Report

A Centrosomal Localization Signal in Cyclin E Required for Cdk2-Independent S Phase Entry

See allHide authors and affiliations

Science  29 Oct 2004:
Vol. 306, Issue 5697, pp. 885-888
DOI: 10.1126/science.1103544

Abstract

Excess cyclin E–Cdk2 accelerates entry into S phase of the cell cycle and promotes polyploidy, which may contribute to genomic instability in cancer cells. We identified 20 amino acids in cyclin E as a centrosomal localization signal (CLS) essential for both centrosomal targeting and promoting DNA synthesis. Expressed wild-type, but not mutant, CLS peptides localized on the centrosome, prevented endogenous cyclin E and cyclin A from localizing to the centrosome, and inhibited DNA synthesis. Ectopic cyclin E localized to the centrosome and accelerated S phase entry even with mutations that abolish Cdk2 binding, but not with a mutation in the CLS. These results suggest that cyclin E has a modular centrosomal-targeting domain essential for promoting S phase entry in a Cdk2-independent manner.

The cyclin E–Cdk2 complex has long been considered to have roles essential for the G1 to S phase transition of the cell cycle (1). Expression of dominant-negative forms of Cdk2 inhibits DNA synthesis, as does microinjection of neutralizing antibodies to Cdk2 or cyclin E (24). Inhibition of centrosome duplication by Cdk inhibitors is reversed by Cdk2 complexed with either cyclin E or cyclin A (57). However, mice lacking Cdk2 are viable (8, 9), and mice lacking both cyclins E1 and E2 have no defects in embryonic mitotic cell cycles (10). The double cyclin E knockout mouse, however, dies during embryogenesis from failure of trophoblasts in the placenta to endoreplicate and become polyploid. Cells from double cyclin E knockout mice are also resistant to oncogene-mediated transformation (10), which is often associated with polyploidy. As Cdk2-knockout mice have no defects in forming placentas (8, 9), the function of cyclin E that is essential for polyploidy may be independent of Cdk2. Although expression of either wild-type cyclin E (7) or a nondegradable mutant (11) did not cause over-duplication of centrosomes, a nondegradable cyclin E mutant promoted extra rounds of DNA synthesis and polyploidy in mammalian tissue culture cells (11).

Nuclear accumulation of cyclin E occurs after passage through the restriction (R) point during the G1 to S phase transition (12). The R point is defined as the point at which 50% of the cells in the population will enter S phase even after serum is withdrawn. To examine localization of cyclin E from G1 through S phase, we first established that the R point in Chinese hamster ovary (CHO) cells occurred 6 hours after serum-stimulation (Fig. 1A). Cells were fixed with methanol, which is commonly used for staining centrosomal components, then immunostained with antibodies to cyclin E and γ-tubulin, a well-characterized centrosomal component (13). Cyclin E was localized to the centrosome from G0 phase (14) up to 4 hours after serum addition (Fig. 1B). After passage through the R point, cyclin E accumulated in the nucleus as reported previously (12), and most cells still showed centrosomal localization of cyclin E (Fig. 1B). We verified biochemically the localization of cyclin E on centrosomes purified by sucrose gradient centrifugation (13). Immunoblotting for the centrosomal components centrin and γ-tubulin showed that centrosomes were contained primarily in fractions 3 through 6 (Fig. 1C). Cyclin E was detected in the same fractions by immunoblotting (Fig. 1C); other immunoreactive bands may represent degraded or posttranslationally modified forms of cyclin E. Full-length cyclin E (52 kD) was seen in fraction 4, where centrin was most abundant, indicating that cyclin E likely localizes to the centriole rather than to the pericentriolar material. After nocodazole treatment to depolymerize microtubules, cyclin E still localized to the centrosome (14). Immunostaining for γ-tubulin and the Myc epitope (13) confirmed that ectopic Myc-tagged rat cyclin E was also targeted to the centrosome (Fig. 1D). Therefore, these data indicate that cyclin E localizes to the centrosome from G1 through S phase in mammalian cells.

Fig. 1.

Centrosomal localization of cyclin E in CHO cells. (A) Determination of the R point in CHO cells. Cells were serum-starved for 48 hours and then incubated in medium containing 10% fetal bovine serum. To monitor S phase, cells at each time point were treated with BrdU for 20 min and stained with antibody to BrdU (squares). Serum was removed at each indicated time point, and cells were treated again with BrdU 12 hours after initial serum addition (triangles). The vertical red line at 6 hours indicates the R point. (B) Centrosomal localization of cyclin E before and after the R point. Cells at the indicated times were methanol-fixed and stained with antibodies to cyclin E and γ-tubulin. Arrowheads mark the centrosome. Scale bar, 10 μm. γ-tub, γ-tubulin; DAPI, 4′,6-diamidino-2-phenylindole. (C) Immunoblotting of centrosome fractions and total lysates with antibodies to cyclin E, γ-tubulin (γTb), and centrin. Lysates from asynchronous cells were fractioned on a 40 to 70% sucrose gradient, and fractions were collected from the bottom of the tube. (D) Centrosomal localization of Myc-tagged rat cyclin E. Asynchronous cells were stained with antibodies to Myc and γ-tubulin 20 hours after transfection. Arrowheads mark the centrosome. Scale bar, 10 μm.

To identify which domain of cyclin E targets it to the centrosome, we transfected truncated mutants of Myc-tagged cyclin E and immunostained the cells with antibodies to Myc and γ-tubulin. The N-terminal truncation mutant, Myc–cyclin E(231–396), but not Myc–cyclin E(238–396), was detected on the centrosome (Fig. 2A). Moreover, expression of Myc–cyclin E(231–396), but not Myc–cyclin E(238–396), inhibited DNA synthesis, as detected with an antibody to bromodeoxyuridine (BrdU) after formaldehyde fixation, which is not suitable for centrosomal staining (13) (Fig. 2B). Detailed analysis of truncation mutants revealed that the sequence 231 to 250 in cyclin E is a centrosomal localization signal, which we designate as the CLS (Fig. 2C). Truncation of the CLS in Myc-tagged cyclin E mutants resulted not only in failure to localize to the centrosome but also in reduced capacity to inhibit DNA synthesis (Fig. 2C). This suggests that the CLS is not only essential for centrosomal targeting of cyclin E but also has critical effects on S phase entry.

Fig. 2.

Identification of a centrosomal localization signal (CLS) in cyclin E. (A) Localization of cyclin E. Cells were stained with antibodies to Myc and γ-tubulin 20 hours after transfection of Myc–cyclin E(231–356) or Myc–cyclin E(238–396). Arrowheads mark the centrosome. Scale bar, 10 μm. (B) Effects on DNA synthesis. Twenty hours after transfection of Myc–cyclin E(231–396) or Myc–cyclin E(238–396), cells were treated with BrdU, formaldehyde-fixed, and stained with antibodies to Myc and BrdU. Scale bar, 10 μm. (C) Deletion analysis of Myc–cyclin E mutants. The centrosomal localization of the indicated mutants was observed by double immunostaining for Myc and γ-tubulin as in (A). –, +, and ++ represent 0%, less than 50%, and more than 90% of cells showing centrosomal localization, respectively. These data identified the region 231 to 250 as the CLS (shown in red). Cells in S phase (the bar graph) were detected by staining for incorporated BrdU 20 hours after transfection as in (B). We counted 200 cells for each mutant. Control transfection of GFP-tagged Myc (center) had no effect on S-phase entry compared to untransfected cells (not shown). The data are shown as the mean ± SEM of three independent experiments.

Alignment of the CLS sequence in various mammalian species is shown in Fig. 3A; ∼50% of the residues are identical and 70% are similar. Staining with antibodies to green fluorescent protein (GFP) and γ-tubulin (13) showed that the GFP-tagged CLS sequence alone localized to the centrosome (Fig. 3B). When several conserved residues in the CLS were mutated to alanine (S234A, W235A, N237A, and Q241A, producing a quadruple GFP-CLS mutant designated SWNQ-A), the peptide did not localize to the centrosome (Fig. 3, A and B). Similar to the effects of the truncated cyclin E mutants (Fig. 2C), expression of a wild-type, but not mutant, GFP-CLS reduced the fraction of cells undergoing DNA synthesis from 34% to less than 10% (Fig. 3C). In cells from the double cyclin E knockout mouse, the loading of the minichromosome maintenance protein MCM2 onto chromatin is deficient during the G0 → G1 transition (10). Staining with an antibody to MCM2 (15) demonstrated that the expressed CLS did not interfere with loading of MCM2 onto chromatin (fig. S1), suggesting that the reduced DNA synthesis of GFP-CLS-expressing cells reflects inhibition in G1 of the onset of S phase. Considering that the CLS itself can be targeted to the centrosome (Fig. 3B, left), it seemed possible that endogenous cyclin E was prevented from localizing to the centrosome by competition with the expressed CLS. We verified that centrosomal localization of cyclin E was lost in cells expressing a wild-type, but not a mutant, CLS (Fig. 3D). This suggests that the CLS is sufficient for targeting cyclin E to the centrosome and that such targeting is required for stimulation of DNA synthesis by cyclin E. However, it cannot be excluded that CLS expression also affects centrosomal localization of other proteins, because continuously cycling cells from mice lacking cyclin E have no defect in DNA synthesis (10). In Xenopus egg extracts, it is well established that DNA synthesis can be promoted by either cyclin E or cyclin A (16). Indeed, cyclin A has a putative CLS and is unable to bind the centrosome when the cyclin E CLS is expressed (fig. S1).

Fig. 3.

The CLS inhibits S phase entry and prevents endogenous cyclin E from localizing to the centrosome. (A) The CLS is conserved. The CLS in rat cyclin E is aligned with cyclin E from other mammalian species. Identical residues are shown in the red rectangles. Red letters indicate mutations in SWNQ-A. (B) GFP-CLS localizes to the centrosome. Cells were methanol-fixed and stained with antibodies to GFP and to γ-tubulin 20 hours after transfection with GFP-CLS or GFP-CLS(SWNQ-A). Arrowheads mark the centrosome. Scale bar, 10 μm. (C) Expression of GFP-CLS inhibits DNA synthesis. Twenty hours after transfection with GFP, GFP-CLS, or GFP-CLS(SWNQ-A), cells were treated with BrdU and stained with antibodies to GFP and BrdU. The fraction of GFP-positive cells in S phase was determined as in Fig. 2C. Scale bar, 10 μm. (D) Expression of GFP-CLS prevents endogenous cyclin E (CycE) from localizing to the centrosome. Cells were stained with antibodies to GFP and cyclin E 20 hours after transfection of GFP-CLS or GFP-CLS (SWNQ-A). Arrowheads mark the centrosome. Scale bar, 10 μm.

We determined whether cyclin E with a mutant CLS (SWNQ-A) binds Cdk2 and has associated histone H1 kinase activity. Myc-tagged cyclin E was immunoprecipitated, and association with Cdk2 was assessed by immunoblotting and immune-complex kinase assays (13). Both wild-type cyclin E and cyclin E with the mutant SWNQ-A CLS bound Cdk2 and had associated histone H1 kinase activity (Fig. 4A). On the other hand, several cyclin E mutants—cyclin E(R131A) (17, 18), cyclin E(1–342), and cyclin E(S180D)—did not bind Cdk2 and had no associated histone H1 kinase activity (Fig. 4A), and the S180D cyclin E mutant still localized to the centrosome in a CLS-dependent manner (Fig. 4B). This suggests that the centrosomal localization of cyclin E depends on the CLS but not on its binding to Cdk2. It is well established that overexpression of cyclin E accelerates entry of cells into S phase (19). This acceleration provides a means to test CLS function on S phase promotion by cyclin E. Both wild-type cyclin E and mutants unable to bind Cdk2 (1–342 and S180D) increased the fraction of cells in S phase, but only if they had an intact CLS (Fig. 4, C and D) (14). This indicates that the CLS, but not Cdk2 binding, is required for cyclin E itself to accelerate entry into S phase.

Fig. 4.

The CLS, but not Cdk2 binding, is required for promotion of DNA synthesis. (A) Cdk2 binding and histone H1 kinase activity. Lysates from cells transfected with Myc-tagged wild-type cyclin E (WT) or with Myc-tagged cyclin E mutants (SWNQ-A, R131A, 1–342, and S180D) were immunoprecipitated (IP) with antibodies to Myc and analyzed for binding to Cdk2 and for associated histone H1 kinase activity (13). (B) Centrosomal localization. Cells were stained with antibodies to Myc and γ-tubulin 20 hours after transfection of Myc–cyclin E(S180D), Myc–cyclin E(SWNQ-A), or Myc–cyclin E(S180D/SWNQ-A). Arrowheads mark the centrosome. Scale bar, 10 μm. (C) Accelerated S phase entry by expression of cyclin E. Either Myc-tagged wild-type cyclin E (WT), a Myc-tagged cyclin E mutant unable to bind Cdk2 (S180D), or CLS-deficient cyclin E mutants (SWNQ-A and S180D/SWNQ-A) were transfected into CHO cells. Twenty hours after transfection, cells were treated with BrdU and stained with antibodies to GFP and BrdU. Scale bar, 10 μm. (D) Quantification of the fraction of GFP-positive cells in S phase in (C) was carried out as described for Figs. 2C and 3C.

Ablation of centrosomes in late G2 phase leads to an arrest at the G1-S boundary of the next cell cycle (20, 21). This arrest has been suggested to represent either a new checkpoint that monitors centrosome number or a role for centrosomes in the normal process of initiating S phase. The effect of CLS expression on DNA synthesis suggests that the requirement for centrosomes for entry into S phase may reflect the necessity of cyclin E, cyclin A, or other unknown proteins to bind the centrosome through the CLS. Nuclear localization of cyclin E does not seem to be affected by the CLS (Fig. 4C, SWNQ-A), even though it has conserved repeats of hydrophobic residues typical of nuclear export signals (Fig. 3A). Neither leptomycin B nor mutation of all the hydrophobic residues in the CLS to alanine or acidic amino acids affected centrosomal or extranuclear localization of cyclin E (14). Staining after formaldehyde fixation showed that both CLS-deficient cyclin E(SWNQ-A) and cyclin E(231–396), which is unable to bind Cdk2, accumulate in the nucleus (Fig. 4C, right, and Fig. 2B, left). Thus, neither the CLS nor Cdk2 binding is required for nuclear localization of cyclin E. Moreover, cyclin E(S180D) promoted DNA synthesis (Fig. 4D), even though it was cytoplasmic (Fig. 4C, S180D), suggesting that nuclear localization of cyclin E is not required for accelerating entry into S phase.

We have shown that the CLS in cyclin E is required for acceleration of S phase entry by a mechanism that does not require binding to Cdk2. Previous studies, including those on cyclin E knockout and Cdk2 knockout mice, have indicated that cyclin E has Cdk2-independent roles essential for polyploidy in trophoblasts and for oncogenic transformation (810, 18, 22). Excess cyclin E protein is often detected in tumor cells exhibiting polyploidy, which can be induced by constitutive expression of cyclin E (11); moreover, it has positive correlations with metastasis and low survival rates in patients (23). Thus, the CLS in cyclin E may be essential for transformation through promotion of S phase entry independently of Cdk2.

Supporting Online Material

www.sciencemag.org/cgi/content/full/306/5697/885/DC1

Materials and Methods

Fig. S1

References and Notes

References and Notes

View Abstract

Navigate This Article