Delayed cyclin A and B1 degradation in non ... - Semantic Scholar

36 downloads 0 Views 996KB Size Report
Entry into mitosis throughout the eukaryotic Phyla is con- trolled by the serine-threonine protein kinase p34cdc2-cyclin. B, a component of the maturation ...
2599

Journal of Cell Science 108, 2599-2608 (1995) Printed in Great Britain © The Company of Biologists Limited 1995

Delayed cyclin A and B1 degradation in non-transformed mammalian cells Franck Girard*, Anne Fernandez and Ned Lamb† Cell Biology Unit, Centre de Recherche de Biochimie Macromoléculaire, Centre National de la Recherche Scientifique, 1919 Route de Mende, BP5051, 34033 Montpellier cedex, France *Present address: Cell Biology, Biozentrum, Klingelbergstrasse 70, CH-4056 Basel, Switzerland †Author for correspondence

SUMMARY Cyclins A and B are known to exhibit significant differences in their function, cellular distribution and timing of degradation at mitosis. On the basis of observations in marine invertebrates and Xenopus, it was proposed that cyclin destruction triggers cdc2 kinase inactivation and anaphase onset. However, this model has recently been questioned, both in Xenopus and in budding yeast. In this report, we present evidence for delayed degradation of both cyclins A and B1 in non-transformed mammalian cells. Indeed, by means of indirect immunofluorescence and confocal microscopy, we show that cyclins A and B1 are present up to anaphase in REF52, Hs68,

INTRODUCTION Entry into mitosis throughout the eukaryotic Phyla is controlled by the serine-threonine protein kinase p34cdc2-cyclin B, a component of the maturation promoting factor or M-phase promoting factor (MPF) (reviewed by Nurse 1990). Apart from cyclin B binding, cdc2 kinase activation requires the dephosphorylation of Thr14 and Tyr15 residues, catalysed by the cdc25 protein tyrosine phosphatase (Dunphy and Kumagai, 1991; Gautier et al., 1991; Millar et al., 1991; Strausfeld et al., 1991), and the phosphorylation of Thr161 by CAK (Cdk activating kinase), one component of which is homologous to the Xenopus MO15 gene product (Fesquet et al., 1993; Poon et al., 1993; Solomon et al., 1993). Exit from mitosis requires cdc2 kinase inactivation, which is achieved in part by cyclin degradation (Murray et al., 1989). p34cdc2 is also found associated with cyclin A, a protein required for both DNA replication (Girard et al., 1991; Pagano et al., 1992) and mitosis (Pagano et al., 1992). Degradation of both cyclin A and B is controlled by a common mechanism, involving ubiquitin-mediated proteolysis and a short N-terminal sequence on cyclins known as the destruction box (Glotzer et al., 1991). Truncated cyclin B, lacking the 90 N-terminal residues including the destruction box, failed to be degraded (Glotzer et al., 1991; Lorca et al., 1991; Luca et al., 1991), and when introduced into frog egg extracts or HeLa cells prevented MPF inactivation (Murray et al., 1989; Gallant and Nigg 1992). In mammalian cells, cyclins A and B show different subcellular localizations. Cyclin A is

human primary fibroblasts and NRK epithelial cells. In marked contrast, cyclin A is shown to be degraded within metaphase and cyclin B just at the transition to anaphase in HeLa and two transformed cell lines, derivatives of normal NRK and REF52. These results further support the notion that cyclin destruction might be not correlated with anaphase onset in normal cells and highlight a significant difference in the fate of mitotic cyclins between transformed and non-transformed cells. Key words: cyclin A, cyclin B1, mitosis, degradation

found exclusively in the nuclear compartment from G1/S transition onwards (Girard et al., 1991; Pines and Hunter, 1991), whilst cyclin B accumulates in the cytoplasm during interphase and is redistributed into the nucleus at prophase before nuclear envelope breakdown (Pines and Hunter, 1991; Bailly et al., 1992). In HeLa cells, cyclin A is reported to be degraded at metaphase, and cyclin B later at the metaphase-anaphase transition (Pines and Hunter, 1991; Pagano et al., 1992). These kinetics of cyclin degradation are also observed during early embryonic cleavage divisions in Xenopus and marine invertebrates (Minshull et al., 1989, 1990; Murray et al., 1989). In Drosophila, both cyclins remain incompletely degraded up to telophase in the early syncytial embryo (Maldonado-Codina and Glover, 1992), which is in contrast to their complete cyclical degradation later in development (at the time of cellularization) (Lehner and O’Farrell, 1989, 1990; Whitfield et al., 1990). It was shown more recently that the cyclin levels remain constant in the first 8 syncytial cycles in Drosophila (Edgar et al., 1994). These observations have led to a model in which cyclin degradation is thought to trigger MPF inactivation, and as a consequence anaphase onset and exit from mitosis (Murray et al., 1989; Glotzer et al., 1991). However, the universality of this model has been questioned. Indeed, Holloway et al. (1993), using Xenopus egg extracts that can assemble a functional mitotic spindle, have shown that sister chromatid separation can be uncoupled from p34cdc2 kinase inactivation. Addition of a non-degradable form of cyclin B to these extracts, whilst inducing high p34cdc2 kinase activity, does not inhibit anaphase onset. Furthermore, addition of an

2600 F. Girard, A. Fernandez and N. Lamb N-terminal fragment of cyclin B containing the destruction box delays anaphase onset in a dose-dependent manner, probably by competing with endogenous cyclin for the ubiquitin proteolysis machinery. Together, these results clearly demonstrate that cyclin degradation and MPF inactivation are not prerequisites for anaphase onset, and suggest that ubiquitin-mediated proteolysis of non-cyclin protein(s) might be required for proper sister chromatid separation. This was also the conclusion drawn by Surana et al. (1993), who showed that in budding yeast anaphase can occur in the presence of high levels of CDC28/CLB kinase in cdc15 mutants and also in the presence of non-degradable CLB2. In this report, we present evidence suggesting that, in nontransformed mammalian cells, cyclin degradation is not correlated with the onset of anaphase. By comparing the intracellular distribution of both cyclins A and B in several mammalian cell lines, we show that normal cells are characterized by cyclin degradation occurring within anaphase. In contrast, HeLa cells and two transformed derivatives of normal NRK and REF52 cells are shown to display patterns of cyclin degradation similar to those previously described, with cyclin A being degraded during metaphase and cyclin B later, at the metaphase to anaphase transition. MATERIALS AND METHODS Cells Human foreskin fibroblasts Hs68 (ATCC CRL1635), rat embryo fibroblasts REF52 (McClure et al., 1982), HeLa cells, WT6 (SV40transformed REF52 cells) (McClure et al 1982), normal rat kidney cells NRK49F (ATCC CRL1570), KNRK (Kirsten Ras-transformed NRK cells) (ATCC CRL1569) and human primary fibroblasts were cultured at 37°C in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 6-10% fetal calf serum (FCS) (Gibco), 1 mM glutamine (Imperial Laboratories) and antibiotics, on plastic dishes or acid-washed glass coverslips as described (Girard et al., 1991). For nocodazole treatment, Hs68 and HeLa cells were cultured in T25 flask (growing surface of 25 cm2) at 37°C. Hs68 were made quiescent by 30 hours of serum deprivation, and restimulated by serum refeeding for 20 hours. HeLa cells were synchronized at the G1/S boundary by treatment with 0.5 mM hydroxyurea (Sigma). Serum-synchronized Hs68 and hydroxyurea-treated HeLa cells were refed and treated for 12 hours with 0.5 µM nocodazole (Sigma). Mitotic-blocked cells were recovered by shaking, washed in PBS and immediately lysed in SDS sample buffer. Antibodies Affinity-purified anti-human cyclin A antibody has been described previously (Girard et al., 1991). Human cyclin B1 cDNA (Pines and Hunter 1990) was inserted in peT3d expression vector at the NcoI/BamHI restriction sites. Cyclin B1 protein was overexpressed in Escherichia coli and purified from inclusion bodies following the protocol already described for cyclin A. New Zealand rabbits were injected with 0.5 mg of purified cyclin B1 in complete Freund’s adjuvant. Subsequently, they were injected every two weeks with the same protein preparation in incomplete Freund’s adjuvant. The cyclin B immunoreactivity of the resulting antisera was analysed by western blotting, ELISA and indirect immunofluorescence. Polyclonal antisera were affinity-purified using cyclin B1 protein coupled to CNBr-activated Sepharose, exactly as described for cyclin A antibodies (Girard et al., 1991). Western blot Protein extracts from nocodazole-blocked HS68 and HeLa cells were

probed for the presence of cyclins A and B1 by western blotting as described (Girard et al., 1991). Immunofluorescence and confocal microscopy Indirect immunofluorescence experiments were performed with asynchronous or serum-synchronized cells, growing on acid-washed glass coverslips or plastic dishes. After a brief wash in PBS, cells were fixed in 3.7% formalin in PBS for 5 minutes, and subsequently extracted in −20°C acetone for 1 minute. Coverslips were incubated in PBS-BSA (1%) for 5-10 minutes, then reacted with primary antibodies: polyclonal anti-cyclin A or anti-cyclin B1, mixed with monoclonal antitubulin DMA1A (Blose et al., 1984) in PBS-BSA for 60 minutes at 37°C. After washing in PBS, biotinylated goat anti-rabbit IgG (Amersham) was incubated for an additional 60 minutes. Thereafter, cells were washed in PBS, and the final incubation was done in fluorescein-conjugated goat anti-mouse IgG (Cappel Organon Teknika) mixed with Streptavidin-Texas Red (Amersham) for 30 minutes. Coverslips were washed and mounted in 15% (w/v) Airvol 205, 33% glycerol in PBS. Hoechst dye (Sigma) was added for 1 minute in the final wash step before mounting. Images were acquired using a Leica confocal laser scanning microscope, equipped with a krypton-argon dual wavelength laser. Images were scanned at 512×512 pixel resolution with 12 bits per pixel (2048 levels of grey/pixel). DNA staining was visualized with the normal fluorescence light source using a CCD low light sensitivity camera. All images were collated on the VME bus of the CLSM before transfer and deconvolution on a Silicon Graphics Iris Indigo workstation. Images were scanned on the CLSM using the same level of recursive scanning (each line of an image was successively scanned 8 times and an average line produced). Images were processed using Image Glide software (S. Guihot S. and N. Lamb, unpublished data), and figures were produced using Iris Showcase 3.2. Final figures were printed on a Kodak Colorease thermal sublimation printer. For quantitative analysis, optical sections were obtained for all the images in a series for comparison at the same settings for microscope pinhole, baseline etc. Pixel mass was calculated using Imgcalc (Ned Lamb, unpublished) on image data files which had not been subject to processing.

RESULTS Both cyclins A and B1 accumulate in nocodazoleblocked cells It has been previously reported that in HeLa cells treated with the microtubule-destabilizing drug nocodazole (Pagano et al., 1992), Xenopus or clam embryo extracts treated with colchicine (Minshull et al., 1990), or Drosophila embryos treated with either nocodazole and colchicine (Whitfield et al., 1990), cyclin A is degraded whereas cyclin B accumulates at high levels. In order to compare these patterns of cyclin degradation with those occurring in non-transformed mammalian fibroblasts, we have analysed the levels of cyclins A and B proteins in nocodazole-blocked HeLa or human fibroblasts HS68. Cells were synchronized at the G1/S boundary by treatment with hydroxyurea and, after release from hydroxyurea block by several washes, incubated for 12 hours with nocodazole. Nocodazole-blocked cells are in pseudometaphase state, and are easily recoverable by shaking. The efficiency of the drug treatment was followed by staining the DNA with Hoechst and by immunofluorescent tubulin staining (data not shown). It is important to note that the nocodazoletreated cells show a complete disassembly of their microtubule network and no formation of a mitotic spindle. Therefore, the

Mitotic cyclin degradation 2601 Fig. 1. Accumulation of both cyclins A and B in nocodazole-blocked Hs68 and HeLa cells. Nocodazole-blocked Hs68 (lanes 1 and 3) and HeLa cells (lanes 2 and 4) were run in 12.5% SDS-PAGE, transferred onto nitrocellulose and probed for the presence of cyclin A (lanes 1 and 2) or cyclin B1 (lanes 3 and 4) as described in Materials and Methods. Molecular masses are given in kilodaltons.

stage at which the cells are recovered from such a nocodazole block would be more adequately described as ‘pseudo- or premetaphase’ (i.e. prometaphase), rather than true metaphase. Fig. 1 shows a western blot of these nocodazole-blocked cells, probed with affinity-purified anti-cyclin A (lanes 1 and 2) or anti-cyclin B1 antibodies (lanes 3 and 4). Both p58 cyclin A and p62 cyclin B1 are present in nocodazole-blocked HS68 fibroblasts (lanes 1 and 3) and HeLa cells (lanes 2 and 4), with levels comparable to that seen in late G2-phase cells (not shown). These data are in contrast to data from a previous report of Pagano et al. (1992), and show that there is no difference in the persistence of cyclin A and B in nocodazoleblocked cells between transformed HeLa cells and normal fibroblasts.

Both cyclins A and B1 are present at anaphase in non-transformed mammalian fibroblasts We have previously shown that cyclin A synthesis starts at the G1/S transition, and that the protein is localized exclusively in the nucleus during interphase (Girard et al., 1991). In contrast, cyclin B1, the synthesis of which starts at the end of S-phase, accumulates in the cytoplasmic compartment in interphase (Pines and Hunter 1991; Bailly et al., 1992). Since we found that both cyclins accumulate in nocodazole-blocked HS68 and HeLa cells, we have more carefully examined, by indirect immunofluorescence and confocal microscopy, the cellular localization of these proteins during mitosis, in order to determine the respective timing of their degradation. Immunofluorescence experiments consisted of double staining for tubulin (to monitor the stage of mitosis) and either cyclin A or B1. Different cell lines were used: human foreskin fibroblasts (HS68), rat embryo fibroblasts (REF52) and human primary fibroblasts (see Materials and Methods). Since we found similar cyclin A staining patterns whatever fibroblastic cells were used, only images from HS68 cells are shown (Fig. 2). Mitotic stage was assessed by both tubulin immunostaining (see legend to Fig. 2) and Hoechst DNA staining (not shown). Cyclin A is found in the nucleus in early prophase, before nuclear envelope breakdown has occurred (Fig. 2A,F). Careful examination of cyclin A immunostaining in parallel with DNA staining suggests that most, if not all, cyclin A proteins are excluded from the condensed chromatin at this stage (Fig. 3A-

Fig. 2. Cellular distribution of cyclin A in mitotic Hs68 fibroblasts. Human foreskin Hs68 fibroblasts were fixed in formalin/acetone, and immunostained for both cyclin A (A-E) and tubulin (F-J), with affinity-purified anti-human cyclin A and monoclonal anti-tubulin DMA1A, respectively. (A,F) Early prophase; (B,G) metaphase; (C,H) anaphase A; (D,I) anaphase B; (E,J) telophase and surrounding G2-phase cells. Bar, 5 µm.

2602 F. Girard, A. Fernandez and N. Lamb

cyclin A

DNA

overlap

Fig. 3. Cyclin A and chromatin are not co-localized in prophase cells. HS68 fibroblasts were grown for 24 hours in the presence of 5bromodeoxyuridine (BrdU) before fixation and double staining for the incorporated BrdU and cyclin A as described (Girard et al., 1991). Shown are optical sections seen on a confocal microscope of the chromatin staining in the FITC wavelength (521 nm green) (B), the staining for cyclin A obtained with Texas Red (A) and the two images superimposed (C). Arrowheads show regions where mutual exclusion of the two stainings is very clear. The optical sections are approximately 280 nm thick and the relative discordance in the Z axis between the optical plane in red and green wavelengths has been corrected both at the level of the objective and during the image deconvolution process.

C). Fig. 3C showing the overlapping of cyclin A and DNA staining, reveals clear areas of complementarity in the distribution of the two stains (arrowed), and only very limited overlap (yellow pixels). In this figure, to allow for confocal scanning of the DNA staining (scanning in the UV/blue wavelength of the common DNA dyes, DAPI and Hoechst, is not possible with an argon/krypton laser and there would also be considerable discontinuity in the Z axis between sections obtained at 400 and 568 nm), we used 24 hour prelabelling of the DNA in vivo, with the thymidine analogue, 5-bromodeoxyuridine (BrdU), which was subsequently stained using a monoclonal anti-BrdU antibody (as described by Girard et al., 1991). This procedure allows high intensity staining of the chromatin and gives better quality images than the currently available green or red DNA fluo dyes. Cyclin A immunoreactivity persists at high levels in metaphase cells (Fig. 2B,G) and is clearly absent from the condensed chromosomes. We found higher immunoreactivity in a zone surrounding the metaphase plate, suggesting a transient association of cyclin A with the mitotic spindle at this stage. Early anaphase cells (Fig. 2C,H) are characterized by only a slight decrease in cyclin A staining, which persists at significant levels (compare with late anaphase and telophase cells in Fig. 2D and E). This persistent staining has been detected in all early anaphase cells observed (anaphase A, when the sister chromatids separate). Finally, cyclin A immunoreactivity is abruptly lost late in anaphase (anaphase B, when the cells move slightly apart, Fig. 2D,I), and no longer detectable at telophase (Fig. 2E,J). The quantification analysis of the fluorescent staining made from these confocal scans is represented in Fig. 4B. As mentioned in Materials and Methods, quantitative analysis was performed on unprocessed images obtained in the same series of scans, with the same microscope settings for each section. Cyclin B1 distribution in mitotic human primary fibroblasts is shown in Fig. 4. The anti-cyclin B1 antibodies (see Materials and Methods) we used in the present study gave the same

results as already published, and do not cross-react with other cyclins, as judged by two-dimensional immunoblotting and ELISA (not shown). We have found that this antibody reacts in immunofluorescence experiments only against human cyclin B1 and does not cross-react with the rat protein. Fibroblasts in late G2-phase are characterized by a cytoplasmic accumulation of cyclin B1 as previously described (Pines and Hunter, 1991). As cells enters mitosis, most of the cyclin B1 immunoreactivity is found redistributed into the nucleus (Fig. 4A and F shows an early prophase cell, before nuclear envelope breakdown). At metaphase, cyclin B1 staining spreads out throughout the cell, with a major fraction associating with the mitotic spindle (Fig. 4B and G). Cyclin B1 staining is still clearly present in anaphase A cells (Fig. 4C and H); anaphase B cells are characterized by a dramatic decrease in cyclin staining (Fig. 4D and I), which falls to nil in telophase cells (Fig. 4E and J). The extent of cyclin B1 degradation at these different stages of mitosis has been quantified from the confocal scans shown in Fig. 4, and represented in Fig. 5. This confirms the precise timing of cyclin B degradation as occurring during the transition from early to late anaphase and at the beginning of telophase. HeLa cells display differences in timing of cyclin degradation Since we found major differences in the kinetics of cyclin A and B degradation in fibroblasts, when compared with previous reports on HeLa cells (Pines and Hunter, 1990, 1991; Pagano et al., 1992), we performed immunofluorescence experiments in HeLa cells, with the same anti-cyclin A and anti-cyclin B1 antibodies as used in the experiments described above. Figs 6 and 7 illustrate, respectively, cyclin A and cyclin B1 immunostaining during the different stages of mitosis (see legend to Figs 6 and 7). We have found no differences between HeLa and fibroblasts in the pattern of cyclin staining during interphase and early prophase (Fig. 6A-B and Fig. 7C-D show early

Mitotic cyclin degradation 2603

Fig. 4. Cellular distribution of cyclin B1 in mitotic human fibroblasts. Human primary fibroblasts (see Materials and Methods) were fixed in formalin/acetone, and immunostained for both cyclin B1 (A,B,C,D,E) and monoclonal anti-tubulin DMAlA (F-J), respectively. Bar, 5 µm. (A,F) Prophase; (B,G) metaphase; (C,H) anaphase early; (D,I) anaphase late; (E,J) telophase.

2000

averaged pixel mass

cyclin B cyclin A

1000

0 meta

ana early

ana late

stages of mitosis

telo

prophase cells stained for cyclin A and cyclin B1, respectively). Prophase cells are characterized by an accumulation of cyclin B1 at the centrosomes (Fig. 7C), as previously reported (Pines and Hunter 1991; Bailly et al., 1992). However, in contrast with the results from fibroblasts presented above, metaphase HeLa cells are distinguished by a substantial decrease in cyclin A staining (Fig. 6C-D), whilst cyclin B1 remains visible at high levels (Fig. 7E-F). We have found that approximately 50% of metaphase HeLa cells show clearly no cyclin A staining (Fig. 6E-F), whilst all metaphase cells were positive for cyclin B1. Finally, immunoreactivity of both cyclin A and B1 was absent in anaphase and telophase cells (Fig. 6G-J for cyclin A, and Fig. 7E and A, for cyclin B1). Fig. 5. Pixel quantificaion of immunostaining for cyclin A and B during exit from mitosis in normal fibroblasts. The precise levels of fluorescence inside cells at the different stages of mitosis were determined by calculating the overall fluorescence in the cell and dividing it by the number of pixels (cell surface area). Shown are the values obtained for cyclin B and cyclin A at different times in the mitotic process.

2604 F. Girard, A. Fernandez and N. Lamb

Fig. 6. Cellular distribution of cyclin A in mitotic HeLa cells. Randomly growing HeLa cells were fixed in formalin/acetone, and immunostained for both cyclin A (B,D,F,H,J) and tubulin (A,C,E,G,I), with affinity-purified anti-human cyclin A and monoclonal anti-tubulin DMA1A, respectively. Bars, 5 µm. (AB) Prophase; (C-D) metaphase (arrow) and prophase; (E-F) metaphase (arrow); (G-H) anaphase (arrow); (I-J) telophase cells (arrows).

These observations are fully consistent with previous reports (Pines and Hunter, 1991; Bailly et al., 1992; Pagano et al., 1992), and show that, in HeLa cells, cyclin A degradation takes place during metaphase and slightly precedes cyclin B1 destruction, which occurs at the metaphase to anaphase transition. These kinetics clearly differ from those observed above in non-transformed fibroblasts, in which cyclin degradation was delayed to late anaphase. Cyclin A degradation occurs within metaphase in transformed cells Cyclin A immunostaining was also performed in normal epithelial cells, in order to determine whether delayed cyclin A degradation is a feature specific to normal fibroblasts or is cell-type independent. Experiments were performed using normal rat kidney cells (NRK49F), as described in Materials and Methods, and the results are presented in Fig. 8. Cyclin A accumulates at high levels in metaphase cells (Fig. 8E). Cyclin A immunoreactivity persists at relatively similar levels early in anaphase (Fig. 8F and G). Progression through anaphase is accompanied by a sudden decrease in cyclin A immunostaining (shown in Fig. 8H is an anaphase B cell, in which cyclin A immunoreactivity falls to levels only slightly higher than those observed in telophase or G1-phase cells), and is completely absent at telophase (not shown). We conclude that

normal epithelial cells display the same pattern of cyclin A degradation as normal fibroblasts, with cyclin A being completely degraded only at the end of anaphase. Cyclin B1 immunostaining was not performed in these cells, since our anti-human cyclin B1 antibody does not cross-react with the rat protein. Given that: (i) delayed cyclin degradation appears to be a characteristic of normal cells (of either fibroblast or epithelial type); and (ii) HeLa cells are highly transformed, we have therefore investigated cyclin A degradation in two other transformed cell lines, which are the transformed counterparts of the normal fibroblasts and epithelial cells that we have studied above: SV40-transformed REF52 cells (WT6) and Kirsten Ras-transformed NRK (KNRK). Immunofluorescence experiments were conducted under exactly the same conditions as used for normal cells. We found that most KNRK and WT6 metaphase cells were positive for cyclin A (Fig. 8I and K, respectively), although, as previously described for HeLa, a proportion of metaphase cells (10-20%) showed complete absence of cyclin A immunoreactivity. As cells proceed to anaphase, cyclin A immunoreactivity falls to undetectable levels, both in KNRK (Fig. 8J) and WT6 (Fig. 8L), in agreement with the situation observed in HeLa cells (Fig. 8M,N,O,P shows the Hoechst staining for the DNA corresponding, respectively, to Fig. 8I,J,K and L). We conclude

Mitotic cyclin degradation 2605

Fig. 7. Cellular distribution of cyclin B1 in mitotic HeLa cells. HeLa cells were fixed in formalin/acetone, and immunostained for both cyclin B1 (A,C,E) and tubulin (B,D,F), with affinity-purified polyclonal anti-human cyclin B1 and monoclonal anti-tubulin DMA1A, respectively. Bar, 5 µm. (A-B) Late G2-phase cell and telophase (t, arrow); (C-D) prophase (p) and telophase (t). Arrows in C show the two centrosomes, stained with anti-cyclin B1 antibody; (E-F) prometaphase (pm), metaphase (m) and anaphase (a). Arrow in E-F points to a G1-phase cell, showing an absence of cyclin B1 immunoreactivity. Surrounding cells, in G2-phase, show positive cytoplasmic staining for cyclin B1.

therefore that in both transformed cell lines, and in marked contrast with their normal counterparts, cyclin A degradation occurs within metaphase. DISCUSSION In this report, we present evidence for delayed cyclin degradation in non-transformed mammalian cells. Using immunostaining and confocal microscopy, we show that: (i) cyclins A and B1 are still present at significant levels up to anaphase, in REF52, HS68, human primary fibroblasts and normal rat kidney cells; (ii) HeLa cells and two other transformed cell lines show major differences in cyclin degradation, with cyclin A being totally degraded within metaphase and cyclin Bl later at the transition to anaphase; and (iii) cyclin A appears to be destroyed before cyclin B1, whatever the cell type.

The presence of both cyclins up to anaphase in normal cells, from either fibroblastic or epithelial origin, contrasts with previous reports in HeLa cells (Pines and Hunter 1990; 1991; Bailly et al., l992; Pagano et al., l992), clam or Xenopus embryos (Minshull et al., 1989, 1990) and cellularized Drosophila embryos (Lehner and O’Farrell 1990, 1991; Whitfield et al., 1990). We have demonstrated that this was not an artefact due to our anti-cyclin antibodies by showing that, in our hands, HeLa cells display a pattern of cyclin degradation similar to those previously reported (Pines and Hunter 1991; Pagano et al., 1992). In their report, Pagano et al. noted that, in most metaphase cells, cyclin A staining was undetectable, although a minor fraction (5-10%) showed persistent staining. We have found that approximately 50% of metaphase HeLa cells were positively stained for cyclin A, the remaining 50% showing undetectable staining (see Fig. 6D and F). This is fully consistent with the observations made in Drosophila,

2606 F. Girard, A. Fernandez and N. Lamb

Fig. 8. Comparison of cyclin A immunostaining in normal and transformed cells. Normal rat kidney cells (NRK-49F), Ras-transformed rat kidney cells (KNRK) and SV40-transformed REF52 cells (WT6) were fixed in formalin/acetone, and immunostained for cyclin A, tubulin and DNA (see Materials and Methods). Bars, 5 µm. (A-H) Cyclin A (E,F,G,H) and tubulin (A,B,C,D) immunostaining of mitotic NRK cells, respectively, in metaphase (A,E), early anaphase (B,F and C,G), and late anaphase (D,H). (I-L) cyclin A immunostaining (I,J,K,L) and corresponding DNA staining (M,N,O,P) of mitotic KNRK (I-J) and WT6 (K-L) cells. Arrows point to metaphase cells in I and K, and anaphase cells in J and L.

in which metaphase cells were shown to exhibit differences in cyclin A staining intensities, with cells highly stained, halfstained or unstained (Lehner and O’Farrell 1989). Consistent with our immunofluorescence experiments, we found that both cyclins A and B1 accumulate at high levels in nocodazoleblocked HeLa and HS68 cells (in ‘pre-metaphase’; which is also reported for cyclin A by Bailly et al., 1992; and Fig. 4). We therefore conclude that in HeLa cells, cyclin A destruction takes place in metaphase and clearly before the transition to anaphase.

In both fibroblasts and HeLa cells, we have observed that cyclin B1 is not associated with the condensed chromosomes at metaphase, in contrast to the observations of Pines and Hunter in HeLa cells (1991). These authors also suggested that cyclin A might be transiently associated with condensed chromatin early in prophase. Our confocal analysis comparing cyclin A and DNA stainings suggests, rather, that most, if not all, cyclin A proteins are excluded from the chromatin at this stage (Fig. 3), although we cannot exclude the possibility that a minor fraction might be attached to the chromosomes, and

Mitotic cyclin degradation 2607 be undetectable because of its association with the chromatin. We have no explanation for these differences, but we believe this is most likely due to the use of different anti-cyclin antibodies or subtle variations in the fixation protocols. Confirming our observations, we found that methanol fixation gave essentially the same results as formalin/acetone fixation in mitosis, for both cyclins A and B1 (although a weaker and more punctate staining pattern is obtained in interphase for cyclin A). In agreement with others (Pines and Hunter, 1991; Bailly et al., 1992), we found that cyclin B1 is associated with centrosomes, which is clearly not the case for cyclin A in HeLa and fibroblasts (compare Fig. 6B and Fig. 7C). Cyclin B1 also strongly associates with the mitotic spindle at metaphase (for a detailed analysis and comments on the significance of this association, see Bailly et al., 1992). Interestingly, this accumulation of cyclin B1 at the mitotic spindle is not visible early in anaphase fibroblasts (whilst the rest of the cell is clearly stained, see Fig. 4C). We also observed that a zone surrounding the metaphase plate was highly stained for cyclin A in metaphase HS68. This might correspond to a transient association of cyclin A with the mitotic apparatus, although such an association is not as clear as that for cyclin B1. We have observed that cyclin A degradation occurred slightly before that of cyclin B1, consistent with cyclin B, but not cyclin A, being involved in triggering cyclin degradation (Félix et al., 1990; Luca et al., 1991). This was particularly evident in HeLa cells (see Figs 6 and 7), and we have observed that this must also be the case in normal cells. Indeed, cyclin B1 staining in anaphase A cells always appears to be less reduced in intensity than that of cyclin A (see Fig. 5). Our results clearly show that in normal cells, anaphase is initiated in the presence of significant levels of both cyclins A and B1. This further supports the recent notion that cyclin degradation and cdc2 kinase inactivation might be uncoupled from anaphase onset, as observed in Xenopus (Holloway et al., 1993) and budding yeast (Surana et al., 1993). The effects of overexpressing non-degradable forms of mitotic cyclins are fully consistent with this model. Indeed, overexpressing wildtype cyclin B or a non-degradable form leads to an arrest in telophase in Saccharomyces cerevisiae (Surana et al., l993) and in late anaphase in Drosophila embryos (Rimmington et al., 1994) and Xenopus egg extracts (Holloway et al., 1993). In human HeLa cells, the expression of stable cyclin B2 does not induce a classical metaphase arrest, and does not seem to prevent chromosome separation (Gallant and Nigg, 1992). Altogether, these data suggest that cyclin degradation and cdc2 kinase inactivation are required for exit from mitosis, rather than for initiating anaphase as initially suspected. It was proposed that ubiquitin-dependent proteolysis of proteins distinct from cyclins A and B might be necessary to trigger sister chromatid separation, although the identity of these proteins is still unknown (Holloway et al., 1993). Due to technical reasons, we have not analysed whether the presence of cyclins A and B1 is correlated with the persistence of cdc2 kinase activity up to anaphase in non-transformed cells. This point will need to be addressed, to investigate the role(s) of the cyclin A- and cyclin B-associated cdc2 kinases during anaphase. Since cyclin B appears to be involved in regulating the behaviour of the mitotic spindle in metaphase, it is possible that it exerts a similar function also during early anaphase. The question which remains outstanding from this study is

the significance of the delayed cyclin degradation observed in normal non-transformed cells. Differences in the patterns of cyclin degradation have already been observed in the Drosophila embryo. Indeed, in the first 8 division cycle of the syncytial embryo, both cyclin levels remained constant (Edgar et al., 1994), and were found not to be totally degraded up to telophase in the early syncycial embryo (Maldonado-Codina and Glover, 1992). In contrast, later in development at the time of cellularization, the pattern of cyclin degradation appears identical to that observed in HeLa cells (Lehner and O’Farrell, 1989, 1990; Whitfield et al., 1990), which suggests that cyclin degradation might be developmentally regulated. Our findings extend these observations, by showing that differential regulation of cyclin degradation might also exist between normal and transformed cells. Since MAP kinase appears to be involved in the cell cycle checkpoint, ensuring that anaphase is not initiated before spindle assembly is complete (Minshull et al., 1994), it will be interesting to investigate whether differences in the MAP kinase activation pathway between normal and transformed cells might be linked to the differences in the cyclins degradation patterns that we describe here. We thank Drs J. Pines and T. Hunter for the gift of human cyclin B1 cDNA, Prof. M. Dorée for helpful discussions, Dr J. C. Cavadore for the preparation of anti-cyclin B1 antibody, Dr N. Basset-Seguin for the gift of human primary fibroblasts, and N. Lautredou for technical assistance. This work was supported by grants from the AFM, the ARC (no. 6140) and FRM (to F.G.).

REFERENCES Bailly, E., Pines, J., Hunter, T. and Bornens, M. (1992). Cytoplasmic accumulation of cyclin Bl in human cells: association with a detergentresistant compartment and with the centrosomes. J. Cell Sci. 101, 529-545 Dunphy, W. and Kumagai, A. (1991). The cdc25 protein contains an intrinsic phosphatase activity. Cell 67, 189-196 Edgar, B. A., Sprenger, F., Duriono R. J., Leopold, P. and O’Farrell, P. H. (1994). Distinct molecular mechanisms regulate cell cycle timing at successive stages of Drosophila embryogenesis. Genes Dev. 8, 440-452; Félix, A. M., Labbé, J. C., Dorée, M., Hunt, T. and Karsenti, E. (1990). Triggering of cyclin degradation in interphase extracts of amphibian eggs by cdc2 kinase. Nature 346, 379-382 Fesquet, D., Labbé, J. C., Derancourt, J., Capony, J. P., Galas, S., Girard, F., Lorca, T., Shuttleworth, J., Dorée, M. and Cavadore, J. C. (1993). The MO15 gene encodes the catalytic subunit of a protein kinase that activates cdc2 and other cyclin-dependent kinases (CDKs) through phosphorylation of Thrl61 and its homologues. EMBO J. 12, 3111-3121 Gallant, P. and Nigg, E. A. (1992). Cyclin B2 undergoes cell-cycle dependent nuclear translocation and, when expressed as a non-destructible mutant, causes mitotic arrest in HeLa cells. J. Cell. Biol. 117, 213-224 Gautier, J., Solomon, M. J., Booher, R. N., Bazan, J. F. and Kirschner, M. W. (1991). cdc25 is a specific tyrosine phosphatase that directly activates p34cdc2. Cell 67, 197-211 Girard, F., Strausfeld, U., Fernandez, A. and Lamb, N. (1991). Cyclin A is required for the onset of DNA replication in mammalian fibroblasts. Cell 67, 1169-1179 Glotzer, M., Murray, A. W. and Kirschner, M. W. (1991). Cyclin is degraded by the ubiquitin pathway. Nature 349, 132-138 Holloway, S. L., Glotzer, M., King, R. W. and Murray, A. W. (1993). Anaphase is initiated by proteolysis rather than by inactivation of Maturation Promoting Factor. Cell 73, 1393-1462 Lehner, C. F. and O’Farrell, P. H. (1989). Expression and function of Drosophila cyclin A during embryonic cell cycle progression. Cell 56, 957968 Lehner, C. F. and O’Farrell, P. H. (1990). The roles of Drosophila cyclins A and B in mitotic control. Cell 61, 535-547 Lorca, T., Galas, S., Fesquet, D., Devault, A., Cavadore, J. C. and Dorée,

2608 F. Girard, A. Fernandez and N. Lamb M. (1991). Degradation of the proto-oncogene product p39mos is not necessary for cyclin proteolysis and exit from meiotic metaphase: requirement for a Ca2+-calmodulin dependent event. EMBO J. 10, 20872093 Luca, F. C., Shibuya, E. K., Dohrmann, C. E. and Ruderman, J. V. (1991). Both cyclin A ~60 and cyclin B ~97 are stable and arrest cells in M-phase, but only cyclin B ~97 turns on cyclin destruction. EMBO J. 10, 4311-4320 Maldonado-Codina, G. and Glover, D. M. (1992). Cyclins A and B associate with chromatin and the polar regions of spindles, respectively, and do not undergo complete degradation at anaphase in syncitial Drosophila embryos. J. Cell Biol. 116, 967-976 McClure, D. B., Hightower, M. J. and Topp, W. C. (1982). Effect of SV40 transformation on the growth factor requirements of the rat embryo cell line REF52 in serum free medium. Cold Spring Harbor Conf. Cell Prolif. 9, 345364 Millar, J. B. A., McGowan, C. H., Lanaers, G., Jones, R. and Russell, P. (1991). p80cdc25 is the tyrosine phosphatase that activates p34cdc2 kinase in fission yeast. EMBO J. 10, 43014309 Minshull, J., Pines, J., Golsteyn, R., Standart, N., Mackie, S., Colman, A., Blow, J., Ruderman, J. V., Wu, M. and Hunt, T. (1989). The role of cyclin synthesis, modification and destruction in the control of cell division. J. Cell. Sci. Suppl. 12, 77-97 Minshull, J., Golsteyn, R., Hill, C. S. and Hunt, T. (1990). The A- and B-type cyclin associated cdc2 kinases in Xenopus turn on and off at different times in the cell cycle. EMBO J. 9, 2865-2875 Minshull, J., Sun, H., Tonks, N. K. and Murray, A. W. (1994). A MAP kinase-dependent spindle assembly checkpoint in Xenopus egg extracts. Cell 79, 475-486 Murray, A. W., Solomon, M. J. and Kirschner, M. W. (1989). The role of cyclin synthesis and degradation in the control of maturation promoting factor. Nature 339, 280-286 Nurse, P. (1990). Universal control mechanism regulating onset of M-phase. Nature 344, 503-508

Pagano, M., Pepperkok, R., Verde, F., Ansorge, W. and Draetta, G. (1992). Cyclin A is required at two points in the human cell cycle. EMBO J. 11, 961971 Pines, J. and Hunter, T. (1990). Human cyclin A is adenovirus ElA-associated protein p60 and behaves differently from cyclin B. Nature 346, 760-763 Pines, J. and Hunter, T. (1991). Human cyclins A and B are differentially located in the cell and undergo cell cycle-dependent nuclear transport. J. Cell Biol. 115, 1-17 Poon, R. Y. C., Yamashita, K., Adamczewski, J. P., Hunt, T. and Shuttleworth, J. (1993). The cdc2-related protein p40MO15 is the catalytic subunit of a protein kinase that can activate p33cdk2 and p34cdc2. EMBO J. 12, 3123-3132 Rimmington, G., Dalby, B. and Glover, D. M. (1994). Expression of Nterminally truncated cyclin B in the Drosophila larval brain leads to mitotic delay at late anaphase. J. Cell Sci. 107, 2729-2738 Solomon, M. J., Wade-Harper, J. and Shuttleworth, J. (1993). CAK, the p34cdc2 activating kinase, contains a protein identical or closely related to p40MO15. EMBO J. 12, 3133-3142 Surana, U., Amon, A., Dowzer, C., McGrew, J., Byers, B. and Nasmyth, K. (1993). Destruction of the CDC28/CLB mitotic kinase is not required for the metaphase to anaphase transition in budding yeast. EMBO J. 12, 1969-1978 Strausfeld, U., Labbé, J. C., Fesquet, D., Cavadore, J. C., Picard, A., Sadhu, K., Russell, P. and Dorée, M. (1991). Dephosphorylation and activation of a p34cdc2-cyclin B complex in vitro by human cdc25 protein. Nature 351, 242-245 Whitfield, W. G. F., Gonzalez, C., Maldonado-Codina, G. and Glover, D. M. (1990). The A-and B-type cyclins of Drosophila are accumulated and destroyed in temporally distincts events that define separable phases at the G2-M transition. EMBO J. 9, 2563-2572

(Received 23 February 1995 - Accepted 31 March 1995)