Functional Substitution of a Eukaryotic Glycyl-tRNA Synthetase with an Evolutionarily Unrelated Bacterial Cognate Enzyme

Two oligomeric types of glycyl-tRNA synthetase (GlyRS) are found in nature: a α2 type and a α2β2 type. The former has been identified in all three kingdoms of life and often pairs with tRNAGly that carries an A73 discriminator base, while the latter is found only in bacteria and chloroplasts and is almost always coupled with tRNAGly that contains U73. In the yeast Saccharomyces cerevisiae, a single GlyRS gene, GRS1, provides both the cytoplasmic and mitochondrial functions, and tRNAGly isoacceptors in both compartments possess A73. We showed herein that Homo sapiens and Arabidopsis thaliana cytoplasmic GlyRSs (both α2-type enzymes) can rescue both the cytoplasmic and mitochondrial defects of a yeast grs1 - strain, while Escherichia coli GlyRS (a α2β2-type enzyme) and A. thaliana organellar GlyRS (a (αβ)2-type enzyme) failed to rescue either defect of the yeast mull allele. However, a head-to-tail αβ fusion of E. coli GlyRS effectively supported the mitochondrial function. Our study suggests that a α2-type eukaryotic GlyRS may be functionally substituted with a α2β2-type bacterial cognate enzyme despite their remote evolutionary relationships.


Introduction
Aminoacyl-tRNA synthetases (aaRSs) establish the genetic code by coupling amino acids specifically to their cognate tRNAs. The resultant aminoacyl-tRNA is then delivered by an elongation factor to ribosomes for protein synthesis. The anticodon of a tRNA reads the codon of mRNA by forming Watson-Crick base pairing. Prokaryotes contain a basic set of 18-20 aaRs, but eukaryotes contain more, as they possess a second compartment (mitochondria) capable of protein translation [1][2][3]. For example, yeast possesses two distinct sets of aaRSs, one localized to the cytoplasm and the other to mitochondria. Each set is specific for cognate tRNAs within its respective cellular compartment, and it is sequestered from isoacceptors that are confined in other compartments. Cytoplasmic and mitochondrial isoforms of an aaRS with a given amino acid specificity are most often encoded by two distinct nuclear genes, regardless of the cellular compartments in which they are localized. However, four Saccharomyces cerevisiae genes, ALA1 (which encodes alanyl-tRNA synthetase) [4], GRS1 (which encodes glycyl-tRNA synthetase [GlyRS]) [5], HTS1 (which encodes histidyl-tRNA synthetase) [6], and VAS1 (which encodes valyl-tRNA synthetase) [7], specify both the cytoplasmic and mitochondrial functions. This dual-functional feature has been conserved in homologues of these genes in almost all yeast species studied [8][9][10]. In addition, the yeast GUS1 gene also specifies both the cytoplasmic and mitochondrial functions; its protein product acts as a glutamyl-tRNA synthetase in the cytoplasm and as a non-discriminating glutamyl-tRNA synthetase in mitochondria [11].
On the basis of conserved sequence motifs, the quaternary structure, and aminoacylation function, aaRSs can be divided into two classes of 10 enzymes each [12,13]. Class I enzymes possess two signature sequences, HIGH and KMSKS, while class II enzymes contain three degenerate motifs_motifs 1, 2, and 3. In addition, class I enzymes first couple amino acids to the 29-OH of the terminal adenylate residue of tRNA before transferring it to the 39-OH, while class II enzymes directly couple it to the 39-OH. Normally, orthologous enzymes that couple the same amino acid to isoaccepting tRNAs share convincing sequence similarities in their catalytic core domains and are grouped into the same class (I or II), suggesting that they descend from a common ancestor. However, three exceptions to this rule exist: lysyl-tRNA synthetase, GlyRS, and phenylalanyl-tRNA synthetase. In the case of lysyl-tRNA synthetase, both class I-and class II-type enzymes were found [14], while in the case of GlyRS, two oligomeric forms were identified: a a 2 b 2 heterotetramer and a a 2 homodimer [15,16]. Despite the fact that both forms of GlyRS possess a class IIdefining architecture, they drastically differ in size and sequence [17]. As a result, they are believed to have different origins [18]. To date, the a 2 b 2 -type enzymes are found only in bacteria and chloroplasts, while the a 2 -type enzymes have been recovered from all three domains of life. In the case of phenylalanyl-tRNA synthetase, neither oligomeric structure nor modular organization is conserved over phylogeny [19].
Results obtained from study of Escherichia coli, S. cerevisiae, and Thermus thermophilus tRNAs Gly suggest that the major identity elements of tRNA Gly reside in the discriminator base (N73), the first three base pairs of the acceptor stem (1:72, 2:71, and 3:70), and C35 and C36 in the anticodon loop [3]. These identity elements essentially dictate the efficiency of tRNA Gly aminoacylation by its cognate enzyme [20]. The most striking difference between bacterial and eukaryotic tRNA Gly isoacceptors is the discriminator base, which is nearly always a U in bacteria and an A in eukaryotic cytoplasm. For example, in E. coli, N73 is a U, while in Arabidopsis thaliana, N73 is a U in the mitochondrialencoded tRNA Gly and an A in the nuclear-encoded cytosolic tRNAs Gly [21]. Typically, a a 2 b 2 -type enzyme is coupled with U73-containing tRNA Gly , while a a 2 -type enzyme pairs with A73containing tRNA Gly [15]. However, this once-strong correlation between the oligomeric structure of GlyRS and the type of the discriminator base is compromised by the discovery that the glycine enzyme of the bacterium T. thermophilus possesses a a 2 structure yet pairs with U73-containing tRNAs Gly [22].
The yeast S. cerevisiae possesses two distinct nuclear GlyRS genes, GRS1 and GRS2. The former encodes both cytoplasmic and mitochondrial forms of GlyRS through alternative initiation of translation [5], while the latter is virtually silent and is dispensable for cell survival under normal conditions [23,24]. As it turned out, GRS2 is actually a stress-inducible gene, and its protein product retains a substantial level of aminoacylation activity in vitro [25]. Since all yeast tRNA Gly isoacceptors carry the same discriminator base_A73_it is not surprising to find that a single gene, GRS1, can support functions in both compartments. Notably, the first three base pairs of the acceptor stem are completely different between the cytoplasmic and mitochondrial tRNA Gly isoacceptors. To explore how important the discriminator base of yeast tRNAs Gly really is for recognition by a a 2 -type glycine enzyme and whether an evolutionarily unrelated a 2 b 2 -type glycine enzyme can functionally substitute for the a 2 -type yeast enzyme in vivo, several bacterial and eukaryotic GlyRS genes were cloned, and their abilities to rescue a yeast GRS1 knockout strain were investigated.

Two distinct oligomeric forms of GlyRS
Two distinct oligomeric forms of GlyRS are found in nature, a a 2 form and a a 2 b 2 form (summarized in Table 1). These two forms exhibit low homology in sequence and size. For example, a subunit of E. coli GlyRS (EcGlyRS) [a a 2 b 2 -type enzyme] shares significant homology (,53% identity) with a moiety of A. thaliana GlyRS2 (AtGlyRS2) [a (ab) 2 -type enzyme], but is highly diverged (,15% identity) from a subunit of S. cerevisiae or Homo sapiens GlyRS (ScGlyRS or HsGlyRS) [both a 2 -type enzymes] (Fig. 1A). Conversely, T. thermophilus GlyRS (TtGlyRS) [a a 2 -type enzyme] is more closely related to a subunit of eukaryotic GlyRSs (a 2 -type enzymes) [.20% identity] than to a or b subunit of E. coli GlyRS or A. thaliana GlyRS2 (,15% identity) (Fig. 1A). In addition, the number of amino acid residues in each subunit of the representative enzymes differs drastically, ranging from 303 to 739 ( Figure  S1). In a a 2 -type enzyme, the class II-defining signature motifs are located in the N-terminal moiety, while the anticodon-binding domain (ABD) is located in the C-terminal moiety. In a a 2 b 2 or (ab) 2 -type enzyme, the class II-defining signature motifs reside in the N-terminus of a subunit, while the ABD resides in the Cterminus of b subunit ( Figure S1).
Many eukaryotes, such as S. cerevisiae, Drosophila melanogaster, Caenorhabditis elegans, Bombyx mori, and Homo sapiens, possess a dualfunctional GlyRS gene. Two distinct protein isoforms are expected to be generated from each of these genes via alternative initiation of translation [5]. A longer form carries a mitochondrial targeting signal (MTS) at its amino-terminus and is imported into (and thus functional in) mitochondria, while a short form that lacks such a signal is confined (and thus functional) in the cytoplasm. The scenario that occurs in A. thaliana appears to be more sophisticated. Two distinct GlyRS enzymes are found in this plant: AtGlyRS1, which possesses a a 2 structure, and AtGlyRS2, which possesses a (ab) 2 structure (Table 1 and Figure S1) [26]. AtGlyRS1 is distributed in both the cytoplasm and mitochondria but is active only in the cytoplasm. In contrast, AtGlyRS2 is confined and functional in both mitochondria and chloroplasts. The N-terminal peptide of this protein serves not only as a MTS but also as a plastidial targeting signal. Unlike E. coli GlyRS, which possesses a a 2 b 2 -heterotetrameric structure and is encoded by an operon that contains two open reading frames in tandem separated by a ninenucleotide spacer, A. thaliana GlyRS2 possesses a unique (ab) 2dimeric structure and is encoded by a single nuclear gene.

The discriminator base N73 of glycine tRNAs
Identity elements of a tRNA are typically located at the acceptor stem and the anticodon loop. In the case of tRNA Gly , identity elements comprise the discriminator base (N73), the top three base pairs of the acceptor stem (1:72, 2:71, and 3:70), and the last two nucleotides of the anticodon [C35 and C36] (Fig. 1B) [3,20]. Even though the discriminator base N73 has been shown to be critical for recognition by GlyRS, bacterial and eukaryote cytoplasmic tRNAs Gly contain different nucleotides at this position. The discriminator base is always a U in bacteria and an A in eukaryotic cytoplasm (Table 1). However, the mitochondrial tRNAs Gly of many eukaryotes, such as S. cerevisiae, Sch. pombe, C. elegans, D. melanogaster, B. mori, and H. sapiens, contain A73 instead of U73 (Table 1), which may contribute to the finding that the same GlyRS enzyme can charge both cytoplasmic and mitochondrial tRNA Gly isoacceptors in these organisms [5,23]. In contrast, in A. thaliana, N73 is a U in the mitochondrial-encoded tRNA Gly and is a U or a C in the chloroplast-encoded tRNA Gly isoacceptors [21]. As a result, A. thaliana possesses two structurally distinct GlyRS enzymes_one responsible for the cytoplasm and the other for organelles. On the other hand, not all glycylation systems can fit in the two paradigms_a 2 /A73 and a 2 b 2 /U73. One such example is the GlyRS enzyme of T. thermophilus, which possesses a a 2 -type structure, yet pairs with U73-containing tRNAs Gly (Table 1). Most strikingly, this bacterial enzyme can efficiently charge both A73 and U73-containing tRNAs Gly in vitro [27].

Rescue of a yeast grs1strain by various bacterial GlyRSs
As T. thermophilus GlyRS can efficiently charge yeast tRNAs in vitro [27], we wondered whether this enzyme can functionally substitute for yeast GRS1 in vivo. As a comparison, E. coli GlyRS, which cannot effectively charge yeast tRNAs, was also included in the assays. To start with, the genes that encode T. thermophilus and E. coli GlyRSs were cloned in pADH or its derivative with a MTS (pWYH8). The resultant constructs were transformed into a yeast grs1strain, and the abilities of the transformants to grow on 5-FOA and YPG were tested. Fig. 2 shows that TtGlyRS failed to rescue the growth defects of the knockout strain on 5-FOA and YPG, but fusion of an MTS to the bacterial enzyme enabled the fusion enzyme to restore the growth phenotype of the null allele on YPG (numbers 3 and 4 in Fig. 2B,C). As expected, the wild-type E. coli GlyRS failed to substitute for either activity of yeast GRS1 in vivo (numbers 7 and 8 ). Unexpectedly, an engineered E. coli GlyRS with a and b chains fused head-to-tail into a single polypeptide, upon being imported into mitochondria, effectively rescued the growth defect of the null allele on YPG (numbers 5 and 6). This result positively suggests that the engineered E. coli enzyme can charge yeast mitochondrial tRNAs to a level that is sufficient for sustaining normal mitochondrial function.
We next examined whether the aforementioned bacterial genes were properly expressed in yeast. The constructs to be tested were transformed into a yeast strain, INVSc1, and soluble protein extracts were prepared from the resultant transformants and analyzed via Western blotting using an anti-His 6 tag antibody. As shown in Fig. 2D, little TtGlyRS protein was detected by the Western blotting, while MTS-TtGlyRS was expressed at a level that is ,20-fold lower than that of yeast GlyRS1 (which was also expressed under the control of an ADH promoter) (numbers 3 and 4 in Fig. 2D). As expression of these two genes, TtGlyRS and MTS-TtGlyRS, was controlled by the same promoter, it appeared that this bacterial enzyme is much more stable in the mitochondria. This effect might, in turn, define its functional potential in yeast (Fig. 2C). As for expression of the engineered E. coli GlyRS, both EcGlyRS(ab) and MTS-EcGlyRS(ab) were properly expressed, with a level that is ,25-fold lower than that of yeast GlyRS1 (numbers 5 and 6). The expression patterns of the wild-type E. coli GlyRS were somewhat more complicated. Only a subunit was detected when a and b subunits of E. coli GlyRS were co-expressed in the yeast cytoplasm. In contrast, only b subunit was detected when both subunits were imported into the yeast mitochondria via fusion to an MTS (numbers 7 and 8). Consequently, no active a 2 b 2 complex could be formed in either compartment of the transformants, which might provide an excuse for the negative phenotype of the wild-type bacterial enzyme in the assays (Fig. 2B,C).

Rescue of a yeast grs1strain by various eukaryotic GlyRSs
As GlyRSs in eukaryotic cytoplasm are invariably coupled with tRNAs Gly that carry an A73 discriminator base (Table 1), we next tested whether the genes that encode H. sapiens and A. thaliana GlyRSs can functionally substitute for yeast GRS1 in vivo. To this end, the genes that encode human GlyRS and Arabidopsis GlyRS1 and GlyRS2 were independently cloned in pADH, and the abilities of the resultant constructs to rescue a yeast GRS1 knockout strain were tested on 5-FOA and YPG. As shown in Fig. 3, A. thaliana GlyRS1 successfully rescued the growth defects of the knockout strain on both 5-FOA and YPG, suggesting that this enzyme can charge yeast cytoplasmic and mitochondrial tRNA Gly isoacceptors to a substantial level. Removal of the putative MTS (N-terminal amino acid residues 1-39) from AtGlyRS1, resulting in AtGlyRS1(DMTS), effectively eliminated its mitochondrial function, but it slightly enhanced its cytoplasmic activity (number 3). This finding is consistent with our assumption that the MTSdeleted AtGlyRS1 enzyme exclusively acts in the cytoplasm. Further removal of the WHEP domain from AtGlyRS1 completely abolished its rescue activities (number 4). In contrast with AtGlyRS1, the full-length or the MTS-deleted AtGlyRS2 enzyme failed to support either function in the assays (numbers 5 and 6).
We next tested human GlyRS for its cross-species rescue activities. The full-length human enzyme feebly supported the growth of the grs1strain on 5-FOA, but it efficiently rescued the growth defect of the knockout strain on YPG (number 7). Removal of the MTS from HsGlyRS selectively impaired its mitochondrial activity (number 8). Most remarkably, removal of both the MTS and the WHEP domain from HsGlyRS led to a truncated enzyme that robustly supported the growth of the yeast knockout strain on both 5-FOA and YPG (number 9).
To gain further insight into the expression levels of these eukaryotic enzymes in yeast, Western blotting using an anti-His 6 tag antibody was carried out. As shown in Fig. 3D, AtGlyRS1 and AtGlyRS1(DMTS) were properly expressed in yeast, with a level that is ,25-fold lower than that of yeast GlyRS1. However, deletion of both the MTS and the WHEP domain knocked off its expression (number 4). This finding provides a rational basis as to why AtGlyRS1(DMTSDWHEP) failed to support either activity in the rescue assays. Similarly, both AtGlyRS2 and At-GlyRS2(DMTS) were properly expressed in yeast, with a level that is ,25-50-fold lower than that of yeast GlyRS1 (numbers 5 and 6). Thus, the negative phenotype of these two constructs in complementation is likely to be the effect of poor cross-species glycylation rather than defective protein expression. Contrary to the scenarios of Arabidopsis GlyRS1 and GlyRS2, only a faint protein band was detected for the full-length and the MTS-deleted human GlyRS enzymes (numbers 7 and 8). Interestingly, even with such a low level of protein expression, human GlyRS could support the growth of the yeast null allele (numbers 7 and 8 in Fig. 3C). Most strikingly, deletion of both the MTS and the WHEP domain from HsGlyRS drastically enhanced its protein expression level [,25-fold lower than that of yeast GlyRS1] (number 9), which may partially account for the increased cytoplasmic activity (Fig. 3B). It is still a puzzle as to why this truncated enzyme regained a mitochondrial activity. Perhaps, a second, cryptic MTS is embedded inside its internal sequence, and such a signal can be functional only when the protein is highly expressed. Another possibility is that the second cryptic MTS is masked by the WHEP domain and exposed upon removal of the latter.

Cellular distributions of human GlyRS in yeast
The MTS used in targeting the bacterial enzymes to mitochondria was grafted from the mitochondrial precursor form of yeast valyl-tRNA synthetase, and its mitochondrial targeting potential has been repeatedly tested and confirmed [23,28]. To evaluate whether the putative MTS of human GlyRS also efficiently acts as a targeting signal in yeast, a fluorescence microscopic analysis was carried out. A DNA sequence that encodes the green fluorescence protein (GFP) was amplified by a PCR and inserted in-frame into an appropriate site at the 39 end of HsGlyRS and HsGlyRS(DMTSDWHEP), yielding HsGlyRS-GFP and HsGlyRS(DMTSDWHEP)-GFP, respectively. The resultant fusion constructs were then transformed into INVSc1, and cellular distributions of these proteins were examined via fluorescence microscopy. Expression of these GFP fusion proteins was driven by a constitutive ADH promoter. As shown in Fig. 4A, the full-length human GlyRS was largely confined in mitochondria (see HsGlyRS-GFP), while the truncated form of human GlyRS was predominantly localized in the cytoplasm [see HsGlyRS (DMTSDWHEP)-GFP]. Due to constraints of the imaging technology used, the possibility that a minor portion of the GFP fusion proteins was targeted to other unintended cellular compartments could not be ruled out. Nevertheless, this assay explicitly demonstrates the mitochondrial targeting potential of the native MTS of human GlyRS in yeast. Thus, the higher cytoplasmic rescue activity of the WHEP-deleted human GlyRS may be the combinatorial effect of higher protein expression and preferential cytoplasmic localization of the protein. A fractionation assay further confirmed that HsGlyRS-GFP can be imported into mitochondria, but HsGlyRS(DMTSDWHEP)-GFP cannot (Fig. 4B).

Aminoacylation of yeast mitochondrial tRNA Gly by E. coli GlyRS
To compare the relative aminoacylation activities of the wildtype and the fusion E. coli GlyRSs, recombinant His 6 -tagged GlyRS enzymes were purified from E. coli transformants via Ni-NTA column chromatography. As a reference, yeast GlyRS1 was also purified and tested. As shown in Fig. 5A, the engineered bacterial enzyme efficiently charged its cognate tRNAs, with efficiency ,2-fold lower than that of its wild-type counterpart. In contrast, yeast GlyRS1 only poorly charged E. coli tRNAs. To check whether the bacterial fusion enzyme is more active than its wild-type counterpart in glycylating yeast mitochondrial tRNA Gly , yeast tRNA m Gly (mitochondrial-encoded mitochondrial tRNA Gly ) was prepared by in vitro transcription. As a control, yeast tRNA n Gly (nuclear-encoded cytoplasmic tRNA Gly ) was also in vitro-transcribed and tested. As shown in Fig. 5, yeast GlyRS1, which efficiently glycylated unfractionated yeast tRNAs (Fig. 5B) and in vitro-transcribed yeast tRNA n Gly (Fig. 5C), feebly charged in vitrotranscribed yeast tRNA m Gly (Fig. 5D). Evidently, tRNA m Gly was a relatively poor substrate for glycylation by yeast GlyRS1, with efficiency at least 30-fold lower than that of tRNA n Gly under the conditions used. As a result, only an almost negligible glycylation activity toward tRNA m Gly was detected for the wild-type and the fusion E. coli enzymes (Fig. 5D). Hence, the fusion approach per se did not confer the bacterial enzyme a higher glycylation activity toward yeast tRNA m Gly . On the contrary, the observed rescue activity may merely reflect the ability of the fusion polypeptide to form a stable and active (ab) 2 complex in yeast mitochondria (Fig. 2), and such a low activity toward yeast tRNA m Gly might be sufficient for sustaining normal mitochondrial function.

Discussion
AaRSs belong to an ancient group of essential translation enzymes that establish the genetic code by coupling amino acids specifically to their cognate tRNAs. However, contrary to longstanding beliefs that a housekeeping gene is highly and constitutively expressed to satisfy the need of viable cell growth, a fairly low level of aminoacylation activity (30 to 50-fold lower than normal) is sufficient for conferring a near wild-type growth phenotype to a yeast strain defective in a specific aaRS activity. Examples of this type include yeast ALA1 [29], GLN4 [28,30], GRS1 [23,25], and VAS1 genes [31]. Such a feature may also contribute to the feasibility of cross-species complementation reported herein (Figs. 2 and 3). Despite the fact that T. thermophilus GlyRS can efficiently charge yeast cytoplasmic tRNA Gly in vitro [27], it failed to provide the cytoplasmic function in yeast, presumably due to misfolding or instability of the protein in the yeast cytoplasm (Fig. 2). Nevertheless, fusion of an MTS to this bacterial enzyme not only enabled the fusion protein to be imported into mitochondria but also enhanced its steady-state protein level. As a result, the MTS fusion enzyme successfully conferred a respiratory phenotype to a yeast strain that was defective in mitochondrial GlyRS activity (Fig. 2). Thus, a protein that is stable in one cellular compartment may be unstable in another, which, in turn, defines its functional potential in vivo.
It was previously shown that fusion of a and b chains of E. coli GlyRS gives rise to a functional enzyme, with a substantial level of aminoacylation activity [32]. Consistent with this observation, the fusion enzyme reported herein retains an aminoacylation activity ,2-fold lower than that of its wild-type counterpart (Fig. 5). This result underscores the possibility that the fusion enzyme adopts an active conformation, presumably a (ab) 2 structure. In contrast with previous observations that a a 2 b 2 -type GlyRS cannot recognize tRNAs Gly with an A73 discriminator base, the bacterial fusion enzyme reported herein effectively rescued the mitochondrial defect of a yeast GRS1 knockout strain (Fig. 2), suggesting that it can glycylate the yeast mitochondrial tRNA Gly to a level that is sufficient for maintaining normal mitochondrial function. Paradoxically, the fusion enzyme does not possess a glycylation activity toward yeast mitochondrial tRNA Gly higher than that of its wildtype counterpart in vitro (Fig. 5). Moreover, the fusion enzyme failed to rescue the cytoplasmic defect of the null allele (Fig. 2) despite the fact that yeast cytoplasmic tRNAs Gly share more identity elements with E. coli tRNAs Gly than does their mitochondrial isoacceptor (Fig. 1B). Perhaps, a higher level of aminoacylation activity (and protein translation) is required for the cytoplasm to sustain viable cell growth.
In the yeast S. cerevisiae, a single GlyRS gene specifies both the cytoplasmic and mitochondrial functions [5]. A similar scenario was found in the GlyRS genes of Sch. pombe, C. elegans, D. melanogaster, B. mori, and H. sapiens [23]. As it happens, the discriminator base N73 of both the nuclear-and mitochondrialencoded tRNA Gly isoacceptors (tRNA n Gly and tRNA m Gly , respectively) is an A in these eukaryotes (Table 1). Such a coincidence appears to be an outcome of co-evolution between GlyRS and its cognate tRNAs. In contrast, the scenario that occurs in A. thaliana is much more complicated. In A. thaliana, N73 is an A in the nuclear-encoded tRNAs Gly , a U in the mitochondrial-encoded tRNA Gly , and a U or a C in the chloroplast-encoded tRNAs Gly [21]. Nonetheless, a nuclear-encoded tRNA Gly (with A73) is imported from the cytosol into the mitochondria of this plant for function. Correspondingly, A. thaliana possesses two distinct GlyRS enzymes: AtGlyRS1 (a a 2 -dimeric enzyme) and AtGlyRS2 [a (ab) 2 -dimeric enzyme] [26]. AtGlyRS1 prefers A73-containing tRNA Gly , while AtGlyRS2 was predicted to glycylate both the organelle-encoded tRNAs Gly and the imported tRNA Gly present in mitochondria [26]. Contrary to this prediction, our results show that only AtGlyRS1 can functionally substitute for yeast GRS1 in vivo (Fig. 3). It has yet to be determined whether AtGlyRS2 really charges the imported tRNA Gly (with A73).
GARS is a dual-functional human GlyRS gene that specifies both the cytoplasmic and mitochondrial functions [33]. Recent studies revealed that human GlyRS is involved in pathways beyond aminoacylation. This protein specifically binds to the poliovirus internal ribosome entry site, thereby facilitating the 59 end-independent internal initiation of translation [34]. In addition, human GlyRS can be secreted from macrophages in response to Fas ligand released from tumor cells, and in turn participate in defense against ERK-activated tumor formation [35]. Moreover, specific mutations in GARS lead to a neurodegenerative disorder, Charcot-Marie-Tooth disease type 2D (CMT2D) [36]. Our study suggests that yeast may serve as a platform for functional analysis of the human GARS gene. Despite the fact that the full-length human GlyRS was poorly expressed in yeast, it successfully rescued both the cytoplasmic and mitochondrial defects of a yeast GRS1 knockout strain, albeit with a weaker cytoplasmic rescue activity. This weak rescue activity could be improved via removal of its N-terminal 120 residues; the resultant enzyme [HsGlyRS(DMTSDWHEP)] had a higher protein expression level and robustly rescued both defects of the yeast knockout strain (Fig. 3). Thus, the WHEP-truncated form of human GlyRS may be preferable over the full-length enzyme for characterizing the CMT2D-causing mutants in yeast. However, regardless of the detailed interpretation, our study suggests that the functional correlation between the oligomeric structure of GlyRS and the discriminator base of tRNA Gly may be looser than previously anticipated [20].

Plasmid construction
Cloning of the GRS1 gene of S. cerevisiae into pADH (a highcopy-number yeast shuttle vector with a constitutive ADH promoter, followed by a multiple cloning site that contains PstI-EagI-SpeI-NdeI-XhoI restriction sites and a short sequence coding for a His 6 tag) was previously described [5,23]. To clone the GARS gene of Homo sapiens, a DNA sequence extending from base pair +1 to +2217 relative to ATG1 of this gene was amplified by a reverse transcription-polymerase chain reaction (RT-PCR) as a NdeI-XhoI fragment using a pair of gene-specific primers, digested with the restriction enzymes NdeI and XhoI, and cloned into the NdeI/ XhoI sites of pADH. Cloning of truncated versions of the human GARS gene followed a similar protocol. To introduce a mitochondrial targeting signal (MTS) into pADH, a DNA sequence that encodes the N-terminal amino acid residues 1-46 of the mitochondrial precursor form of yeast valyl-tRNA synthetase (base pairs +1,+138 relative to ATG1) was amplified by a PCR as an XbaI-SpeI fragment, and it was inserted in-frame into the SpeI site of pADH. The resultant construct pWYH8 can then be used to express an MTS fusion protein under the control of an ADH promoter. Cloning of the genes that encode Thermus thermophilus GlyRS and Arabidopsis thaliana GlyRS1 and GlyRS2 followed a similar strategy.
To co-express a and b subunits of E. coli GlyRS in yeast, the open reading frames that encode these two subunits were cloned in a dual-promoter vector, pCI55. pCI55 is a high-copy-number yeast shuttle vector with a GAP (followed by a multiple cloning site that contains XbaI-BamHI-KpnI-SacI-SalI restriction sites and a short sequence coding for a His 6 tag) and an ADH promoter (followed by a multiple cloning site that contains PstI-EagI-SpeI-NdeI-XhoI restriction sites and a short sequence coding for a His 6 tag). The open reading frames that encode these two subunits were respectively amplified by a PCR as an SpeI-XhoI and an NdeI-XhoI fragment, and they were respectively cloned in the XbaI/ SalI and NdeI/XhoI sites downstream of the GAP and ADH promoters. Insertion of an MTS into the dual-promoter vector followed a protocol described earlier. To fuse a and b chains into a single polypeptide, the open reading frames that encode these two subunits were respectively amplified by a PCR as an SpeI-NdeI and an NdeI-XhoI fragment and simultaneously cloned into pADH or pWYH8 (resulting in pCI35 and pCI124, respectively). The head-to-tail fusion gene constructed in this way encodes a long polypeptide with a tetrapeptide linker (a-Lys-Glu-Ala-His-b).
To generate the green fluorescence protein (GFP) fusion constructs HsGlyRS-GFP and HsGlyRS(DMTSDWHEP)-GFP, the open reading frame that encodes GFP was amplified by a PCR as an XhoI-SalI fragment, digested with the restriction enzymes XhoI and SalI, and inserted in-frame into the XhoI site of pCI29 or pWYH230 (resulting in pCI195 and pCI196, respectively).

Rescue assay for the cytoplasmic function of GRS1
The yeast GRS1 knockout strain, RJT3/II-1, was previously described [24]. Complementation assays for cytoplasmic GlyRS activity were carried out by introducing a test plasmid that carries the gene of interest and a LEU2 marker into RJT3/II-1, and the ability of the transformants to grow in the presence of 5-FOA was determined. Starting from a cell density of 1.0 A 600 , cell cultures were 5-fold serially diluted, and 10-ml aliquots of each dilution were spotted onto the designated plates containing 5-FOA. Plates were incubated at 30uC for 3,5 days. The transformants evicted the maintenance plasmid with a URA3 marker in the presence of 5-FOA, and thus could not grow on the selection medium unless a functional cytoplasmic GlyRS was encoded by the test plasmid.
Rescue assay for the mitochondrial function of GRS1 RJT3/II-1 was co-transformed with a test plasmid (carrying a LEU2 marker) and a second maintenance plasmid (carrying a TRP1 marker) that only expresses the cytoplasmic form of GlyRS (due to a mutation in the initiator codon of the mitochondrial form). In the presence of 5-FOA, the first maintenance plasmid (carrying a URA3 marker) was evicted from the co-transformants, while the second maintenance plasmid was retained. Thus, all cotransformants survived 5-FOA selection, due to the presence of the cytoplasmic GlyRS derived from the second maintenance plasmid. The mitochondrial phenotypes of the co-transformants were further tested on YPG plates at 30uC, with results documented on day 3 following plating. Because a yeast cell cannot survive on glycerol without functional mitochondria, the co-transformants did not grow on the YPG plates unless a functional mitochondrial GlyRS was generated from the test plasmid.

Western blotting
The protein expression patterns of the constructs used in the complementation assays were determined via a chemiluminescence-based Western blot analysis. INVSc1 was first transformed with the constructs of interest, and total protein extracts were prepared from the transformants. Aliquots of the protein extracts (2 or 50 mg) were loaded onto a mini gel (size: 8610 cm) that contains 10% polyacrylamide and electrophoresed at 100 V for ,1 h. Following electrophoresis, the resolved proteins were transferred using a semi-dry transfer device to a PVDF membrane in a buffer containing 30 mM glycine, 48 mM Tris base (pH 8.3), 0.037% sodium dodecylsulfate, and 20% methanol. The membrane was probed with an HRP-conjugated anti-His 6 tag antibody, and then exposed to x-ray film following the addition of appropriate substrates.

Aminoacylation assay
Aminoacylation reactions were carried out at 25uC in a buffer that contains 50 mM HEPES (pH 7.5), 50 mM KCl, 15 mM MgCl 2 , 5 mM dithiothreitol, 10 mM ATP, 0.1 mg/ml bovine serum albumin (BSA), 100 mM unfractionated yeast tRNA (Boehringer Mannheim, Germany) or 5 mM in vitro-transcribed yeast tRNA Gly , and 24 mM glycine (4 mM 3 H-glycine; Moravek Biochemicals, Brea, CA). The specific activity of 3 H-glycine used was 35.0 Ci/mmol. Purification of GlyRS and in vitro transcription of tRNA followed protocols described earlier [28,37]. Determination of the active protein concentrations by active site titration was as previously described [38]. Reactions were quenched by spotting 10 ml aliquots of the reaction mixture onto Whatman filters (Maidstone, UK) soaked in 5% trichloroacetic acid and 1 mM glycine. Filters were washed three times for 15 min each in icecold 5% trichloroacetic acid prior to liquid scintillation counting. Data were obtained from three independent experiments and averaged. Error bars indicate (62 x standard deviation).