Unmasking the S-Phase-Promoting Potential of Cyclin B1

See allHide authors and affiliations

Science  09 May 2003:
Vol. 300, Issue 5621, pp. 987-990
DOI: 10.1126/science.1081418


In higher eukaryotes, the S phase and M phase of the cell cycle are triggered by different cyclin-dependent kinases (CDKs). For example, in frog egg extracts, Cdk1–cyclin B catalyzes entry into mitosis but cannot trigger DNA replication. Two hypotheses can explain this observation: Either Cdk1–cyclin B fails to recognize the key substrates of its S-phase–promoting counterparts, or its activity is somehow regulated to prevent it from activating DNA synthesis. Here, we show that Cdk1–cyclin B1 has cryptic S-phase–promoting abilities that can be unmasked by relocating it from the cytoplasm to the nucleus and moderately stimulating its activity. Subcellular localization of vertebrate CDKs and the control of their activity are thus critical factors for determining their specificity.

DNA replication and entry into mitosis are triggered by different cyclin-dependent kinase (CDK)–cyclin complexes in vertebrate cells. Thus, in Xenopus egg extracts, Cdk2–cyclin E or Cdk1–cyclin A can promote entry into S phase, whereas Cdk1–cyclin B does not (1, 2). Conversely, Cdk1–cyclin B promotes entry into mitosis (2), but Cdk2–cyclin E does not. The origins of this specificity are unclear, especially considering that a single B-type cyclin can promote both S phase and mitosis in fission and budding yeast (35), and the addition of cyclin A to Xenopus egg extracts can trigger either DNA replication or mitosis, depending on the amount added (2, 6). An important difference between vertebrate and yeast mitosis is that mitosis in yeast takes place largely within the nucleus.

The inability of Cdk1–cyclin B to support replication might stem from its failure to recognize the targets of S-phase–promoting CDK-cyclin complexes as substrates (7). Alternatively, Cdk1–cyclin B may have an innate potential to promote replication that is masked by mechanisms that regulate its protein kinase activity in time and subcellular space. Two potentially relevant restraints are the switchlike response of Cdk1–cyclin B activity to increases in cyclin B concentration (810) and the predominantly cytoplasmic localization of the vertebrate B-type cyclins (1114). We found that overcoming these two restraints by redirecting Cdk1–cyclin B1 to the nucleus, and adding Cdc25 phosphatase to allow its stable activation at intermediate levels, is sufficient to enable this form of Cdk1 to support DNA replication.

Egg extracts contain sufficient Cdk2–cyclin E to support replication of added chromatin without further additions. We removed (15) more than 90% of endogenous cyclin E and associated Cdk2 with Sepharose beads that were covalently coupled to GST-p21N [a fusion protein comprising glutathione S–transferase (GST) linked to the N-terminal Cdk inhibitor portion of p21] (Fig. 1A). This reduced the replication to 10 to 15% of that observed in control extracts (Fig. 1B). This depletion protocol did not substantially affect the concentration of Cdk1, and the replication could be fully restored by the addition of recombinant GST–cyclin A1 (16), whereas addition of GST–cyclin B1 caused the extracts to enter mitosis.

Fig. 1.

Nuclear targeting of cyclin B1 allows it to support DNA replication. (A) Cyclin E and Cdk1 content in interphase egg extracts after incubation with Sepharose beads coated with either GST or GST-p21N. (B) Replication of chromosomal DNA in mock and cyclin E–depleted extracts. (C) Distribution of endogenous cyclin E and 35S-labeled cyclin E–B1 between the cytoplasmic and nuclear fractions of egg extracts. (D) Localization of Myc-tagged GST–cyclin E–B1 expressed in mouse cells. Myc-tagged cyclin is shown in red; Hoechststained nuclei are shown in blue. (E) Localization of Myc-tagged GST–cyclin B1ΔN120 expressed in mouse cells. (F) Nuclear morphology and Cdc27 mobility 100 min after the addition of various concentrations of cyclin B1ΔN120 and cyclin E–B1 to a cyclin E–depleted extract. Entry into mitosis is marked by chromosome condensation and the conversion of the Cdk1–cyclin B1 substrate Cdc27 to a slow-migrating form. (G) Histone H1 kinase activity of cyclin E–depleted extracts 100 min after the addition of various concentrations of cyclin B1ΔN120 and cyclin E–B1 (in arbitrary units). (H) DNA replication assayed 85 min after the addition of various concentrations of GST–cyclin E–B1 or GST–cyclin B1ΔN120 to cyclin E–depleted egg extracts. (I) Density of newly synthesized (radiolabeled) DNA in bromodeoxyuridine triphosphate–supplemented mock-depleted or cyclin E–depleted egg extracts plus and minus 80 nM cyclin E–B1. Extracts contained 3 μM Cdc25B to permit cyclin E–B1 to trigger replication but not entry into mitosis (Fig. 3B). The products of semiconservative DNA replication have a density of ∼1.75 g/ml. (J) DNA replication in cyclin E–depleted extracts supplemented with various concentrations of cyclin B1ΔN120 or a version targeted to nuclei by addition of the basic NLS from SV40 large T antigen.

Cyclin B1 is normally kept in the cytoplasm by a nuclear export signal (NES) that is located between residues 108 and 117 (13, 1719). Because cyclin E can only support replication as a nuclear protein (14), the inability of cyclin B1 to support DNA replication might be explained by its failure to dwell in the nucleus. A GST–cyclin B1ΔN120 fusion protein retains the essential Cdk-activating portions of cyclin B1, but lacks its NES and accumulates in nuclei, albeit slowly (19) (Fig. 1E). To accelerate nuclear import, we introduced the nuclear localization sequence (NLS) from cyclin E into the N terminus of cyclin B1 (14, 15). The resulting GST–cyclin E–B1 chimera concentrated within the nuclear fraction of egg extracts to a similar degree as endogenous cyclin E (Fig. 1C). GST–cyclin E/B1 chimeras also localized to the nucleus in transfected mouse cells (Fig. 1D), whereas GST–cyclin B1ΔN120 localizes predominantly to the cytoplasm (Fig. 1E). Similar results were observed in Xenopus oocytes (16).

Various concentrations of GST–cyclin E-B1 or GST–cyclin B1ΔN120 were added to egg extracts that lacked endogenous cyclin E, and the consequences were examined after an 80-min incubation. The addition of up to 120 nM concentrations of either form of cyclin B1 resulted in little or no activation of histone H1 kinase activity, but when a concentration of 160 nM or more was added, robust kinase activity was detected (Fig. 1G), and the extracts entered mitosis, as judged by nuclear envelope breakdown, chromatin condensation, and modification of Cdc27, a well-established substrate of Cdk1–cyclin B (Fig. 1F). No concentration of GST–cyclin B1ΔN120 restored replication, but substantial DNA synthesis occurred in extracts that were supplemented with GST–cyclin E-B1 (Fig. 1H). Density substitution experiments indicated that the DNA synthesis induced by cyclin E–B1 was the product of semiconservative DNA replication (Fig. 1I). Replication was also stimulated by a recombinant GST–cyclin B1ΔN120 protein containing the basic NLS from SV40 large T antigen (20) (Fig. 1J), suggesting that the addition of nuclear targeting signals to cyclin B1 was sufficient to enable it to support replication.

The concentrations of GST–cyclin E–B1 that restored replication overlapped with those that evoked mitosis, however, so we examined the timing of replication and onset of mitosis in supplemented egg extracts in more detail. The addition of 200 nM GST–cyclin E–B1 to a cyclin E–depleted extract restored the majority of the replication that was observed in a control mock-depleted extract at the 60-min time point (Fig. 2A), but higher or lower concentrations were ineffective. The kinetics of histone H1 kinase activation (Fig. 2B) revealed a substantial lag before full activity was achieved when 200 nM cyclin E–B1 was added; higher concentrations of cyclin E–B1 activated the protein kinase more quickly. DNA replication in the extract containing 200 nM cyclin E–B1 stopped shortly after 1 hour (Fig. 2C), coincident with the entry into mitosis, as judged by nuclear envelope breakdown and chromosome condensation (Fig. 2D). Extracts supplemented with higher concentrations of cyclin E–B1 entered mitosis more rapidly and replicated very little of their DNA (Fig. 2, C and D), whereas the mock-depleted extract remained in interphase (Fig. 2D) and continued to synthesize DNA (Fig. 2C). These data indicate that the addition of appropriate concentrations of cyclin E–B1 induced a gradual rise in Cdk1 activity that first triggered replication and then provoked entry into mitosis, which terminated further DNA synthesis.

Fig. 2.

Intermediate concentrations of cyclin E–B1 sequentially trigger replication and mitosis. (A) Effect of various cyclin E–B1 concentrations on chromosomal DNA replication (assayed 60 min after cyclin addition) in cyclin E–depleted egg extracts. (B) Kinetics of histone H1 kinase activation upon addition of cyclin E–B1 to cyclin E–depleted egg extracts. Units are arbitrary. (C) Kinetics of DNA replication in mock-depleted (control) or cyclin E–depleted extracts supplemented with the indicated concentrations of cyclin E–B1. (D) Kinetics of entry into mitosis in cyclin E–depleted extracts supplemented with mock-depleted (control) or cyclin E–B1. As in Fig. 1, histone H1 kinase activation correlates with the onset of mitosis.

The nonlinear relationship between cyclin B concentration and histone H1 kinase activity (Fig. 1G) is thought to arise because the enzymes that regulate Cdk1–cyclin B kinase activity are also its substrates. Cdk1–cyclin B acts to suppress its inhibitors, the Wee1 and Myt1 kinases, and it stimulates its activators, the Cdc25 phosphatases (2123). Wee1 and Myt1 suppress the kinase activity of low concentrations of Cdk1–cyclin B until cyclin B levels rise beyond a critical threshold value that fires the autocatalytic activation mechanism. By forcing a B-type cyclin into nuclei, we have apparently opened a window during which its associated kinase activity could trigger the firing of replication origins before catalyzing the onset of mitosis. We therefore attempted to short-circuit the feedback loop that tends to restrict Cdk1–cyclin B1 activity to either very low or very high levels, hoping to establish an intermediate range of cyclin E–B1 concentrations that could promote DNA replication but not trigger mitosis.

To overcome the inhibitory effects of endogenous Wee1 and Myt1, we added high concentrations (3 μM) of a variant of Cdc25B phosphatase containing only the catalytic domain (24). This stimulated the histone H1 kinase activity evoked by added GST–cyclin E–B1 at low concentrations (Fig. 3A) and reduced the concentration of cyclin E–B1 required for maximum DNA replication from 200 to 80 nM (Fig. 3B). The addition of Cdc25 did not reduce the concentration of cyclin E–B1 required to induce nuclear envelope breakdown (16). Thus, with added Cdc25B, the concentrations of cyclin E–B1 that best restored replication were not sufficient to trigger entry into mitosis. The addition of Cdc25B also allowed the NES-deficient cyclin B1ΔN120 to support some DNA replication (Fig. 3C) but did not restore replication in extracts that were supplemented with any concentration of GST–cyclin B1ΔN82, which retains its NES (Fig. 3D). This result presents evidence against the possibility that heterologous sequences providing nuclear-localization determinants might also confer specificity for replication-promoting substrates, and implies that nuclear localization of the active kinase is the key.

Fig. 3.

Addition of Cdc25 potentiates the ability of nuclear-targeted cyclin B1 proteins to support chromosomal DNA replication. (A) Effects of Cdc25B (3 μM) (solid circles) on activation of Cdk1 by cyclin E–B1 in cyclin E–depleted egg extracts. (B) Effect of Cdc25B (3 μM) on the ability of cyclin E–B1 to support DNA replication in cyclin E–depleted extracts. (C) Effect of Cdc25B addition on the ability of cyclin B1ΔN120 to restore DNA replication in cyclin E–depleted egg extracts. (D) Effects of Cdc25B addition on the ability of cyclin B1ΔN82 to restore DNA replication in cyclin E–depleted egg extracts. Cyclin B1ΔN82 retains the NES of cyclin B1 and does not accumulate in nuclei (13, 16).

These data show that Cdk1–cyclin B1 can trigger DNA replication if a suitable level of activity is targeted to the nucleus. Cdk2–cyclin E, in contrast, appears unable to provoke entry into mitosis, even when present in large amounts (2, 14, 25). This pattern is consistent with what has previously been observed in fission and budding yeast, in which the major mitotic B-type cyclins are, under some circumstances, able to compensate for the absence of their S-phase–promoting counterparts (35, 7), whereas the converse is not true: Overexpression of Sphase cyclins (Clb5/6 in budding yeast or Cig2 in fission yeast) cannot replace the major mitotic cyclins (2628). A variety of mechanisms normally prevent Cdk1–cyclin B1 from supporting S-phase events. Apart from transcriptional control and the APC/C-mediated degradation of cyclin B during G1 phase, nuclear export signals in vertebrate cyclin B1 keep it out of the nucleus, and Wee1/Myt1-mediated inhibitory phosphorylation of Cdk1 maintains a low level of kinase activity until S phase is complete. These controls presumably evolved to increase the fidelity with which the genome is replicated and segregated to daughter cells. Avoiding premature entry into mitosis is vital, especially if initiation of M phase prevents completion of S phase, as seems to be the case in vertebrate cells (29). The nuclear export–imposed block on Cdk1–cyclin B1 supporting replication extends the range of situations in which the localization of Cdkcyclins has been shown to influence their function (7). Localization determinants on the budding yeast G1 cyclins Cln2 and Cln3 contribute to their different functions (30), as do the respective microtubule- and Golgi-targeting N termini of human cyclins B1 and B2 (31). Although differences in substrate specificity between different Cdk-cyclin complexes undoubtedly do exist (7), our results suggest that access to substrates in time and space plays a critical role in determining the function of particular Cdk-cyclin complexes.

Supporting Online Material

Materials and Methods


References and Notes

Stay Connected to Science

Navigate This Article