In eukaryotes, release factors 1 and 3 (eRF1 and eRF3) are recruited to promote translation termination when a stop codon on the mRNA enters at the ribosomal A-site. However, their overexpression increases termination efficiency only moderately, suggesting that other factors might be involved in the termination process. To determine such unknown components, we performed a genetic screen in Saccharomyces cerevisiae that identified genes increasing termination efficiency when overexpressed. For this purpose, we constructed a dedicated reporter strain in which a leaky stop codon is inserted into the chromosomal copy of the ade2 gene. Twenty-five antisuppressor candidates were identified and characterized for their impact on readthrough. Among them, SSB1 and snR18, two factors close to the exit tunnel of the ribosome, directed the strongest antisuppression effects when overexpressed, showing that they may be involved in fine-tuning of the translation termination level.
TRANSLATION termination is the step that liberates the newly synthesized polypeptide from the ribosome before recycling the translational machinery. Three triplets—UAA, UAG (Weigert and Garen 1965), and UGA (Brenner et al. 1967)—were identified as nonsense stop codons and shown to serve in vitro as signals for the release of polypeptide from the ribosome (Takanami and Yan 1965). The misincorporation of an amino acid at the stop codon occurs at a frequency of ∼10−4 and is called readthrough. The efficiency of this termination is modulated by cis and trans factors. In general, release factors efficiently recognize the termination codons, but in certain instances, near-cognate transfer RNAs (tRNAs) overcompete and lead to readthrough. tRNA decoding of a stop codon occurs more frequently when the stop codon is surrounded by a context that modifies the competition for stop codon recognition between a release factor and near-cognate tRNA (Salser 1969; Fluck and Epstein 1980; Engelberg-Kulka 1981). In Saccharomyces cerevisiae, both 5′ and 3′ sequences play a role in translation termination (Bonetti et al. 1995; Namy et al. 2001; Tork et al. 2004). Several studies point to different elements that could be involved in the 5′ effect in S. cerevisiae: (i) the tRNA located on the ribosomal P-site (Mottagui-Tabar and Isaksson 1998), (ii) the mRNA structure shape due to the nucleotide sequence at the P site that could alter decoding through distortion of the ribosome structure (Tork et al. 2004), and (iii) the chemical property of the amino acid at the penultimate position. Previous analyses have shown that the nucleotides 3′ of the stop have a predominant role on readthrough efficiency and that the 5′ context effect is dependent on the 3′ context (Skuzeski et al. 1991; Bonetti et al. 1995; Howard et al. 1996; Mottagui-Tabar and Isaksson 1998; Cassan and Rousset 2001; Namy et al. 2001). In particular, the nucleotide immediately following the stop is highly biased in prokaryotes and eukaryotes and it has been proposed that the stop signal could involve four nucleotides (nt) (Brown et al. 1990). Several studies have pointed to at least three nt upstream and six nt downstream of the stop to be involved in determining readthrough efficiency (Bonetti et al. 1995; Namy et al. 2001). Aminoglycosides can increase readthrough and have been shown to suppress premature stop mutations in several animal and cultured cell models (Bedwell et al. 1997; Barton-Davis et al. 1999; Manuvakhova et al. 2000; Bidou et al. 2004). These observations have opened the possibility of treating patients who bear nonsense mutations with aminoglycoside antibiotics to express full-length protein. Given the numerous human diseases caused by nonsense mutation (Krawczak et al. 2000), it is thus imperative to determine the precise mechanism of translation termination in eukaryotes.
In eukaryotic cells, termination necessitates the recruitment of the release factors eRF1 and eRF3 by the ribosomal machinery at the A-site. eRF1 is involved in stop codon recognition but fully efficient termination needs interaction with the GTPase eRF3. In 1994, Frolova and coworkers showed that SUP45 protein of S. cerevisiae belongs to a highly conserved eukaryotic protein family and corresponds most likely to the yeast eRF1 (Frolova et al. 1994). That assignment was subsequently experimentally demonstrated by Stansfield et al. (1995b). eRF1 comprises three domains: the N-terminal domain involved in stop codon recognition (Bertram et al. 2000; Song et al. 2000; Chavatte et al. 2001) and the M domain that contains a GGQ motif highly conserved throughout evolution (Frolova et al. 1999), which is responsible for peptidyl transferase hydrolytic activity. These two domains form the functionally active “core” (Frolova et al. 2000). The third domain in eRF1, the C-terminal domain, is involved in the interaction with the protein phosphatase PP2A (Andjelkovic et al. 1996) and with eRF3 (Stansfield et al. 1995b; Zhouravleva et al. 1995). eRF3, encoded by SUP35 in S. cerevisiae, is made up of three domains. The N-terminal and M domains are not essential for viability and termination (Ter-Avanesyan et al. 1993). In S. cerevisiae, the N terminus is asparagine and glutamine rich and underlies the conformational changes of eRF3 to proteinase-resistant aggregates, leading to the [PSI+] phenotype (see review in Patino et al. 1996; Paushkin et al. 1996; Chernoff 2001; Cosson et al. 2002). [PSI+] cells present a defect in translation termination characterized by an omnipotent nonsense suppression phenotype (Liebman and Sherman 1979). The C-terminal domain carries GTPase activity (Frolova et al. 1996), which is essential for viability and termination and interacts with eRF1 and Upf1 (Stansfield et al. 1995b; Weng et al. 1996; Czaplinski et al. 1998). Recently, Salas-Marco and Bedwell (2004) showed that eRF3 mutants with a reduced GTPase activity lead to a decreased translation termination efficiency. Recent results suggest that a stable interaction between eRF1 and the stop codon in the A-site stimulates eRF3 GTP hydrolysis, which leads to efficient release of the polypeptide from the ribosome by eRF1 (Salas-Marco and Bedwell 2004; Alkalaeva et al. 2006). In spite of genetic, biochemical, and crystallographic analyses of eRF1 and eRF3, questions about the translational termination mechanism remain. In particular, several factors have been demonstrated to interact with the termination process, either directly through contacts with release factors or indirectly, as demonstrated by genetic experiments. This is the case for the Upf1p factor that physically interacts with release factors eRF3 and eRF1. The two other Upf factors (Upf2p and Upf3p) are also connected with translational termination through a mechanism not well identified (Weng et al. 1996; Czaplinski et al. 1998; Wang et al. 2001). An interaction of eRF1 with PABp has also been shown in Xenopus and human cells (Cosson et al. 2002) and could help recycling of translational components. In addition, Itt1p (Urakov et al. 2001) and PP2A (Andjelkovic et al. 1996) have been described to interact with eRF1, but without clue on the mechanism of translational termination mediated by these interactions. Several observations also suggest a link between termination and the cytoskeleton. Sla1p is involved in the cytoskeleton and has been found to interact with the N-terminal domain of eRF3 (Bailleul et al. 1999). Actin mutants have been associated with increased readthrough on the UAA stop codon (Kandl et al. 2002), and a microtubule binding protein of the spindle pole body Stu2p has been identified in a genetic screen for factors modulating translational termination efficiency (Namy et al. 2002).
Apart from the above-mentioned proteins, one can envision that other factors able to modulate the termination process remain to be discovered. Indeed, overexpression of yeast eRF factors, Sup45p and Sup35p, increases translational termination efficiency no more than 2.6-fold (Stansfield et al. 1995b; Williams et al. 2004). To identify antisuppressors limiting near-cognate, tRNA-mediated suppression, we developed a screen for factors that would increase translational termination when overexpressed (multicopy antisuppressors). For this purpose, we used a strain that carries an allele of the ADE2 gene, interrupted by an in-frame UAG stop codon surrounded by sequences known to promote a readthrough level high enough to obtain white colonies. We screened for candidate DNA fragments able to confer a red color to the colonies. Among those, SSB1 and snR18 sequences were found repeatedly and have been shown to actually decrease the readthrough level. The mechanism of SSB1-induced readthrough decrease has been further characterized.
MATERIALS AND METHODS
Yeast strains and media:
The S. cerevisiae strains used for this work are OL556 (MATa/MATα, cdc25-5/cdc25-5 his3/his3 leu2/leu2 trp1/TRP1 rca1/rca1 ura3/ura3) (Boy-Marcotte et al. 1996), 74D694 (MATa ade1-14 trp1-289 leu2-3,112 his3-200 ura3-52)[psi−]; [psi+] (Derkatch et al. 1998), MT557/3b (MATα ade2-1 sup45-2 leu2-3,112 ura3-1 his5-2) (Stansfield et al. 1995a), and FS1 (MATα, ade2-592 lys2-201 leu2-3,112 his3-200 ura3-52) (Namy et al. 2001).
The modified FS1strain used in the screen was constructed as follows: From the ADE2 gene and its promoter cloned in a centromeric URA3 vector (pFL38), a readthrough sequence derived from tobacco mosaic virus (TMV) (GGAACACAATAGCAGTTACAG) was cloned in the unique HpaI restriction site located within the coding sequence of the ADE2 gene (Namy et al. 2001). A homologous recombination in the FS1 strain at the ADE2 locus was performed with this vector linearized by enzymatic restriction. The recombined white clones were selected on complete medium depleted in adenine due to the recovery of the activity of the synthesized Ade2p protein. The correct integration was verified by sequencing of the genomic allele.
The strains were grown in minimal media supplemented with the appropriate amino acids to allow maintenance of the different plasmids after transformation. Yeast transformations were performed by the lithium acetate method (Ito et al. 1983). Color screening was performed on plates containing a drop-out medium, complete supplemented medium (CSM) (Bio 101), with all amino acids and 10 mg/liter adenine. The color intensity was checked after incubation for 5 days at 30°. 5-FOA was added at a final concentration of 1.5 mg/ml to select the loss of URA3 plasmids.
Plasmids and molecular biology methods:
A yeast genomic DNA library was kindly provided by François Lacroute. It was constructed by partial restriction of genomic DNA by SauIIIA from the S288c strain, and then fragments were ligated into the BamHI site of the pFL44L multicopy vector (Bonneaud et al. 1991). pAC derivatives were constructed by cloning the fragment of interest in the unique MscI site between LacZ and Luc open reading frames (ORFs) of pAC99 (Stahl et al. 1995; Bidou et al. 2000).
The identification of the candidate genes was obtained by release of plasmid DNA from yeast, as already described by Hoffman and Winston (1987), and used to transform Escherichia coli strain DH5α. Plasmid DNA was extracted from transformants, and boundaries of the insert were sequenced using −21M13 and M13 reverse primers. This allowed us to determine the coordinates of the genomic region and to identify the ORFs and genes present on the insert by comparison with data from the Saccharomyces Genome Databank.
The construction of the mutated SSB1 coding sequence was realized as follows: A mutagenesis on pUC-SSB1cds using a high-fidelity Taq DNA polymerase Pfu from Stratagene (La Jolla, CA) was done for SSB435, a couple of oligonucleotides, 435w (CAAGAGAAGAACCTTTACTA CAGTCGCTG ACAACCAAACCACCGTTC) and 435c (GAACGGTGGTTTGGTTGTCAG CGACTGTAGTAAAGGTTCTTCTCTTG); for SSB436, 436w (GAGAAGAACCTTTACT ACATGTAGTGACAACCAAACCACCGTTCAATTCCC) and 436c (GGGAATTGAACG GTGGTTTGGTTGTCACTACAT GTAGTAAAGGTTCTTCTC); and for SSBCA, CAw (CCATCAAGAGAAGAACCTTTACTACAGTCAGTGACAACCAAACCACCGTTCAAT TCCC) and CAc (GGGAATTGAACGGTGGTTTGGTTGTCACTGACTGTAGTAAAGGT TCTTCTCTTGATGG). The mutated pUC-SSB1cds vectors, after control of the sequence of the mutated region using as primer of sequence the oligonucleotide SSBseq2, have been digested by BglII and AgeI restriction enzymes to be cloned at the same sites in the pUC-SSB. The sequence of the mutated pUC-SSB was verified. Then the SSB-mutated sequences under its own promoter were cloned in pFL44L vector following the same procedure as for wild-type SSB1 under its own promoter.
The construction of SUP45 on multicopy vector was realized from the pSP35-45 with SUP35 and SUP45 under control of their own promoter (Bidou et al. 2000). The SUP35 and SUP45 with their promoter were inserted into the multicopy pHS8 vector at the PvuII restriction site and were called pHS35-45. The SUP45 with its own promoter was purified from agarose gel after digestion of pHS35-45 by XbaI and cloned at the same restriction site into the pHS8 vector, and this vector was called pHS-SUP45.
The construction of the eEF1Bα coding sequence under the cyc1 promoter was realized as follows: Total RNA from the Fy S. cerevisiae strain was extracted from 5 ml of exponential yeast culture (Schmitt et al. 1990) treated by 10 units of RNase free DNase I (Boehringer Mannheim, Indianapolis) at 37° for 1 hr. DNase I was inactivated by heating at 90° for 5 min, as recommended by the manufacturer. RNA was reverse transcribed with random primer with a Superscript II kit (Invitrogen, San Diego) for amplification of eEF1Bα coding sequence with high-fidelity Taq DNA polymerase Pfu from Stratagene using eEF1Bα AUG (ATGGCATCCACCGATTTCTC) and eEF1Bα UAA (TTATAATTTTTGCATAGCAG) as primers. The amplimer was cloned in pUC19 vector at the HincII restriction site and a recombinant vector called pUC-eEF1Bαcds was sequenced using −21M13, M13 reverse primers. The eEF1Bα coding sequence cloned in pUC-SSB1cds vector was cut by PstI and Ecl136II restriction enzymes to be cloned at the same restriction site in pCM189 vector. The eEF1Bα coding sequence under the cyc1 promoter was then cloned in the pFL44L vector at the SmaI site by enzymatic restriction of the pCM-eEF1Bαcds vector with Eco47III and HindIII filled by Klenow enzyme. Recombinant clones in the right orientation without an intron were verified by amplification, enzymatic restriction, and sequencing of the junction site.
The small nucleolar RNA (snoRNA) snR18 was also cloned under the cyc1 promoter as the eEF1Bα coding sequence but in the first step using snRw (TAAGCATCCACCGATTTCTCCAAGATTG) and snRc (TTAGGTTGAACCATCTGGAGAATTTCTGGG) to amplify genomic DNA from the Fy S. cerevisiae strain with high-fidelity Taq DNA polymerase Pfu from Stratagene. All constructs were verified by sequencing the region of interest using the Big Dye terminator kit and were migrated on an ABI310 automatic sequencer (Applied Biosystems, Foster City, CA).
Quantification of readthrough efficiency:
Luciferase and β-galactosidase activities were assayed in the same crude extract as previously described (Stahl et al. 1995). All the quantification was the median of at least five independent measurements. The efficiency is defined as the ratio of luciferase activity to β-galactosidase activity. To establish the relative activities of β-galactosidase and luciferase, the ratio of luciferase activity to β-galactosidase activity from an in-frame control plasmid was taken as a reference. Efficiency of readthrough, expressed as percentage, was calculated by dividing the luciferase/β-galactosidase ratio obtained from each test construct by the same ratio obtained with an in-frame control construct (Bidou et al. 2000).
The genetic screen used was based on the ability to monitor termination efficiency through the expression of the ADE2 gene, which encodes the P-ribosyl-amino-imidazole-carboxylase (EC 18.104.22.168), responsible for the degradation of the red pigment amino imidazole ribotide. The screen was performed in a FS1 strain where the ade2 gene is interrupted by an in-frame UAG stop codon derived from the TMV leaky context (for the strain construction, see materials and methods). This context promotes 15% of readthrough that leads to an expression of Ade2p sufficient to degrade its red substrate and obtain white colonies (Namy et al. 2001). This strain was transformed with a S. cerevisiae genomic library cloned on the multicopy pFL44L vector and plated on minimal medium supplemented with CSM containing a minimum quantity of adenine (10 mg/liter), allowing healthy growth with the optimization of red color. This allowed us to isolate “antisuppressor” factors in a single step. Of 52,000 transformants, 26 displayed a red color after 5 days at 30°. To check whether the antisuppressor phenotype of the clones was due to the presence of an overexpressed gene, they were plated in the presence of 5-FOA, which selects for cells that have lost the plasmid. All of the isolated candidates, except no. 17, reversed the phenotype after 1 week; this isolate was kept to serve as a negative control in further experiments. This result demonstrates that, for the vast majority of the candidates, the effect was dependent on the continuous presence of the vector. This point is important since it ruled out the involvement of a cytoplasmic factor, which might have been induced by an overexpressed gene. For each of the 25 confirmed candidates, sequencing of the fragment boundaries was performed, allowing identification of the inserted genomic fragment. The complete list of the genes present on these 25 fragments is presented in Table 1. Seven are known to be involved in different aspects of translation: ribosomal protein, translation termination factor, translation initiation factor, elongation factor, tRNA, and Hsp70 chaperone.
Many irrelevant factors might modify the pigment accumulation in cells. Among these are the enzymes early in the adenine biosynthesis pathway, proteins involved in vacuole permeability where the pigment accumulates, factors involved in controlling the efficiency of translation initiation, etc. To identify factors actually involved in translation termination, we used an independent reporter system and quantified readthrough efficiency in the presence or absence of the candidate plasmids.
These vectors (pAC) carry a dual lacZ-luc reporter interrupted by a unique cloning site at the junction of the two coding sequences where in-frame stop codons in different contexts are inserted. The SV40 promoter, known to be active in both S. cerevisiae and mammalian cells (Camonis et al. 1990), drives the expression. β-Galactosidase that originates from translation upstream of the stop codon is used as an internal control, recapitulating the different levels where expression could be modulated. The firefly luciferase activity depends on translation downstream of the stop codon and allows precise quantification of readthrough (Stahl et al. 1995; Bidou et al. 2000). Under these conditions, the ratio of firefly luciferase to β-galactosidase activities reflects the readthrough efficiency without interference from other levels of control. To obtain absolute readthrough levels, the values obtained with the test constructs are normalized against results from a similar dual lacZ-luc reporter gene where the stop is replaced with a sense codon. Each of the 26 vectors was cotransformed, with a pAC vector bearing a UAG stop codon, into the FS1 strain. For each cotransformation, five independent assays with two independent clones were performed. The readthrough level was quantified and compared to that obtained in the presence of an empty pFL44L vector. As shown in Table 1, the readthrough level decreases in the presence of each of the 25 confirmed candidates. To assess if the difference of readthrough level between candidates and strain FS1 transformed by the empty pFL44L vector is significant, we performed a nonparametric statistical test (Mann–Whitney). Candidate no. 17, like the empty pFL44L cloning vector, did not affect readthrough, which validated the screen. According to the Mann–Whitney test, a difference of at least 20% on termination translation between the candidate and the empty vector is needed to consider an effect as significant. Two candidates do not exhibit a significant difference in readthrough efficiency compared to the negative controls: isolate no. 3 carrying the transcriptional activator gene MSN1 and isolate no. 12 bearing the HEK2 gene involved in translation initiation. On the other hand, isolate no. 7 harboring SUP35 (eRF3), isolate no. 9 carrying RPL20B, isolate no. 18 carrying tRNAser (IGA), the five isolates carrying the SSB1 gene, and the two isolates carrying the eEF1Bα gene display a statistically significant difference. Altogether, 12 of the 25 candidates directed a significant decrease of readthrough efficiency, which indicates that the screen based on ADE2 activity was highly stringent.
eEF1Bα is the β-subunit of the eukaryotic translation elongation factor 1 (eEF1), which is highly conserved both functionally and structurally among species (Le Sourd et al. 2006). In yeast cells, eEF1 is a heterotrimer containing three units, responsible for binding the amino-acylated tRNA to the ribosomal A-site and also participating in the proofreading of the codon–anticodon match; eEF1A is a classic G protein involved in the GTP-dependent binding of amino-acylated tRNA, and the eEF1Bα subunit, associated with the γ-subunit, functions as a guanine exchange factor in vitro and catalyzes the exchange of GDP for GTP on eEF1A to recycle it. The function of eEF1B has been described as critical in regulation of eEF1A activity, translational fidelity, translation rate, and cell growth (Le Sourd et al. 2006). The eEF1Bα gene, in addition to encoding the eEF1Bα translation factor, contains an intervening sequence encoding the small nucleolar RNA snR18. To determine whether the increased expression of eEF1Bα protein or snR18 is involved in the decrease of readthrough, we have cloned two different versions of the coding region in a multicopy vector pFL44L under the strong Cyc1 promoter: (i) the open reading frame of eEF1Bα without the intron and (ii) the intron containing snR18 surrounded by only 80 nt of the eEF1Bα coding sequence and lacking the ATG. These constructs were transformed into the FS1 parental strain. As shown in Figure 1, only the construct carrying snR18 was able to restore the antisuppression phenotype. The readthrough efficiencies directed by these two constructs was quantified using the three stop codons in the same surrounding context cloned in the pAC vector as targets. Strains cotransformed with the two pFL44L constructs were compared to strains cotransformed with the empty vector. Results presented in Table 2 show that, with the construct expressing only the snR18 matured from the eEF1Bα intron, there is a significant decrease of readthrough (31–20%, P = 0.03) on the UAG stop codon and a slight decrease of readthrough on the UAA and UGA stop codons (11–10%, P = 0.045 and 15–10%, P = 0.025, respectively). Interestingly, a strong increase of readthrough on UAA and UGA stop codons is observed upon overexpression of the eEF1Bα open reading frame without an intron (11–27%, P = 0.012 and 15–38%, P = 0.006, respectively). This is reminiscent of previous observations by Carr-Schmid et al. (1999). Altogether, we have observed that the increase of termination, especially for the UAG stop codon, involves the small nucleolar RNA snR18, and not eEF1Bα.
As mentioned above, the five candidates carrying the SSB1 gene and YDL228c were all able to efficiently decrease the readthrough level (from 35 to 51%) when cotransformed with pAC-TMG (compared to the empty vector pFL44L). To establish if this effect could be attributed to SSB1 or YDL228c overexpression, we cloned the SSB1 coding sequence in a centromeric vector under the strong Cyc1 promoter (pCMSSB1) and tested the readthrough efficiency in the presence of this construct in different S. cerevisiae strains. Results presented in Figure 2 show that overexpression of SSB1 reproduces the effect observed with the entire DNA fragment, although translational termination increases to a greater extent when expressed from the multicopy vector than from a centromeric vector (2- and 1.6-fold, respectively, in the FS1 strain). Overexpression of SSB1 from a multicopy vector has a significant effect on termination in 74D694 [psi−] and FS1 strains (1.3- to 2-fold, respectively), but surprisingly not in the [psi+] state in the 74D694 strain (Figure 2). This may be related to different levels of accumulation of the Sup35 protein in these two strains.
To better characterize the effect of SSB1 protein, we tested a panel of recoding targets corresponding to different stop codons and surrounding sequences in the FS1 strain: the UAG stop codon targets corresponding to the TMV termination context, a derived sequence (TMG) where CAA on each side of the stop were replaced by CAG, and the MoMuLV context including the UAG stop and the downstream pseudoknot. In all cases, a 2-fold increase of translation termination is observed upon SSB1 overexpression (Table 2). The effect of SSB1 overexpression on termination was also examined in the three stop codons in the TMV context. Although an effect is observed in all cases, its extent varies, from a 1.5-fold increase with UGA to a 4-fold increase with UAA.
We also tested the effect of overexpressing the paralogous SSB2 gene under its own promoter cloned in the multicopy vector pYESSB2 and compared the same vector expressing SSB1 (kindly provided by S. Rospert) on the same set of readthrough targets. As shown in Table 2, the effect on the level of translation termination was significantly less pronounced for Ssb2p than for Ssb1p.
As mentioned above, overexpression of release factors in yeast has been shown to have only a moderate effect on translation termination. To compare the extent of the effect directed by Ssb1p and release factor overexpression, we cloned both SUP45 and SUP35 genes on the same multicopy vector and quantified the readthrough efficiency. As for Ssb1p, a threefold increase was obtained upon Sup35p and Sup45p overexpression on the UAA stop codon in the TMV readthrough context. This confirms the relatively weak effect of the overexpression of both factors on termination.
To evaluate whether Ssb proteins are actually involved in the termination process, we used the MT556/3b strain that carries a sup45 thermo-sensitive allele and determined whether Ssb1p overexpression could revert the phenotype. The readthrough level was quantified at 30° in the presence or absence of the overexpressed release factor eRF1 or/and the Ssbp chaperones. The Sup45 thermo-sensitive S. cerevisiae strain was first transformed with multicopy URA3 vectors carrying either SUP45 alone or SUP45 and SUP35. The MT556/3b strain was then transformed with pYESSB1 or pYESSB2 or the empty pFL44L vector. Figure 3 shows that a very high readthrough level is obtained with all three stop codons in the TMV context in the MT557/3b strain (27% on UGA, 33% on UAA, and 69% on UAG stop codon). The readthrough efficiency decreases ∼10-fold for all stop codons in the presence of overexpressed wild-type Sup45p protein, but is not affected by Ssb1p or Ssb2p overexpression (Figure 3). We then examined the thermo-sensitive phenotype. At 30°, all transformants grow on liquid culture with a generation time of 2 hr. At 37°, only cells transformed with SUP45, with SUP45 and SUP35, or with SSB1 are able to grow, but no growth is detected in cells transformed with pFL44L or SSB2. The generation time is 8 hr with SSB1 and 2 hr with SUP45 at nonpermissive temperature (Table 3). Thus, overexpression of Ssb1p, but not of Ssb2p, allows a partial recovery of the thermo-resistance in the Sup45 thermo-sensitive strain (Figure 4). To determine if suppression of the thermo-sensitive phenotype is a general chaperone effect of Ssb1p, we tested in the same manner the ability of Ssb1p to revert thermo-sensitivity of a cdc25 thermo-sensitive mutant in the strain OL556. We observe no reversion of thermo-sensitive phenotype in this strain in presence of Ssb1p overexpression (data not shown).
The difference in the ability of Ssb1p and Ssb2p to revert the sup45p thermo-sensitivity allows determining if one or more of the amino acids that differ between the two proteins could play a role in the ability to revert the thermo-sensitivity. An aspartate is found in Ssb1p and a glutamine is found in Ssb2p at position 49 in the ATPase domain. Three others are found in the polypeptide-binding domain and correspond to methionine at position 413, cysteine at position 435, and alanine at position 436 in Ssb1p compared to isoleucine, valine, and serine at the corresponding positions in Ssb2p (Figure 5). Methionine and isoleucine are neutral amino acids, but cysteine in SSB1 could form a disulfide bridge, which is not possible in Ssb2p. The cysteine in position 435 was mutated to valine or the alanine in position 436 was mutated to serine in Ssb1p protein. A third mutant was constructed with both mutations. Overexpression of any of these mutated forms is not able to revert the thermo-sensitive phenotype (Figure 4).
In this work, we identify eEF1Bα and SSB1 genes as able to increase translation termination efficiency when overexpressed. Since the eEF1Bα gene that encodes the β-subunit of the eukaryotic translation elongation factor 1 also carries an intron encoding the snR18 snoRNA, we uncoupled expression of these two factors and showed that the effect on termination is directed by the snR18. In the following sections, we shall discuss the results obtained with the two genes independently, although they may have similar mode(s) of action since both potentially affect the peptide exit tunnel region of the ribosome.
The maturation of preribosomal RNA in the translational machinery involves a large number of cleavage events, which frequently follow alternative pathways. In addition, ribosomal RNAs (rRNAs) are extensively modified, with the methylation of the 2′-hydroxyl group of sugar residues and conversion of uridines to pseudouridines being the most frequent modifications, although the extent of the modification event (i.e., the proportion of modified ribosomes) is unknown and possibly variable (Grosjean 2005). In particular, the degree of modification has been shown to vary with growth temperature in certain Archaea, plants, and trypanosomes (Brown et al. 2003; Omer et al. 2003; Uliel et al. 2004). In humans, it has been shown that the 5.8S rRNA is hypo-2′-O-methylated in neoplastic tissues (Munholland and Nazar 1987). Both cleavage and modification reactions of pre-rRNAs are assisted by a variety of an abundant class of trans-acting RNAs, snoRNAs, which function in the form of ribonucleoprotein particles (snoRNPs). The majority of snoRNAs act as guides directing site-specific 2′-O-ribose methylation or pseudouridine formation by base pairing near target sites. Eukaryotic rRNAs display a complex pattern of ribose methylations. Ribose methylations of eukaryotic rRNAs are each guided by a cognate small RNA, belonging to the family of box C/D antisense snoRNAs, through transient formation of a specific base pairing at the rRNA modification site. Over 100 RNAs of this type have been identified to date in vertebrates and the yeast S. cerevisiae. Many snoRNAs are produced by unorthodox modes of biogenesis, including salvage from introns of pre-mRNAs, as for snR18 in yeast, or from non-protein-coding transcripts. In yeast, however, numerous snoRNAs are generated from independent transcription units.
The snoRNA snR18 is 102 nucleotides long and guides the methylation at the sites corresponding to the A647/C648 positions of the 25S rRNA (Lowe and Eddy 1999). We have shown here that overexpression of snR18 induces increased termination efficiency. Different mechanisms may account for this observation. If the effect is direct, it could act through a hyper-modification of the A647/C648 position. This would imply that, in the normal situation, not all ribosomes are methylated at this position and that modified ribosomes are more efficient terminators than unmodified ones. Although the proportion of ribosomes methylated at A647/C648 is not known, it has been shown for other positions that only a fraction of ribosomes are modified (Grosjean 2005). Alternatively, the effect of snR18 overexpression might be indirect. A possibility would be that it acts through titration of a general factor(s)—perhaps even the methylase Nop1p—involved in the biogenesis or action of methylation snoRNPs, which could result in producing fewer snoRNP complexes, less active snoRNP complexes, and/or less stable snoRNP complexes. Such an effect might lead to hypomethylation of other position(s). The precise role of the different methylated nucleotides in the rRNA has not yet been deciphered, precluding further speculations.
While trying to identify the portion of the eEF1Bα gene region involved in the effect on readthrough, we made the interesting observation that overexpression of the eEF1Bα gene, devoid of its snR18-encoding intron, actually decreases termination efficiency. This is in full agreement with the work of Carr-Schmid et al. (1999) who have previously shown that eEF1Bα mutants exhibit an antisuppressor phenotype. They interpreted this effect as a more efficient competition for recognition of the stop codon by release factors due to an increased ratio of release factor to active eEF1A. This interpretation is strongly supported by the results presented here. Finally, it might be significant that two factors acting in an opposite way on termination are coexpressed as a single RNA, preventing an imbalance in termination efficiency.
We show that the SSB1 gene is able to increase translation termination efficiency when overexpressed. This increase is effective on the three stop codons, although to different extents, and on several stop codon contexts. Ssb1p is one of the Hsp70 homologs present in the S. cerevisiae genome. It is closely related to its Ssb2p paralogous gene. Ssb1p and Ssb2p share identical function and a similar level of expression; they differ by four amino acids (Boorstein et al. 1994). Rakwalska and Rospert (2004) showed previously that the lack of functional Ssb1/2 in yeast caused severe problems in translational fidelity, which were strongly enhanced by paromomycin and correlated with growth inhibition. Since SSB1 and SSB2 transcript levels are regulated independently of those of genes encoding ribosomal proteins (see Discussion in Muldoon-Jacobs and Dinman 2006), not all ribosomes would be associated with functional chaperone complement, and overexpression of Ssb proteins would improve the functioning of the ribosome.
The quantification of readthrough in the strain with a conditional-lethal mutant allele of SUP45 (sup45-2) demonstrates an extremely high level of readthrough on the UAG stop codon and 30% readthrough on UAA and UGA stop codons in the same surrounding environment. Ssb1p overexpression does not decrease the readthrough level in the SUP45 thermo-sensitive strain but specifically allows a partial recovery of the thermo-resistance phenotype, which suggests that this effect could be associated with the chaperone role of the Ssb1p protein in the folding of the Sup45p protein. It would, however, be insufficient to allow a significant effect on the termination capacity of the protein, since the mutation specifically affects this activity. Interestingly, overexpression of Ssb2p protein was unable to reverse the Sup45p thermo-sensitivity phenotype. The proteins differ by only four amino acids. Of these four amino acids, three are located in the polypeptide-binding domain. We analyzed the role of cysteine 435 and alanine 436 found in the Ssb1p protein by mutating them to the amino acids found in the corresponding residues of the Ssb2p protein. The mutant Ssb1p proteins are not able to reverse the thermo-sensitivity of the mutant SUP45 strain. This demonstrates that the polypeptide-binding domain is responsible for the specific effect of Ssb1p on the thermo-sensitive phenotype of the strain with allele sup45-2. Since the effect was recapitulated by mutation of only the cysteine residue, it could be dependent on the formation of a specific disulfide bridge with misfolded Sup45p. Although we cannot exclude an additional effect of the polymorphism located in the ATPase domain, this effect should be limited, since the mutants of the polypeptide-binding domain recapitulate the observed difference between the two isoforms.
Possible involvement of the polypeptide exit tunnel of the ribosome in translation termination:
Ssbp is associated with the ribosome when it is actively synthesizing proteins (Nelson et al. 1992), and it interacts with both the ribosome and directly with the nascent chain as it emerges from the ribosome. Ssb1p and Ssb2p are actually in close proximity to a variety of nascent polypeptides (Pfund et al. 1998; Gautschi et al. 2002; Rospert et al. 2002). This suggests that Ssbp functions as a chaperone for polypeptide chains during translation. Such a role in emerging polypeptides could be to facilitate the successful folding of newly synthesized proteins (Beckmann et al. 1990; Frydman et al. 1994; Hardesty et al. 1995; Eggers et al. 1997). An alternative explanation of the role of the Ssb chaperone would be through a direct role in the decoding process. As proposed by Muldoon-Jacobs and Dinman (2006), Ssb chaperone activity might help nascent peptides to back up into the exit tunnel and thus would participate in the efficient accommodation of the aminoacyl tRNA in the ribosomal A-site. Whatever the mechanism involved, it is significant that the absence of the RAC complex components, Ssz1 and zuotin, decreases translational accuracy, reinforcing the role of the ribosome-exit-tunnel-associated chaperone in decoding (Rakwalska and Rospert 2004). Whether overexpression of Ssz1 and zuotin also increases termination efficiency would be interesting to determine.
Remarkably, positions A647/C648 that are methylated by the snoRNA snR18 were included in the sequences identified as approaching the surface around the lumen of the polypeptide exit tunnel in the large ribosomal subunit (Nissen et al. 2000) (see Figure 6). Although the precise mechanism of this effect could not be inferred from the study reported here, the fact that both Ssb1p and snR18 are somehow linked to the exit tunnel might be significant regarding the termination mechanism.
We thank members of our laboratory for numerous stimulating discussions. This research was supported by grants from the Association pour la Recherche sur le Cancer (grant 3849 to J.-P.R.) and the Association Française contre les Myopathies (grants 9584 and 10683 to J.-P.R.).
Communicating editor: A. Nicolas
- Received January 11, 2007.
- Accepted April 27, 2007.
- Copyright © 2007 by the Genetics Society of America