Overexpression of eIF4E in Saccharomyces cerevisiae causes slow growth and decreased alpha-factor response through alterations in CLN3 expression.

The association of G(1) cyclins and Cdc28/cyclin-dependent protein kinase catalyzes the cell cycle entry (Start) in budding yeast. Activation of Start is presumed to be triggered by a post-transcriptional increase in Cln3 during early G(1). Cells arrested by mating pheromone show a loss of cyclin-dependent protein kinase activity caused by transcriptional shutoff of cyclins and/or inhibition by Far1. We report that overexpression of eIF4E (Cdc33), a rate-limiting translation initiation factor, causes an increase in CLN3 mRNA translation, which results in increased expression of CLN2 and in slow growth and decreased alpha-factor response. This phenotype was abrogated in a Deltacln3 or Deltacln2 background. We isolated the transcription factor MBP1 as a multicopy suppressor of the growth and alpha-factor response defects. Furthermore, elevated MBP1, a transcriptional regulator of cyclins, altered the transcriptional start site in CLN3 mRNA, shifting it 45 nucleotides upstream of the normal. This lengthened 5'-untranslated region likely reduces translation efficiency and down-regulates CLN3 protein synthesis, thereby correcting for the excess translation promoted by elevated Cdc33. In addition, the CLN2 mRNA level returned to normal. We propose that regulation of translation initiation by Cdc33 plays a pivotal role in the activation of Start and cell cycle progression in budding yeast.

The association of G 1 cyclins and Cdc28/cyclindependent protein kinase catalyzes the cell cycle entry (Start) in budding yeast. Activation of Start is presumed to be triggered by a post-transcriptional increase in Cln3 during early G 1 . Cells arrested by mating pheromone show a loss of cyclin-dependent protein kinase activity caused by transcriptional shutoff of cyclins and/or inhibition by Far1. We report that overexpression of eIF4E (Cdc33), a rate-limiting translation initiation factor, causes an increase in CLN3 mRNA translation, which results in increased expression of CLN2 and in slow growth and decreased ␣-factor response. This phenotype was abrogated in a ⌬cln3 or ⌬cln2 background. We isolated the transcription factor MBP1 as a multicopy suppressor of the growth and ␣-factor response defects. Furthermore, elevated MBP1, a transcriptional regulator of cyclins, altered the transcriptional start site in CLN3 mRNA, shifting it 45 nucleotides upstream of the normal. This lengthened 5-untranslated region likely reduces translation efficiency and down-regulates CLN3 protein synthesis, thereby correcting for the excess translation promoted by elevated Cdc33. In addition, the CLN2 mRNA level returned to normal. We propose that regulation of translation initiation by Cdc33 plays a pivotal role in the activation of Start and cell cycle progression in budding yeast.
The translation initiation factor eIF4E 1 recognizes the 7-methylguanosine-containing cap of mRNA in the first step of mRNA recruitment for translation (1). eIF4E functions as a subunit of eIF4F, the complex that unwinds secondary structure at the 5Ј-UTR of mRNAs in an ATP-dependent reaction.
This function is likely needed during scanning for exposing and locating the translation start site (2)(3)(4). The low abundance of eIF4E/F appears to be limiting for the translation of some mRNAs, particularly those with structured 5Ј-UTRs or upstream open reading frames (5)(6)(7). Overexpression of eIF4E in rodent and human cells causes deregulated cell growth and malignant transformation (8 -10). Furthermore, eIF4E is elevated in several common human malignancies such as breast and head and neck carcinomas (11,12). Conversely, reducing the level of eIF4E by antisense RNA inhibits the malignant properties of several carcinoma cell lines (13,14). One explanation for eIF4E-mediated transformation is that the normally repressed translation of some cell cycle regulators, proto-oncogenes, or growth factor transcripts becomes preferentially increased by excess eIF4E. In support of this hypothesis, elevated eIF4E increases the translation of c-myc (10), cyclin D1 (15), FGF2 (16), VEGF (17), ODC (18 -20), and RNR2 (21) in mammalian cells.
In Saccharomyces cerevisiae, eIF4E was identified as a cell division cycle gene (CDC33) essential for viability (22,23). Evidence that CDC33 is involved in the G 1 to S transition of the cell cycle came from the isolation of a temperature-sensitive strain, cdc33-1, that arrests with unbudded cells at the nonpermissive temperature (22,24). CDC33 may also play a regulatory function during DNA synthesis by regulating the synthesis of RNR2, the small subunit of ribonucleotide reductase (21). Therefore, proper levels of Cdc33 seem to play a critical role in regulating cell cycle progression through G 1 and during S phase in budding yeast.
Yeast cells must pass through a specific point in mid-G 1 , called Start, before committing to cell division. This G 1 to S transition is characterized by the appearance of the bud and is analogous to the restriction point in mammalian cells (25)(26)(27). At Start, cells must integrate both external environmental signals (nutrients and pheromones) and internal signals (size) to determine whether to commit to mitosis, differentiation, or stationary phase. Once a cell passes Start, it is committed to three major events: DNA replication, bud emergence, and duplication of spindle pole bodies. Time spent in G 1 and the passage through Start is regulated by interaction of the cyclin-dependent protein kinase Cdc28 with the early G 1 cyclins, Cln1, Cln2, and Cln3 (28,29). Cln3/Cdc28 is thought to be the earliest activator of Start, and its activity triggers the transcriptional activation of CLN1, CLN2, and S phase-specific genes like RNR1, RNR2, CLB5, and CLB6 (28, 30 -32). CLN3 is expressed throughout the cell cycle with a transient post-transcriptional increase in early G 1 (30). Therefore, CLN3 is a good candidate for coordinating the control of early cell cycle events, likely through translational control and subsequent degradation.
Cln3-dependent activation of downstream genes depends on two transcriptional complexes, SBF (composed of Swi4/Swi6) and MBF (composed of Mbp1/Swi6). Swi4 and Mbp1 provide the DNA recognition function, while Swi6 enhances their binding (33)(34)(35)(36)(37). SBF binds to SCB sequences (Swi4/Swi6 cell cycle box, CACGAAA) and is thought to activate the G 1 cyclin genes CLN1, CLN2, PCL1, PCL2, and HO endonuclease (38,39). The MBF complex activates through an MCB sequence (Mul1 cell cycle box, ACGCGTNA). This sequence is found in the promoter of many genes, including CLB5, CLB6, SWI4, and the DNA synthesis genes POL1, TMP1, RNR1, and RNR2 (28, 32, 35, 37, 40 -42). MCB binding activity is periodic, peaking at G 1 /S (35), but unlike the activation of CLN genes, the mechanism linking the onset of Start with periodic induction of MCB-responsive genes has not been resolved (37). Moreover, there is some overlap in the transcriptional activation of SCB-containing genes by Mbp1 and of MCB-containing genes by Swi4. It has been shown that Mbp1 binds to SCB elements, and binding of Swi4 to SCB is efficiently competed by MCB oligonucleotides (35,40). We present evidence that CLN3 is translationally up-regulated by elevated Cdc33, and we propose a role for Cdc33 in the activation of Start. In addition, we present evidence that Mbp1 is a suppressor of the phenotype caused by overexpression of Cdc33.

MATERIALS AND METHODS
Yeast Strains and Plasmids-Strains used in this study are listed in Table I. Three plasmid constructs were obtained for CDC33 overexpression. Plasmid Yep24 contains a ClaI genomic fragment that encompasses the CDC33 ORF with 639 bases of 5Ј-and 1668 bases of 3Јflanking sequence. A PstI-HindIII digest of the Yep24 vector generated a fragment that includes the ORF with 619 bases 5Ј and 522 bases 3Ј of the flanking sequence. This fragment was cloned into the Yep181 plasmid vector. pYes/GS/CDC33, which contains the eIF4E ORF only under the control of a GAL1 promoter, was purchased from Invitrogen.
The strains 366, 785, and 1036 (⌬cln1, ⌬cln2, and ⌬cln3) in the strain 4741 (Research Genetics) were donated by Dr. Neal Mathias. Strains BVB367, BVB786, and BVB1037 were constructed using the empty pYes/GS vector. Strains BVB368, BVB787, BVB1038 contain the pYes/ GS/CDC33 vector. The plasmid ycplac33, containing the full-length MBP1 (41), was a gift from Noel Lowndes and was used for the construction of plasmid Yep425/MBP1. A HindIII-SpeI fragment containing the MBP1 ORF and promoter sequence was ligated at the same sites into the 2 plasmid pRS425.
Growth Rates of CDC33-overexpressing Strains-Cells carrying either s (BVB311) or pYes/GS/CDC33 plasmids (BVB312) were grown to early log phase at the appropriate temperature in either liquid culture medium or solid medium lacking uracil and containing 2% raffinose. 2% galactose was added to liquid cultures where indicated to induce overexpression of eIF4E. Doubling time was determined by direct counting and by spectrophotometry. For some experiments, it took ϳ96 h to reach the desired density, although in most cases the cells were grown to an OD of 1 in raffinose and subsequently shifted to galactose for three or four generations.
Suppressors of CDC33 Overexpression-BVB312 cells were transformed with the Yep13 vector or a high copy number Yep13 library containing random yeast chromosomal DNA (43) by the lithium acetate/ polyethylene glycol method and plated on medium containing 2% raffinose and 2% galactose lacking uracil and leucine. Cells were grown at 16°C to score for suppressors of the slow growth phenotype from CDC33 overexpression. In the first screen, ϳ4 ϫ 10 3 transformants that grew on Leu Ϫ Ura Ϫ plates were replica-plated on 2% galactose at 16°C. Plasmids were isolated from 5-ml cultures using Spheroplast Buffer (1 M sorbitol, 0.1 M sodium acetate, 6 mM EDTA, pH 7.0), 3 mg of zymolase, and 40 l of ␤-mercaptoethanol. Cells were then disrupted in lysis buffer containing 2% SDS, 100 mM Tris-HCl, and 100 mM EDTA, and plasmid DNA was precipitated with EtOH. Plasmids were amplified in Escherichia coli strain KC8 by electroporation. Colonies were selected for growth on M9 plates lacking leucine for selection of the plasmid. Plasmid DNA was then sequenced using forward primer CACTATC-GACTACGCGATCA and reverse primer ATGCGTCCGGCGTAGA in the multiple cloning site. Sequencing was performed at Iowa State University DNA Sequencing Facility.
Mating Factor Sensitivity Assays-For halo assays, exponentially growing cells were spread on synthetic medium containing 2% raffinose/galactose, and paper disks containing the indicated amounts of ␣-factor were placed on the surface. Plates were incubated for 2-3 days at 30°C.
Northern Analysis-Total RNA was extracted by the hot phenol method (43). 10 -30 g of RNA was denatured with glyoxal, run on a 1.5% agarose gel, and transferred to a membrane (Immobilon-Nyϩ, Millipore). The RNA was immobilized on the membrane by UV light. Subsequently membranes were prehybridized for 4 h at 45°C in 7% SDS, 0.1 mg/ml salmon sperm DNA, 2.5 M Na 2 HP0 4 , pH 7.2, 10 mM EDTA and then hybridized overnight at 55-65°C with the appropriate riboprobe. Blots were washed with 2ϫ SSC (300 mM NaCl, 30 mM sodium-citrate), 0.1% SDS at room temperature. Probes were constructed by polymerase chain reaction using the primers listed in Table  II, which included a T7 promoter sequence on the antisense strand. Probes were synthesized using T7 polymerase and by incorporating [␣-P 32 ]UTP. The HSC82 probe was made by single strand polymerase chain reaction amplification using [␣-P 32 ]dATP. The membranes were exposed onto a developing screen for 16 -24 h and scanned using a Molecular Dynamics (Sunnyvale, CA) PhosphorImager. The transcripts were quantified with ImageQuaNT (version 4.0) software, and all bands were normalized to the HSC82 transcript.
Polysome Isolation-Cells were briefly treated with cycloheximide at a final concentration of 150 g/ml to freeze the ribosomes. Cells were crushed with glass beads in ice-cold homogenization buffer (10 mM Tris, pH 7.4, 100 mM NaCl, 30 mM MgCl 2 , 50 g/ml cycloheximide, 200 g/ml heparin in diethyl pyrocarbonate-treated H 2 O), and debris were removed by centrifugation at 30,000 ϫ g at 4°C for 20 min. Polysomes pYes2/GS/CDC33 ura3⌬, leu2⌬, his3⌬1, met1-500 eIF4E and ␣-Factor Insensitivity were fractionated by sucrose gradient sedimentation at 100,000 ϫ g for either 1 or 4 h to pellet heavy polysomes (P100) as described previously (21). Fractionation of the gradient and filtration through a slot-blotter were done as described previously (21). Westerns-Protein samples were prepared by collecting cells from early logarithmic growing cultures. Cells were washed once, and cold trichloroacetic acid to a final concentration of 10% was added. 0.5 ml of glass beads was added, and the suspension was vortexed for 7 min and centrifuged for 1 min at 10,000 ϫ g. The pellet was washed four times with acetone and then twice with 100% ethanol. Pellets were dried, and protein was dissolved in 50 l of 5 M urea, boiled for 2 min, and then supplemented with 250 l of 2% SDS, 125 mM Tris-HCI, pH 6.8. 40 -60 g of protein from each sample were separated on a 12% SDS gel and transferred by semidry transfer (Trans-Blot SD, Bio-Rad) to nitrocellulose membranes. Membranes were probed with the appropriate primary and horseradish peroxidase-conjugated secondary antibodies, and detection was with 3,3Ј-diaminobenzidine or ECL (Amersham Pharmacia Biotech).
Primer Extension-Total RNA was extracted as described in "Materials and Methods." Poly(A) ϩ RNA was prepared with the Oligotex mRNA isolation kit (Qiagen, Valencia, CA) with 400 g of total RNA for each sample. cDNA was synthesized using Superscript II Reverse Transcriptase (Life Technologies, Inc.). A 20-base primer for the CLN3 promoter (Ϫ320 to Ϫ301; 5Ј-TCCTCAGAAATCCATTTGAC-3Ј) was 5Јend-labeled with [␥-P 32 ]dATP to a specific activity of 10 6 cpm/pmol. 250 ng of mRNA, 500 M dNTPs, 2 units of RNase Out (Life Technologies, Inc.), 20 M dithiothreitol, and 2 pmol of labeled primer were incubated at 42°C for 1 h in a reaction volume of 25 l. The cDNA product was ethanol-precipitated, denatured, and analyzed on a 4% polyacrylamide sequencing gel.
Fluorescence Microscopy, Fluorescence-activated Cell Sorting Analysis, and Budding Index-Cellular DNA was visualized by staining with 4,6-diamidino-2-phenylindole. Microscopic visualization was performed on an Olympus (Provis model Ax70TRF) microscope. Budding analysis was performed by scoring a minimum of 10 3 cells. Fluorescence-activated cell sorting analysis was performed on yeast cells stained with propidium iodide as described previously (44). Cells were harvested, washed, sonicated, and fixed overnight in 70% ethanol at 4°C and then treated with RNase A for 1 h at 50°C before staining and analysis.

CDC33 Overexpression Causes a Slow Growth
Phenotype-To study the possible link between protein synthesis rates and cell cycle regulation, we examined the phenotype of cells overexpressing eIF4E/Cdc33, a factor that is limiting for the translation initiation of some mRNAs (45). Three different CDC33 plasmids, including a galactose-inducible vector (pYes/ GS/CDC33), were constructed and introduced into the strain BVB306. We confirmed that all three CDC33 vectors expressed a functional Cdc33 protein since they rescued growth at 37°C of the temperature-sensitive strain cdc33-1 (data not shown). Strain BVB310 harbors Yep24, BVB308 contains Yep181, and strain BVB312 contains pYes/GS/CDC33. The expectation was that the different plasmids would result in different levels of CDC33 expression. The BVB312 strain grew substantially slower than the vector control strain (BVB311) at all temperatures but particularly at 16°C when grown on galactose (Fig.  1, A and C). At 16°C the BVB312 doubling time was about 600 versus 190 min for the control, BVB311; an example of liquid culture growth at 30°C is also shown (Fig. 1C, lower panel). No growth inhibition was observed under repressing conditions on dextrose (not shown). Similar but smaller differences in growth rates were seen with the BVB308 and BVB310 strains (210 and 230 min doubling time at 16°C, respectively). We examined the level of Cdc33 protein in the overexpressing strains by Western blot (Fig. 1B). The level of overexpressed CDC33 varied between 3-and 10-fold with the highest level found with the galactose-inducible vector (Fig. 1B, strain BVB312). Note that in this strain the overexpressed Cdc33 migrated more slowly than the endogenous protein due to the presence of the inframe tags. The endogenous Cdc33 also appeared elevated possibly due to a stabilization phenomenon from the exogenous protein. We conclude from this that the slow growth phenotype is dose-dependent with higher levels of eIF4E protein resulting in a more dramatic phenotypic difference.
Since BVB312 gave the strongest phenotype and the effect is inducible, we used it throughout the rest of our work. We hypothesized that the slow growth phenotype elicited by CDC33 overexpression is due to untimely translation of messages that are normally repressed and encode cell cycle regulators.
Overexpression of CDC33 Results in Accumulation of Cells in S Phase and a Loss of ␣-Factor Sensitivity-At Start the cell must attain a critical rate of protein synthesis before committing to DNA replication. Cells that have abnormally elevated translation rates, due to overexpressed Cdc33, may enter earlier in the cycle. We tested the effect of overexpression of CDC33 on the cell cycle by comparing the DNA content in BVB311 and BVB312 cells grown in the presence of galactose using flow cytometry (Fig. 2A). The BVB311 strain showed a normal distribution of G 1 and G 2 /M cells with 1N and 2N DNA content. The BVB312 strain showed a reduced fraction of 1N and 2N cells and increased distribution of cells with intermediate DNA content, indicative of cells accumulating in S phase. This increase in the S phase population was confirmed by comparing it with the distribution of cells arrested in S phase with hydroxyurea ( Fig. 2A, HU). Other abnormalities in the cell cycle were observed in the budding pattern (Table III). Cells from early log phase cultures did not differ in the number of budded cells. As expected, the number of budded cells decreased as the BVB311 culture approached stationary phase. However, a substantial number of cells overexpressing CDC33 had buds at late log phase, indicating that many cells proceeded past Start even at high saturation density. BVB312 cells in early to mid-log phase also had a number of cells that contained multiple buds (Table III) much like the wee phenotype (46). Interestingly, a wee phenotype was obtained by increasing the stability of Cln1, -2, and -3 proteins through deletion of "PEST" motifs (46). The BVB311 strain did not present multiple buds. Cells were also examined for viability by trypan blue staining. The BVB312 strain did not differ from BVB311 in the fraction of dead cells as the cultures approached stationary phase. The results from the budding index and fluorescence-activated cell sorting distribution suggest that overexpression of CDC33 causes a premature entry of cells in Start, resulting in subsequent cell replication anomalies. Abnormal regulation of CLN expression seemed the likely culprit.
Altered Response to ␣-Factor-If CLN genes are elevated due to overexpressed CDC33, then these cells may have a reduced response to the mating pheromone. Therefore, equal numbers of wild type and CDC33-overexpressing cells were spread onto galactose plates, and disks with serially diluted amounts of ␣-factor were added to the plates. Results shown in Fig. 2B demonstrated that cells overexpressing CDC33 have a 10-fold decreased response (growth arrest) to ␣-factor. Activation of the mating signal transduction pathway interferes with the execution of Start, whereas increased expression of any of the cyclins promotes cell division even in the presence of ␣-factor. The decrease in ␣-factor sensitivity in BVB312 may therefore represent an increase in expression of one or more of the G 1 cyclins.
Loss of CLN2 or CLN3 Restores Normal Growth and ␣-Factor Sensitivity-It seemed possible that altered CLN3 expression could result in the altered phenotype since it is an activator of Start, and CLN3 was previously proposed to be regulated translationally (47). To determine whether the phenotype from CDC33 overexpression requires Cln1, -2, and -3, we examined the CDC33 overexpression in knockouts for each gene. Isogenic strains containing knockouts for the early cyclins Cln1, -2, or -3 were transformed with plasmid pYes/GS/CDC33. Halo assays were carried out as before (Fig. 2C). Deletion of any of the Cln proteins restored ␣-factor sensitivity in cells overexpressing CDCD33, suggesting that the pheromone insensitivity is mediated through changes in the expression of the Cln proteins. FIG. 1. A, the overexpression of CDC33 results in slow growth. Wild type (BVB306) cells were transformed with either pYes/GS vector (BVB311) or pYes/ GS/CDC33 (BVB312) and streaked on selective solid medium containing 2% galactose to induce expression of Cdc33. The plate was incubated at 16°C for 8 days. B, Cdc33 protein expression. 50 g of cell extract was separated on a 12% polyacrylamide gel, transferred to a nitrocellulose membrane, and probed with Cdc33 antiserum. The slower migrating band indicated by an arrow in lanes 4 and 5 is the hemagglutinin-tagged Cdc33. C, wild type (BVB306) or cells transformed with pYes/GS (BVB311) or pYes/GS/ CDC33 (BVB312) were grown in Ura Ϫ liquid medium at 30°C overnight. Cultures were diluted to an OD 600 of 0.25 and placed in culture at 16°C in medium containing 2% galactose and raffinose, and aliquots were taken at selected times points. The graph reflects results from three independent experiments.

eIF4E and ␣-Factor Insensitivity
We next determined the effect of CDC33 on growth. The Cln deletion strains grow more slowly than the BVB306, so that the overexpression of CDC33, if it behaved as in BVB306, would be expected to result in extremely slow growth. Instead overexpression of CDC33 in strains lacking either Cln2 or Cln3, but not Cln1, showed improved growth on galactose when compared with controls (Fig. 3). Therefore, overexpression of CDC33 improved growth of the strains deleted for Cln2 or Cln3 rather being inhibitory as in the wild type BVB306 strain. The results from these knockout experiments suggest that the altered phenotype caused by elevated Cdc33 is through changes in expression of CLN3 and/or CLN2.
The Distribution of CLN3 mRNA Is Altered in Polysomes of CDC33-overexpressing Cells-To test the hypothesis that the translation of Cln3 is altered in CDC33-overexpressing cells, we examined the distribution of CLN3 mRNA in the polysomes of BVB311 and BVB312. mRNAs that are translationally repressed engage few or no ribosomes and thus have a slow sedimentation rate. Cell extracts from BVB311 and BVB312 were centrifuged through a sucrose gradient, fractionated, and applied to a membrane with a slot-blotter. Fig. 4 shows that in control cells (BVB311) the CLN3 mRNA sedimented predominantly with the monosomes and small polysomes. In contrast, the CLN3 mRNA was found mostly in large polysomes in BVBV312. The same pattern was observed for CLN2 mRNA, although the change in distribution was smaller. In contrast, the profile of CLN1 and HSC82 was identical for the two strains. Both mRNAs were efficiently utilized and loaded with polyribosomes. Integrity of the transcripts was confirmed by Northern blot of specific gradient fractions (not shown). These experiments indicate that the polysomal redistribution of CLN3 mRNA by overexpression of CDC33 is rather specific as neither distribution of CLN1 nor HSC82 was altered in BVB312. This increased utilization of CLN3 mRNA in BVB312 cells is expected to result in elevated Cln3 protein and in alterations in growth and sensitivity to pheromone.

Mbp1 Suppresses the Slow Growth and ␣-Factor
Resistance from CDC33 Overexpression-To further understand the mechanism of the CDC33 overexpression phenotype, we screened for high copy suppressors. The BVB312 strain was transformed with an expression library, and clones were screened for restored growth (equivalent to BVB311) on galactose-containing medium. From this set of transformants, 58 formed large colonies in 2 days and, upon isolation of the plasmid and retransformation, showed healthy growth on galactose medium at 16°C. An empty Yep13 vector was used as a negative growth control, and these cells could only form microcolonies after 7 days of culture. As a positive control, we overexpressed CAF20, which encodes a protein that binds to and inhibits Cdc33 and thus should repress the phenotype of CDC33 overexpression. Indeed, the Yep/CAF20 plasmid restored normal growth of BVB312 cells at 16°C (data not shown). Of the 58 colonies A, DNA content analysis by flow cytometry. Cultures were grown in liquid selective medium containing 2% raffinose/galactose, and samples were taken at midlog phase. HU, hydroxyurea. B, CDC33 overexpression decreases ␣-factor sensitivity. Cells from strain BVB311 and BVB312 were compared for sensitivity to mating pheromone using halo assays. Disks containing the indicated amounts of ␣-factor were added to plates containing lawns of the indicated strains. C, halo assays for ␣-factor sensitivity to CDC33 in strains containing knockouts of CLN1, CLN2, and CLN3. Isogenic strains containing individual knockouts for CLN1, CLN2, or CLN3 were transformed with pYes/GS/CDC33 and cultured as lawns on plates containing 2% raffinose/galactose. Disks contained the indicated amount of ␣-factor.

TABLE III
Budding pattern Samples were grown at 30°C overnight in 2% raffinose and then diluted to an OD of 0.2 in 2% raffinose/galactose. Aliquots were removed at the designated OD and examined by microscopy. Percentages were derived from three independent samples with a total of 3 ϫ 10 3 cells counted from each aliquot. eIF4E and ␣-Factor Insensitivity examined, 24 did not show any appreciable reduction in the level of overexpressed Cdc33 (data not shown). These were chosen for subsequent analysis. Of these, the most interesting suppressor was the transcription factor and cell cycle regulator Mbp1, and subsequent work was focused on this. Fig. 5A shows the growth recovery of BVB312 from increased expression of Mbp1 at 16°C on a 2% galactose plate. Western blot analysis showed that in this strain the level of Mbp1 was increased 2.5-fold over wild type (Fig. 5C, inset). A growth curve of cells grown in liquid culture (2% galactose) confirmed the restoration of near normal growth (Fig. 5C). The generation time for BVB312 is about 600 min at 16°C, whereas it was decreased substantially after MPB1 overexpression (220 min) compared with 190 min for the control strain, BVB311.
The ␣-factor sensitivity in the Mbp1-suppressed strain was also completely restored (Fig. 5B). One possible explanation was that elevated Mbp1 suppresses the growth defect and ␣-factor insensitivity of BVB312 cells by restoring the early cyclins levels back to normal.
Cnl2 and Cln3 Proteins Are Elevated in BVB312 and Return to Normal with MBP1 Overexpression-We examined the protein levels for several genes involved in G 1 /S progression in strains BVB311, BVB312, and BVB312/MBP1. Two independent experiments were used to prepare the extracts, and the results from representative blots are shown in Fig. 6. In strain BVB312, Cln3 protein was elevated by 2.5-fold (Fig. 6), consistent with the previous result from polysome analysis. Cln2 was also increased 3-fold in BVB312. We also examined the levels of cyclins Cln1, Clb5, and Clb6 and the transcription factors Swi4 and Swi6. We saw no change in protein expression for Cln1, Clb6, or Swi4 (data not shown), but Clb5 and Swi6 showed an increase of ϳ5-fold. We also examined the levels of these proteins in the Mbp1-suppressed strain. Cln3 levels were reduced back to normal levels (Fig. 6). Cln2 was also reduced but still slightly elevated compared with BVB311. The level of Clb5 was unaffected by Mbp1, while the transcription factor Swi6 returned to normal level.
Cdc33-induced Alteration of SCB Gene Expression and Suppression by Mbp1-We have proposed that excess Cdc33 leads FIG. 3. Deletion of CLN2 and CLN3, but not CLN1, restores normal growth of BVB312 cells. Cells from isogenic stains containing knockouts for CLN1, CLN2, and CLN3 were transformed with either pYes/GS (BVB786, BVB1037, or BVB367, respectively) or pYes/ GS/CDC33 (strains BVB787, BVB1038, or BVB368) and grown in containing 2% galactose at 30°C overnight. Cultures were diluted to an OD 600 of 0.22 (early log phase growth) and incubated at 16°C, and aliquots were taken at selected times points. The graphs reflect results from three independent experiments. FIG. 4. Distribution of CLN1, CLN2, and CLN3 mRNAs in the heavy polysomes (P100). Polysomes were fractionated as described under "Materials and Methods," and aliquots from fractions were applied to a membrane with a slot-blotter.
FIG. 5. Suppression of CDC33 slow growth and ␣-factor insensitivity by MBP1 overexpression. A, cells transformed with pYes/ GS, pYes/GS/CDC33, or pYes/GS/CDC33 and pRS425-MBP1 were streaked on selective solid medium containing 2% galactose to induce expression of Cdc33. The plate was incubated at 30°C for 96 h. B, halo assays using cells overexpressing Mbp1 (BVB315) that show restored ␣-factor sensitivity. C, growth in liquid medium containing 2% galactose at 16°C. The graph reflects results from three independent experiments. In the inset, the expression of Mbp1 was determined by Western blot. eIF4E and ␣-Factor Insensitivity to elevated Cln3 protein and subsequent transcriptional activation of several downstream genes. To test this, we compared the expression of some of these genes in BVB311 (control) cells and strains BVB312 and BVB312/MBP1 (Fig. 7A). RNA was isolated from cells grown in liquid culture medium with galactose during mid-log phase (OD 600 ϭ 0.8) at 16°C. All the results were normalized for loading to HSC82, a constitutively expressed housekeeping gene. The expressed transcripts of the BVB311 strain were given the value of 1, and all other conditions are referred to these. Fig. 7A, lane 3, shows that CLN3 mRNA was not increased in BVB312, confirming that the increase in Cln3 protein is largely translational (compare with the result in Fig. 4). In the strain overexpressing MBP1, the normal CLN3 transcript was slightly reduced compared with BVB312 and B311 cells. However, in this strain we unexpectedly detected an additional, slightly larger transcript that was present at ϳ20% of the normal transcript. The addition of ␣-factor slightly reduced the level of CLN3 mRNA (Fig. 7A,  lane 6), but the ratio of larger to normal transcript increased. The CLN2 mRNA was strongly reduced in the BVB311 treated with ␣-factor. In contrast, it remained elevated in the CDC33overexpressing strain with and without ␣-factor, which could explain in part the failure to arrest the cell cycle. More than a 3-fold decrease was seen when MBP1 was overexpressed (Fig.  7A, lane 5), and the level dropped to almost undetectable with ␣-factor (Fig. 7A, lane 6). We also examined mRNA levels for CLN1 and SWI4, which are SCB-responsive genes and hence are dependent on Cln3 levels. As expected, CLN1 mRNA was slightly elevated in BVB312, and the level was reduced with ␣-factor. The level of SWI4 mRNA showed little variation in the different strains with and without addition of ␣-factor. To control for pheromone activity, we monitored the expression of FUS1, which is transcriptionally activated by ␣-factor. The BVB311 strain showed strong induction of FUS1 with ␣-factor as anticipated. Interestingly, FUS1 was also induced in BVB312, although this strain does not arrest the cell cycle at this concentration of pheromone.
Overexpression of MBP1 Alters the Transcription Start Site in CLN3 mRNA-Given the additional RNA species detected by Northern blots, we hypothesized that the CLN3 5Ј-UTR is extended in the MBP1-overexpressing cells. We thus examined the length of CLN3 mRNA by primer extension. Cell cultures from each strain (BVB311, BVB312, and BVB312/MBP1) were induced with galactose for 6 h, and 400 g of total RNA was used to isolate polyadenylated RNA. Equal counts of a radiolabeled probe for the CLN3 5Ј-UTR were used for primer extension and examined on a 4% sequencing gel. mRNA isolated from the ⌬cln3 strain was used as a negative control, and primer extension of 85 bases of the HSC82 5Ј-UTR was used as a normalization control. A single extension product was seen for BVB311 cells (Fig. 7B, lane 1) positioned at Ϫ364 bases upstream of the translation initiation start site (ϩ1 AUG). This location is consistent with the known site selection from the TATAA box at Ϫ439 (Fig. 7C). The strain BVB312 showed the same extension product. In addition, a larger transcript was detected, which is positioned at Ϫ409 from the translation initiation site (Fig. 7B, lane 2). Accordingly this message begins 30 base pairs downstream of the TATAA box and adds 45 bases to the normal 5Ј-UTR (see diagram in Fig. 7C). This message is expressed at low level in the BVB312 strain and was not detected previously by Northern (Fig. 7A). In BVB312/MBP1 there was a decrease in normal transcript (initiated at Ϫ364) and a substantial increase in the larger message (initiated at Ϫ409). We propose that increased Mbp1 promotes the shift from the normal start site at Ϫ364 to the Ϫ409 site. This extended form of the message could be translated less efficiently and reduce the Cln3 protein elevation obtained by CDC33 overexpression (Fig. 6A, compare BVB312 with BV312/MBP1).

DISCUSSION
Overexpression of the limiting translation factor eIF4E in mammalian cells results in malignant transformation and cell cycle alterations, presumably by enhancing the translation of growth regulators including cyclin D1 (45). In this work, we have shown that elevated expression of Cdc33 (yeast eIF4E) causes a slow growth phenotype, particularly at low temperature. In addition, many cells remain budded as the culture approaches the late log phase and frequently present multiple buds during exponential growth. This phenotype is indicative of premature progression through Start and into S phase, subsequently resulting in a deficiency of some components needed during DNA replication. This results in accumulation of cells in S phase. Enforced entry into Start is also indicated by an override of the cell cycle arrest signal by mating pheromone. All strains were grown in media containing 2% galactose at 30°C. Cultures were diluted to an OD 600 of 0.20, and where indicated 10 g/ml ␣-factor was added. This concentration of ␣-factor is sufficient to achieve full arrest of strain BVB311 but not of BVB312. B, primer extension to map the transcription start site in CLN3 mRNA. Results were normalized to an HSP82 transcript primer extension. C, sketch of the promoter region for CLN3. Both transcription start sites are depicted as well as the TATA box at position Ϫ439. A consensus site for MBF binding is denoted at Ϫ743 with the nucleotide sequence given.

eIF4E and ␣-Factor Insensitivity
The slow growth and loss of pheromone response are likely a result of the translational increase in Cln3 (47) and the subsequent induction of downstream genes CLN2, CLB5, and SWI6, which are normally responsive to increases in CLN3 expression (48). The insensitivity to ␣-factor is most likely due to higher than normal levels of Cln3 and Cln2 proteins since the phenotype is abolished in ⌬cln3 or ⌬cln2 background.
Overexpressing eIF4E preferentially increases the translation of repressed mRNAs, particularly those with a long and structured 5Ј-UTR or containing upstream ORFs (49). The dependence of Cln3 protein synthesis on elevated eIF4E is likely due to its extraordinarily long (for yeast) 5Ј-UTR. In addition, it contains a short upstream ORF (47), which was suggested to inhibit translation of CLN3. Translational control of CLN3 at Start is also supported by the fact that overexpression of CLN3 in the cdc33-1 strain rescues the cells from G 1 arrest at nonpermissive temperature (24). This indicates that in the mutant a specific loss of CLN3 translation when Cdc33 is partially inactivated is the primary cause for the G 1 arrest. In support of this, a LacZ reporter preceded by the CLN3 5Ј-UTR was poorly expressed in cdc33-1 (24). Conversely, increased expression of CLN3 in a wild type background causes a shortened G 1 phase, small cell size, and slow growth (46,50), which is very similar to the phenotype we have observed by CDC33 overexpression. We propose a mechanism whereby excess Cdc33 elevates Cln3 synthesis and, in turn, activates an SCB-dependent response that is sufficient to suppress ␣-factor G 1 arrest and activate Start prematurely. CLN2 is activated through an SCB element, and its expression is increased in cells overexpressing CDC33.
We further identified MBP1, a component of the transcriptional activator MBF, as a multicopy suppressor of the slow growth and ␣-factor-insensitive phenotype. A 2.5-fold increase in Mbp1 was sufficient to reduce the level of Cln3 and Cln2 proteins back to normal levels. This was likely through altered transcription of CLN3, resulting in the appearance of a 5Јextended transcript, and a reduction of Cln3 protein. This set of events seems responsible for implementation of other downstream events such as the return of CLN2 and SWI6 mRNA back to normal levels.
Little is known about the regulatory elements for the transcription of CLN3. Some reports indicate that transcription may be regulated by a glucose-responsive element within its promoter (51). However, transcriptional changes caused by variations in the carbon source were reported to be slight. Therefore, CLN3 expression is mostly modulated by changes in translation and degradation. We attribute the suppression by MBP1 to a transcriptional alteration whereby the start site selection was moved 45 bases further upstream of normal. A sequence at Ϫ743 in the CLN3 promoter is in perfect agreement with an MCB consensus. We predict that this site is sufficient for Mbp1 binding and the consequent alteration in start site selection. A prediction of secondary structure with M-Fold (48) showed that the lengthened 5Ј-UTR could fold into a hairpin structure with a ⌬G of Ϫ41 kcal/mol, which in yeast would be very inhibitory for translation. We hypothesize that the lengthened 5Ј-UTR would reduce the translation of CLN3 mRNA, thereby offsetting the increased translational efficiency obtained by excess Cdc33. Finally our work clearly uncovered the fact that CLN3 is transcriptionally regulated by Mbp1. It seems likely that the Ϫ409 initiated transcript expressed in BVB312 at low level is also MBF-dependent, particularly considering that Swi6, the partner of Mbp1, is also elevated 5-fold in this strain.