A fission yeast general translation factor reveals links between protein synthesis and cell cycle controls

The replication and segregation of chromosomal DNA are the two most important events of the cell cycle, and in eukaryotes the initiation of both these processes is triggered by activation of one or more cyclin-dependent kinases (or CDKs). Whilst higher eukaryotes have multiple CDKs that regulate cell cycle progression, a single CDK regulates both S-phase and mitosis in the yeasts Schizosaccharomyces pombe and Saccharomyces cerevisiae, though other CDKs play a role in certain circumstances (Espinoza et al., 1994; Measday et al., 1994; Nishizawa et al., 1998). In both yeasts, B-type cyclins promote entry into both S-phase and mitosis. Three B-cyclins are responsible for the activation of Cdc2 during the fission yeast cell cycle. Cig2 plays the major role in promoting entry into DNA replication (Fisher and Nurse, 1996; Martin-Castellanos et al., 1996; Mondesert et al., 1996; Obara-Ishihara and Okayama, 1994) and the levels of Cig2 and its associated Cdc2 kinase activity peak around G₁/S (Martin-Castellanos et al., 1996; Mondesert et al., 1996). Cdc13 is essential for the onset of mitosis (Booher and Beach, 1988; Hayles et al., 1994; Moreno et al., 1989) and Cig1 represents a third B-cyclin whose precise role is unclear (Bueno et al., 1991) though it is an important regulator of the stability of the Rum1 CDK inhibitor (Benito et al., 1998; Correa-Bordes et al., 1997).

Since Cdc2 can only function in association with a cyclin partner, periodicity in the levels of B-cyclins provides an important mechanism by which Cdc2 kinase activity is regulated during the cell cycle. Whilst cdc13 mRNA levels are constant throughout the cell cycle (J. Hayles, unpublished data), Cdc13 protein accumulates from S-phase to the end of mitosis, when it is degraded by the anaphase promoting complex (or APC; Moreno et al., 1989; Yamano et al., 1996). In contrast, Cig2 protein and mRNA levels both peak around G₁/S, when the level of Cdc13 protein is low (Connolly and Beach, 1994; Martin-Castellanos et al., 1996; Mondesert et al., 1996; Obara-Ishihara and Okayama, 1994).

Although proteolysis is clearly of major importance for the regulation of cyclin periodicity, cyclin expression can also be regulated at the level of translation. The translation of Drosophila cyclin B mRNA is inhibited during early development (Dalby and Glover, 1993), preventing accumulation of cyclin B until just before proliferation resumes in the germ cells. In budding yeast, expression of the G₁ cyclins CLN2 and CLN3 is also inhibited by means of translational repression under certain conditions (Gallego et al., 1997; Hall et al., 1998; Philpott et al., 1998; Polymenis and Schmidt, 1997), thereby linking translation to regulation of progression through the G₁ phase of the cell cycle.

Here we characterise mutations in a fission yeast general translation factor, that give rise to phenotypes suggesting defects in B-cyclin expression. The B-cyclins Cdc13 and Cig2 are especially sensitive to mutation of this factor, suggesting that translation may be relevant to the regulation of B-cyclin expression. We show that translation of both Cdc13 and Cig2 is inhibited in response to nitrogen starvation, when cells become arrested in G₁ as a prelude to mating and meiosis. Our results suggest that inhibition of B-cyclin translation may play a role in this important cell cycle transition.

All our strains are congenic with the Schizosaccharomyces pombe 972h⁻ strain, and all basic growth and media conditions were as described by Moreno et al. (1991). The cig2Δ and rum1Δ strains were described by Obara-Ishihara and Okayama (1994), and Moreno and Nurse (1994). Since rum1Δ and nuc2-663 strains are sterile, the ded1-1D5 rum1Δ and ded1-1D5 nuc2-663 strains were made by protoplast fusion. The cig2::HA strain was described by Martin-Castellanos et al. (1996).

Experiments in liquid culture were performed using EMM medium, and exponential cultures were grown to a density of 2-4×10⁶ cells/ml at the start of each experiment.

To induce expression of the nmt1 promoter, cells growing in the presence of 5 μg/ml thiamine were washed twice and then resuspended in fresh medium lacking thiamine, at a dilution calculated to produce a cell density of 2-4×10⁶ cells/ml at the time of peak expression from the promoter.

The ded1⁺ gene was cloned by complementation of the t.s and c.s defects of the ded1-1D5 and ded1-61 mutants, using the fission yeast genomic library described by Barbet et al. (1992), and the cDNA library generated by Bruce Edgar and Chris Norbury (Moreno and Nurse, 1994). Restriction maps of the clones isolated from the genomic library suggested the presence of the same gene in each plasmid, and the minimal complementing fragment was then sequenced on both strands. The sequence of a cDNA showed that the ded1⁺ gene lacks introns, and the cDNA sequence was submitted to the EMBL/GenBank/DDBJ database with the accession number AJ237697. In order to confirm that we had isolated the ded1⁺ gene, and not a suppressor, we subcloned the sup3-5 nonsense suppressor into a vector containing an ded1-rescuing genomic fragment, and transformed this plasmid into an ded1 mutant containing the ade6-704 allele, which is rescued by sup3-5. White colonies were selected, and integration by homologous recombination at the site corresponding to the ded1-rescuing fragment was confirmed by Southern blotting. The integrants were then back-crossed to a wild-type ade6-704 strain, and the ability of all progeny of this cross to grow at both high and low temperatures shows that the rescuing plasmid had integrated at the ded1⁺ locus.

A plasmid expressing a myc-tagged version of Ded1 was constructed by amplifying the ded1⁺ cDNA with NdeI sites at the first ATG and at the end of the gene, before cloning into pREP41MycHis (Craven et al., 1998), such that Ded1 is expressed from the medium strength nmt1 promoter with two copies of the c-myc epitope and six His residues at the N terminus. Expression of this fusion protein rescues the c.s phenotype of ded1-61, showing that the tagged Ded1 is functional.

To make pREP1 DED1 (S. cerevisiae), the budding yeast DED1 open reading frame was amplified by PCR from genomic DNA isolated from the wild-type budding yeast strain W303-1a, using oligos that introduced NdeI sites at the ATG of the DED1 ORF, and also at the 3′ end of the gene. The sequence of the oligos was as follows: 5′ oligo GTCTATCACATATGGCTGAACTGAGCGAACAAGTGCAAAATT-TAAG; 3′ oligo ATCTGAATCATATGTCACCACCAAGAAGAGT-TGTTTGAACCACCGCTG. The amplified product was then digested with NdeI and subcloned into the NdeI site of pREP1, so that expression of DED1 was placed under the control of the nmt1 promoter.

pREP3X-DBP2 was made by subcloning the XhoI-NsiI fragment containing a DBP2 cDNA from pIG69 (Iggo et al., 1991), into XhoI-SmaI digested pREP3X. pREP3X-vasa was made by subcloning an XhoI-SmaI fragment containing the vasa gene from Bluescript-vasa (Hay et al., 1988) into XhoI-SmaI digested pREP3X.

The cig2 cDNA constructs described in Fig. 7 were amplified by PCR and subcloned into vectors containing the medium strength nmt1 promoter (Basi et al., 1993). cDNA I corresponds exactly to the cig2 cDNA sequence described by Obara-Ishihara and Okayama (1994). cDNA II is identical except that it lacks the first 32 nucleotides of cDNA I. cDNA III represents the cig2 cDNA without any UTRs, and was kindly provided by Hiroyuki Yamano.

Protein extracts for western blotting were made by TCA extraction, as described previously (Foiani et al., 1994). For western blot analysis the following rabbit polyclonal antibodies were used: SP4 anti-Cdc13 at a dilution of 1/2000, MOC8 anti-Cig2 at a dilution of 1/500, anti-Cdc21 antibody at a dilution of 1/500. In addition, two mouse monoclonal antibodies were used: 100.4 anti-Cdc2 at a final concentration of 5 μg/ml (kindly provided by Hiroyuki Yamano), and anti-α-tubulin (Sigma T5168) at a dilution of 1/10000. The secondary antibodies were either Protein A-HRP or anti-mouse IgG-HRP conjugates, used at a dilution of 1/10000. Detection was performed using the enhanced chemiluminescence procedure (Amersham ECL kit).

Total RNA was prepared as described (Moreno et al., 1991) and 10 μg of each sample was resolved on a formaldehyde gel. Probes were labelled using the ‘Prime-a-Gene’ kit (Promega). An NdeI-NcoI cig2 cDNA fragment was used to detect the cig2 mRNA, and an HindIII genomic fragment of the ura4 gene was used to detect the ura4 mRNA.

For the experiments described in Figs 7C and 8, Cig2 and α-tubulin protein levels were quantitated using a phosphoimager after ECF detection (Amersham). For each sample, the Cig2 protein level was expressed relative to the level of α-tubulin, to provide an internal loading control. In order to avoid errors due to signal saturation the samples were serially diluted before immunoblot analysis.

To quantitate cig2 mRNA levels by northern blot analysis, signals were quantitated directly using a phosphoimager, and expressed relative to ura4 mRNA levels, which again served as an internal loading control.

About 10⁷ cells were spun down for each sample and fixed in 70% ethanol before storing at 4°C. Samples were processed for flow cytometry as described (Sazer and Sherwood, 1990), and analysed with a Becton-Dickinson FACScan or FACSCalibur. In order to stain cells for fluorescence microscopy with the DNA binding dye 4′,6-diamidino-2-phenylindole (DAPI), ethanol fixed cells were rehydrated in water and treated as described previously (Moreno et al., 1991).

A wild-type strain was transformed with pREP41MycHis-Ded1, and expression from the medium strength nmt1 promoter induced as described above. The localization of Myc-Ded1 was determined by immunofluorescence as described previously (Maiorano et al., 1996), using the 9E10 anti-Myc monoclonal.

In order to label newly synthesised proteins with [³⁵S]methionine, cells were grown in minimal medium to log phase before adding 1 mCi ProMix (Amersham) to 10⁸ cells in a volume of 10 ml. Incubation was continued for 40 minutes at 25°C or 24 minutes at 36°C. The labelled cells were then harvested and washed once with ice-cold STOP buffer (Moreno et al., 1991), before freezing in liquid nitrogen. Protein extracts were then prepared in HB buffer (Moreno et al., 1991).

For immunoprecipitation, 2 μl of the SP4 polyclonal antibody (anti-cdc13) or 12CA5 monoclonal antibody (Boehringer, anti-HA-Cig2) were bound to 20 μl Protein A-Sepharose beads (Pharmacia) for two hours at 4°C, before washing the beads with PBS. The beads were then incubated with 1 mg labelled extracts for one hour at 4°C, washed four times with HB buffer, and boiled in Laemmli buffer.

In a screen for novel fission yeast cell cycle genes (Grallert and Nurse, 1996) we isolated a temperature-sensitive (t.s.) mutant, ded1-1D5, with a phenotype suggesting defects in B-cyclin/Cdc2 function. Upon shifting an asynchronous culture of the ded1-1D5 mutant to 36°C, we found that cells are unable to enter S-phase and mitosis and so become arrested with a single interphase nucleus with 1C or 2C DNA contents (Fig. 1). Most cells arrest before mitosis, since G₂ is the predominant phase of the fission yeast cell cycle in nitrogen rich media, but if cells are synchronised in G₁ by nitrogen starvation, subsequent inactivation of Ded1 prevents G₁ progression upon return to nitrogen-rich medium (not shown). The G₁ arrest of the ded1-1D5 mutant is dependent upon expression of the rum1⁺ gene (Fig. 1B), just as previously reported for cdc2^t.s. mutants (Labib et al., 1995), and Ded1 function is not required to complete S-phase upon release from a hydroxyurea block (not shown). Inactivation of Ded1 therefore produces similar effects to inactivation of B-cyclin/Cdc2 kinases, blocking cell cycle progression both in G₁ phase before Start and in G₂ phase before mitosis.

In an independent screen, we isolated additional alleles of ded1 with phenotypes that again suggest defects in B-cyclin expression or function. We found that deletion of the gene encoding the B-type cyclin Cig2 specifically suppresses t.s. defects in a subset of proteins required for DNA replication (Table 2), and we reasoned that mutations inhibiting the expression or function of B-cyclins might produce a similar effect. We therefore screened for other mutations able to suppress the same DNA replication mutants – cdc18-K46, cdc19-P1 and cdc21-M68 – and isolated two new alleles of ded1 (see Fig. 2A,B; the details of the screen will be described elsewhere – K.L. and S.K., in preparation). Both of the new alleles, ded1-61 and ded1-78, support cell growth at high temperatures but cause a cold-sensitive phenotype at 20°C (Fig. 2C). This suggests that partial inactivation of Ded1 mimics the effects of cig2 deletion, while more severe inactivation of Ded1 blocks cell growth. In this case, the effects of cig2 deletion should also be reproduced when the t.s. allele, ded1-1D5, is grown at a semi-restrictive temperature. As shown in Fig. 2D, the lethality of the cdc19-P1 mutation at 34.5°C is suppressed in the presence of the ded1-1D5 mutation. Therefore, partial inactivation of the Ded1-1D5 protein at 34.5°C produces a similar phenotype to deletion of the cig2 gene. In addition, a synthetic lethal phenotype is produced when either the t.s. or c.s. alleles of ded1 are combined with cdc2-33 (not shown). Taken together, these experiments show that mutation of ded1 produces similar phenotypes to inactivation of cyclin B/cdc2 kinase activity.

Genetic analysis showed that the ded1 gene is unlinked to any previously characterised regulator of B-cyclin/Cdc2 function, and we considered the possibility that mutation of Ded1 may affect the expression of B-cyclins or of Cdc2. An asynchronous culture of the ded1-1D5 strain was grown at 25°C together with a wild-type control, and samples were taken after shifting to 36°C, in order to prepare protein extracts. As shown in Fig. 3A, immunoblot analysis revealed that although the levels of Cdc2, Cdc21/Mcm4 and α-tubulin were unchanged at 36°C, the levels of both the Cig2 and Cdc13 B-type cyclins fell dramatically. Both cig2 and cdc13 mRNA levels were unchanged at 36°C (not shown). Importantly, the decrease in B-cyclin levels did not require degradation by the APC, since Cdc13 and Cig2 still disappear when a nuc2-663 ded1-1D5 double mutant is shifted to 36°C (Fig. 3B), in conditions where the APC is inactivated (Goebl and Yanagida, 1991; Hirano et al., 1988). The disappearance of Cig2, which is normally only expressed around G₁/S, is not simply a consequence of cell cycle arrest in G₁ and G₂, since a similar decrease is not seen when a cdc2-M26 t.s. mutant is shifted to 36°C (Fig. 3C). In addition, Ded1 inactivation does not cause a dramatic reduction in the level of Cdc18 (Fig. 3A), another highly unstable protein that is expressed around G₁/S. Our results suggest that B-cyclin expression at some post-transcriptional step is particularly sensitive to inactivation of Ded1.

We cloned the ded1⁺ gene by complementation of both the t.s. and c.s. alleles with fission yeast cDNA and genomic libraries. Conceptual translation of the ded1 ORF predicts a protein containing all the conserved motifs typical of RNA helicases (Schmid and Linder, 1992). During the course of this work, the same sequence was published as sum3, a high-copy suppressor of specific checkpoint and stress responses (Forbes et al., 1998), moc2, a high-copy suppressor of sterility (Kawamukai, 1999), and the same gene was also identified in a third screen (Hsing-Yin Liu and Nancy Walworth, unpublished data).

Having consulted with these groups we have agreed to adopt the name ded1, since the encoded protein is 60% identical to the budding yeast Ded1 protein (Forbes et al., 1998). It has recently been shown that Saccharomyces cerevisiae Ded1 has RNA helicase activity in vitro (Iost et al., 1999) and a general role in the initiation of translation in vivo (Chuang et al., 1997; De La Cruz et al., 1997). Budding yeast and fission yeast Ded1 proteins are also very similar to other cytoplasmic RNA helicases required for translation, such as the Vasa/PL10 subfamily of RNA helicases, essential for germ cell formation in a wide variety of animals (Gee and Conboy, 1994; Gruidl et al., 1996; Hay et al., 1988; Leroy et al., 1989).

We found that expression of either the vasa or budding yeast DED1 genes can rescue the t.s. phenotype of the fission yeast ded1-1D5 mutant, in contrast to expression of the DBP2 gene, which encodes a budding yeast member of the p68 family of nuclear RNA helicases (Fig. 4A). This shows that ded1⁺, vasa and DED1 encode functionally homologous proteins, and strongly suggests that fission yeast Ded1 is also an RNA helicase with a role in translation. Consistent with this view, we find that the Ded1 protein is cytoplasmic and appears to be excluded from the nucleus (Fig. 4B).

To investigate the efficiency of protein synthesis in ded1 mutants, we examined the incorporation of [³⁵S]methionine into newly synthesised proteins during a short pulse given at different times after shifting to the restrictive temperature. General protein synthesis is profoundly inhibited when an asynchronous culture of the cold-sensitive ded1-61 mutant is shifted from the permissive temperature of 32°C to the restrictive temperature of 20°C (Fig. 5A). A fivefold reduction in general translation is also seen when the t.s. ded1-1D5 mutant is shifted from 25°C to 36°C (Fig. 5B). This indicates that fission yeast Ded1 is a general translation factor analogous to Ded1 in budding yeast.

Our data suggest that synthesis of the B-type cyclins Cig2 and Cdc13 is particularly sensitive to the efficiency of the translational machinery. This may explain why partial inactivation of Ded1 produces a similar effect to deletion of the cig2 gene. To test this idea, we compared the translation of Cig2 and Cdc13 with that of other proteins, under conditions where partial inactivation of Ded1 suppresses defects in the initiation step of DNA replication. The ded1-1D5 mutant was grown at 25°C in parallel with a wild-type strain, and both cultures were then shifted to 34.5°C, since at this temperature the lethality of the cdc19-P1 mutant is suppressed (Fig. 2D). At both temperatures, cells were given a short pulse of [³⁵S]methionine. Cell extracts were made, and the incorporation of radiolabel into immunoprecipitates of Cig2 and Cdc13 was compared with incorporation into total protein. Fig. 6 shows that general protein synthesis increased almost 2-fold in both the wild-type and ded1-1D5 strains at 34.5°C. In addition, the translation of Cig2 and Cdc13 was not impaired when the wild-type strain was shifted from 25°C to 34.5°C. However, incorporation of [³⁵S]methionine into Cig2 was reduced by 60% in the ded1-1D5 mutant at 34.5°C, while incorporation into Cdc13 was reduced by 40%. This experiment shows that translation of Cig2, and to a lesser extent Cdc13, is especially sensitive to mutation of Ded1, in comparison to the majority of other proteins.

If the ability of ded1 mutants to suppress defects in the initiation step of DNA replication is indeed due to reduced translation of Cig2, increased expression of Cig2 should prevent ded1 mutants from suppressing such defects. It has previously been shown, by 5′ and 3′ RACE PCR, that the cig2⁺ mRNA has large 5′ and 3′ UTRs, each of around 1 kb (Obara-Ishihara and Okayama, 1994). Such UTRs often contain elements that limit the efficiency of translation. Although the full sequence of the cig2 UTRs has yet to be described, the sequence of a cig2 cDNA containing shorter UTRs has been reported (Obara-Ishihara and Okayama, 1994). We compared this cDNA (cDNA I in Fig. 7A) with two shorter versions, either lacking a short upstream open reading frame in the the 5′ UTR (cDNA II in Fig. 7A), or lacking all cig2 UTR sequences (cDNA III in Fig. 7A), and asked whether expression of these cDNAs prevented suppression of cdc21-M68 and cdc18-K46 by mutation of ded1. As shown in Fig. 7B, suppression of cdc21-M68 at 35°C by ded1-78 was partially blocked by expression of cig2 cDNA I, and more completely inhibited by expression of cDNAs II and III. A similar result was obtained when we examined the ability of ded1 c.s. alleles to suppress cdc18-K46 in the presence of the cig2 cDNAs (not shown). Therefore, reduced expression of Cig2 is likely to be an important reason why defects in the initiation step of DNA replication are suppressed by partial inactivation of Ded1.

Expression of cDNA III, lacking all cig2 UTR sequences, produces a 4-fold increase in the Cig2 protein/mRNA ratio, relative to cDNA I (Fig. 7C). This is unlikely to be due to saturation of Cig2 degradation by the APC, since the level of Cdc13 does not increase correspondingly (Fig. 7D). This suggests that the efficiency of Cig2 translation is increased by removal of the UTRs. Consistent with this view, the decrease in Cig2 protein levels upon inactivation of Ded1 at 36°C is progressively reduced as the UTR sequences are removed from the cig2 cDNA (Fig. 8).

The sensitivity of Cig2 protein levels to perturbations of the translation machinery led us to consider whether Cig2 expression is regulated at the level of translation during the fission yeast life cycle. A variety of evidence suggests that translational regulation of the cig2 mRNA might play a role under conditions of nitrogen starvation, when the fission yeast cell switches from the mitotic cell cycle to conjugation and the meiotic cycle. In contrast to many other proteins such as Cdc2, both Cig2 and Cdc13 disappear in starving cells (Wu and Russell, 1997; Yamaguchi et al., 1997; see Fig. 9A), and this is important for the cell to arrest in G₁ as a prelude to conjugation (Kominami et al., 1998; Martin-Castellanos et al., 1996; Obara-Ishihara and Okayama, 1994; Yamaguchi et al., 1997). Although Cig2 protein disappears upon nitrogen starvation of wild-type cells, expression of the cig2⁺ mRNA is strongly induced (Obara-Ishihara and Okayama, 1994). This raises the possibility that proteolysis may not be sufficient to explain the timely decrease in the level of Cig2 protein as cells switch from the mitotic cell cycle to conjugation and meiosis. In order to test whether translation of B-cyclins is inhibited upon nitrogen starvation, we exposed cells to a short pulse of [³⁵S]methionine at different times after transfer into medium lacking a nitrogen source. General protein synthesis was compared with de novo synthesis of Cig2 and Cdc13, by SDS-PAGE analysis of total protein samples together with immunoprecipitates of Cig2 and Cdc13, followed by phosphoimager detection of the incorporation of radiolabel into newly made proteins. The rate of total [³⁵S]methionine incorporation increased somewhat upon nitrogen starvation (Fig. 9B,C), probably due to a depletion of the cellular methionine pools. Nevertheless, incorporation of [³⁵S]methionine into Cig2 and Cdc13 decreases between 2 and 4 hours after shifting to nitrogen-free medium (Fig. 9D), despite the general increase seen for the majority of proteins. This shows that the translation of the Cig2 and Cdc13 B-type cyclins is indeed impaired upon nitrogen starvation. Inhibition of B-cyclin translation occurs with similar kinetics to the disappearance of Cdc13 and Cig2 proteins (Fig. 9A) and is therefore likely to contribute to their timely disappearance as cells exit the mitotic cell cycle. Moreover, the level of Cig2 protein remains high upon starvation if the cig2 gene is expressed in the absence of the native UTRs (Fig. 10A). This is not due to saturation of APC mediated proteolysis, since Cdc13 disappears on schedule (Fig. 10A). Failure to downregulate the level of Cig2 protein is associated with a 10-fold inhibition of conjugation and meiosis (Fig. 10B; Obara-Ishihara and Okayama (1994) also reported a three fold reduction in conjugation efficiency upon increased Cig2 expression), emphasising the importance of the inhibition of Cig2 function for sexual differentiation in fission yeast.

We have identified a functional homologue of the budding yeast Ded1 RNA helicase, which is required for general translation together with the eIF4A RNA helicase (Chuang et al., 1997; De La Cruz et al., 1997). Fission yeast Ded1 is also required for protein synthesis and our ded1 alleles are likely to provide useful experimental tools, since fission yeast strains allowing rapid inactivation of the translation machinery have not previously been isolated. Although the precise role of Ded1 in translation remains to be determined, it is likely to be involved in the initiation step, since budding yeast ded1 mutants show reduced levels of polysomes (Chuang et al., 1997; De La Cruz et al., 1997) and overexpression of the DED1 gene rescues the lethality associated with defects in either the cap-binding protein Cdc33, or in the RNA helicase Tif1/eIF4A (De La Cruz et al., 1997); for general reviews of translational initiation see (Gray and Wickens, 1998; McCarthy, 1998; Sachs et al., 1997). Defects in fission yeast Ded1 function can also be rescued by expression of the vasa gene, encoding a highly related RNA helicase from Drosophila (Hay et al., 1988; Liang et al., 1994). Vasa is required in vivo to overcome translational repression of genes such as nanos (Gavis et al., 1996). This inhibition is mediated by a motif in the 3′UTR of the nanos mRNA, that is predicted to have a complex secondary structure (Gavis et al., 1996). It is possible that Ded1 homologues in budding and fission yeasts may also be required to overcome the inhibitory effects of such motifs upon translation.

The ded1⁺ gene was previously identified as a multicopy suppressor of a checkpoint defect (Forbes et al., 1998), or of sterility (Kawamukai, 1999), though a role for Ded1 in translation was not demonstrated in these studies. Overexpression of ded1⁺ (called sum3⁺ in Forbes et al., 1998) was found to prevent cells with high levels of the Cdc25 tyrosine phosphatase from entering mitosis before the completion of S-phase. Elevated expression of ded1⁺ also prevented the advancement of mitosis that normally results from osmotic shock, without affecting transcription of genes induced by the stress-specific MAP kinase pathway (Forbes et al., 1998). Overexpression of a second putative translation factor produces the same effect (Humphrey and Enoch, 1998). Although it is not clear how these effects are produced, it will be interesting to see whether translation is regulated by the checkpoints that link cell cycle progression to genomic integrity.

We have shown that levels of the B-cyclins Cdc13 and Cig2 are greatly reduced upon inactivation of Ded1, even if the APC is inactivated at the same time. Another unstable protein, Cdc18, is not affected to the same degree. While Ded1 appears to be a general translation factor, our data indicate that the synthesis of Cig2 and Cdc13 is particularly sensitive to a reduction in the general efficiency of translation. For Cig2, our data suggest that elements in the UTRs of the cig2 mRNA may limit the efficiency of Cig2 translation, and it is tempting to speculate that Ded1 may be required to overcome the inhibitory effects of such sequences. We are currently characterising the UTRs of the cig2 mRNA in more detail, in order to define which elements have a role in regulating Cig2 translation. Regarding Cdc13, we note that a recent study has shown that the 5′ UTR of the cdc13 mRNA is predicted to have a complex structure, analogous to that of the 5′UTR of the cdc25 mRNA, which may again serve to limit the efficiency of translation (Daga and Jimenez, 1999). This study also showed that the levels of both Cdc25 and Cdc13 are sensitive to a modest reduction in the efficiency of translation initiation activity.

There is no evidence that the rate of translation of Cig2 and Cdc13 changes through the cell cycle in proliferating cells. Indeed, it has previously been shown that the rate of Cdc13 translation is similar in G₁ and G₂ phases (Correa-Bordes et al., 1997). Cig2 protein and mRNA levels peak around G₁/S (Connolly and Beach, 1994; Martin-Castellanos et al., 1996; Mondesert et al., 1996; Obara-Ishihara and Okayama, 1994), and regulated transcription and proteolysis may be sufficient to explain these changes in cycling cells. However, the translation of both Cdc13 and Cig2 is inhibited upon nitrogen starvation (Fig. 9), when the disappearance of B-cyclins is crucial for exit from the mitotic cell cycle and entry into conjugation and meiosis. This identifies a third mechanism by which Cdc2/cyclin B kinase activity is regulated during this important cell cycle transition, in addition to cyclin B proteolysis (Kominami et al., 1998; Kumada et al., 1995; Stern and Nurse, 1998), and CDK-inhibition by Rum1 (Correa-Bordes and Nurse, 1995; Kominami et al., 1998; Moreno and Nurse, 1994; Stern and Nurse, 1998). The mechanism by which B-cyclin translation is inhibited upon nitrogen starvation remains to be determined, although we find that removal of the cig2 UTRs prevents the reduction in Cig2 protein levels upon starvation (Fig. 10). We do not know whether regulation of Ded1 plays any role in this process, and we find that the translation of most other proteins is less affected in starving cells. However, it is interesting to note that the level of Ded1 is reduced upon nitrogen starvation (H.-Y. Liu and N. Walworth, personal communication), and this may impair the translation of mRNAs that are particularly sensitive to Ded1 function. It is also possible that translational regulation of Cig2 expression may involve multiple mechanisms, since two cig2 mRNAs are induced upon nitrogen starvation, and one of these lacks a poly(A) tail (Obara-Ishihara and Okayama, 1994). Poly(A)⁻ mRNAs are poorly translated, and subsequent polyadenylation at some point during meiosis, or in the germinating spore, could provide a means of rapidly making new Cig2 protein, as an efficient trigger to the onset of S-phase. This would then be analogous to translational regulation of certain mRNAs by polyadenylation during Xenopus and Drosophila early development (Salles et al., 1994; Wickens, 1990).

Translational regulation of B-type cyclins has been shown to be important in higher eukaryotes for both cell cycle control and development. Cyclin B mRNA is concentrated in the posterior pole of the Drosophila oocyte, and becomes incorporated into the pole cells that are the progenitors of the germ-line, but is not translated until around the time that mitosis resumes in the developing gonad (Dalby and Glover, 1992). In many animals, translation of B-type cyclins is stimulated after germinal vesicle breakdown during oocyte maturation, and it has been shown in starfish that this is important for the repression of DNA replication between the two meiotic divisions (Picard et al., 1996). Translational regulation of D-type cyclins may also be important for cell cycle control, when mammalian cells exit G₀ and resume proliferation. Overexpression of the cap-binding protein eIF4E can lead to oncogenic transformation of mammalian fibroblasts (Lazaris-Karatzas et al., 1990); reviewed by Hentze (1995), and this may be related to increased translation of specific mRNAs such as cyclin D1 (Koromilas et al., 1992; Rosenwald et al., 1993). It will be of considerable interest to establish the extent to which translational control provides a general linkage between the regulation of cell growth and division.

We thank Lilian Linbergsengen for her help with 2-D gel electrophoresis, and are grateful to Tony Carr, Richard Iggo, D. J. Jamieson, Thomas A. Jongens, Cristina Martín-Castellanos, Chris Norbury and Hiroyuki Yamano for plasmids. We thank Sergio Moreno, Hiroto Okayama and Takashi Toda for strains; and Jaime Correa-Bordes, Sergio Moreno, Bodo Stern and Hiroyuki Yamano for generously providing antibodies. We thank Tamar Enoch, Jacky Hayles, Tim Humphrey, Makoto Kawamukai and Nancy Walworth for communicating unpublished data, Domenico Maiorano for helpful discussions, and Hiroyuki Yamano for critically reading the manuscript. Particular thanks go to John Diffley for allowing K.L. to finish experiments relating to this work in his laboratory. This work was funded by the Norwegian Cancer Society, the Cancer Research Campaign, the Wellcome Trust, the European Union (contract ERB-MRX-CT970125) and the Imperial Cancer Research Fund.