Abstract
Rsm28p is a dispensable component of the mitochondrial ribosomal small subunit in Saccharomyces cerevisiae that is not related to known proteins found in bacteria. It was identified as a dominant suppressor of certain mitochondrial mutations that reduced translation of the COX2 mRNA. To explore further the function of Rsm28p, we isolated mutations in other genes that caused a synthetic respiratory defective phenotype together with rsm28Δ. These mutations identified three nuclear genes: IFM1, which encodes the mitochondrial translation initiation factor 2 (IF2); FMT1, which encodes the methionyl-tRNA-formyltransferase; and RMD9, a gene of unknown function. The observed genetic interactions strongly suggest that the ribosomal protein Rsm28p and Ifm1p (IF2) have similar and partially overlapping functions in yeast mitochondrial translation initiation. Rmd9p, bearing a TAP-tag, was localized to mitochondria and exhibited roughly equal distribution in soluble and membrane-bound fractions. A small fraction of the Rmd9-TAP sedimented together with presumed monosomes, but not with either individual ribosomal subunit. Thus, Rmd9 is not a ribosomal protein, but may be a novel factor associated with initiating monosomes. The poorly respiring rsm28Δ, rmd9-V363I double mutant did not have a strong translation-defective phenotype, suggesting that Rmd9p may function upstream of translation initiation, perhaps at the level of localization of mitochondrially coded mRNAs.
TRANSLATION initiation appears to be a key point of regulation in the expression of mitochondrial genes in Saccharomyces cerevisiae. Both the level and the location of protein synthesis within the organelle are strongly influenced by membrane-bound mRNA-specific translational activators that recognize target sites in the leaders of mitochondrially coded mRNAs (Rödel 1997; Sanchirico et al. 1998; Steel and Bussoli 1999; Green-Willms et al. 2001; Fiori et al. 2003). mRNA features necessary for the selection of translation start sites include both the initiation codon itself and other features of the mRNA (Folley and Fox 1991; Mulero and Fox 1994; Bonnefoy and Fox 2000). However, the mechanisms by which mitochondrial translation is initiated are poorly understood, owing largely to the absence of in vitro systems derived from the organelles. Furthermore, the extraordinary divergence of mitochondrial genetic systems from their eubacterial ancestors and from each other (Gray et al. 2004) limits the degree to which mechanisms can be inferred by the identification of components homologous to those of bacteria. Therefore, genetic analysis is an important tool for the further study of translation initiation in mitochondria.
The mitochondrially coded COX2 mRNA contains within its open reading frame antagonistic signals that affect translation efficiency: a positive-acting sequence within the first 15 codons (Bonnefoy et al. 2001) and inhibitory sequence elements further downstream (Williams and Fox 2003). Mutations in the positive-acting sequence strongly reduce translation of the cox2 mRNA and produce nonrespiratory growth phenotypes due to cytochrome oxidase deficiency. It is not known whether the mutations affect initiation, elongation, or both. These cox2 mutations can be suppressed by compensating mutations in the COX2 reading frame, overproduction of the COX2 mRNA-specific translational activator Pet111p, overproduction of the large subunit mitochondrial ribosomal protein MrpL36p, and by a dominant chromosomal mutation that alters the structure of the small subunit mitochondrial ribosomal protein Rsm28p (Bonnefoy et al. 2001; Williams et al. 2004, 2005).
Rsm28p, which has no detectable homology to bacterial ribosomal proteins, is required for fully efficient translation of at least the COX1, COX2, and COX3 mRNAs as judged by expression of a reporter gene inserted into each mitochondrial locus (Williams et al. 2005). However, it is not essential for mitochondrial translation since rsm28Δ mutants are able to grow on nonfermentable carbon sources, albeit with reduced efficiency. The dominant suppressor mutation, RSM28-1, is an internal in-frame deletion of 67 codons that appears to increase or alter the activity of the protein, improving expression of the cox2 mutant mRNAs. Interestingly, RSM28-1 also weakly suppresses both cox2 and cox3 initiation codon mutations (Williams et al. 2005). These findings suggest that Rsm28p could have a positive role in translation initiation that is enhanced by the internal deletion.
To examine further the function of Rsm28p we have screened for additional mutations that enhance the translation defect caused by rsm28Δ, thereby producing synthetic respiratory defective growth phenotypes. Two of the genes identified in this screen encode mitochondrial translation initiation factor 2 (IF2) and the mitochondrial methionyl-tRNA-formyltransferase, strongly suggesting that Rsm28p in fact does have a role in translation initiation. The third gene identified in this screen, RMD9, had not previously been ascribed to any defined cellular process but is now implicated in mitochondrial gene expression.
MATERIALS AND METHODS
Yeast strains, media, and genetic methods:
S. cerevisiae strains relevant to this study are listed in Table 1. Strains used were isogenic or congenic to D273-10B (American Type Culture Collection, ATCC, no. 25627), except for YSC1178-7500474. Yeast were cultured in either complete medium (1% yeast extract, 2% bacto-peptone, 50 mg adenine/liter) or synthetic complete media (0.67% yeast nitrogen base supplemented with appropriate amino acids) containing 2% glucose, 2% raffinose, or 3% ethanol/3% glycerol. Standard genetic techniques were performed as described (Sherman et al. 1974; Fox et al. 1991; Guthrie and Fink 1991). Nonrespiring mutant strains were tested for the presence of wild-type mtDNA (rho+) by mating to rho0 strains DA1rho0 or DL2rho0 and scoring growth of the resulting diploids on a nonfermentable carbon source.
Strains used in this study
Screen for mutations creating a synthetic Pet− phenotype in the presence of rsm28Δ:
The wild-type RSM28 gene with ∼500 flanking bp on both sides was isolated by PCR amplification from strain NAB97. The resulting fragment was cleaved with XbaI and inserted into XbaI-cleaved pTSV31A, a high-copy 2μ ADE3 URA3 plasmid kindly provided by J. Pringle, to generate pCB8. Strain CAB67 was transformed with pCB8, and the transformant was subjected to mutagenesis with ethylmethane sulfonate as described (Lawrence 1991). Mutagenized cells were plated for single colonies on complete nonfermentable medium (YPEG) and incubated at 30° for 5–6 days. Nonsectored red colonies were picked and restreaked to YPEG (no. 1). These streaks were printed to complete fermentable medium (YPD) to allow plasmid loss, and the YPD plates were printed to medium containing 5-fluoroorotic acid. Following growth on 5-fluoroorotic acid, the cells lacking the plasmid pCB8 were printed to YPEG (no. 2). Putative mutants grew on YPEG (no. 1), before plasmid loss, but not on YPEG (no. 2), after plasmid loss. Putative mutants lacking pCB8 were tested for rho+ by mating to DL2rho0. rho+ putative mutants were next mated to the rsm28Δ rho0 strain NAB109rho0 to test for the presence of a recessive nuclear mutation causing a synthetic Pet− phenotype: five putative mutants gave Pet+ diploids in this cross, suggesting they had new recessive mutations. These diploids were sporulated, and in each case Pet+ segregated 2:2. MATa spores bearing each mutation were crossed back to each of the original mutants, and the phenotypes of the resulting diploids were scored. This complementation analysis indicated that the mutations identified three distinct genes.
Isolation and identification of genes interacting with rsm28Δ:
Functional genes corresponding to the mutations causing synthetic Pet− phenotypes were isolated by transformation of three mutant strains with libraries of wild-type S. cerevisiae DNA. Pet+ transformants were screened by PCR for the absence of plasmid-borne RSM28. FMT1 was identified in transformants of CAB74 bearing genomic fragments inserted into YCP50 (Rose et al. 1987). IFM1 was identified in transformants of CAB75 bearing genomic fragments inserted into pFL44L (Stettler et al. 1993). RMD9 was the gene isolated in transformants of CAB76 bearing a cDNA inserted into pFL61 (Minet et al. 1992). RMD9 was also isolated from a library of genomic fragments in YEP24 (Green-Willms et al. 1998). Chromosomal mutations were identified by PCR amplification of mutant genes from genomic DNA and sequence analysis of the total amplification products. Sequencing of the FMT1 gene from wild-type strains of the D273-10B (ATCC 25657) and S288C (Goffeau et al. 1996) genetic backgrounds revealed an error in the original reference genomic sequence (Goffeau et al. 1996). A 1-nucleotide insertion in the C-terminal coding sequence relative to the database sequence extended the predicted polypeptide from 393 to 401 residues (GenBank accession no. AY490279).
Mitochondrial isolation, subfractionation, and protein analyses:
Mitochondria were isolated and purified on Nycodenz gradients from yeast cells grown on complete medium containing raffinose as previously described (Glick and Pon 1995). Mitochondrial ribosomes were extracted from purified mitochondria as previously described (Williams et al. 2005) and layered onto a 39-ml continuous 15–30% sucrose gradient containing 100 mm NH4Cl, 10 mm Tris, 10 mm Mg acetate pH 7.4, 7 mm β-mercaptoethanol, 0.2% Triton X-100, 0.5 mm PMSF, and one complete protease inhibitor mini tablet without EDTA (Roche, Indianapolis). Gradients were centrifuged, fractionated, and subjected to SDS gel electrophoresis and Western blotting as previously described (Williams et al. 2005). Rmd9p-TAP was detected by incubation of the blots with peroxidase-anti-peroxidase soluble complex (Sigma, St. Louis). Mrp7p and Mrp13p were detected using mouse monoclonal antibodies (Fearon and Mason 1988; Partaledis and Mason 1988) as previously described (Williams et al. 2005).
In vivo pulse labeling with 35S-methionine in the presence of cycloheximide was performed as described (Fox et al. 1991) with the following modifications. Cells were grown to saturation in liquid 1% yeast extract, 2% bacto-peptone, 2% raffinose and then transferred to synthetic complete medium lacking Met [0.67% yeast nitrogen base, 0.08% CSM-Met (BIO 101, Vista, CA), 2% raffinose]. After labeling for 30 min the cells were chased with unlabeled 2.5 mm methionine for 30 min before isolation of crude mitochondria.
RESULTS
Isolation of mutations in three genes that cause synthetic Pet− phenotypes with rsm28Δ:
To explore the function of Rsm28p, we took advantage of the fact that it is dispensable for respiratory growth by looking for mutations in other genes that would cause a synthetic Pet− (respiratory defective) phenotype together with an rsm28Δ∷LEU2 mutation. Starting with an rsm28Δ, ade2, ade3, ura3 strain (CAB67) containing a multicopy plasmid bearing RSM28, ADE3, and URA3 (pCB8), we used a modification of the sectored colony screen (Bender and Pringle 1991) to identify mutants that could not grow on nonfermentable medium if the plasmid was lost (materials and methods). The screen yielded five independent recessive nuclear mutations that caused Pet− growth phenotypes only in an rsm28Δ background. These five mutations identified three complementation groups (materials and methods): two groups with two linked mutations each and one group with the remaining mutation.
Three rsm28Δ strains (CAB74, CAB75, and CAB76), each containing a mutation from one of the three complementation groups, were transformed with libraries of wild-type DNA fragments (materials and methods). Pet+ transformants were isolated, and those containing plasmids bearing RSM28 were identified by PCR and discarded. Characterization of the remaining complementing plasmids, and further analysis, revealed that this screen had identified FMT1, IFM1, and RMD9 as genes interacting with RSM28.
The dispensable mitochondrial methionyl-tRNA-formyltransferase, Fmt1p, is essential for respiratory growth in the absence of Rsm28p:
Overlapping genomic clones that complemented the respiratory defect of strain CAB74 (materials and methods) all contained the gene FMT1, encoding the mitochondrial methionyl-tRNA-formyltransferase (Li et al. 2000). We confirmed that this candidate gene was indeed the active locus by isolating a plasmid (pEHW255) bearing only the wild-type FMT1 gene and showing that it too complemented when transformed into CAB74. Finally, sequencing of this gene amplified from CAB74 genomic DNA revealed a frameshift mutation truncating the normally 401-amino-acid protein after residue 335 (Table 2). To generate a true null allele, we constructed an fmt1Δ∷URA3 complete deletion. As previously reported (Li et al. 2000), the absence of Fmt1p had virtually no effect on respiratory growth (Figure 1). However, an rsm28Δ, fmt1Δ double-mutant haploid, EHW468, failed to respire, although the cells remained stably rho+ (Figure 1). Since a complete block in mitochondrial translation destabilizes mtDNA, producing rho− mutants (Myers et al. 1985), this result indicates that residual mitochondrial translation occurs in the absence of both Rsm28p and Fmt1p. Nevertheless, the absence of formylated methionine on the mitochondrial initiator tRNA enhances the modest translational defect caused by the lack of Rsm28p.
Mutations in FMT1 cause a synthetic respiratory defect with rsm28Δ. Strains with the relevant genotypes indicated in the sector diagram were streaked to complete glucose medium and then printed to complete nonfermentable (ethanol–glycerol) and fermentable (glucose) media, followed by incubation at 30° for 6 and 1 days, respectively. The same streaks were also mated to a rho0 tester strain (DL2rho0) and then printed to complete nonfermentable medium to reveal mtDNA maintenance (crossed to rho0 ethanol–glycerol). Strains were, clockwise from upper left, NB80, EHW467, CAB67, EHW468, CAB74, EHW469 (Table 1).
Substitutions that caused synthetic respiratory defects in the absence of Rsm28p
Absence of Rsm28p sensitizes mitochondrial translation to mutations altering mitochondrial translation initiation factor 2:
Overlapping genomic clones that complemented the respiratory defect of strain CAB75 (materials and methods) all contained the gene IFM1, encoding the mitochondrial homolog of bacterial translation initiation factor 2 (Vambutas et al. 1991). We confirmed that this candidate gene was indeed the active locus by isolating a plasmid (pCB12) bearing only the wild-type IFM1 gene and showing that it too complemented when transformed into CAB75. Finally, sequencing of this gene amplified from CAB75 genomic DNA by PCR revealed the presence of a missense mutation, ifm1-Q234K (Table 2). Ifm1p (mitochondrial translation initiation factor 2) is required for normal levels of mitochondrial translation and for respiratory growth (Vambutas et al. 1991; Tibbetts et al. 2003). Since strains carrying only the ifm1-Q234K mutation can grow on nonfermentable medium, this missense mutation must alter or reduce, but not destroy, Ifm1p function.
Null mutants entirely lacking Ifm1p retain the ability to translate mitochondrially coded mRNAs, albeit at greatly reduced rates, and can therefore maintain rho+ mtDNA (Tibbetts et al. 2003). We constructed by transformation an ifm1Δ∷URA3 complete deletion mutant, CAB78, and confirmed that it retained mtDNA. Furthermore, CAB78 crossed to wild-type DUL1 yielded tetrads containing Ura−, Pet+, rho+ spores and Ura+, Pet−, rho+ spores in a 2:2 ratio, as expected. To test the phenotype of ifm1Δ∷URA3, rsm28Δ∷LEU2 double null mutants we crossed CAB78 to the rsm28Δ∷LEU2 strain CAB67. In this case, every ifm1Δ∷URA3 spore was rho−, regardless of whether it was RSM28 or rsm28Δ∷LEU2. Thus, it appears that reduced levels of Rsm28p in the heterozygous rsm28Δ/RSM28 diploid cells that formed these haploid spores prevented the spores lacking IFM1 from maintaining rho+ mtDNA, presumably due to reduced translation. This quantitative genetic interaction of null mutations is more severe than the original synthetic defective phenotype observed with the ifm1-Q234K missense mutation and confirms that Rsm28p and Ifm1p are likely to have roles in the same process.
In bacteria, initiation factor 2 stimulates binding of initiator fMet-tRNAfMet to the ribosomal small subunit, in a reaction partially dependent on formylation of the charged tRNA (Laursen et al. 2005). We therefore expected that the absence of formylation caused by an fmt1Δ mutation would have no synergistic effect on an ifm1Δ mutant. Indeed, fmt1Δ, ifm1Δ double mutants were Pet−, rho+, a phenotype indistinguishable from that of the ifm1Δ single mutant.
Partial suppression of the ifm1Δ respiratory growth defect by the dominant RSM28-1 mutation:
The synthetic defective interactions between the rsm28Δ null mutation and both fmt1 and ifm1 mutations indicate that Rsm28p may play a role in yeast mitochondrial translation initiation. To test this hypothesis further, we asked whether the dominant hypermorphic allele, RSM28-1, originally selected as a suppressor of cox2 translational defects (Bonnefoy et al. 2001; Williams et al. 2005), might also suppress the respiratory growth defect caused by the lack of mitochondrial initiation factor 2, Ifm1p. An ifm1Δ∷URA3 strain was crossed to an RSM28-1 strain, and the ability of haploid progeny in tetrads to grow on nonfermentable carbon sources was scored (Figure 2). Every tetrad contained two spores with the ifm1Δ∷URA3 mutation, but half of these spores exhibited weak respiratory growth, indicating suppression by the unlinked RSM28-1 mutation. The control cross of the ifm1Δ∷URA3 strain to a wild-type RSM28 strain produced a normal 2:2 segregation of the ifm1Δ∷URA3 Pet− phenotype, as expected (Figure 2). These results support the notion that Rsm28p has a role in mitochondrial translation initiation.
The dominant mutation RSM28-1 partially suppresses the ifm1Δ∷URA3 mutation. The ifm1Δ∷URA3 strain CAB78 was crossed with the RSM28-1 strain EHW227 (ifm1Δ × RSM28-1) and the wild-type RSM28 strain NB40-36a (ifm1Δ × RSM28). The diploids were induced to sporulate and 20 tetrads were dissected from each cross on complete glucose medium. After growth of the spore clones the plates were replicated to complete nonfermentable medium and incubated at 30° for 4 days. Five tetrads representative of the 20 analyzed from each cross are shown.
A missense mutation in RMD9, a gene of unknown function, causes respiratory deficiency in strains lacking Rsm28p:
The respiratory defect of strain CAB76, carrying a mutation in the third complementation group causing synthetic respiratory deficiency with rsm28Δ, was complemented by a plasmid from a bank of yeast cDNAs inserted into the expression vector pFL61 (Minet et al. 1992) (materials and methods). This plasmid contained the reading frame corresponding to RMD9 (YGL107C), a gene first identified as “required for meiotic nuclear division” (Enyenihi and Saunders 2003). This gene was also present on several overlapping genomic fragments that complemented the CAB76 respiratory defect.
Sequencing of the RMD9 locus from strain CAB76 revealed the presence of a missense substitution, rmd9-V363I (Table 2). This synthetic defective allele does not inactivate the gene since a strain carrying only this mutation was respiratory competent, while an rmd9Δ∷URA3 allele we constructed caused a tight nonrespiratory phenotype. (This tight respiratory defect explains the “required for meiotic nuclear division” since yeast cells must be respiratory competent to sporulate.) In the D273-10B strain background used in this study, the rmd9Δ∷URA3 mutation caused cells to become rho−, suggesting that Rmd9p is essential for mtDNA maintenance and therefore possibly for overall mitochondrial translation. However, in strains whose mtDNA lacks all known introns, the absence of Rmd9p only partially reduced the stability of mtDNA (Nouet et al. 2007). Therefore, while essential for respiratory growth, Rmd9p is not absolutely essential for residual mitochondrial gene expression.
We further examined the phenotype of the rsm28Δ, rmd9-V363I double mutant by labeling mitochondrial translation products in vivo in the presence of cycloheximide (Figure 3A). By this assay, the double mutant had modestly reduced translation relative to wild-type and both single-mutant strains. Interestingly, labeling of the cytochrome oxidase subunits Cox1p, Cox2p, and Cox3p was reduced relative to labeling of apo-cytochrome b in the double mutant. This somewhat specific reduction in cytochrome c oxidase was confirmed by spectral analysis of cytochromes in whole cells: while cytochromes a + a3 were undetectable in the double mutant, cytochrome b absorbance was still evident (Figure 3B).
Mitochondrial protein synthesis and cytochrome spectra of the rsm28Δ, rmd9-V363I double mutant. (A) Mitochondrial translation products were labeled with 35S-methionine in the presence of cycloheximide, and crude mitochondria were isolated (materials and methods). The strains were RSM28, RMD9 (PJD1); rsm28Δ, RMD9 (CAB67); rsm28Δ, rmd9-V363I (CAB76); and RSM28, rmd9-V363I (CAB104), as indicated. Samples were applied to a 15% polyacrylamide–SDS gel, which was dried and autoradiographed. The major mitochondrial translation products are indicated. (B) Low-temperature cytochrome spectra were recorded after addition of dithionite to whole cells grown on complete galactose medium at 28°, as described (Chiron et al. 2005). Absorption maxima for cytochromes a + a3, b, c1, and c are 602, 558, 552, and 546 nm, respectively. Strains were the same as in A.
Rmd9p is a mitochondrial membrane protein at least partially associated with mitochondrial ribosomes:
A large-scale study of fusion protein location (Huh et al. 2003) and proteomic analysis of yeast mitochondria (Sickmann et al. 2003) found Rmd9p in mitochondria. Submitochondrial analysis of functional myc-epitope-tagged Rmd9p revealed that it was peripherally associated with the inner surface of the inner membrane (Nouet et al. 2007). We examined the location of Rmd9p in mitochondria purified from a strain bearing a chromosomally integrated RMD9∷TAP fusion gene that placed the TAP-tag at the C terminus (Ghaemmaghami et al. 2003). This strain exhibited normal respiratory growth indicating that Rmd9p-TAP was functional. Analysis of soluble and membrane fractions (Glick 1995) of these mitochondria indicated that Rmd9p-TAP was roughly evenly distributed between them (our unpublished data). This behavior was reminiscent of mitochondrial ribosomal proteins (McMullin and Fox 1993; Williams et al. 2005).
To ask whether Rmd9p-TAP was associated with mitochondrial ribosomal subunits we subjected detergent-solubilized mitochondria to sucrose density gradient sedimentation under standard, high-salt (500 mm NH4Cl), conditions. Analysis of the gradient fractions revealed that the bulk of the Rmd9p-TAP sedimented slowly and showed evidence of proteolytic degradation. While some Rmd9p-TAP sedimented into the gradient, there were no distinct peaks (our unpublished results). This result strongly indicated that Rmd9p-TAP is not a true ribosomal protein, but did not rule out a weaker ribosomal association that could be detectable at lower salt concentrations (Datta et al. 2004). We therefore subjected solubilized mitochondria to gradient sedimentation under low-salt (100 mm NH4Cl) conditions (Figure 4). In this case we reproducibly observed a faint but distinct peak of rapidly sedimenting Rmd9p-TAP at a position where both the small ribosomal subunit marker protein Mrp13p and the large subunit marker Mrp7p cosedimented. This result indicates that at the lower salt concentration some yeast mitochondrial monosomes remain intact and that a small fraction of the Rmd9p-TAP is loosely associated with those monosomes. There was no indication that Rmd9p-TAP is specifically associated with either separated ribosomal subunit. In addition, some Rmd9p-TAP, but neither ribosomal protein marker, sedimented to the bottom fractions of the gradient. The nature of this species remains to be determined.
A small fraction of Rmd9p-TAP cosediments with mitochondrial ribosomes (monosomes). Ribosomes were extracted from purified mitochondria of strain YSC1178-7500474 (expressing Rmd9p-TAP) and sedimented through a continuous 15–30% sucrose gradient containing 0.1 m NH4Cl. Gradient fractions were precipitated and analyzed by Western blotting to detect Rmd9p-TAP, the large ribosomal subunit protein Mrp7p (Fearon and Mason 1988), and the small subunit protein Mrp13p (Partaledis and Mason 1988). Top and bottom denote the orientation of the gradient. The arrow indicates the peak fraction of intact monosomes.
DISCUSSION
While the function of mitochondrial small ribosomal subunit protein Rsm28p is not essential for mitochondrial translation or respiratory growth, previous work had suggested that it plays a role in general translation initiation and/or early steps in elongation (Bonnefoy et al. 2001; Williams et al. 2005). This study of mutations that cause synthetic respiratory defects in the absence of Rsm28p has provided strong evidence for a role in an early initiation step by identifying interacting genes encoding mitochondrial translation initiation factor 2 and the methionyl-tRNA-formyltransferase. We also identified a new gene product, Rmd9p, that, on the basis of our results and those of Nouet et al. (2007), is likely to participate in mRNA maturation and localization, as well as in translation initiation at the surface of the inner membrane. The genetic interactions observed in this study are summarized in Table 3.
Phenotypes of single and double mutants
In bacteria, translation initiation factors 1, 2, and 3 (IF1, IF2, and IF3) participate in the assembly of active initiation complexes containing the initiator fMet-tRNAfMet, mRNA, the small ribosomal subunit, and the large ribosomal subunit (reviewed in Laursen et al. 2005). IF2 is a GTP/GDP-binding protein that interacts directly with the ribosomal small subunit and the initiator fMet-tRNA and positions the tRNA in the ribosomal P site. IF1 and IF3 appear to promote dissociation of ribosomal subunits prior to initiation, assist IF2 in proofreading the fMet-tRNAfMet-initiation codon interaction, and transition the ribosome to the decoding mode.
The synthetic defective interaction between RSM28 and IFM1 can be rationalized if one assumes that Rsm28p and Ifm1p (IF2) have partially overlapping, mutually reinforcing roles in establishing productive initiation complexes and/or accurate positioning of the fMet-tRNAfMet on the mRNA and the ribosomal small subunit. The recessive missense mutation synthetically defective with rsm28Δ, ifm1-Q234K, affects a highly conserved residue in the G domain of IF2, corresponding to Escherichia coli IF2 codon 478 (Laursen et al. 2003, 2005). We do not know whether this mutation affects the presumed GTPase activity of the protein. However, ifm1-Q234K presumably reduces mitochondrial IF2 activity to the point that Rsm28p becomes necessary to promote translation initiation at a level sustaining respiratory growth. In the complete absence of both Rsm28p and Ifm1p, cells lose functional mtDNA (become rho−), indicating a complete loss of mitochondrial translation (Myers et al. 1985).
Supporting the notion that Rsm28p and Ifm1p have mutually supporting roles, we also observed a “positive” genetic interaction between RSM28 and IFM1. The dominant RSM28-1 mutation, which is an internal in-frame deletion of 67 codons that was isolated as a suppressor that improves expression of translationally defective COX2 mRNAs, appears to have increased Rsm28p activity (Williams et al. 2005). Interestingly, RSM28-1 partially suppressed the respiratory growth defect of an ifm1Δ mutation, suggesting that this apparently hypermorphic allele improved mitochondrial translation initiation in the absence of Ifm1p (IF2). Bacterial IF2 has also been reported to have an additional chaperone-like activity (Caldas et al. 2000), raising the possibility that Rsm28p might also play a role in protein folding.
Formylation of the charged initiator fMet-tRNAfMet strengthens its interaction with bacterial IF2 (Sundari et al. 1976) and, to a lesser extent, with yeast mitochondrial IF2 (Garofalo et al. 2003). E. coli mutants lacking the formyltransferase grow very poorly but are viable (Guillon et al. 1992), while yeast fmt1 mutants lacking the mitochondrial formyltransferase exhibit essentially normal respiratory growth (Li et al. 2000; Tibbetts et al. 2003). However, in the absence of Rsm28p, formylation becomes required to support a level of mitochondrial translation sufficient to sustain respiratory growth. Perhaps the absence of formylation in fmt1 mutants causes a reduction in effective mitochondrial IF2 activity, thereby producing a synthetic defect with rsm28Δ that mimics the one produced by ifm1-Q234K. This notion predicts that a double ifm1Δ, fmt1Δ mfsutant would have a phenotype (Pet−, rho+) similar to that of the ifm1Δ single mutant, a prediction that we have confirmed.
Apparent orthologs of bacterial IF2 and S. cerevisiae Ifm1p can be found in most eukaryotes and presumably perform critical functions in mitochondrial translation. One interpretation of our results is that the small subunit ribosomal protein Rsm28p has some functions in budding yeast mitochondrial initiation that are played by IF1 and/or IF3 in bacteria (Laursen et al. 2005) and IF3 in mammalian mitochondria (Bhargava and Spremulli 2005). Analysis of homologous genes in the genomes of other fungal species is largely consistent with this possibility. No genes encoding potential mitochondrial proteins homologous to bacterial IF1 have been identified in eukaryotes (Koc and Spremulli 2002), although genes coding divergent forms of this short protein could go undetected. Divergent homologs of bacterial IF3 are present in mammals and in the fission yeast Schizosaccharomyces pombe (Bhargava and Spremulli 2005). No clear homologs are present in budding yeasts, although a possible candidate of questionable significance has been reported to exist in the S. cerevisiae genome (Koc and Spremulli 2002). In contrast, genes encoding proteins homologous to Rsm28p can be identified in genomes of species closely related to S. cerevisiae, such as Kluyveromyces lactis, Candida glabrata, and Ashbya gossypii, but appear to be absent in S. pombe. However, neither RSM28 homologs nor genes encoding IF3 homologs are detectable in the genomes of C. albicans or Neurospora crassa.
The third gene identified in our synthetic defective screen with rsm28Δ was RMD9. Rmd9p is conserved among budding yeasts, but N. crassa and S. pombe appear to lack it. It is not closely related to any proteins of known function. In the D273-10B (ATCC no. 25657) strain background employed in this study, deletion of RMD9 caused cells to lose rho+ mtDNA.
RMD9 has also been isolated independently in a screen for high-copy suppressors of a temperature-sensitive oxa1 mutation (Nouet et al. 2007). Oxa1p is a highly conserved integral inner membrane protein that participates in the membrane insertion of mitochondrially encoded proteins (Bauer et al. 1994; Bonnefoy et al. 1994; He and Fox 1997; Hell et al. 1997; Kermorgant et al. 1997). The homologous bacterial protein, YidC, has similar functions with respect to the plasma membrane (Luirink et al. 2005). The Oxa1p C-terminal domain is exposed on the matrix side of the membrane and interacts with mitochondrial ribosomes, apparently facilitating cotranslational insertion of newly synthesized proteins (Jia et al. 2003; Szyrach et al. 2003; Ott et al. 2006).
Despite the genetic interactions of RMD9 with both the ribosomal protein gene RSM28 and OXA1, the roles of Rmd9p in mitochondrial gene expression remain poorly defined. Nouet et al. (2007) localized epitope-tagged Rmd9p to the inner surface of the inner membrane, where it could participate in localized translation initiation, possibly in conjunction with Oxa1p and mRNA-specific translational activators (Sanchirico et al. 1998; Naithani et al. 2003). They found that Rmd9p is not essential for all mitochondrial translation since rmd9Δ mutants containing mtDNA that lacked introns were respiratory deficient but able to remain rho+. Interestingly, these rmd9Δ, rho+ strains exhibited profoundly lowered levels of mitochondrially encoded mRNAs for respiratory complex subunits, consistent with a role for Rmd9p prior to translation initiation, perhaps in mRNA processing, localization, and/or stabilization. Similar roles have been ascribed to the mitochondrial proteins Nam1p (Groudinsky et al. 1993; Wallis et al. 1994) and Sls1p (Bryan et al. 2002; Rodeheffer and Shadel 2003).
Our isolation of the recessive rmd9-V363I missense mutation as an enhancer of the rsm28Δ phenotype suggests that one function of Rmd9p could be participation in mitochondrial translation initiation, an hypothesis supported by the fact that a fraction of Rmd9p was associated with assembled mitochondrial monosomes during sedimentation of solubilized organelles in low salt. However, Rmd9p is not a ribosomal protein. Furthermore, analysis of mitochondrial translation products in the rsm28Δ, rmd9-V363I double mutant revealed only a modest decrease in protein synthesis, and this effect was more pronounced for the three cytochrome c oxidase subunits than for cytochrome b. These data are consistent with the possibility that the mutant protein Rmd9p-V363I causes a modest defect in delivery of mitochondrially coded mRNAs to their proper location on the inner membrane and reduces the efficiency of translation initiation, an effect that is exacerbated by loss of RSM28 and the resulting dependence of mitochondrial translation initiation on IF2 alone.
In view of the known functions of the other genes isolated in our screen, IFM1 and FMT1, we propose that the budding yeast-specific protein Rsm28p functions in mitochondria in the assembly of active translation initiation complexes capable of directing nascent chains into the inner membrane. Rmd9p is likely to function in delivering mRNAs to initiation complexes and in the initiation process itself. The participation of these unique budding yeast proteins in this otherwise highly conserved process (Lomakin et al. 2006) is apparently a reflection of the high degree of evolutionary divergence observed among mitochondrial genetic systems (Gray et al. 2004).
Acknowledgments
We thank F. Lacroute for the generous gift of genomic and cDNA libraries. E.H.W. was a Howard Hughes Medical Institute Predoctoral Fellow. This work was supported by grants from the Association Française contre les Myopathies (to N.B.) and the National Institutes of Health (grant GM29362 to T.D.F.).
Footnotes
↵1 Present address: Department of Developmental Biology, Stanford University School of Medicine, Stanford, CA 94305.
Communicating editor: M. D. Rose
- Received August 9, 2006.
- Accepted December 19, 2006.
- Copyright © 2007 by the Genetics Society of America
