In vivo localisation of fission yeast cyclin-dependent kinase cdc2p and cyclin B cdc13p during mitosis and meiosis

Cyclin-dependent kinases (CDKs)¹ are required for the onset of S phase and M phase of the mitotic cell cycle (reviewed by Nurse, 2000). In higher eukaryotes, distinct CDKs associate with several cyclins to control the G₁/S transition, whereas cdk1/cyclin B is specifically required to control the G₂/M transition (reviewed by Nigg, 1995). In the fission yeast Schizosaccharomyces pombe, cdc2p/cdc13p complex controls the G₂/M transition of the cell cycle, whereas cig2p B-type cyclin associates with cdc2p to promote the G₁/S transition, although this can be accomplished by the cig1p and cdc13p B-type cyclins (Fisher and Nurse, 1996; Mondesert et al., 1996). Fission yeast possesses two other cyclins, puc1p and pas1p, which are related to the Cln1-3p G₁ cyclins of Saccharomyces cerevisiae (Forsburg and Nurse, 1991; Tanaka and Okayama, 2000). Puc1p may regulate the G₁ phase progression in response to cell size (Martin-Castellanos et al., 2000), whereas pas1p associates with a second non-essential CDK, pef1p, to activate the res2p-cdc10p transcriptional complex at the START point of the cell cycle (Tanaka and Okayama, 2000).

CDKs are not only required for the onset of S phase and M phase of the mitotic cell cycle but also are involved in the regulation of centrosomes and microtubule (MT) dynamics. In Xenopus egg extracts, CDKs induce a change in MT dynamics and steady-state length (Verde et al., 1990). Cytostatic factor-arrested Xenopus egg extracts (which contain active cdc2 kinase bound to cyclin), but not interphase extracts, can convert fission yeast spindle pole bodies (SPBs, the equivalent of animal centrosomes) to a nucleation competent state (Masuda et al., 1992). However, purified cdc2/cyclin B1 complex was unable to do so, suggesting that other factors besides cdc2 kinase are also required for SPB activation (Masuda et al., 1992). In multicellular eukaryotes, duplication of the centrosome requires cdk2/cyclin E activity (reviewed by Whitehead and Salisbury, 1999; Meraldi et al., 1999; Okuda et al., 2000). Cdk1 localises to spindle MTs and metazoa centrosomes via association with MT-associated proteins (MAPs), whose phosphorylation induces a change in MT dynamics at the onset of mitosis (reviewed by Andersen, 2000). In fission yeast, Alfa et al. showed by immunofluorescence that cdc2p and cdc13p are localised to the SPBs during mitosis (Alfa et al., 1990). In a more recent paper, cdc13-GFP fusion was shown to localise on the SPBs and the spindle from prophase to metaphase (Yanagida et al., 1999).

S. pombe cdc2p is also required during the meiotic cell cycle for premeiotic DNA synthesis, the second division and, very likely, the first meiotic division (reviewed by Murakami and Nurse, 2000). Once karyogamy and premeiotic DNA replication have occurred, the meiotic prophase nucleus shows an elongated morphology, called a `horse-tail', and oscillates back and forth between the cell poles (reviewed by Hiraoka, 1998). During this horse-tail movement, telomeres are clustered at the SPB in a bouquet-like arrangement at the leading end of the tail while centromeres from the three pairs of duplicated chromosomes are separated from the SPB (reviewed by Hiraoka, 1998). When this movement stops, the first meiotic division starts, leading to reductional segregation of homologous chromosomes, followed by separation of sister chromatids in meiosis II (reviewed by Bickel and Orr-Weaver, 1996).

In this study, we used GFP- and YFP-tagged proteins to investigate the in vivo localisation of cdc2p/cyclin B in the fission yeast S. pombe and to establish its significance for both mitotic and meiotic cell cycle regulation.

Growth, transformation and genetic manipulation of S. pombe were performed using methods described previously (http://www.bio.uva.nl/pombe/). Cells were routinely grown at 32°C in either YE medium with supplements (YES) or minimal EMM2 medium with or without addition of 15 μM thiamine to control the nmt1 promoter. Mating and sporulation of cells was induced on EMM2 medium without NH₄Cl. The S. pombe strains used in this study are listed in Table 1.

The cdc2-YFP, cdc2-CFP and cdc13-YFP genes were cloned into the SalI/SmaI-digested pREP5 plasmid, which contains the full-strength thiamine-repressible nmt1 promoter and the sup3-5 marker (Maundrell, 1993). The primers used for PCR amplifications are listed in Table 2. The cdc2⁺ (primers 1 and 2) and cdc13⁺ (primers 3 and 4) genes were amplified by PCR. The EYFP gene was isolated from the pEYFP plasmid (Clontech, cat. number 6004-1). The CFP gene was amplified by PCR on the pECFP plasmid (Clontech, cat. number 6075-1) (primers 5 and 6). The YFP and CFP fragments were digested with NotI, blunted with Klenow and digested with BamHI. The cdc2⁺ and cdc13⁺ ORFs were cloned as SalI/BamHI-digested fragments. Plasmids with Y(C)FP fusions were integrated at the own gene locus in S. pombe cells, generating strains AD143 and AD112 in which the fusions are controlled by the endogenous gene promoters (see Fig. 1B). In AD185 strain, the nmt1prom-cdc2 gene was replaced by the kan^R gene using a method described previously (Bähler et al., 1998). For the cloning of cdc13▵81-YFP (lacking the first 81 amino acids) into pREP5 plasmid, the cdc13▵81 (primers 7 and 8) and the YFP (primers 9 and 6) genes were amplified by PCR and processed as described above. The same strategy was used to clone cdc13▵81-YFP and cdc13-YFP into pREP45 (medium-strength nmt1 promoter). In the pREP5:: cdc13▵81 plasmid, the cdc13▵81 gene was amplified by PCR (primers 7 and 10) and cloned into SalI/BamHI-digested pREP5 plasmid. For N-terminal YFP-tagging of cig2p, the YFP (primers 5 and 11) and cig2⁺ (primers 12 and 13) genes were amplified by PCR. YFP was digested with SalI and KpnI and ligated to the SalI/BamHI-digested pREP45 plasmid together with the KpnI/BglII-digested fragment of cig2⁺. Plasmids pREP5::cdc13▵81-YFP, pREP45::cdc13▵81-YFP, pREP5::cdc13▵81 and pREP45::YFP-cig2 were integrated into the genome of strains AD217, AD203, AD259 and AD179, respectively. The transcription of all four genes is repressed by thiamine. C-terminal GFP-tagging of dis1p was performed using a method described previously (Bähler et al., 1998), where the endogenous gene is replaced by a fusion between the GFP(S65T) and the dis1⁺ genes and the transcription is controlled by the own gene promoter. Visualization of GFP-α2-tubulin was achieved by transformation of yeast with the pEG5 plasmid as described (Ding et al., 1998).

The techniques used have been described previously (http://www.bio.uva.nl/pombe/). For co-immunoprecipitation assays, 2 mg of cell extract proteins were incubated with anti-cdc2p polyclonal serum C2 (1:50, Simanis and Nurse, 1986EF48). Proteins were separated on a 12% SDS-polyacrylamide gel (Laemmli, 1970EF25) and blotted onto Immobilon™-P membrane (Millipore). The antibodies used were anti-cdc13p mAb 6F (1:500; kind gift from Hayles and Steel) and anti-cdc2p mAb Y63 (1:500, Yamano et al., 1996EF57). Immunoreactive bands were detected using ECL (Amersham).

Cells were grown exponentially, pelleted and spread onto a slide covered with a ∼1 mm-thick layer of solid medium (2% low-melting agarose, 2% glucose and 225 μg/ml of adenine, histidine, uracil and leucine). In these conditions, cells were able to grow and divide for at least 3 hours at room temperature (RT, 23-25°C). Live fluorescence microscopy (LFM) was done at RT with a Zeiss Axioplan microscope, using a 100×, 1.3 oil immersion lens. GFP and YFP were excited with a mercury lamp, using a HQ 480/40 filter. A HQ 535/50 filter was used for fluorescence emission and images were captured with a Hamamatsu CCD camera C5985 and processed with the Adobe PhotoShop 5.5 software (Adobe Systems, San Jose, CA). The same procedure was used for the simultaneous observation of either cen1-GFP and cdc2-YFP (Fig. 3) or GFP-α2-tubulin and cdc2-YFP (Fig. 5D). For individual observation of cdc2-CFP and cdc13-YFP in the same cell (Fig. 2D), we used a Zeiss LSM 510 laser-scanning confocal microscope and a 63×, NA 1.4 oil immersion lens. CFP was excited at 458 nm and YFP at 514 nm. Fluorescence emission was observed using a BP475-525 filter for CFP and a LP530 filter for YFP.

In Fig. 4, cdc25-22 cells were grown overnight at 25°C in EMM2 medium. Cultures were arrested at the G₂/M transition by incubation for 4 hours at 36°C. The cdc13 gene transcription was repressed by addition of 15 μM thiamine after either 3 hours (AD210) or 4 hours (AD245 and AD207) at 36°C. Cells were pelleted and processed for LFM at RT (23-25°C), as described above, or incubated at 25°C and processed for DNA content measurement by FACS as described previously (http://www.bio.uva.nl/pombe/).

The following treatments were applied to the cells before LFM at RT: thermosensitive (ts) strains were grown at 25°C in YES medium and incubated at 36°C for either 4 hours (cdc10-V50 and cdc25-22) or 3 hours (mts2 and cut4-533). Cold-sensitive dis1-203 cells (Ohkura et al., 1988) were grown at 32°C in YES and incubated for 8 hours at 20°C. Cell cycle block with hydroxyurea (HU) was performed by incubation of growing cells with 11 mM HU (Sigma) for 4 hours at 32°C in either YES (AD185 and AD112) or EMM2 medium (AD179). For induction of cdc13▵81(-YFP) expression, cells were grown in YES medium, pelleted, washed and incubated for 16 hours at 32°C in EMM2 medium to allow switching ON of the nmt1 promoter.

In Fig. 8, strains AD185 (cdc2-YFP) and PN745 were mixed on EMM2-NH₄Cl plates for 8 hours at 25°C and observed by LFM. For observation of cdc2-YFP and cen1-GFP in the horse-tail nucleus (Fig. 9), we crossed strains MKY7A-4 (Nabeshima et al., 1998) and AD185 in the same conditions. GFP and YFP fluorescence were observed separately using the procedure described above for CFP/YFP. Even though excitation of the GFP at 458 nm is not optimal, we were able to detect the cen1-GFP signal. In Fig. 10, the cyr1▵sxa2▵ strain expressing cdc2-YFP was grown at 32°C in EMM2 medium before incubation for 8 hours at 30°C in the presence of 0.5 μg/ml P-factor, as described (Chikashige et al., 1997).

To follow the intracellular localisation of cdc2p and the cdc13p B-type cyclin in fission yeast, we constructed C-terminal fusions between the corresponding genes and the yellow fluorescent protein-encoding gene (YFP, reviewed by Tsien, 1998). The fusions were expressed using the inducible nmt1 promoter in the pREP5 plasmid (see Materials and Methods). Although the cdc2-YFP fusion fully complemented the cdc2-33 mutation, the cdc13-YFP construct failed to rescue the cdc13-117 mutation. However, overexpression of the cdc13-YFP fusion resulted in the appearance of short septated cells indicating premature entry into mitosis, suggesting that the fusion had some biological activity (not shown). Further support for this suggestion were obtained by expressing a truncated version of cdc13-YFP (lacking the first 81 amino acids), which resulted in an anaphase block, as described before (reviewed by Yanagida, 1998), and a high level of kinase/cyclin complex (Fig. 1A, lanes 2,4). In addition, co-immunoprecipitation experiments with anti-cdc2p antibody showed that cdc13-YFP and cdc13p interacted with cdc2p with similar efficiency (Fig. 1A, lanes 1,3). Fusion at the N-terminus of both cdc2p and cdc13p proteins gave similar results to the C-terminal fusions (not shown).

The C-terminal fusions in the pREP5 plasmid were integrated at their own locus by homologous recombination, generating strains with the YFP-fusions controlled by the cdc2 and cdc13 gene promoters and with the untagged genes controlled by the nmt1 promoter (Fig. 1B, top, and lanes 1-2,5-6). The nmt1-cdc2 gene was replaced by the kanamycin resistance gene, leaving the cdc2-YFP fusion as the only source of cdc2p in the cell (Fig. 1B, lane 3). The cdc13-YFP-expressing strain grew normally in the presence of thiamine (YES), even though expression of the untagged cdc13p was low compared with wild-type cells grown as control (Fig. 1B, lane 6 vs WT in lane 4), giving further support for partial functionality of cdc13-YFP. Cdc13-YFP was present at WT levels (Fig. 1B, lane 6 vs lane 4). This strain enabled us to monitor cdc13-YFP behaviour in cells behaving like WT, even though they are expressing only very low amounts of untagged cdc13p.

We followed cdc2-YFP and cdc13-YFP localisation in living cells using the cdc2-YFP (AD185) and cdc13-YFP (AD112)-expressing strains grown in YES (see Materials and Methods). In the conditions of our experiments, the nmt1::cdc13 gene expression remained OFF during microscopic observations. Quantification of cdc2-YFP fluorescence revealed that at least 60-70% of the total cdc2-YFP was present in the cytoplasm of G₂ cells (Fig. 2A, top; B), whereas all the cdc13-YFP fluorescence was detected in the nucleus (Fig. 2A, bottom). We believe that the fact that we can detect the cell outlines in cdc13-YFP-expressing cells does not reflect a cytoplasmic localisation of the cyclin since autofluorescence was also observed in control cells that do not express any YFP-fusion (Fig. 2An). The fluorescence of mitochondrial flavoproteins (Kunz et al., 1997) is likely to contribute to the green autofluorescence of yeast cells. The cable-like structures that can be seen in the cytoplasm of some cells in Fig. 2A (bottom) may therefore correspond to the fluorescence of mitochondria aligned with cytoplasmic MTs (interphase MTs and post-anaphase array), as described (Yaffe et al., 1996).

For both cdc2-YFP and cdc13-YFP, nuclear fluorescence was lowest in late mitosis when cdc13-YFP fluorescence was undetectable (Fig. 2Ak,B,C). Nuclear signals gradually increased once mitosis had been completed (Fig. 2Al-m,B). Nuclear fluorescence of both cdc2-YFP and cdc13-YFP was clearly detectable in septated cells known to undergo S phase (Nasmyth et al., 1979) but was not as high as in G₂ cells (Fig. 2A, m vs a and b, B). In late G₂, a bright spot of both cdc2-YFP and cdc13-YFP appeared at the nuclear periphery and split into two dots during early mitosis (Fig. 2Ab-c). These dots appeared to correspond to the SPBs because they separated apart with a spindle between them during mitosis as confirmed by simultaneous observation of cdc2-YFP and and GFP-α2-tubulin (cf. Fig. 5Dp-t). From prophase to metaphase, cdc2-YFP and cdc13-YFP were enriched on the forming spindle and SPBs (Fig. 2Ac-f). At mitotic exit, the fluorescence of both proteins disappeared, first from the middle of the nucleus, then from the spindle, and finally from the SPBs (Fig. 2Ag-i). In anaphase, cdc13-YFP was mainly detected at the nuclear periphery (Fig. 2Ai-j,Cb-h) and completely disappeared by nuclear division (Fig. 2Ak,Ci). Cdc2-YFP nuclear fluorescence was reduced by 75% in anaphase but did not accumulate at the nuclear periphery, and some residual cdc2-YFP fluorescence followed the sister chromatids towards the cell ends (Fig. 2Ai-k, top, B). Observation of the same cell revealed that transition from stage (c) to (h) and (h) to (m) required 4 and 45 minutes, respectively (not shown). The above data indicate that the level of cdc13p and cdc2p in the nucleus change together during the cell cycle. To confirm this, the levels of cdc2p and cdc13p were compared within the same cell by fusing the cdc2⁺ gene to CFP, using the same strategy as that described for the cdc2-YFP strain. Cells at different stages of the cell cycle are presented in Fig. 2D with cdc13-YFP shown in the left panel, cdc2-CFP in the middle panel and merged images in the right panel. In anaphase, cdc2-CFP and cdc13-YFP did not co-localise at the nuclear periphery (Fig. 2Da), whereas in S phase, cdc2-CFP and cdc13-YFP fluorescence was lower than that observed in G₂ cells or in mitosis (Fig. 2D, b vs c and d).

Finally, we investigated the dynamics of cdc2p and cdc13p at mitotic exit. By monitoring cells in real time, we found that both cdc13-YFP and cdc2-YFP leave the mitotic spindle rapidly, in less than 1 minute (Fig. 3A, 3′ vs 2′, B, 2′ vs 1′). Expression of cdc2-YFP in a strain with the centromeric region of chromosome I marked by GFP fluorescence (cen1-GFP; Nabeshima et al., 1998) revealed that cdc2-YFP leaves the mitotic spindle immediately prior to sister chromatid separation (Fig. 3B, 3′ vs 2′). In this experiment, YFP and GFP fluorescence were observed simultaneously and the separation of sister chromatids was detected when the two dots of cen1-GFP moved apart.

Because cdc2-YFP nuclear fluorescence decreases dramatically upon cdc13p degradation at mitotic exit, we next tested whether B-type cyclins are required for the accumulation of cdc2p in the nucleus. Cdc 13p is the essential cyclin at the G₂/M transition and is required for entry into S phase in the absence of cig1p and cig2p (Fisher and Nurse, 1996). Using cig1 and cig2 deletions (Bueno et al., 1991; Obara-Ishihara and Okayama, 1994; Fisher and Nurse, 1996) in combination with a cdc13 thiamine-repressible allele (Hayles et al., 1994; Fisher and Nurse, 1996) and the cdc25-22 ts mutation, we constructed four strains to test our hypothesis (see Materials and Methods). Strains were grown at 25°C without thiamine before shifting the temperature to 36°C for 4 hours to block cells at the G₂/M transition. Thiamine was added either after 3 hours at 36°C for cig2⁺ strains (Fig. 4Ad,Bb) or at the time of release for the cig2-deleted strains (Fig. 4Ae-f,Bc-d) to allow cell progression through mitosis. These conditions produced cells containing solely cdc13p (Fig. 4Ac), cig2p (Fig. 4Ad) or cig1p (Fig. 4Ae), or lacking all of these B-type cyclins (Fig. 4Af). After 30 minutes of release into mitosis, the cdc2-YFP fluorescence was found to be enriched on the mitotic spindle in all strains (Fig. 4Aa) and faded by the onset of anaphase. In the presence of all three B-cyclins, cdc2-YFP fluorescence never completely disappeared from the daughter nuclei at mitotic exit and started to increase again at G₁/S (Fig. 4Ab). The cdc2-YFP nuclear fluorescence profile was similar when cdc13p was the only B-type cyclin present (Fig. 4Ac,Ba). However, when only cig2p was present, the nuclear fluorescence was lower at G₁/S and remained unchanged as cells proceeded through S phase (Fig. 4Ad,Bb). When cig1p was the only cyclin present, the cdc2-YFP fluorescence disappeared immediately after mitosis, and only became detectable again in septated cells after 2 hours of release (Fig. 4Ae). The FACS profile showed a major 1C peak after 1 hour and 40 minutes of release and the following S phase was very slow (Fig. 4Bc). In the absence of all three cyclins, nuclear staining of cdc2p never appeared again (Fig. 4Af). The FACS profile showed a major 1C peak at the time of daughter cell separation and some of the post-mitotic cells took a long time or failed to separate, resulting in a broad FACS profile (Fig. 4Bd).

To confirm the above data, we looked at cdc2-YFP nuclear fluorescence in cdc10-arrested cells, when cdc13p, cig1p and cig2p cyclins are absent (Hayles et al., 1994; Mondesert et al., 1996; Blanco et al., 2000). After incubation for 4 hours at 36°C in a cdc10-V50 ts mutant, nuclear cdc2-YFP fluorescence levels were similar to those in the cytoplasm (Fig. 4C, left). Cytoplasmic cdc2-YFP often appeared as bright blobs, possibly due to aggregation of cdc2-YFP at 36°C as recently reported for GFP (Fukuda et al., 2000) (Fig. 4C, left). This phenomenon has also been observed in WT cells expressing cdc2-YFP after incubation at 36°C and is therefore not related to the ts mutation (not shown). In cdc10-V50-arrested cells, cdc13-YFP was completely absent from the nucleus (Fig. 4C, right). This observation is in contrast with previous work in which cdc13p was detected by immunofluorescence in the nucleus of cdc10-129 cells after 6 hours at 36°C (Booher et al., 1989). This is probably because of leak-through of this allele, and we made similar observations with our cdc13-YFP fusion after prolonged incubation of cdc10-V50 cells at 36°C (not shown).

The above results show that B-type cyclins, cdc13p, cig1p or cig2p, are required to allow cdc2-YFP accumulation in the nucleus after completion of mitosis. In the absence of all three B-type cyclins, the concentration of cdc2-YFP is similar in the nucleus and in the cytoplasm.

We next examined the association between cdc2p/cdc13p and the SPBs. In an asynchronous population, cdc2-YFP and cdc13-YFP were strongly associated with the SPB in 20% of the cells, consistent with an enrichment in late G₂ or early mitosis. However, low cdc2-YFP fluorescence could be observed at the SPB of some binucleate cells after septation has occurred (Fig. 5Aa-e). Therefore, we tested whether we could detect SPB-associated fluorescence in cells arrested in early S phase with HU. In HU-treated cells, cdc13-YFP and cdc2-YFP accumulated strongly in the nucleus and a bright dot was often seen at the nuclear periphery, indicating that the complex is most probably localised to the SPB (Fig. 5Ba,b). Observation of the same cell releasing from an HU block confirmed that the bright dot observed at the nuclear periphery co-localises with the SPB since the dot further split into two dots that separated apart with a spindle between them during mitosis (data not shown). The B-type cyclin cig2p level is also increased in HU-arrested cells (Mondesert et al., 1996), and cig2p appears to be localised to the SPB as revealed by YFP-cig2 LFM (Fig. 5Bc). To address the question of whether the strong accumulation of the cdc2p/cdc13p complex observed at later stages of the cell cycle required the Tyr15-dephosphorylation of cdc2p, we looked at the SPB fluorescence of cells arrested at the G₂/M transition in a cdc25-22 ts mutant (Fig. 5C). After 4 hours incubation at 36°C, cells had a high level of nuclear cdc2-YFP and cdc13-YFP and both proteins became strongly accumulated at the SPB (Fig. 5Ca,b,Da,f), showing that cdc2p/cdc13p complexes with low kinase activity can locate to the SPB. In the G₂/M block, cytoplasmic MTs are still present (Fig. 5Dk, GFP-α2-tubulin), establishing that the enrichment of cdc2p/cyclin B on the SPB occurs prior to the reorganization of the cytoplasmic array of MTs. Upon release into mitosis at 25°C, cdc2-YFP and cdc13-YFP were still enriched on the separating SPBs and on the forming spindle (Fig. 5Db-e,g-j,l-o,q-t).

Our data suggest that a fraction of S. pombe cdc2p is present at the SPB in G₁/S or S phase. More kinase then accumulates on the SPB of G₂ cycling cells. SPB-association was also detected in cells arrested in early S phase and at the G₂/M transition prior to the reorganization of the cytoplasmic array of MTs and mitotic spindle formation when the bright dots of cdc2-YFP and cdc13-YFP separate and spindle MTs form between them.

Next, we looked at the localisation of cdc2-YFP in a stf1.1 mutant. Indeed, Hudson et al. have shown that fission yeast stf1.1 is a semi-dominant mutation that bypasses the requirement for cdc25p-mediated activation of cdc2p at the G₂/M transition (Hudson et al., 1990). Isolation of the cut12 gene then revealed that cut12⁺ is allelic to stf1⁺ (Bridge et al., 1998). As cut12p was found to be localised to the SPB throughout the cell cycle, the authors suggested that stf1.1/cut12.G17V allows the formation of cdc2p/cdc13p complexes at the SPB that would not be Tyr15-phosphorylated by wee1p/mik1p kinases, thereby promoting mitosis in the absence of cdc25p function. We tested whether cdc2-YFP would go prematurely to the SPB of stf1.1 cells but we did not find any significant difference with the WT situation (not shown).

We next addressed what would happen to cdc2p localisation during mitosis if cdc13p was not degraded. The `destructionbox' motif on the cyclin and a functional anaphase-promoting complex (APC) are required for cdc13p degradation and APC mutant cells become blocked in anaphase (reviewed by Yanagida, 1998). In the cut4-533 ts mutant, which lacks a functional APC, cdc2-YFP and cdc13-YFP fluorescence remained associated with the spindle, the SPBs, and throughout the nucleus (Fig. 6c-d). Overexpression of a truncated cdc13p, lacking the 81 first amino acids, including the `destruction box' required for recognition by the APC, resulted in an anaphase block (Yamano et al., 1996) and strong cdc2-YFP and cdc13Δ81-YFP fluorescence remained associated with the spindle, SPBs and nucleus (Fig. 6e-f). In most cells, the mitotic spindle failed to elongate. Deletion of the first 106 amino acids of cdc13p, including a region of cdc13p not present in cig2p, did not impair its ability to bind to mitotic MTs (not shown). Ts mutants of the mts2 subunit of the proteasome complex that become arrested at the metaphase/anaphase transition (Gordon et al., 1993) still accumulated the cdc2p/cdc13p complex in the nucleus and on the mitotic spindle and SPBs (Fig. 6a-b).

These data indicate that interaction of cdc13p with the APC occurs on the spindle and the SPBs during anaphase, and that recognition of cdc13p by the APC is required for cdc2p to leave the spindle at mitotic exit. If cdc13p cannot be recognized by the APC, or if a functional APC is absent, then cdc2p and cdc13p remain associated with mitotic MTs.

The above data suggest that cdc2p contributes to the regulation of MT dynamics in mitosis. In Xenopus, XMAP215 is known to target cdk1 to the mitotic MTs and to be a major MT-stabilizing factor (reviewed by Andersen, 2000). Therefore, we tested whether S. pombe dis1p, a homologue of XMAP215, plays a similar role in fission yeast.

We replaced the dis1⁺ gene by a dis1-GFP fusion in a cdc25-22 background (Materials and Methods) and investigated the localisation of dis1-GFP in mitosis. In agreement with previous studies (Nabeshima et al., 1995), dis1-GFP was present on the metaphase spindle (Fig. 7Ab-e) but, in contrast to the earlier study, did not appear to be enriched on the SPBs during metaphase. By following the same cells in a cdc25-22 block-and-release experiment, dis1-GFP was found to be associated with dots likely to correspond to the centromeres, which become located near the SPBs during anaphase (Fig. 7Ab-f). The extremities of the mitotic spindle often showed more than one dot of dis1-GFP consistent with a localisation on the centromeres (Fig. 7Ag,h). Moreover, we found that dis1-GFP and bub1-GFP, a kinetochore-binding protein (Bernard et al., 1998), had similar localisation patterns in metaphase (not shown). Because of its association with the metaphase spindle, we tested whether dis1p was required for the targeting of fission yeast cdc2p/cyclin complex to mitotic MTs (Fig. 7B) using a cold-sensitive (cs) dis1-203 mutant. The dis1-203 mutation is a nonsense mutation resulting in the formation of a truncated dis1p protein containing only the first 265 amino acids (wild-type dis1p is an 882-amino-acid protein) and therefore lacking the central domain of dis1p, which comprises the cdc2p phosphorylation sites and is thought to be responsible for the binding of dis1p to MTs (Nabeshima et al., 1995). The dis1-203 cs mutant arrests in mitosis with condensed chromosomes scattered along the elongated and disrupted spindle (Ohkura et al., 1988). Cdc2-YFP and cdc13-YFP were found to be present on the SPBs and on the disrupted spindle at 20°C (Fig. 7Ba,b).

These results show that dis1p MAP function is not required for the targeting of cdc2p/cdc13p complex to the mitotic MTs. Dis1-GFP may associate with centromeres in mitosis, in agreement with a role of this protein in the dynamic movement of centromeres during metaphase (Nabeshima et al., 1998EF37).

In S. pombe meiotic prophase, telomeres cluster near the SPB to form a bouquet-like arrangement (reviewed by Hiraoka, 1998) followed by separation of the centromeres from the SPB as cells proceed towards the first meiotic division (Chikashige et al., 1994; Chikashige et al., 1997) (Fig. 8A). A cdc2-YFP-expressing strain was crossed with a cdc2⁺ strain and the cross monitored by LFM (Fig. 8B). At onset of conjugation, cdc2-YFP fluorescence was detected in only one of the two cells (Fig. 8Ba). As conjugation proceeded, the two cells fused, the nuclear fluorescence of the cdc2-YFP-expressing cell increased, and the nucleus of the cdc2⁺ strain became fluorescent, showing that cdc2-YFP could enter the nucleus at this stage of conjugation (Fig. 8Bb). When nuclei adopted a drop-shape, their SPB-associated extremities became very bright (Fig. 8Bc). This was followed by karyogamy when the bright dots migrated from the site of nuclear fusion towards the opposite end of the nuclei, probably due to cdc2-YFP becoming associated with the clustered centromeres as they detached from the SPBs (Fig. 8Bd-f). After completion of karyogamy, `horsetail' movements started and cdc2-YFP was concentrated into 1-3 bright dots in the elongated zygotic nucleus (Fig. 8Bg-k). The same diploid cell was followed for 70 minutes (Fig. 8Bi-s). When horse-tail movements stopped, cdc2-YFP was found to be located in several fuzzy dots throughout the nucleus that disappeared when cdc2-YFP relocalised to the spindle (Fig. 8B1-u). After the first meiotic division had been completed, cdc2-YFP fluorescence reappeared on the SPBs and the spindle of the two nuclei in the second meiotic division (Fig. 8Bv-x). The appearance of cdc2-YFP on the meiosis I spindle was different in appearance to either mitosis or meiosis II. Instead of being first located on the SPBs and then on the spindle, cdc2-YFP appeared all along the spindle as it formed during meiosis I (Fig. 8Bp-s), suggesting distinct pathways of spindle formation, reminiscent of the formation of `acentrosomal' meiotic spindles in many oocytes and in some spermatocytes (reviewed by Merdes and Cleveland, 1997).

In a time-lapse series during horse-tail movements (Fig. 9A), cdc2-YFP fluorescence was concentrated into 2 or 3 dots. To see whether these bright dots were associated with the centromeres, we crossed the cen1-GFP and the cdc2-YFP-expressing strains, distinguishing GFP (Fig. 9Bb) from YFP fluorescence (Fig. 9Ba). Cells in meiotic prophase showed that the cen1-GFP signal overlapped with one of the cdc2-YFP dots in the horse-tail (Fig. 9Bc). The fact that the GFP and YFP signals were not fully superimposed is probably due to the 30 kb distance between the cen1-GFP signal and the centromere of chromosome I. The co-localisation of cen1 with one cdc2-YFP dot suggests that the remaining bright dots of cdc2-YFP correspond to the two other pairs of centromeres.

The above data show that cdc2-YFP enters the nucleus very early in conjugating cells, with cdc2-YFP becoming enriched at the SPB-associated end of the fusing nuclei. Cdc2-YFP then migrates towards the opposite end of the nucleus, becoming associated with the clustered centromeres. In the horse-tail nucleus, cdc2-YFP is enriched in 1-3 bright dots, at least one of which co-localises with the centromeric region of chromosome I.

Finally, we addressed the question of whether the accumulation of cdc2p on the centromeric regions of chromosomes in mating cells occurred prior to or after the detachment of centromeres from the SPB. Chikashige et al. have shown that, in a sxa2Δcyr1Δ mutant responding to P-factor, telomeres cluster together with the SPB and the centromeres to form a bouquetlike arrangement (Chikashige et al., 1997) (Fig. 10, left). Cdc2-YFP was enriched close to a darker portion of the nucleus, which is likely to be the nucleolus (corresponding to chromosome III telomeres; Fig. 10, arrowheads) in sxa2Δcyr1Δ cells incubated for 8 hours at 30°C in the presence of P-factor. Cdc2-YFP was enriched in either a bright single dot or multiple dots very close to each other (Fig. 10, arrows). The fact that these bright dots of cdc2-YFP were not observed at the nuclear periphery but in the nucleoplasm is consistent with a localisation on the centromeres and not on the SPB. Therefore, in the absence of a mating partner, cdc2-YFP is likely to become enriched on the SPB-associated centromeres prior to their detachment from the SPB during mating.

We have followed the in vivo localisation of YFP-tagged cdc2p and cdc13p during the fission yeast cell cycle. Our study is consistent with previous immunofluorescence data obtained during S. pombe mitotic cell cycle (Booher et al., 1989; Alfa et al., 1990) but our in vivo analysis gives more information and reveals new dynamics previously unsuspected during both mitosis and meiosis. During the mitotic cell cycle, cdc13-YFP was detected only in the nucleus, whereas cdc2-YFP fluorescence was found in both the cytoplasm and the nucleus. The level of cdc2-YFP and cdc13-YFP fluorescence in the nucleus changed together during the cell cycle, in agreement with previous immunofluorescence data (Booher et al., 1989). The nuclear level of both was highest from late G₂ to metaphase, dropped in anaphase, and reappeared around the completion of S phase. In early G₁ cells, cdc13-YFP was not detected and cdc2-YFP levels were similar in the nucleus and cytoplasm. As cyclin levels increased at G₁/S, nuclear cdc2-YFP fluorescence also increased. Using cyclin B-depleted strains, we established that the accumulation of cdc2-YFP fluorescence in the nucleus of post-mitotic cells required at least one of the three B-type cyclins, cig1p, cig2p or cdc13p. In a HU block, the cyclin B level is highly increased, resulting in the accumulation of more than 80% of the cdc2-YFP cellular content in the nucleus. The above data show that cdc2p requires the B-type cyclins to accumulate in the nucleus after mitotic exit. One hypothesis is that cdc2p requires the cyclins to enter the nucleus, in agreement with the fact that the cdc2p amino acid sequence does not contain any nuclear localisation signal (NLS), whereas all three S. pombe B-type cyclins contain regions of homology to the NLSs of SV40T antigen (Kalderon et al., 1984). This model suggests that cdc13p associates with cdc2p in the cytoplasm and that the complex is then rapidly transported into the nucleus, explaining why cdc13p could not be detected in the cytoplasm. An alternative hypothesis would be that the actual function of the B-type cyclins is to retain cdc2p in the nucleus. In this model, the uncomplexed cdc2p may be able to enter the nucleus freely and would only associate with the cyclin in the nucleoplasm. Once complexed to the cyclin, the kinase may not be able to get out of the nucleus anymore, unless the cyclin is degraded. In both these models, it appears that the level of cyclin in the cell is the limiting step for nuclear accumulation of cdc2p.

In cycling cells, low amounts of cdc2-YFP were detected on the SPB of septated cells (undergoing S phase), whereas cdc2-YFP, cdc13-YFP and YFP-cig2 accumulate more clearly on the SPB of cells arrested in early S phase with HU. This may be related to the role of cdk2/cyclin A or E complex in higher eukaryote centrosome duplication during G₁-S (reviewed by Whitehead and Salisbury, 1999; Meraldi et al., 1999; Okuda et al., 2000). In late G₂ cycling cells, we found that both cdc2p and cdc13p are much more strongly associated with the SPB, suggesting that the kinase activity required on the SPB of cells entering mitosis is higher. Cdc25p tyrosine-phosphatase activity is not required for the enrichment of cdc2- and cdc13-YFP on the SPB. Upon release of a cdc25-22 ts mutant at 25°C, SPB separation occurred quickly, suggesting that the cdc2p/cdc13p complex is activated on the SPB upon entry into mitosis. This is consistent with the work on activation of the human centrosomal cdk1/cyclin B1 complex (De Souza et al., 2000) and with work in budding yeast showing that separation of duplicated SPBs requires Tyr-19 dephosphorylation of Cdc28 (Lim et al., 1996). We propose that active fission yeast cdc2p regulates in vivo SPB MT-nucleating activity and spindle formation. However, the nature of the signal that triggers accumulation of the cdc2p/cdc13p complex on the SPB of G₂ cells is unknown. We do not think that it is only dependent on cell size since the cdc2-YFP/cdc13-YFP fluorescence observed on the SPB of elongated HU-arrested cells is lower than the one observed in a cdc25-22 block. One hypothesis might be that SPB accumulation of cdc2p/cdc13p requires maturation of the duplicated SPB, which occurs during the G₂ interphase of cycling cells (reviewed by Adams and Kilmartin, 2000).

In mitosis, the cdc2p/cdc13p complex accumulates on the SPBs and the spindle from prophase to metaphase, in agreement with recent data obtained with a cdc13-GFP fusion (Yanagida et al., 1999). In anaphase, we observed that cdc2-YFP and cdc13-YFP disappeared rapidly from the mitotic spindle, immediately prior to sister chromatid separation. This is shortly before the switch in MT dynamics (Mallavarapu et al., 1999) at the onset of anaphase B. Because CDKs from a number of different species regulate MT dynamics in mitosis via the association with MAPs (reviewed by Andersen, 2000), we searched for S. pombe MAPs homologues that could associate with cdc2p/cdc13p on the MT spindle in metaphase. We found homology between S. pombe ORF SPAC23A1.17 and Xenopus XMAP4 (Shiina and Tsukita, 1999) (BLAST score of 10^-11), but the GFP-tagged protein was localised to the septum and the cell ends and not on the mitotic spindle (not shown). Also, dis1p (Nabeshima et al., 1995), which shows high identity with Xenopus XMAP215, a cdk1-associated MT-stabilizing factor (reviewed by Andersen, 2000; Tournebize et al., 2000), was not required for the location of cdc2p/cdc13p to the mitotic MTs. Dis1-GFP was associated with centromeres in metaphase, and because dis1p is phosphorylated on cdc2p consensus sites and is required for centromere movement in metaphase (Nabeshima et al., 1995; Nabeshima et al., 1998), we suggest that cdc2p has an effect on centromere dynamics. It has been suggested that, in metaphase, kinetochore proteins bind loosely to MTs whereas, during anaphase, they become strongly attached (Zhai et al., 1995). Cdc2p might phosphorylate centromeric dis1p during metaphase resulting in low stability of kinetochore MT attachment to the centromere and, on displacement of cdc2p/cdc13p from the spindle, dephosphorylation of dis1p might result in strong binding to the kinetochore MTs followed by sister chromatid separation in anaphase A.

We also showed that, in anaphase A and early anaphase B, cdc13-YFP fluorescence was mainly detected at the nuclear periphery, possibly corresponding to its degradation by the proteasome, since subunits of the 19S regulatory cap have been shown to localise to the inner face of the nuclear membrane in S. pombe (Wilkinson et al., 1998). As this was not observed for the cdc2-YFP protein, the dissociation of the cdc2p/cdc13p complex may need to occur prior to cdc13p degradation. This is consistent with studies showing that, in Xenopus egg extracts, dissociation of the cyclin B-cdk1 complex occurs prior to cyclin B degradation (Nishiyama et al., 2000). When a stable form of cdc13p was overexpressed, both cdc2-YFP and cdc13Δ81-YFP remained associated with the spindle. A similar localisation was observed in a cut4 ts mutant, defective in APC function, suggesting that cdc13p needs to be recognized by the APC to leave the spindle. Interaction between the APC and cdc13p appears to occur on the spindle itself and recognition of cdc13p by the APC appears to be required for cdc2p to leave the spindle. The fact that cdc2p dissociation from the spindle first requires cdc13p dissociation is in agreement with the observation that the association of mammalian cdk1/cyclin B complex with spindle MTs occurs through the interaction of cyclin B with MAPs (Ookata et al., 1995; Charasse et al., 2000). Our data are also consistent with the co-localisation of APC with the centrosomes and the mitotic spindle (reviewed by Peters, 1999; Tugendreich et al., 1995).

In the second part of our work we investigated the localisation of cdc2p during mating, karyogamy and meiosis. In conjugating cells, we found that cdc2-YFP enters the nucleus before karyogamy starts, possibly because of the requirement of cdc2p kinase activity for premeiotic DNA synthesis (reviewed by Murakami and Nurse, 2000). When nuclei adopted the drop-shaped profile, cdc2-YFP was enriched on the cluster of centromeres-SPB-telomeres. As karyogamy proceeded, nuclei fused together at their cdc2-YFP enriched ends, followed by the appearance of one bright dot in the middle of each fusing nucleus, suggesting that cdc2-YFP was associated with the clustered centromeres known to leave the SPB at that stage (Chikashige et al., 1997). In the horse-tail nucleus, cdc2-YFP was in the nucleus and was enriched in 1-3 bright dots. Because one of the dots co-localised with cen1-GFP in the centromeric region of chromosome I, we conclude that the other cdc2-YFP dots colocalise with the centromeres of chromosomes II and III. The enrichment of cdc2-YFP with the cluster of centromeres-SPB-telomeres in shmooing cells does not require a mating partner since a bright cdc2-YFP dot(s) was observed in cyr1Δsxa2Δ cells responding to P-factor. These data suggest that cdc2p plays a role at the centromeres early during the mating process. One possibility would be that cdc2p activity regulates the centromere-SPB detachment that occurs at early stages of karyogamy (reviewed by Hiraoka, 1998). Cdc2p target proteins at the centromeres might need to be phosphorylated throughout meiotic prophase to avoid early reassociation of centromeres with the SPB. Given cdc2p localisation we propose that cdc2p might influence the proper segregation of chromosomes in meiosis I, and we have preliminary evidence that inactivation of cdc2p during meiotic prophase increases the frequency of equational division of sister chromatids (not shown). Surprisingly, cdc13-YFP nuclear fluorescence was much lower in karyogamy and meiosis compared to mitosis (not shown), suggesting a possible association of cdc2p with another cyclin(s) in the nucleus of meiotic cells. Although the fluorescence intensity was low, we detected the association of cdc13-YFP with the centromeres during horse-tail movement of the nucleus in meiotic prophase.

Association of cdc2-YFP with centromeres was no longer detected in the first meiotic division. After the horse-tail movements had stopped, cdc2-YFP relocalised to the spindle at the first meiotic division. Observation of MT-associated cdc2-YFP fluorescence in meiosis I revealed that cdc2-YFP fluorescence appeared all along the spindle while, in mitosis or meiosis II, fluorescence is first enriched on the duplicated SPBs before extending to the elongating spindle. This suggests that regulation of spindle formation in meiosis I might be different from mitotic division and influence the distribution of sister chromatids in meiosis I. This observation suggests the existence of a different pathway of spindle formation in fission yeast meiosis I, and possibly the formation of an `acentrosomal' meiotic spindle as observed previously in many oocytes and in some spermatocytes (reviewed by Merdes and Cleveland, 1997EF33). Thus it appears that modifications in cdc2p localisation play a role in changing chromosomal behaviour in meiosis from that observed in mitosis.

We thank Stephanie Yanow, Mercedes Pardo, Satoko Yamaguchi and Takashi Toda for critical reading of the manuscript, and all members of the ICRF Cell Cycle Laboratory for their help and support. We are grateful to M. Yanagida for the MKY7A-4 strain and to D.-Q. Ding and Y. Hiraoka for the GFP-α2-tubulin plasmid. This work was supported by the European Commission, Biotechnology Programme (A.D.) and the ICRF.