Most factors known to function in mRNA turnover are not essential for cell viability. To identify essential factors, ∼4000 temperature‐sensitive yeast strains were screened for an increase in the level of the unstable CYH2 pre‐mRNA. At the non‐permissive temperature, five mutants exhibited decreased decay rates of the CYH2 pre‐mRNA and mRNA, and the STE2, URA5 and PAB1 mRNAs. Of these, the mutant ts1159 had the most extensive phenotype. Expression of the TIF51A gene (encoding eIF‐5A) complemented the temperature‐sensitive growth and mRNA decay phenotypes of ts1159. The tif51A allele was rescued from these cells and shown to encode a serine to proline change within a predicted α‐helical segment of the protein. ts1159 also exhibited an ∼30% decrease in protein synthesis at the restrictive temperature. Measurement of amino acid incorporation in wild‐type cells incubated with increasing amounts of cycloheximide demonstrated that a decrease in protein synthesis of this magnitude could not account for the full extent of the mRNA decay defects observed in ts1159. Interestingly, the ts1159 cells accumulated uncapped mRNAs at the non‐permissive temperature. These results suggest that eIF‐5A plays a role in mRNA turnover, perhaps acting downstream of decapping.
Differences in mRNA turnover rates can have significant effects on the expression of specific genes and provide a cell with flexibility in bringing about rapid changes in protein levels. The structures and mechanisms involved in the determination of individual mRNA decay rates have begun to be elucidated, particularly in yeast (Caponigro and Parker, 1996; Jacobson and Peltz, 1996). It is now well established that mRNA decay is a precise process dependent on a variety of specific cis‐acting sequences and trans‐acting factors. In addition, there is extensive evidence for an important role for translation in mRNA decay, including experiments demonstrating that: (i) inhibition of translational elongation can reduce mRNA decay rates (Peltz et al., 1992; Beelman and Parker, 1994); (ii) instability elements can be localized to mRNA coding regions (Parker and Jacobson, 1990; Caponigro et al., 1993; Hennigan and Jacobson, 1996; Jacobson and Peltz, 1996); (iii) the activity of some instability elements depends on ribosome translocation up to, or through the element (Parker and Jacobson, 1990; Peltz et al., 1993a; Hennigan and Jacobson, 1996); (iv) some 3′‐untranslated region (UTR) instability elements can influence mRNA translational activity (Kruys et al., 1989; Marinx et al., 1993); (v) factors involved in decay and the decay process are polysome‐associated (Peltz et al., 1993b; Caruccio and Ross, 1994; Atkin et al., 1995, 1997; Zhang et al., 1997); (vi) premature translational termination can enhance mRNA decay rates (Peltz et al., 1993a,b); (vii) factors essential for the rapid decay of nonsense‐containing mRNAs also regulate translational suppression (Leeds et al., 1992; Cui et al., 1996; Weng et al., 1996); and (viii) metabolism of the poly(A) tail is a rate‐limiting step in the decay of several mRNAs yet this structure and its binding protein also have a role in translational initiation (Jacobson, 1996; Sachs et al., 1997).
The general outlines for at least three mRNA decay pathways have been delineated in the yeast Saccharomyces cerevisiae. In two of these pathways, mRNA decay proceeds from decapping to 5′→3′ exonucleolytic digestion (Decker and Parker, 1993, 1994), whereas, in the third pathway, the predominant event is 3′→5′ exonucleolytic digestion (Muhlrad et al., 1995). The initiation of decapping is known to be promoted by shortening of the mRNA poly(A) tail or premature translational termination (Muhlrad and Parker, 1994; Muhlrad et al., 1994). A number of trans‐acting factors with demonstrated or implied roles in these pathways have been identified and include: a 5′→3′ exonuclease (Xrn1p; Hsu and Stevens, 1993), the decapping enzyme (Dcp1p; Stevens, 1988; Beelman et al., 1996; La Grandeur et al., 1998), poly(A)‐binding protein (Pab1p; Adam et al., 1986; Sachs et al., 1986), constituents of a poly(A) nuclease complex (Pan2p and Pan3p; Boeck et al., 1996; Brown et al., 1996), two proteins that regulate decapping (Mrt1p and Mrt3p; Hatfield et al., 1996) and a number of factors that are required specifically for the decay of nonsense‐containing mRNAs (Upf1p, Nmd2p/Upf2p and Upf3p; Peltz et al., 1994; Cui et al., 1995; He and Jacobson, 1995; Lee and Culbertson, 1995; Lee et al., 1995; He et al., 1997).
Only three of these mRNA decay factors (Pab1p, Mrt1p and Mrt3p) are encoded by essential genes (Adam et al., 1986; Sachs et al., 1986; Hatfield et al., 1996). In contrast, many genes involved in other complex cell processes, such as transcription or translation, are essential. In order to identify other essential factors involved in mRNA decay, we screened a yeast library of ∼4000 temperature‐sensitive (ts) mutants for those that were incapable of degrading either wild‐type or nonsense‐containing mRNAs at the restrictive temperature. We have identified several mutants in which a shift to the non‐permissive temperature promotes mRNA stabilization and report here on the ts strain displaying the most extensive impairment of mRNA decay. That strain, ts1159, was found to harbor a mutation in TIF51A, the gene encoding eIF‐5A.
Identification of temperature‐sensitive strains that selectively accumulate the unstable CYH2 pre‐mRNA
The collection of yeast strains used in this study was generated by J.Abelson and colleagues in a study of ts mutants defective in pre‐mRNA splicing. The yeast strain SS330 was treated with ethylmethylsulfonate (EMS), and ∼4000 mutants capable of growth at 24°C, but not at 37°C, were isolated (Vijayraghavan et al., 1989; J.Abelson, personal communication). Northern blots of RNA isolated from each mutant after a 2 h shift to the non‐permissive temperature were then hybridized to an ACT1 probe. This led to the identification of a number of new splicing mutants that accumulated the ACT1 pre‐mRNA at 37°C (Ruby and Abelson, 1991).
The original collection of blots was rescreened to identify mutants in which mRNA was stabilized at 37°C. Earlier work had shown that endogenous substrates of the nonsense‐mediated mRNA decay pathway include inefficiently spliced pre‐mRNAs that enter the cytoplasm (such as the CYH2 pre‐mRNA; He et al., 1993). Mutants incapable of nonsense‐mediated mRNA decay, i.e. strains containing deletions of the UPF1, NMD2 or UPF3 genes, lead to the stabilization and selective accumulation of such RNAs (He et al., 1993; Peltz et al., 1994; He and Jacobson, 1995) and we expected that strains with ts lesions in this pathway would have related phenotypes. The blots were thus hybridized with probes for CYH2 and also for ACT1 transcripts (the latter to minimize the chance of isolating splicing mutants). We identified 96 mutants that accumulated the CYH2 pre‐mRNA, but not the ACT1 pre‐mRNA, at the restrictive temperature (data not shown).
The RNA in the original collection of Northern blots had been prepared only from cells that had been shifted to the non‐permissive temperature for 2 h. In order to directly compare the levels of the CYH2 transcripts in these cells before and after the change in temperature, we prepared RNA from aliquots of these strains grown at 24°C or incubated at 37°C for 45 min. Northern blots of these RNA preparations were probed with radiolabeled CYH2 DNA. Figure 1 shows an example of these blots including seven ts mutants and a splicing mutant (prp2) control. Three mutants, ts1159, ts1189 and ts1197, exhibited an increase in the level of the CYH2 pre‐mRNA after the shift to the non‐permissive temperature and were thus chosen for additional testing. Nine other strains (ts817, ts942, ts970, ts1100, ts1387, ts1581, ts3753, ts3766 and ts3771) were chosen from additional blots by the same criteria (data not shown).
Identification of mutants with temperature‐sensitive defects in mRNA decay
Increases in CYH2 pre‐mRNA abundance could be attributable to changes in the rates of decay or synthesis of that transcript. To distinguish between these possibilities for the selected mutants, and to characterize their mRNA decay defects further, we analyzed the kinetics of turnover of a number of mRNAs in these strains after a shift to the non‐permissive temperature. The cells were grown at 24°C to mid‐log phase and shifted to 37°C for 1 h. Transcription was then inhibited by the addition of thiolutin (Jimenez et al., 1973). RNA was isolated 0, 5, 10, 20 and 35 min after inhibition of transcription and analyzed by Northern blotting, using probes for five transcripts (Herrick et al., 1990). These represented different classes of transcripts found in the cell and included the CYH2 pre‐mRNA and the CYH2, STE2, PAB1 and URA5 mRNAs. The CYH2 pre‐mRNA, as noted above, is a substrate of the nonsense‐mediated mRNA decay pathway, the STE2 and URA5 mRNAs have short half‐lives (<7 min) and in this strain, the PAB1 and CYH2 mRNAs are moderately stable transcripts (Herrick et al., 1990).
Seven of the mutant strains tested in this manner (ts970, ts1197, ts1387, ts1581, ts3753, ts3766 and ts3771) did not show a defect in the decay rates of the mRNAs tested. They were thus deemed not to be mRNA‐turnover mutants and were not characterized further (data not shown). The remaining mutants (ts817, ts942, ts1100, ts1159 and ts1189) exhibited decreased decay rates of some or all of the transcripts tested when compared with the rates measured in the wild‐type parental strain, SS330. Of the latter set of mutants, ts1159 exhibited the most substantial effects on mRNA decay, stabilizing some mRNAs up to 5‐fold (Figure 2).
Identification of the gene mutated in ts1159
The ts growth phenotypes of the five strains that exhibited mRNA decay defects were complemented by a yeast genomic DNA library (cloned in the centromere vector YCp50; Rose et al., 1987) and plasmids capable of inducing temperature‐independent growth in each strain were identified. Transformation of ts1159 cells with the YCp50 library yielded three clones (p1159‐1 to p1159‐3) capable of growth at the restrictive temperature. Cells from the original ts1159 strain, or from that strain transformed with a complementing plasmid, were patched onto rich, synthetic complete (SC) or SC‐ura plates, grown for 3 days at 24°C, and then replica‐plated to two plates, one of which was incubated at 24°C and the other at 37°C, each for 3 days. Expression of the relevant YCp50 clones was shown to restore growth at 37°C to the mutant cells (Figure 3A). Moreover, growth at the higher temperature was shown in two ways to be plasmid dependent: (i) by isolation of the respective plasmids, retransformation into fresh mutant cells and verification of the complementation phenotype; and (ii) by growing complemented cells to saturation in non‐selective media and then showing that all clones that had lost the plasmid also lost the ability to grow at 37°C (data not shown).
In order to verify that the ts growth phenotype was linked to the mRNA decay defect, ts1159 cells with and without a complementing plasmid were grown at 24°C and shifted to 37°C for 1 h. RNA prepared from these cells was analyzed by Northern blotting, using an STE2 fragment as a hybridization probe. There is a 3.3‐fold increase in the level of the STE2 mRNA in ts1159 cells after a shift to the non‐permissive temperature (Figure 3B), but this increase is not evident in cells harboring the complementing 1159‐1 YCp50 plasmid. This indicates that the same plasmid which complemented the ts growth phenotype also complemented the mRNA decay defect. Initial mapping of the three plasmids rescued from the complemented cells using the restriction enzymes EcoRI and SalI showed that they comprised two different plasmids (one represented twice, and the other once; data not shown). The vector–insert junctions of the two plasmids (1159‐1 and 1159‐3) were sequenced, compared with the available databases and found to contain overlapping fragments of chromosome V. The overlapping sequence included three open reading frames (ORFs) of unknown function (yEL033w, yEL035c and yEL038w), a serine tRNA gene, the TIF51A (HYP2) gene (which encodes the putative translation initiation factor eIF‐5A; Kemper et al., 1976; Benne et al., 1978; Schnier et al., 1991), the ANP1 gene (which encodes an endoplasmic reticulum protein; Chapman and Munro, 1994) and the RAD23 gene (involved in DNA repair; Miller et al., 1982).
The ORFs contained in the complementing plasmids were then subcloned separately into the centromere vector pRS316 (Sikorski and Hieter, 1989), along with a substantial length of 5′ and 3′ sequences to ensure inclusion of the regulatory sequences flanking each gene. Each ORF construct was transformed into ts1159 cells and tested for complementation of the growth defect at 37°C. Only one plasmid (pDZ4), expressing the TIF51A ORF, was capable of complementing the growth of the mutant strain at the non‐permissive temperature. ts1159 cells transformed either with the pRS316 vector alone or with the complementing TIF51A plasmid were patched onto SC‐ura plates, grown for 3 days at 24°C, and then replica‐plated to two plates, one of which was incubated at 24°C and the other kept at 37°C, each for 3 days. It is evident from the data shown in Figure 4A that the construct expressing TIF51A restored the growth at 37°C, while the vector alone could not complement the growth defect.
In order to ensure that the complementation of the ts growth phenotype and mRNA decay defect resulted from the same ORF on the original YCp50 clone, ts1159 cells transformed with the empty vector and with pDZ4 were grown at 24°C and shifted to 37°C for 1 h. RNA prepared from these cells was again analyzed by Northern blotting. As can be seen in Figure 4B, there is a 3‐fold increase in the level of the STE2 mRNA in the ts1159 cells transformed with pRS316 after a shift to the non‐permissive temperature (as was seen for the untransformed ts mutant in Figure 3B), but this increase does not occur in the cells harboring the plasmid expressing TIF51A. This indicates that both the growth and mRNA decay defects were complemented by expression of a single gene.
Comparable experiments identified the genes complementing the other four mutants (Table I). SLA2 encodes a protein containing a talin‐like domain, which previously was identified as being involved in membrane cytoskeletal assembly, endocytosis of α‐factor and maintenance of the PMA1 ATPase (Holtzman et al., 1993; Raths et al., 1993; Na et al., 1995). GRC5 (also called QSR1 and QM1) encodes the 60S ribosomal subunit protein L9 (Dick et al., 1997; Nika et al., 1997), and THS1 encodes the yeast threonyl‐tRNA synthetase (Pape and Tzagoloff, 1985). MRT4 (mRNA turnover; yKL009w) is a previously uncharacterized gene which encodes a protein similar to the P0/L10 family of 60S ribosomal proteins. Here we report on the initial characterization of the ts1159 mutant and the tif51A allele found in that strain. Further studies of the other four mutants will be reported elsewhere.
Isolation of the mutated tif51A allele from ts1159
To identify the specific mutation in the ts1159 tif51A allele (which we designate tif51A‐1159), we rescued the mutation by gap repair. The pDZ4 construct contains the 473 nucleotide TIF51A ORF plus 193 nucleotide upstream and 105 nucleotide downstream sequences in the URA3‐based vector, pRS316 (see Materials and methods). We generated two types of pDZ4 plasmids containing 74 or 209 nucleotide gaps at the 3′ end of the gene (with the latter encompassing the former). These plasmids were transformed into ts1159 cells, ura+ colonies were selected, and the repaired plasmids (which now included sequence from the chromosomal copy of tif51A‐1159) were isolated.
Re‐transformation of the wild‐type and repaired alleles into fresh ts1159 cells showed that the repaired tif51A‐1159 allele [pDZ4(g‐r)] could no longer complement the ts growth phenotype of the ts1159 mutant when compared with the complementation observed in cells transformed with the original TIF51A gene (Figure 5A). In order to test for complementation of the mRNA turnover phenotype, cells harboring the pDZ4 wild‐type allele, the repaired plasmid or no plasmid were grown at 24°C, shifted to 37°C for 1 h and then tested for levels of STE2 mRNA by Northern blotting. The results in Figure 5B reiterate those shown in Figure 4B, i.e. the ∼3‐fold increase in STE2 mRNA levels in the ts1159 cells at 37°C is abrogated by the expression of the wild‐type TIF51A allele (pDZ4). However, this increase is not reversed by expression of the gap‐repaired pDZ4(g‐r) construct.
Sequencing of the gap‐repaired plasmids revealed a single T→C change at position 445 of the TIF51A ORF (Figure 5C), which resulted in a serine to proline change at position 149 of the eIF‐5A protein. The remaining chromosomal sequence in the repaired gap is the same as in the construct expressing the wild‐type TIF51A allele, as is the sequence in the rest of the pDZ4 insert. Therefore, we conclude that the lack of complementation of both the growth and mRNA decay phenotypes in the ts1159 cells expressing the gap‐repaired plasmid is due to this single amino acid change.
The results of this experiment and of the complementation experiments shown in Figure 4 support the notion that both the growth and mRNA decay phenotypes are due to the same mutation in the TIF51A gene. Additional evidence for mutation of a single gene comes from the fact that tetrads resulting from sporulation of a cross between the ts1159 strain and the SS328 parent strain segregate 2:2 for ts growth (data not shown).
Protein synthesis properties of the mutant strain
Since the eIF‐5A protein was isolated initially from a ribosomal fraction of rabbit reticulocytes, it was considered a translation initiation factor (Kemper et al., 1976; Benne et al., 1978). It was thus important to investigate whether the ts1159 strain was defective in protein synthesis. Cytoplasmic extracts of ts1159 and wild‐type cells before and after a shift to the non‐permissive temperature were fractionated on sucrose gradients and ribosome distribution was examined. The wild‐type cells display a polysome peak of approximately eight ribosomes at both 24°C and after a 1 h shift to 37°C (Figure 6A and B, respectively), with a slight increase in the amount of 80S subunits after the shift to the restrictive temperature. The profiles in the ts1159 mutant at both temperatures look very similar to those seen in the wild‐type extracts, except that polysomes now peak at approximately six ribosomes (Figure 6C and D). These experiments suggest that any temperature‐induced translation defect in the ts1159 mutant is not extensive and, because of the absence of a substantial shift to smaller polysomes at 37°C, is not likely to have a significant effect on translation initiation.
In order to study the protein synthesis properties of the ts1159 cells in a more quantitative manner, we measured the incorporation of 35S‐labeled amino acids (methionine and cysteine) into cells after a shift to the non‐permissive temperature. Duplicate aliquots of cells were taken at various time points after the shift, incubated with the labeled amino acids, and then precipitated and washed with trichloroacetic acid (TCA). The resulting precipitates were counted and compared with the amount of label incorporated at the time of the shift to 37°C (t = 0). The results of these experiments are shown in Figure 7. It is evident that protein synthesis in the wild‐type cells increases after 60 min at the higher temperature (after a modest drop at 30 min, probably due to the heat shock of the shift to 37°C). The ts1159 (tif51A‐1159) cells display a drop in translation of ∼30% after 60 min at the non‐permissive temperature (the time at which decay rates were measured; Figure 2).
Effect of general inhibition of translation on the accumulation of mRNAs
While the results of the polysome gradients and the amino acid incorporation assays suggest that any translation defects in ts1159 are modest (Figures 6 and 7), they nevertheless raise the question of whether the mRNA decay phenotype observed in this strain is an indirect consequence of impaired protein synthesis. Previous work has shown that complete inhibition of protein synthesis results in a general stabilization of all cellular mRNAs (Herrick et al., 1990; Peltz et al., 1992; Beelman and Parker, 1994). However, the effects on mRNA decay rates of a partial deficiency in translation have not been characterized previously.
In order to investigate further the correlation between the extent of inhibition of protein synthesis and the resulting stabilization of specific mRNAs, we used the drug cycloheximide as a general inhibitor of translation. The SS330 parental strain was incubated with increasing concentrations of cycloheximide (ranging from 0.05–0.8 μg/ml) for 1 h, and then duplicate aliquots were allowed to incorporate 35S‐labeled amino acids. RNA was prepared from the same cultures, and Northern blots of these RNAs were probed with the CYH2 and URA5 probes. The results are shown in Figure 8. It is evident that inhibition of protein synthesis by 30% results in a <2‐fold stabilization of the three transcripts tested. The ts1159 cells, which have a similar defect in protein synthesis, exhibited a 2.4‐fold increase in the half‐life of the CYH2 pre‐mRNA, a 5‐fold increase in that of the URA5 mRNA and a 5.2‐fold increase in the half‐life of the CYH2 mRNA (Figure 2). A similar 4‐ to 5‐fold increase in the abundance of the CYH2 and URA5 mRNAs was only observed in the cycloheximide experiments when the inhibition of protein synthesis reached 80% (Figure 8). These results, taken together with the virtually normal polysome profiles of the ts1159 cells at the non‐permissive temperature, suggest that the mRNA decay defects displayed by ts1159 are at least partially a result of a defect in some metabolic process other than protein synthesis.
The ts1159 cells accumulate uncapped mRNAs at the non‐permissive temperature
The results of the experiments described above suggest that eIF‐5A may play a role in mRNA turnover in yeast. Proteins involved in this pathway could function at any of the previously identified stages of this process, i.e. deadenylation, decapping or exonucleolytic decay (Caponigro and Parker, 1996; Jacobson and Peltz, 1996). It has been shown that while inherently unstable mRNAs decay after degradation of their poly(A) tails (Decker and Parker, 1993), nonsense‐containing RNAs decay with long poly(A) tails (Muhlrad and Parker, 1994). The two pathways then converge at the decapping step, since both types of transcripts are substrates for the same decapping enzyme, Dcp1p (Beelman et al., 1996), and exonuclease, Xrn1p (Muhlrad and Parker, 1994; Hagan et al., 1995).
Since the ts1159 cells exhibited an increase in the stability of both nonsense‐containing (the CYH2 pre‐mRNA) and wild‐type mRNAs (CYH2, STE2, URA5 and PAB1), it is reasonable to propose that the mutation in these cells affects steps in mRNA decay at, or downstream of, decapping. In order to determine which step is affected, we examined the cap status of the mRNA in the mutant cells. RNA prepared from wild‐type and ts1159 cells (grown at 24°C or shifted to to 37°C for 1 h) was immunoprecipitated with antisera directed against the 5′ cap structure. The resulting RNA fractions (input, supernatant and pellet) were subjected to Northern analysis and hybridized to a radiolabeled URA5 probe.
The results of these experiments demonstrate that in wild‐type cells at both temperatures and in ts1159 cells grown at 24°C, 70–75% of the URA5 mRNA was precipitated by the anti‐cap antibody, indicating that only 25–30% of the mRNA in these cells was uncapped (Figure 9A). However, in ts1159 cells that had been shifted to the non‐permissive temperature, the fraction of uncapped mRNA increased to 50%, suggesting that mRNA was accumulating in the mutant cells in an uncapped form. This effect was consistent (±2%) in two separate experiments. The efficacy of the immunoprecipitation procedure was demonstrated by the results from xrn1Δ cells, in which the majority (78%) of the URA5 mRNA was shown to be uncapped, a result previously shown for other mRNAs in cells with this mutation (Hsu and Stevens, 1993; Muhlrad et al, 1994).
Similar results were obtained when the blot shown in Figure 9A was reprobed with a radiolabeled STE2 probe. The fraction of uncapped STE2 mRNA in ts1159 cells increased from 28 to 66% after a shift to the restrictive temperature (Figure 9B). These results indicate that the defect in the ts1159 cells at the non‐permissive temperature generates uncapped mRNAs, suggesting that decapping is occurring in these cells but that the exonucleolytic process may be inhibited.
Most of the factors identified to date as having a role in mRNA decay have proved to be non‐essential for cell growth. Because of the fundamental nature of this process, we investigated whether essential factors have been overlooked. To identify mutants defective in the turnover of both wild‐type mRNAs and those containing premature nonsense codons, we tested a panel of ∼4000 yeast strains that are temperature‐sensitive for growth for accumulation of the CYH2 pre‐mRNA. (In principle, this panel could also have been tested for mutants that increase turnover rate by seeking strains containing lower levels of the test RNA.)
Of 96 strains that exhibited an increase in the amount of the test RNA, five (ts817, ts942, ts1100, ts1159 and ts1189) proved to have decreased decay rates of several mRNAs (Figure 2; D.Zuk and A.Jacobson, in preparation). The ts1159 strain had the most extensive overall effect on the decay of the five RNAs tested, and was chosen for further study. The genes complementing the ts growth of the other four mutants have been identified (Table I) and the initial characterization of these strains is underway (D.Zuk and A.Jacobson, in preparation).
The growth defect of ts1159 cells at the non‐permissive temperature was complemented by a plasmid from a centromere‐based yeast DNA library, and subsequently by expression of the TIF51A gene alone. Expression of this gene also complemented the mRNA decay phenotype of ts1159 cells. The mutation in the TIF51A gene harbored by ts1159 was rescued using a gap repair approach. Subsequent sequencing of the mutant gene showed that it contained a single T→C change at the 3′ end of the ORF, resulting in a serine to proline change at position 149 of the eIF‐5A protein. Since a plasmid expressing the mutant allele could not complement the growth or mRNA defects of the ts1159 cells, we conclude that this mutation in eIF‐5A is solely responsible for the observed phenotypes.
eIF‐5A (originally called eIF‐4D) is a highly conserved protein in all eukaryotic cells analyzed (Gordon et al., 1987) that contains a unique spermidine‐dependent hypusine modification on a specific lysine residue (Cooper et al., 1983). However, the function of this protein remains elusive, even after a decade of intensive study. Based on its isolation from a rabbit reticulocyte ribosomal high salt wash fraction, and on its activity in an in vitro assay for the synthesis of methionyl‐puromycin, eIF‐5A originally was thought to be a translation initiation factor involved in formation of the first peptide bond (Kemper et al., 1976; Benne et al., 1978). Other experiments in mammalian cells are consistent with a role in protein synthesis (Shi et al., 1996, 1997).
However, work by Kang and Hershey (1994) has shown that marked depletion of this protein in yeast cells, while resulting in cell‐cycle arrest, only led to a 30% drop in the first round of protein synthesis. The translation assays performed on ts1159 cells after a shift to the non‐permissive temperature showed a comparable drop in incorporation of labeled amino acids. Furthermore, analysis of sucrose gradients of cytoplasmic extracts from these cells shifted to 37°C for 1 h exhibited only modest impairment of polysome profiles. Since the mutation in ts1159 had an effect comparable with the ‘degron’ construct of Kang and Hershey (1994), it is likely that the former mutation completely inactivates eIF‐5A after a shift to 37°C. In light of the fact that the mutated region of eIF‐5A is predicted to be α‐helical (by the ‘nnpredict’ program; Kneller et al., 1990), it is not surprising that a serine to proline substitution should have such a deleterious effect. Collectively, the data obtained in yeast suggest that eIF‐5A may not play a major role in protein synthesis, but may be involved in other aspects of cell metabolism, such as mRNA turnover or cellular RNA export, with an indirect effect on translation.
The phenotype of ts1159 cells at 37°C suggests that eIF‐5A may play a general role in the turnover of mRNAs. ts1159 exhibited decreased decay rates of both wild‐type mRNAs (CYH2, STE2, URA5 and PAB1) and of a transcript containing a premature nonsense codon (the CYH2 pre‐mRNA). Decay of these two types of RNAs is triggered by different events, with wild‐type RNAs undergoing shortening of their poly(A) tails before decapping and exonucleolytic cleavage (Decker and Parker, 1993), while the decay of nonsense‐containing RNAs is initiated by premature translation termination (Muhlrad and Parker, 1994). However, the two pathways converge at the decapping step, since both types of transcripts are substrates for the same decapping enzyme, Dcp1p (Beelman et al., 1996), and exonuclease, Xrn1p (Muhlrad and Parker, 1994; Hagan et al., 1995). This suggested that the mutation in ts1159 affects steps involved downstream of decapping (Figure 10).
In order to test this hypothesis, we examined the cap status of the URA5 and STE2 mRNAs in the ts1159 cells after a shift to the restrictive temperature. The fraction of uncapped mRNAs increased 2‐fold (from 25 to 50%) in the mutant cells when compared with wild‐type cells or with ts1159 cells grown at the permissive temperature (Figure 9). The simplest interpretation of these results is that eIF‐5A functions downstream of Dcp1p (since decapping occurs) and upstream of Xrn1p (Figure 10).
The acccumulation of uncapped mRNAs in ts1159 cells grown at 37°C not only serves to define the precise step in which eIF‐5A functions in mRNA decay, but also supports the conclusion that the stabilization of mRNAs seen in these cells is not attributable to the indirect effects of partially inhibiting translation. Using cycloheximide to inhibit protein synthesis to the same level observed in ts1159 cells at 37°C did not increase mRNA abundance to the same extent seen in the mutant (Figure 8). In addition to this quantitative difference, mRNAs accumulating in ts1159 cells at 37°C, or as a result of cycloheximide treatment, differ qualitatively since those accumulating in the drug‐treated cells are predominantly capped (Beelman and Parker, 1994; D.Zuk and A.Jacobson, unpublished observations).
Previous studies have implicated human eIF‐5A (heIF‐5A) as a cellular co‐factor for the HIV protein Rev (Ruhl et al., 1993; Bevec et al., 1996). The latter protein is required for export of the viral transcripts from the nucleus to the cytoplasm (Malim and Cullen, 1993; Fischer et al., 1994). This suggests an alternative, albeit more complex, interpretation of our results, i.e. that there is an increase in the export of uncapped mRNAs from the nucleus after a temperature shift of ts1159 cells. These uncapped transcripts would, however, have to be impervious to the Xrn1p exonuclease (since the ts1159 cells are wild‐type for this gene), a result inconsistent with recent experiments demonstrating that uncapped HIS4 mRNAs are indeed substrates of this exonuclease (Lo et al., 1998).
Materials and methods
Yeast strains and media
The collection of ts mutants used in this study was generated from strain SS330 (MATa ade2‐101 his3‐200 tyr1 ura3‐52) by EMS mutagenesis as described in Vijayraghavan et al. (1989). A related strain, SS328 (MATα ade2‐101 his3‐200 lys2 ura3‐52), was used for crosses. SS328, SS330 and the mutants derived from the latter strain were all generously provided by John Abelson. The prp2 ts strain, RL9ZK0, and the W303 xrn1Δ strain were obtained respectively from David Elliot and Robin Ganesan. Cells were grown in rich, synthetic or sporulation media prepared as described in Sherman et al. (1986).
Temperature shifts and measurement of mRNA decay rates
For steady‐state experiments, 25 ml cultures in rich, synthetic complete (SC) or SC‐ura media (as indicated) were grown at 24°C to mid‐log phase (OD600 = 0.6–0.8). The cells were collected by centrifugation, resuspended in 7 ml of fresh medium and shaken at 24°C for 10 min. A 4 ml aliquot to be used for RNA isolation was removed, centrifuged for 10 s, and the pellet frozen quickly in dry ice. The remaining cells were shifted rapidly to 37°C by addition of 3 ml of the same medium pre‐warmed to 56°C and then shaken at 37°C for the indicated time periods, after which a 4 ml aliquot was frozen and stored for subsequent RNA isolation.
For measurement of mRNA decay rates (Parker et al., 1991), 100 ml cultures in SC medium were grown to mid‐log phase, collected by centrifugation, resuspended in 12 ml of fresh medium and shaken at 24°C for 10 min. An equal volume of medium pre‐warmed to 56°C was added and the cultures were shaken at 37°C for 1 h. Transcription was then inhibited by the addition of thiolutin (Jimenez et al., 1973; Herrick et al., 1990; a generous gift from Pfizer Inc) to a final concentration of 4 μg/ml, and 4 ml aliquots were taken for RNA extraction at the indicated times thereafter.
RNA extraction and Northern blot analysis
RNA was isolated by the hot phenol method as described previously (Herrick et al., 1990). Aliquots (15 μg) of each RNA sample were analyzed by Northern blotting, using radiolabeled probes prepared by random priming (Feinberg and Vogelstein, 1983) with a kit from Boehringer Mannheim. mRNA decay rates, expressed as half‐lives (t1/2), were determined by direct counting of the blots with a Betagen blot analyzer, normalization of the data such that time zero after addition of thiolutin equaled 100%, and plotting the data with respect to time on semi‐log axes (Peltz et al., 1992). The transcripts monitored, and the DNA probes used to detect them, included: CYH2 (a 500 bp EcoRI–HindIII fragment from pGEM4ZCYH2 which hybridizes to both the pre‐mRNA and the mRNA; Herrick et al., 1990), STE2 (a 2 kb SalI fragment from pDH9; Herrick et al., 1990), PAB1 (a 3.4 kb EcoRI–SalI fragment from pDH2; Herrick et al., 1990) and URA5 (a 2 kb KpnI–EcoRI fragment from FL44; Herrick et al., 1990).
Complementation by the YCp50 library and ORF subclones
The ts1159 strain was electroporated with 1 μg of the A4 pool of a YCp50 library (7–10×103 plasmids) prepared by Mark Rose (Rose et al., 1987) using the protocol of Becker and Guarente (1991). The electroporated cells were grown on SC‐ura plates containing 1 M sorbitol at 37°C for 5 days. The complementing plasmids were isolated from yeast cells and amplified in Escherichia coli using standard procedures (Sambrook et al., 1989). The resulting plasmids were transformed back into ts1159 using the modified PEG‐LiAc procedure (Soni et al., 1993) and shown to retain complementation of the growth defect. All the complementing plasmids were cut with EcoRI and SalI (which flank the insert in the YCp50 vector) to generate a preliminary map of the inserts. Aliquots (30 μg) of 1159‐1 and 1159‐3 DNAs were sequenced with the SequenaseII kit (Gibco‐BRL) using a primer (made by Operon, CTTCGCTACTTGGAGCCACT) complementary to a region 40 nucleotides upstream of the BamHI site of YCp50 into which the library inserts had been cloned. The sequence obtained was compared with the Saccharomyces Genome Database at Stanford University (http://genome-www.stanford.edu/saccharomyces/) using the GCG BLAST program (Altschul et al., 1990).
The complementing plasmids contained seven ORFs. Using the sequence information, fragments containing a single ORF plus enough sequence of flanking regions to overlap with the neighboring ORFs were cut out of the YCp50 plasmids and cloned into the URA3‐containing vector pRS316 (Sikorski and Hieter, 1989). Each subclone was transformed into ts1159 cells using the modified PEG‐LiAc procedure (Soni et al., 1993). Eight clones from each transformation experiment were patched onto SC‐ura plates and grown at 24°C for 3 days. These plates were replica‐plated to two SC‐ura plates, one of which was kept at 24°C and the other incubated at 37°C for an additional 3 days to test for complementation of the growth defect. As a negative control, cells were transformed with the empty pRS316 vector and analyzed in the same manner. ts1159 was complemented by a plasmid expressing the TIF51A ORF (pDZ4). This plasmid contained a 742 bp RsaI–HaeIII fragment of 1159‐1 cloned into the SmaI site of pRS316, including 193 nucleotides upstream of the TIF51A translational start site and 76 nucleotides downstream of the translational stop site.
Complementation of the RNA decay defect in cells harboring the different plasmids was done by comparing levels of the STE2 mRNA in wild‐type and ts1159 cells grown at 24°C and after a 1 h shift to 37°C. These experiments utilized Northern blot analyses performed as described above for the temperature shifts.
Identification of the tif51A allele in ts1159 by gap repair
The pDZ4 plasmid was cut with HpaI (which cuts five nucleotides 3′ of the translational stop site of the TIF51A ORF) and BstEII (which cuts 69 nucleotides 5′ of the translational stop), or with HpaI and PflmI (which cuts 204 nucleotides 5′ of the translational stop) to generate plasmids containing one of two overlapping gaps in the TIF51A ORF The gapped plasmids were isolated from a 1% agarose gel and transformed into ts1159 cells using the modified PEG‐LiAc procedure (Soni et al., 1993). The repaired plasmids were isolated from the resulting ura+ colonies, sequenced with the SequenaseII kit (Gibco‐BRL), using the M13/pUC forward primer (Gibco‐BRL), which hybridizes to sequences in the pRS316 vector, and compared with the sequence of the original pDZ4 insert. Wild‐type and repaired plasmids were re‐transformed into the ts1159 cells, and complementation of the ts growth and mRNA decay phenotypes was determined as detailed for the complementation by the original clones.
Cells were grown in 200 ml of rich medium at 24°C to an OD600 of 0.6–0.8. For the 24°C experiments, cytoplasmic extracts were then prepared as described previously (Peltz et al., 1992). For temperature shifts, duplicate cultures were centrifuged, resuspended in 40 ml of medium, and shaken at 24°C for 10 min. An equal volume of medium pre‐warmed to 56°C was added and the cells were shaken at 37°C for 1 h. Extracts were prepared from these cultures in the same manner as for the 24°C cultures. The extracts were fractionated on 15–50% sucrose gradients [buffered with 50 mM Tris–acetate pH 7.4, 50 mM NH4Cl, 12 mM MgCl2 and 1 mM dithiothreitol (DTT)] which were centrifuged in a Beckman SW41 rotor at 35 000 r.p.m. for 165 min at 4°C and analyzed by continuous monitoring of A280 (Manrow and Jacobson, 1986).
Measurement of amino acid incorporation
Cells were grown in 100 ml of SC‐met at 24°C to an OD600 of 0.6–0.8, harvested by centrifugation, resuspended in 10 ml of fresh medium, and shaken for 10 min at 24°C. An equal volume of medium pre‐heated to 56°C was added and the cells were shaken at 37°C. Duplicate 1 ml aliquots were removed at the indicated time points and incubated with a mixture of 5 μCi of 35S‐translabel (ICN; 70% methionine and ∼15% cysteine) and 5 μl of 50 mM unlabeled methionine for 4 min at 37°C. Incorporation of the radiolabeled amino acids was monitored by TCA precipitation. TCA (5 ml of a 5% solution) was added to each aliquot, followed by heating at 90°C for 20 min and subsequent incubation on ice. The precipitates were collected on GF/C filters, which were washed with 20 ml of 5% TCA and 20 ml of EtOH, dried under a heat lamp and counted by scintillation spectrometry. Each experiment was repeated at least three times.
Inhibition of translation with cycloheximide
The SS330 wild‐type cells were grown in SC‐met medium to an OD600 of 0.6–0.8, divided in 12 ml cultures containing increasing amounts of cycloheximide (0–0.8 μg/ml) and shaken at 30°C for 1 h. Duplicate 1 ml aliquots from each culture were then incubated with a mixture of 5 μCi of 35S‐Translabel (ICN; 70% methionine and ∼15% cysteine) and 5 μl of 50 mM unlabeled methionine for 4 min at 30°C. Incorporation of the labeled amino acids was measured as described above. The remaining 10 ml of each culture was centrifuged for 10 s, and the pellet frozen quickly in dry ice. RNA prepared from the pellets was analyzed by Northern blotting and hybridization to the labeled URA5 and CYH2 probes (see details in RNA extraction and Northern blot analysis).
Immunoprecipitation of uncapped mRNAs
Immunoprecipitations with anti‐cap antibodies were performed essentially as described in Muhlrad et al. (1994), except that the RNA was released from the protein A–Sepharose beads (Pharmacia) by incubating in 1% SDS for 1 h at 4°C. One μl of the polyclonal anti‐7mG antibodies produced by T.W.Munns (Munns et al, 1992) and graciously provided by E.Lund was used to immunoprecipitate 10 μg of RNA prepared as described above. An equal aliquot of the RNA (input) and the supernatant and pellet fractions were ethanol‐precipitated in the presence of 10 μg of E.coli tRNA (Boehringer Mannheim) and analyzed by Northern blotting (see details above).
We are indebted to John Abelson for generously providing blots, strains, information and hospitality essential for the success of this project. We are also grateful to David Elliot for the prp2 mutant, to Robin Ganesan for xrn1Δ RNA and, in particular, to Elsebet Lund for the generous gift of anti‐7mG antibodies. We thank the members of the Jacobson lab for their comments on the manuscript. This work was supported by Research Project Grant #93‐026‐05‐GMC to A.J. from the American Cancer Society, by a postdoctoral fellowship from the Human Frontiers Science Program Organization (LT‐78/94) and by an Institutional Postdoctoral Training Grant (T32‐HD07312‐14) to D.Z.
- Copyright © 1998 European Molecular Biology Organization