevis homolog of Cdc5 is an in vitro substrate for Cdc2 kinase (31), there is as yet no other evidence to suggest that transcriptional events regulated by Cdc2 are important for G2 progression and mitotic entry.
Characterization of hCdc5 provides a new tool for investigating and manipulating the mammalian cell cycle. Identification of genes regulated by hCdc5 will shed light on the precise mechanism by which Cdc5-like proteins regulate cell cycle progression, specifically G2/M. Moreover, inhibitors of hCdc5 function might prove useful for arresting cell proliferation in human neoplasms. Conversely, overexpression of hCdc5 in concert with other regulatory molecules may allow cellular regeneration of cardiac myocytes or other nondividing tissues.