Divergent Alanyl-tRNA Synthetase Genes of Vanderwaltozyma polyspora Descended from a Common Ancestor through Whole-Genome Duplication Followed by Asymmetric Evolution

Abstract

Cytoplasmic and mitochondrial forms of a eukaryotic aminoacyl-tRNA synthetase (aaRS) are generally encoded by two distinct nuclear genes, one of eukaryotic origin and the other of mitochondrial origin. However, in most known yeasts, only the mitochondrial-origin alanyl-tRNA synthetase (AlaRS) gene is retained and plays a dual-functional role. Here, we present a novel scenario of AlaRS evolution in the yeast Vanderwaltozyma polyspora. V. polyspora possesses two significantly diverged AlaRS gene homologues, one encoding the cytoplasmic form and the other its mitochondrial counterpart. Clever selection of transcription and translation initiation sites enables the two isoforms to be localized and thus functional in their respective cellular compartments. However, the two isoforms can also be stably expressed and function in the reciprocal compartments by insertion or removal of a mitochondrial targeting signal. Synteny and phylogeny analyses revealed that the AlaRS homologues of V. polyspora arose from a dual-functional common ancestor through whole-genome duplication (WGD). Moreover, the mitochondrial form had higher synonymous (1.6-fold) and nonsynonymous (2.8-fold) substitution rates than did its cytoplasmic counterpart, presumably due to a lesser constraint imposed on components of the mitochondrial translational apparatus. Our study suggests that asymmetric evolution confers the divergence between the AlaRS paralogues of V. polyspora.

INTRODUCTION

Accurate decoding of mRNA depends on accurate formation of aminoacyl-tRNA. Aminoacylation is carried out at the expense of ATP by a structurally diverse group of enzymes called aminoacyl-tRNA synthetases (aaRSs). Each aaRS recognizes an amino acid and a particular set of identity elements on its cognate tRNA. The identity elements of a tRNA normally include the sequence or structure of the anticodon loop and acceptor stem (1). Once an aminoacyl-tRNA is formed, it is released from the enzyme and delivered to ribosomes for protein synthesis. Watson-Crick base pairing (A to U and G to C) between the codon of mRNA and the anticodon of tRNA ensures the fidelity of mRNA decoding. Thus, aaRSs control the efficiency and fidelity of protein translation. Eukaryotes, such as mammals, require two sets of aaRSs for translation, one for the cytoplasm and the other for mitochondria (2–4). Plants require a third set for chloroplasts. Eukaryotic aaRSs are all nuclear encoded and are targeted to their respective cellular compartments after they are synthesized in the cytoplasm. In general, the cytoplasmic and mitochondrial isoforms of an aaRS are each specific to their own tRNA isoacceptors because of differential cellular localization and tRNA specificity (5). In addition to aminoacylation, some aaRSs possess noncanonical activities, including mitochondrial intron splicing, cytokine-like activity, transcriptional control, and translational regulation (6). Mutations in aaRSs are often linked to diseases such as Charcot-Marie-Tooth disease type 2D (CMT2D); Alpers encephalopathy; myopathy, lactic acidosis, and sideroblastic anemia (MLASA) syndrome; and hyperuricemia, pulmonary hypertension, renal failure, and alkalosis (HUPRA) syndrome (7–10). Thus, aaRSs have been an attractive target not only for biological research but also for medical research.

According to the endosymbiotic theory, the eukaryotic genes that encode the cytoplasmic and mitochondrial isoforms of an aaRS evolved from two different sources (11). Generally, genes encoding the cytoplasmic form are of eukaryotic origin (i.e., from archaea), while genes encoding its mitochondrial counterpart are of mitochondrial origin (i.e., from bacteria). This theory is largely true for contemporary eukaryotic aaRS genes. However, occasionally, both protein isoforms of an aaRS are specified by a single nuclear gene through alternative initiation of translation. In these instances, only one of the orthologous genes has been retained during evolution, while the other has been lost forever. Four such examples have been found in the yeast Saccharomyces cerevisiae: the alanyl- (AlaRS) (12), glycyl- (GlyRS) (13), histidyl- (14), and valyl-tRNA synthetase (ValRS) genes (15). A similar scenario occurs in the higher-eukaryotic aaRS genes, examples of which are the AlaRS, threonyl-tRNA synthetase, and ValRS genes of Arabidopsis thaliana (16) and the GlyRS gene of Homo sapiens (17). Moreover, the evolution of certain eukaryotic aaRS genes appears to have deviated from the canonical pathway. For example, Schizosaccharomyces pombe contains two distinct ValRS genes of mitochondrial origin, one specifying the cytoplasmic form and the other its mitochondrial equivalent (18); S. cerevisiae contains two distinct GlyRS genes of eukaryotic origin, one dual functional and the other dispensable (19); and Vanderwaltozyma polyspora contains two distinct AlaRS genes of mitochondrial origin, one encoding the cytoplasmic form and the other its mitochondrial isoform (20).

It has been estimated that ∼17% of bacterial genes (21) and ∼0.1% of higher-eukaryotic genes (22) use non-AUG initiator codons. Non-AUG initiator codons refer to triplets that differ from AUG by 1 nucleotide, such as GUG, UUG, AUU, and ACG (23). However, relatively few examples of non-AUG initiation have been found in lower eukaryotes. ALA1 and GRS1 appear to be the first two genes shown to use a naturally occurring non-AUG initiator codon in S. cerevisiae (12, 13). Although translation initiation from an AUG codon is not affected by its surrounding sequence in yeast (24–26), translation initiation from a non-AUG codon is greatly affected by its surrounding sequence. The best sequence context for a non-AUG initiator codon is AAA at its relative −3 to −1 positions (23, 27), and a non-AUG initiator codon requires at least an A at its relative −3 position to be properly functional. A peculiar feature of ALA1 is that the gene uses AUG1 as the translation initiation codon of its cytoplasmic form while using two upstream consecutive ACG codons (ACG/ACG) as the translation initiation codons of its mitochondrial isoform (12). Repetitive initiation from two successive ACG codons slightly enhances initiation activity (∼2-fold) and may represent a subtle mechanism for fine-tuning the efficiency of translation (28). Since the discovery of non-AUG initiation in ALA1 and GRS1, many more examples have recently been recovered in yeasts (18, 29).

As aaRSs are an ancient group of essential enzymes that still remain active in organisms, they are favorable targets for evolutionary studies (30–32). Our previous report revealed that V. polyspora possesses two distinct nuclear AlaRS genes, one encoding the cytoplasmic form and the other its mitochondrial relative. Surprisingly, these two genes, while significantly diverged in sequence, are both of mitochondrial origin (20). In this study, we further explored the expression, function, and evolution of these two isoforms in hopes of gaining further insight into their phylogenetic relationships. The results presented here show that the AlaRS genes of V. polyspora are descended from a dual-functional common ancestor through whole-genome duplication (WGD), and the mitochondrial form possessed appreciably higher substitution rates than did its cytoplasmic counterpart, most probably because lower selection pressure was imposed on constituents of the mitochondrial translation machinery. As a result, these two isoforms were not much alike in protein sequence. A similar scenario occurred in Tetrapisispora phaffii, which also possesses a second AlaRS gene.

MATERIALS AND METHODS

Plasmid construction.

Cloning of the ALA1 gene of S. cerevisiae was previously described (12, 33). To clone the ALA1 gene of V. polyspora (VpALA1) into pRS315-His₆ (a low-copy-number yeast shuttle vector with a LEU2 marker and a His₆ tag sequence), a pair of gene-specific primers was used to amplify the gene via PCR using yeast genomic DNA as the template. The forward primer, with an EagI site, is located 300 bp upstream of the first ATG codon of the open reading frame (ORF), while the reverse primer, with an XhoI site, is located immediately upstream of the stop codon. The 3.3-kb PCR-amplified DNA fragment was digested with EagI and XhoI prior to cloning into pRS315-His₆. Cloning of VpALA2, T. phaffii ALA1 (TpALA1), and TpALA2 into pRS315-His₆ followed a similar protocol. These clones were expressed under the control of their native promoters.

To make the fusion construct S. cerevisiae VAS1 (ScVAS1)-VpALA1, a presequence of ScVAS1 (bp −300 to +138 relative to ATG1) was amplified by PCR and fused in frame to the ORF of VpALA1 that had been cloned in pRS315-His₆. ScVAS1-VpALA2, ScVAS1-TpALA1, ScVAS1-TpALA2, VpALA1-VpALA2, and VpALA2-VpALA1 were constructed in a similar fashion. To make VpALA2-ScALA1^OF (the superscript “OF” means out of frame) fusion constructs, a presequence of VpALA2 (bp −300 to −1, −300 to −22, −300 to −43, or −300 to −67) was amplified by PCR as an EagI-SpeI fragment and fused out of frame to the ORF of ScALA1^OF (bp −1 to +2874) that had been cloned in pRS315-His₆. Note that the ORF of ScALA1^OF starts from bp −1 instead of +1.

Cloning of the VpALA1 and VpALA2 ORFs into pADH (a high-copy-number yeast shuttle vector with a LEU2 marker, an ADH promoter, and a His₆ tag sequence) (34) or pGAL1 (a high-copy-number yeast shuttle vector with a LEU2 marker, a GAL1 promoter, and a His₆ tag sequence) followed a similar protocol. Genes cloned in pADH were used for protein purification and rescue assays, while genes cloned in pGAL1 were used for cycloheximide (CHX) chase assays (35).

Rescue assays for cytoplasmic AlaRS activity.

The yeast ALA1 knockout strain CHT1 (MATa his3Δ200 leu2Δ1 lys2-801 trp1Δ101 ura3-52 ala1Δ::TRP1) was maintained by a plasmid carrying the ALA1 gene and a URA3 marker (36). Rescue assays of cytoplasmic AlaRS activity were performed by introducing a test plasmid carrying the gene of interest and a LEU2 marker into CHT1, and the abilities of the transformants to grow in the presence of 5-fluoroorotic acid (5-FOA) were determined. Starting from a cell density (A₆₀₀) of 2.0, cell cultures were 3-fold serially diluted, and 20-μl aliquots of each dilution were spotted onto designated plates containing 5-FOA. The plates were incubated at 30°C for 3 days. The transformants evicted the maintenance plasmid in the presence of 5-FOA and thus could not grow on the selection medium unless a functional cytoplasmic AlaRS was encoded by the test plasmid.

Rescue assays for mitochondrial AlaRS activity.

CHT1 was cotransformed with a test plasmid and a second maintenance plasmid (carrying a HIS3 marker) that expresses only the cytoplasmic form of AlaRS (due to a mutation in its non-AUG initiator codon). In the presence of 5-FOA, the first maintenance plasmid (carrying a URA3 marker) was evicted from the cotransformants while the second maintenance plasmid was retained. Thus, all cotransformants survived 5-FOA selection due to the presence of the cytoplasmic AlaRS derived from the second maintenance plasmid. The mitochondrial phenotypes of the cotransformants were further tested on yeast extract-peptone-glycerol (YPG) plates at 30°C, and the results were documented on day 3 following plating. Because a yeast cell cannot survive on glycerol without functional mitochondria, the cotransformants did not grow on YPG plates unless a functional mitochondrial AlaRS was produced from the test plasmid.

RT-PCR.

To determine the relative levels of V. polyspora ALA1 and ALA2 mRNAs, a semiquantitative reverse transcription (RT)-PCR experiment was carried out following the protocols provided by the manufacturer (Invitrogen, Carlsbad, CA). Total RNA was isolated from cells grown in yeast extract-peptone-dextrose (YPD) or yeast extract-peptone-galactose (YPGal) medium and then treated with DNase I to clear contaminating DNA. Aliquots (2 μg) of the RNA were then reverse transcribed into single-stranded cDNA using an oligo(dT) primer. After RNase H treatment, the single-stranded cDNA products were used as the template for PCR amplification. Two pairs of gene-specific primers, GSP1/GSP2 and GSP3/GSP4, were used to amplify the cDNA fragments of VpALA1 and VpALA2, respectively. GSP1 and GSP2 are complementary to nucleotides +1998 to +2022 (AGTTCCATTAGATTTGGCTAAGAC) and +2111 to +2135 (TCCATTCTTCATTGGATGGGTTAG) of VpALA1, respectively, while GSP3 and GSP4 are complementary to nucleotides +2004 to +2028 (TGTCCAGCTAAATGAAGCCAACCG) and +2106 to +2140 (ATGAATTTGTTTCAGTAGGATCTT) of VpALA2, respectively. A real-time RT-PCR experiment using the same sets of primers was subsequently followed to obtain more quantitative data. Quantitative data were obtained from three independent experiments and averaged.

Western blot analysis.

The protein expression patterns of the VpALA1 and VpALA2 constructs were determined by a chemiluminescence-based Western blot analysis. CHT1 was first transformed with the constructs of interest, and total protein extracts were prepared from the transformants. Aliquots of the protein extracts (50 μg) were loaded onto a minigel (8 by 10 cm) containing 10% polyacrylamide and electrophoresed at 100 V for 1 to 2 h. Following electrophoresis, the resolved proteins were transferred using a semidry transfer device to a polyvinylidene difluoride (PVDF) membrane in buffer containing 30 mM glycine, 48 mM Tris base (pH 8.3), 0.037% sodium dodecyl sulfate (SDS), and 20% methanol. The membrane was probed with an anti-His₆ tag antibody (Invitrogen), followed by a horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG antibody (Invitrogen), and then exposed to X-ray film following the addition of appropriate substrates.

Fluorescence microscopy.

Constructs of interest were first transformed into INVSc1 (MATa his3Δ1 leu2 trp1-289 ura3-52; MATα his3Δ1 leu2 trp1-289 ura3-52) (Invitrogen), and the resultant transformants were grown to an optical density at 600 nm (OD₆₀₀) of ∼0.6 in selective medium (synthetic defined medium lacking leucine). We pretreated the cells with 4′,6-diamidino-2-phenylindole (DAPI) (0.5 μg/ml) or MitoTracker (∼300 nM) (Invitrogen, Carlsbad, CA) for 30 min. The samples were then examined by fluorescence microscopy (Axio observer.A1; Carl Zeiss, Inc.) using a 100× objective at 25°C, and images were captured with a charge-coupled-device (CCD) camera (Axiocam MRm; Carl Zeiss, Inc.). Nuclear and mitochondrial tracks and merged images were generated with AxioVision release 4.8 software and then subjected to two-dimensioinal (2D) deconvolution with AutoQuant X2.

RESULTS

Differential regulation of gene expression.

V. polyspora possesses two distinct nuclear AlaRS genes, VpALA1 and VpALA2. VpALA1 encodes the cytoplasmic form, while VpALA2 encodes its mitochondrial isoform (Fig. 1A). As these two isoforms function in their respective cellular compartments, we wondered whether their expression is separately regulated to satisfy the needs of each compartment. To explore this hypothesis, V. polyspora was grown in two different media, YPD (a fermentation medium that contains glucose as the sole carbon source) and YPGal (a mitochondrion-enriched medium that contains galactose as the sole carbon source). The relative levels of their specific mRNAs were then determined by semiquantitative RT-PCR. The expected sizes of the PCR-amplified products were 137 and 136 bp for VpALA1 and VpALA2, respectively.

FIG 1.

Differential regulation of gene expression. (A) 5′-end sequences of VpALA1 and VpALA2. The sequence extends from nucleotides −180 to +60 relative to ATG1. For clarity, the initiator codons of VpALA1 and VpALA2 are underlined, and the MTS of VpAlaRS2 is shaded. “m” denotes translation initiation from a non-ATG codon. Also shown is the transcription initiation site of VpALA2 at nucleotide position −112 relative to ATG1. *, stop codon. (B) RT-PCR. (Left) PCR amplification of the VpALA1 and VpALA2 fragments using genomic DNA (gDNA) as the template. (Right) Relative levels of endogenous VpALA1 and VpALA2 mRNAs were determined by RT-PCR. Total RNA was isolated from cells grown in YPD or YPGal. RNA (control) denotes PCR amplification using RNA as the template. (C) Quantitative RT-PCR. The error bars indicate standard deviations.

As shown in Fig. 1B, similar levels of the VpALA1 and VpALA2 DNA fragments were amplified by their respective primers using genomic DNA of V. polyspora as the template (Fig. 1B, left), suggesting that the two sets of primers anneal to their respective targets with similar efficiencies. The same sets of primers were then used to amplify the cDNA fragments of the two genes, which had been reverse transcribed from a total RNA preparation of V. polyspora using oligo(dT) as the primer. Figure 1B shows that similar levels of the VpALA1 cDNA fragment were amplified, regardless of whether the yeast was grown in YPD or YPGal, suggesting that VpALA1 maintains similar levels of gene expression in both media (Fig. 1B, right). In contrast, VpALA2 showed a slightly higher transcription level in YPGal than in YPD. Further determination using quantitative RT-PCR (qRT-PCR) showed that VpALA1 had similar transcription levels in both media, but VpALA2 had a 2-fold-higher transcription level in YPG than in YPD (Fig. 1C; see Fig. S1 in the supplemental material). This result supports our hypothesis that the two genes are separately regulated in the cell.

Clever selection of transcription and translation initiation sites for V. polyspora ALA2.

As two stop codons, TGA (−2) and TAG (−9), are close to ATG1 of VpALA1, it is logical for the gene to choose ATG1 as its translation start codon. However, it remained a puzzle to us why the ATG1 codon of VpALA2 cannot serve as an alternative translation initiation site following “leaky scanning” (37–39). Even when the upstream non-ATG initiator codons are inactivated by point mutations, this ATG1 still cannot be recognized as a translation start site (20). To obtain more clues to this puzzle, a strategy called 5′ rapid amplification of cDNA ends (5′-RACE) was carried out to pinpoint the transcription initiation sites of the gene. As it turned out, VpALA2 had a single transcription initiation site at a position 112 bp upstream of ATG1 (Fig. 2A). As the 40S ribosomal subunit is prone to falling off an mRNA template during scanning, this long leader sequence may contravene operation of the scanning mechanism.

FIG 2.

Selection of transcription and translation initiation sites for V. polyspora ALA2. (A) 5′ rapid amplification of cDNA ends. The transcription initiation site of VpALA2 was determined by 5′ RACE using total RNA of V. polyspora as the template. (B) Schematic summary of the VpALA2-ScALA1^OF fusion constructs and their rescue activities. A presequence of VpALA2 with various deletions was fused out of frame to the ORF of ScALA1, and the abilities of the resultant constructs to rescue the growth defects of CHT1 were tested on 5-FOA and YPG. + and −, positive and negative complementation, respectively; Mit, mitochondrial; Cyt, cytoplasmic. For clarity, out-of-frame (○) and in-frame (●) initiator codons relative to the ORF of ScALA1 are labeled. (C) Rescue assays for cytoplasmic and mitochondrial AlaRS activities. dT₃₀ denotes 30 dT residues. The numbers refer to the constructs shown in panel B.

To investigate this possibility, a presequence of VpALA2 with various 3′ ends (bp −300 to −1, −300 to −22, −300 to −43, or −300 to −67) was fused out of frame to the ORF of ScALA1, yielding various VpALA2-ScALA1^OF fusion constructs. Note that an extra base pair was positioned 5′ to the ORF of ScALA1 (Fig. 2B), causing the upstream non-ATG initiator codons of VpALA2 to be out of frame with the ORF of ScALA1. Thus, ATG1 of ScALA1 became the only functional initiator codon in these fusion constructs. The abilities of the resultant constructs to rescue the cytoplasmic and mitochondrial defects of CHT1 (an ala1⁻ strain of S. cerevisiae) were tested. Two controls, pCHT369 and pCHT437, were constructed to verify the experimental design. In pCHT369, a presequence of ScALA1 (bp −300 to −2) was fused in frame with the ORF of ScALA1, while in pCHT437, a presequence of ScVAS1 (bp −300 to +138) was fused out of frame with the ORF of ScALA1. As expected, pCHT369 provided both the cytoplasmic and mitochondrial functions, while pCHT437 provided only the cytoplasmic function (Fig. 2B).

Figure 2C shows that pCHT433 provided neither cytoplasmic nor mitochondrial activity, suggesting that the leader sequence of VpALA2 somehow silenced the translation of the downstream ORF. However, the silencing effect vanished when the leader sequence was partially deleted (−21 to −1, −42 to −1, or −66 to −1). These fusion constructs showed positive cytoplasmic rescue activity (Fig. 2B, pCHT439, pCHT435, and pCHT434). Undoubtedly, a portion of the scanning ribosome bypassed the upstream non-AUG initiator codons (in the leader sequence of VpALA1 mRNA) and safely reached the AUG1 initiator codon of ScALA1. Most remarkably, replacing part of the deleted sequence in pCHT435 with dT₃₀ (30 dT residues) successfully restored its silencing effect (Fig. 2B, pCHT445), lending further support to the hypothesis that the silencing effect of the leader sequence is length dependent and non-sequence specific.

Stability of V. polyspora AlaRS1 and AlaRS2 in vivo.

To compare the relative levels of protein stability of VpAlaRS1 and VpAlaRS2 in their native and reciprocal compartments, we performed a CHX chase assay as previously described (35). VpALA1, VpALA2, and their derivatives were cloned into pGAL1, and the resultant constructs were transformed into an S. cerevisiae strain, INVSc1. Cultures of the transformants were induced with galactose for 4 h, followed by addition of CHX to terminate protein synthesis. The CHX-treated cells were harvested at various intervals following the induction. Protein extracts were prepared for Western blot analyses using an anti-His₆ tag antibody. As shown in Fig. 3, VpAlaRS1 and VpAlaRS2 were both very stable in their native compartments (Fig. 3, left). Remarkably, the two proteins retained similar levels of stability when they were forced into the reciprocal compartments, that is, VpAlaRS1 in mitochondria and VpAlaRS2 in the cytosol (Fig. 3, right). Less than 10% of the proteins were degraded in their native and reciprocal compartments even after 8 h of CHX treatment. This result clearly indicates that VpAlaRS1 and VpAlaRS2 can be stably expressed and maintained in their native and reciprocal compartments.

FIG 3.

Stability of V. polyspora AlaRS1 and AlaRS2 in vivo. (Top) Transformants carrying a plasmid-borne VpALA1 or VpALA2 gene were grown to a cell density (A₆₀₀) of ∼0.6 and then induced with galactose for 4 h before addition of CHX. The induced cells were harvested at various intervals following treatment with CHX and lysed. T₀, T_0.5, T₁, T₂, T₄, and T₈ denote 0, 0.5, 1, 2, 4, and 8 h, respectively, after induction. AlaRS and phosphoglycerate kinase (PGK) (as a loading control) are labeled. (Bottom) Quantitative data for the relative levels of AlaRSs. The error bars indicate standard deviations.

Differential localizations and functions of V. polyspora AlaRS1 and AlaRS2.

To examine the cellular distributions of the proteins derived from the above-mentioned fusion constructs, a DNA sequence encoding green fluorescent protein (GFP) was inserted in frame at the 3′ ends of the genes, resulting in various GFP fusion constructs. As genetic tools for V. polyspora have not yet been fully developed, here, we used S. cerevisiae as a platform for the assay. The GFP fusion constructs were individually transformed into INVSc1, and the cellular localization of the GFP fusion proteins in the transformants was examined with fluorescence microscopy. As shown in Fig. 4C, VpAlaRS2-GFP was exclusively localized in mitochondria (images 5), while VpAlaRS1-GFP was evenly distributed in the cytosol (images 2). Adding or removing a mitochondrial targeting signal (MTS) from the two fusion proteins swiftly switched their distributions (images 4 and 7). In contrast, fusion of a presequence of ScVAS1 (bp −300 to +138) to the ORF of VpALA1 or VpALA2 resulted in the production of two protein isoforms, one (without an MTS) localized in the cytosol and the other (with an MTS) in mitochondria (images 3 and 6). Thus, VpAlaRS1 per se is localized in the cytosol, but it could be forced into the mitochondria by insertion of an MTS. Conversely, VpAlaRS2 is naturally a mitochondrial protein, but it can be retained in the cytosol by removal of its MTS.

FIG 4.

Differential localizations and functions of V. polyspora AlaRS1 and AlaRS2. A presequence of ScVAS1, VpALA1, or VpALA2 was fused to the ORFs of VpALA1 and VpALA2, and the abilities of the fusion constructs to rescue the growth defects of CHT1 were tested on 5-FOA and YPG. (A) Schematic summary of the fusion constructs and their rescue activities. (B) Rescue assays for the cytoplasmic and mitochondrial AlaRS activities. (C) GFP assay. The subcellular distributions of the respective GFP fusion proteins were determined by fluorescence microscopy. A mitochondrial tracker and DAPI were used to label mitochondria and nuclei, respectively. (B and C) The numbers refer to the corresponding GFP fusion constructs shown in panel A.

To examine whether the two isoforms of VpAlaRS can function properly when they are localized to the reciprocal compartments, various fusion genes were cloned in pRS315, and their rescue activities were assayed in CHT1. Consistent with our previous report, VpALA1 and VpALA2, respectively, rescued the cytoplasmic and mitochondrial defects of CHT1 (Fig. 4A and B, rows 2 and 5) (20). Fusion of a presequence of VpALA1 (bp −300 to −1) to the ORF of VpALA2 yielded a fusion construct that selectively rescued the cytoplasmic defect of the knockout strain (row 7), while fusion of a presequence of VpALA2 (bp −300 to −1) to the ORF of VpALA1 yielded a fusion construct that selectively rescued the mitochondrial defect (row 4). In contrast, fusion of a presequence of ScVAS1 (bp −300 to +138) to the ORF of VpALA1 or VpALA2 yielded a dual-functional gene (rows 3 and 6). Two protein isoforms with different amino termini were predicted to be generated from each of the two fusion constructs, one initiating from the ATG1 initiator codon of ScVAS1 and the other from the ATG1 codon of VpALA1 or VpALA2. The longer form (with an MTS) acted in mitochondria, while the shorter form (without an MTS) acted in the cytoplasm. Thus, VpAlaRS1 and VpAlaRS2 can properly function in the reciprocal compartments provided that an MTS is inserted into or removed from them.

Compartmental barriers are overcome with protein overexpression.

Since VpAlaRS1 and VpAlaRS2 can be stably expressed and function in the reciprocal compartments by addition or removal of an MTS, we wondered whether the two genes could be made dual functional simply by manipulating their protein expression levels. To explore this possibility, the ORFs of the two genes were individually cloned in pADH, and the protein expression levels and functions of the resultant constructs were analyzed. As shown in Fig. 5, cloning of VpALA1 and VpALA2 in pADH significantly enhanced their protein expression levels, which in turn enabled the VpAlaRS isoforms to function in both compartments (Fig. 5, 3 and 5). Surprisingly, VpALA2 cloned in pRS315, while positive for complementation of the cytoplasmic activity, had a protein expression level that was undetectable by Western blotting (Fig. 5C, lane 4). In contrast, VpALA1 cloned in pRS315 had a moderate level of protein expression (lane 2). Thus, the compartmental barriers could be overcome by overexpression of VpALA1 or VpALA2.

FIG 5.

Compartmental barriers are overcome with protein overexpression. Various VpALA1 and VpALA2 constructs were transformed into CHT1, and the abilities of the constructs to rescue the growth defects of the knockout strain were tested on 5-FOA and YPG. (A) Schematic summary of the ALA1 and ALA2 constructs and their rescue activities. (B) Rescue assays for cytoplasmic and mitochondrial AlaRS activities. (C) Western blotting using an anti-His₆ antibody. (Top) AlaRS. (Bottom) PGK (as a loading control). (D) Fractionation and Western blotting. (Top) The cytoplasmic and mitochondrial fractions were isolated from transformants harboring various VpALA1 and VpALA2 constructs and analyzed by Western blotting using an anti-His₆ tag antibody. (Middle and bottom) As internal controls, the cytoplasmic and mitochondrial fractions of the extracts were probed with anti-PGK and anti-Hsp60 antibodies. (B to D) The numbers refer to the constructs shown in panel A.

To take a closer look at the biological relevance of protein expression levels in relation to the functions of the VpAlaRS isoforms, the cytoplasmic and mitochondrial fractions of transformants harboring various VpALA1 and VpALA2 constructs were individually extracted according to the protocols described by Guthrie and Fink (40), and their relative AlaRS levels were determined by Western blotting. As shown in Fig. 5D, VpAlaRS1 and VpAlaRS2, when expressed from pRS315, were localized in the cytosol and mitochondria, respectively (Fig. 5D, compare lanes 2 and 4), and thus could rescue the cytoplasmic and mitochondrial defects of CHT1 (Fig. 5B, compare rows 2 and 4). On the other hand, these two forms, when overexpressed from pADH, could overcome the compartmental barrier (Fig. 5D, compare lanes 3 and 5) and thus could function in both compartments (Fig. 5B, compare rows 3 and 5).

Differential localization and function of T. phaffii ALA1 and ALA2.

In addition to V. polyspora, another yeast species, T. phaffii, also possesses two distinct AlaRS genes (TpALA1 and TpALA2). As it turned out, the proteins deduced from the ORFs of these two genes (TpAlaRS1 and TpAlaRS2) shared only 59.3% identity (Fig. 6A). Similar to the scenario for VpAlaRS, TpAlaRS1 possessed higher similarity to dual-functional yeast AlaRSs, such as Ashbya gossypii AlaRS (AgAlaRS) and ScAlaRS (72.0 to 78.8% identity), than did its mitochondrial counterpart, TpAlaRS2 (56.6 to 59.4% identity). However, unexpectedly, TpAlaRS2 did not closely resemble VpAlaRS2 (56.3% identity). A. gossypii is a yeast species that has not gone through WGD, and its AlaRS was used here as a reference. To explore the functional potentials of the two genes, they were individually cloned into pRS425 with their native promoters, and their abilities to rescue CHT1 were tested. Figure 6 shows that TpALA1 and TpALA2 restored the growth phenotypes of the null allele on 5-FOA and YPG, respectively (Fig. 6B and C, rows 2 and 4), suggesting that they perform cytoplasmic and mitochondrial functions, respectively. Moreover, fusion of a presequence of ScVAS1 (bp −300 to +138) to the ORFs of the two genes enabled the fusion constructs to function in both compartments (Fig. 6B and C, rows 3 and 5). Thus, TpAlaRS1 and TpAlaRS2 can also function in the reciprocal compartments provided that an MTS is inserted into or removed from them.

FIG 6.

Differential localizations and functions of T. phaffii ALA1 and ALA2. (A) Identities among various yeast AlaRS sequences. (B) Schematic summary of the ALA1 and ALA2 constructs and their rescue activities. (C) Rescue assays for cytoplasmic and mitochondrial AlaRS activities. Various ALA1 and ALA2 constructs were transformed into CHT1, and the abilities of the constructs to rescue the growth defects of the knockout strain were tested on 5-FOA and YPG. The numbers refer to the constructs shown in panel B.

Phylogenetic relationships of yeast AlaRSs.

We have previously shown that VpALA1 and VpALA2 and their eukaryotic homologues are clustered in a monophyletic branch, which displays higher affinity with the bacterial branch than with the archaeal branch, suggesting that all eukaryotic AlaRS genes are of mitochondrial origin and that VpALA1 and VpALA2 are descended from a common predecessor (20). To better illustrate possible historical relationships among yeast AlaRS homologues, representative AlaRS sequences were retrieved from 11 yeasts. Eight of them had gone through WGD, and three of them had not. To ensure the reliability of sequence alignment, only the conserved core active sites were retained. For example, only 53% and 51% of the sequences of VpAlaRS2 and TpAlaRS2, respectively, were reserved for further analyses. ClustalW (41) was then used to align the sequences obtained. The phylogeny was constructed using the maximum-likelihood method based on the Tamura-Nei model (only the 1st and 2nd positions were included) (42) with 10,000 bootstrap replications (43), as implemented in MEGA6 (44). We found that T. phaffii also has two AlaRS paralogues. The phylogenetic tree obtained revealed that VpAlaRS2 and TpAlaRS2 were clustered together with 100% bootstrap support but were not clustered with their own paralogues, VpAlaRS1 and TpAlaRS1, respectively (Fig. 7A). This result suggests that the two paralogous pairs were duplicated earlier than the speciation event between V. polyspora and T. phaffii. It is reasonable to speculate that the duplication event corresponds to the WGD. However, due to the limited length of AlaRS genes, the information used for phylogeny construction is insufficient, and therefore, not all clusterings are supported by high bootstrap values. Moreover, we could not rule out the possibility that the clustering of VpAlaRS2 and TpAlaRS2 was due to a long-branch attraction effect (45). Therefore, we adopted synteny analysis to clarify the origins of these two AlaRS paralogues.

FIG 7.

Phylogenetic relationships of yeast AlaRSs. (A) Phylogenetic tree of yeast AlaRS genes constructed using the maximum-likelihood method based on the Tamura-Nei model (only the 1st and 2nd positions are included) with 10,000 bootstrap replications. The possible whole-genome duplication event is indicated by WGD, and the non-WGD species are marked with asterisks. Accession numbers: Naumovozyma dairenensis, XP_003670443; Naumovozyma castellii, XP_003676658; Kazachstania africana, XP_003957527; Saccharomyces cerevisiae, NP_014980; Candida glabrata, XP_445923; Tetrapisispora blattae, XP_004177771; Vanderwaltozyma polyspora (VpAlaRS1, XP_001643652; VpAlaRS2, XP_001642960); Tetrapisispora phaffii (TpAlaRS1, XP_003686912; TpAlaRS2, XP_003688259); Lachancea thermotolerans, XP_002552804; Ashbya gossypii, NP_984459; and Kluyveromyces lactis, XP_455190. (B) Synteny of yeast AlaRS genes. X denotes loss of the homologue in the designated chromosome. For clarity, ALA1 and ALA2 are highlighted.

The annotated genomes of V. polyspora, T. phaffii, A. gossypii, and S. cerevisiae were thus compared, with A. gossypii (which has not experienced WGD) and S. cerevisiae (which has experienced WGD but retained only one copy of the AlaRS gene) serving as the references. The synteny map shown here contained a total of 15 genes, including four upstream and 10 downstream of the AlaRS homologue of A. gossypii (Fig. 7B). We then checked what happened to the duplicated genes in these yeasts during evolution. Although one copy of the duplicated AlaRS genes was absent from the chromosome of S. cerevisiae, retention of some of its neighboring genes still permitted rapid identification of the corresponding region. As shown in Fig. 7B, in some cases, only one copy of the duplicated genes was preserved (e.g., TEA1, KRE5, POP5, and PRP45), while in others, both copies were deleted (e.g., MTW1 and MYO2 in V. polyspora) or preserved (e.g., FUN19). It thus appears that retention or deletion of the redundant genes is independent for each gene in a given organism. This synteny relationship confirms that the two chromosomes of S. cerevisiae, V. polyspora, or T. phaffii compared are all sister chromosome pairs, and it strongly supports the idea that yeast ALA1 and ALA2 were duplicated in the WGD event. Based on this finding and others, a proposed evolutionary history of yeast AlaRS is shown in Fig. 8, which details the deletion, duplication, and gain or loss of function of the AlaRS homologues and their relationships with the speciation events of representative yeasts.

FIG 8.

Proposed evolutionary history of yeast AlaRS. Some yeast species have never experienced WGD, while others have. In S. cerevisiae and C. glabrata, only one of the duplicated ALA1 paralogues is retained after WGD. In contrast, in V. polyspora and T. phaffii, both copies are retained after WGD and are later converted into single functional paralogues by loss-of-function mutations. The AlaRS genes of eukaryotic and mitochondrial origins are shaded in blue and red, respectively. Cyt, cytoplasmic form; Mit, mitochondrial form.

DISCUSSION

S. cerevisiae possesses a dual-functional AlaRS gene that encodes both cytoplasmic and mitochondrial forms of AlaRS through alternative use of two in-frame initiator codons (12). These two isoforms carry essentially the same polypeptide sequence, except that the mitochondrial form contains an MTS at its amino terminus. As it turns out, many other yeast species also possess a dual-functional AlaRS gene (33). V. polyspora and T. phaffii are so far the only two exceptions; they each possess two distinct AlaRS homologues (ALA1 and ALA2) (Fig. 6). ALA1 and ALA2 of V. polyspora or T. phaffii could substitute for the cytoplasmic and mitochondrial functions, respectively, of S. cerevisiae ALA1 (Fig. 4 and 6). VpALA2 had a higher transcription level in a mitochondrion-enriched medium than in a fermentation medium, while VpALA1 had similar transcription levels in both media. Such a feature nicely fits the roles of ALA1 and ALA2 as genes encoding a cytoplasmic and a mitochondrial translation enzyme, respectively. While the cytoplasmic and mitochondrial forms of VpAlaRS were exclusively localized in their respective cellular compartments under normal growth conditions, the isoforms could also function in the reciprocal compartments by addition or removal of an MTS (Fig. 4 and 6). This result is consistent with our previous finding that purified recombinant VpAlaRS1 and VpAlaRS2 can charge cytoplasmic and mitochondrial tRNA^Ala isoacceptors with comparable efficiencies in vitro (20). In this regard, it is interesting that, except for a few exceptions in the mitochondrial tRNA^Ala (46, 47), the primary identity element of tRNA^Ala—G3-U70 base pair—is conserved in all known tRNA^Ala isoacceptors (48, 49) (see Fig. S2 in the supplemental material). Perhaps that is why cross-species and cross-compartmental substitution can be feasibly achieved for the yeast ALA1 gene (Fig. 4 and 6) (33). As a result, even the human AlaRS gene can rescue a yeast ALA1 knockout strain (36).

Many higher-eukaryotic genes encode distinct protein isoforms via alternative transcription and translation or leaky scanning (37–39). Similar mechanisms also operate in yeast. For example, yeast GRS1 uses leaky scanning to generate two isoforms of GlyRS from a single transcript (13), while yeast ALA1 uses alternative transcription and translation to generate two isoforms of AlaRS (12). While the AUG1 codon of VpALA2 is located at a position equivalent to that of the AUG1 initiator codon of VpALA1 or ScALA1, translation of the gene is exclusively initiated from two upstream non-AUG codons (Fig. 1 and 4). The question arose as to why AUG1 of VpALA2 cannot be used as a translation start site even when the upstream non-AUG initiator codons are inactivated by point mutations (20). Because only one transcript was derived from VpALA2 (Fig. 2), the possibility of alternative transcription and translation seems unlikely for the gene. In addition, like the 5′ untranslated regions of many yeast mRNAs (50), the leader sequence of VpALA2 mRNA was rich in AU (74%). Moreover, analysis of the leader sequence of VpALA2 mRNA using the Mfold Web server (51) failed to identify any stable secondary structure that was strong enough to arrest ribosomal scanning. Perhaps, the long leader sequence of VpALA2 mRNA tires out the scanning ribosome, thereby preventing it from reaching AUG1 (Fig. 2). Evidence supporting this view came from the observation that shortening the leader sequence enabled the downstream AUG1 to function as a start codon and replacing the deleted segment of the leader sequence with oligo(dT) successfully restored its silencing effect (Fig. 2). Thus, the silencing effect of the leader sequence appears to be length dependent and nonspecific to sequence. Despite all this, a previous report argued that leader length is irrelevant to the efficiency of ribosomal scanning (52). More thorough studies are necessary to reach a more solid conclusion.

Increasing evidence supports the idea that the AlaRS homologues in mitochondrion-containing eukaryotes are of mitochondrial origin (53, 54). In contrast, the AlaRS homologues recovered from the amitochondriate protists Giardia lamblia and Trichomonas vaginalis are clustered into a sister group that shows higher affinity for the archaeal branch (53), suggesting that these homologues have evolved from a different origin. Conceivably, the phylogeny of eukaryotic AlaRSs is far more sophisticated than anticipated. In this regard, our study reported here is of particular interest. Although the common ancestor of present-day yeasts should have one dual-functional AlaRS of mitochondrial origin, WGD provided an opportunity for yeast AlaRS to further evolve. While for most yeast species, like S. cerevisiae, only one of the duplicated AlaRS paralogues was retained after the WGD event, for some other yeast species, like V. polyspora and T. phaffii, the two duplicated AlaRS paralogues, AlaRS1 and AlaRS2, undertaking the cytoplasmic and mitochondrial functions, respectively, were both preserved by subfunctionalization (55, 56). We further found that the two AlaRS2 orthologues (VpAlaRS2 versus TpAlaRS2) had higher synonymous (1.6-fold) and nonsynonymous (2.8-fold) substitution rates than the two AlaRS1 orthologues (VpAlaRS1 versus TpAlaRS1). Since the cytoplasmic function of AlaRS is essential for yeasts while the mitochondrial function is not, it is reasonable to speculate that the selective constraints of a dual-functional AlaRS mainly result from its cytoplasmic role. There might be a trade-off for AlaRS to function efficiently in either the cytoplasm or mitochondria. Therefore, similar selective constraints may act on cytoplasmic and dual-functional AlaRSs but not on mitochondrial AlaRSs. The accelerated evolutionary rate of AlaRS2 is likely due to the relaxed constraints on its mitochondrial role. The pattern of rate asymmetry had been largely attributed to neofunctionalization in yeast species (57). Here, we have presented an example of asymmetric evolution generated by subfunctionalization. Further studies on these whole-genome-duplicated yeasts may shed more light on how duplicated paralogues may evolve.