Cross-species and Cross-compartmental Aminoacylationof Isoaccepting tRNAs by a Class II tRNA Synthetase*

Received for publication, February 27, 2006, and in revised form, August 7, 2006 Published, JBC Papers in Press, August 23, 2006, DOI 10.1074/jbc.M601869200

Hsiao-Yun Huang‡, Yu Kuei‡, Hen-Yi Chao‡, Shun-Jia Chen‡, Lu-Shu Yeh§1, and Chien-Chia Wang‡2

From the ‡Department of Life Science, National Central University, Jung-li, Taiwan 32001 and §Department of Life Science,Tzu-Chi University, Hua-lien, Taiwan 97041

It was previously shown thatALA1, the only alanyl-tRNA syn-thetase gene in Saccharomyces cerevisiae, codes for two func-tionally exclusive protein isoforms through alternative initia-tion at two consecutive ACG codons and an in-framedownstreamAUG.We reported here the cloning and character-ization of a homologous gene from Candida albicans. Func-tional assays show that this gene can substitute for both the cyto-plasmic and mitochondrial functions of ALA1 in S. cerevisiaeand codes for two distinct protein isoforms through alternativeinitiation from two in-frame AUG triplets 8-codons apart.Unexpectedly, although the short form acts exclusively in cyto-plasm, the longer form provides function in both compart-ments. Similar observations are made in fractionation assays.Thus, the alanyl-tRNA synthetase gene of C. albicans hasevolved an unusual pattern of translation initiation and proteinpartitioning and codes for protein isoforms that can aminoacy-late isoaccepting tRNAs fromadifferent species and fromacrosscellular compartments.

Typically there are 20 aminoacyl-tRNA synthetases in pro-karyotes, one for each amino acid (1–4). These enzymes eachcatalyze the formation of an aminoacyl-tRNA by attaching aparticular amino acid to the 3�-end of its cognate tRNA, withaccompanying hydrolysis of ATP to AMP and pyrophosphate.The activated amino acid, i.e. aminoacyl-tRNA, is then trans-ferred to ribosome for protein synthesis. In eukaryotes proteinsynthesis occurs not only in the cytoplasm but also inorganelles, such as mitochondria and chloroplasts (5). Com-partmentalization of the protein synthesis machinery withinthe cytoplasm and organelles of eukaryotes leads to isoaccept-ing tRNA species that are distinguished by nucleotidesequence, subcellular location, and enzyme specificity. Thus,eukaryotes such as yeast commonly have two genes that encodedistinct sets of proteins for each aminoacylation activity, onelocalized to the cytoplasm and the other to the mitochondria.Each set aminoacylates the isoaccepting tRNAs within its

respective cell compartment. Except for some algae (6), all ami-noacyl-tRNA synthetases are encoded by nuclear genes regard-less of the cell compartments to which they are confined.In contrast tomost known eukaryotic tRNA synthetases, two

Saccharomyces cerevisiae genes, HTS1 (the gene encoding his-tidyl-tRNA synthetase) (7) and VAS1 (the gene encoding valyl-tRNA synthetase (ValRS)) (8), specify both the mitochondrialand cytosolic forms through alternative initiation from two in-frame AUG codons. Each of these genes encodes mRNAs withdistinct 5�-ends. Some of these mRNAs have their 5�-endslocated upstream of the first AUG codon, whereas others havetheir 5�-ends located between the first and second AUGcodons. The mitochondrial form of the enzyme is translatedfrom the first AUG on the “long” messages, whereas the cyto-solic form is translated from the second AUG on the “short”messages. As a consequence, the mitochondrial enzymes havethe same polypeptide sequences as their cytosolic counterparts,except for a short amino-terminal mitochondrial targetingsequence. The transit peptide is subsequently cleaved awayupon import into the mitochondria. Because the two isoformsare targeted to different subcellular compartments, they cannotsubstitute for each other in vivo (7–8). A similar scenario hasbeen observed for the genes that encode themitochondrial andcytoplasmic forms of Arabidopsis thaliana alanyl-tRNA syn-thetase (AlaRS),3 threonyl-tRNA synthetase, and ValRS (9).

Recently it was shown that ALA1, the only gene coding forAlaRS in S. cerevisiae, also encodes distinct protein isoforms(10–11). Although the cytoplasmic form is initiated from acanonical AUG triplet, its mitochondrial counterpart is initi-ated from two successive in-frameACG triplets that are located23 codons upstream of the AUG initiator, i.e. ACG(�25)/ACG(�24). These two forms function exclusively in theirrespective compartments and, thus, cannot substitute for eachother under normal conditions. A similar scenario has beenobserved in GRS1 (12), the only active yeast gene coding forglycyl-tRNA synthetase. Because to date examples of nativenon-AUG initiation are still rare in low eukaryotes (10–12), wewondered whether a similar mechanism of translation initia-tion has been conserved in the AlaRS genes of other yeastsduring evolution. In addition, we wondered whether the AlaRSgene of a closely related yeast species, such asCandida albicans,also provides function in both compartments and whether it

* This work was supported in part by National Science Council (Taiwan) GrantNSC 94-2311-B-008-009 (to C.-C. W.) and by Institute of Nuclear EnergyResearch, Atomic Energy Council (Taiwan) Grant 94-2001-INER-EE-009 (toC.-C. W.). The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby marked“advertisement” in accordance with 18 U.S.C. Section 1734 solely to indi-cate this fact.

1 To whom correspondence may be addressed. Tel.: 886-3-856-5301 (ext.7541); Fax: 886-3-857-2526; E-mail: susanyeh@mail.tcu.edu.tw.

2 To whom correspondence may be addressed. Tel.: 886-3-426-0840; Fax:886-3-422-8482; E-mail: dukewang@cc.ncu.edu.tw.

3 The abbreviations used are: AlaRS, alanyl-tRNA synthetase; ADH, alcoholdehydrogenase; 5-FOA, 5-fluoroorotic acid; ValRS, valyl-tRNA synthetase;YPG, yeast extract-peptone-glycerol; RACE, rapid amplification of cDNAends; RT, reverse transcription.

THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 281, NO. 42, pp. 31430 –31439, October 20, 2006© 2006 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in the U.S.A.

31430 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 281 • NUMBER 42 • OCTOBER 20, 2006

by guest on February 4, 2020http://w

ww

.jbc.org/D

ownloaded from

can surmount the species barrier and charge the tRNAs of S.cerevisiae. It is our hope that results obtained from this studycould provide further insight into the bifunctional nature of aparticular nuclear gene and the diversity of mechanisms bywhich protein isoforms can be partitioned between two distinctcompartments. In thework described here we presented exper-imental evidence that an ALA1 homologue of C. albicans (des-ignated here as CaALA1) can rescue both the cytoplasmic andmitochondrial defects of a S. cerevisiae ala1� strain. Similar tothe ALA1 gene in S. cerevisiae, two protein isoforms with dis-tinct amino termini are alternatively generated from this gene;however, no non-canonical initiators are involved in this case.Instead, these isoforms are initiated from two in-frame AUGtriplets. Even more unexpectedly, whereas the short form thatis initiated from the second AUG is confined in the cytoplasm,the longer form that is initiated from the first AUG is dual-targeted and, thus, bifunctional. The implications of theseobservations will be further discussed in the context of co-evo-lution of tRNAs and their cognate tRNA synthetases.

EXPERIMENTAL PROCEDURES

Construction of Plasmids—Cloning of CaALA1 from C. albi-cans followed standard protocols (13). The wild-type CaALA1sequence (base pairs �300��2910 relative to ATG1) wasamplified by PCR and cloned into pRS315 (a low copy numberyeast vector) or pRS425 (a high copy number yeast vector). Ashort sequence coding for a His6 tag or FLAGwas subsequentlyinserted in-frame into the 3�-end of the CaALA1 open readingframe. Various point mutations, such as ATG1/ATA2 to TCT/AGA and ATG9 to GCG, were introduced into the wild-typeclone following standard protocols (the number 1 in ATG1refers to the codon position in the open reading frame). TocloneCaALA1 in pADH (a high copy number yeast vector withan ADH promoter), a segment of CaALA1 DNA containingbase pairs�40��2910 relative to ATG1was amplified by PCRas an EagI/XhoI fragment and cloned into appropriate sites ofthis vector.Cloning of CaALA1-VAS1c constructs followed a strategy

described earlier (12). Basically, various CaALA1 sequences(�370��102 bp, �370��126 bp, or �370��159 bp) werePCR-amplified as EagI-SpeI fragments and fused in-frame tothe EagI/SpeI sites 5� to VAS1c (the open reading frame codingfor the cytoplasmic formofValRS) cloned in a low copy numbervector pRS315, resulting in various CaALA1-VAS1c constructsin which the ATG initiator for the cytoplasmic form of ValRShad been mutated.Mapping the 5�-Ends of CaALA1 Transcripts—Identification

of the 5�-ends of CaALA1 transcripts was carried out with5�-RACE (rapid amplification of cDNA ends; Invitrogen).Briefly, total RNA isolated from C. albicans was first treatedwith alkaline phosphatase to remove the 5�-phosphate groupfrom truncated mRNA and non-mRNA and then with tobaccoacid pyrophosphatase to remove the 5�-cap from intact full-length mRNA. An RNA oligonucleotide was subsequentlyfused to the 5�-end of the decapped mRNA with RNA ligase.The 5�-end modified mRNA was transcribed with SuperScriptIII reverse transcriptase into first strand cDNAs using an “anti-sense” CaALA1-specific primer that was annealed to a region

630-bp downstream of ATG1. The reaction mixture wastreated with RNaseH, and the first strand cDNAproducts werethen amplified via PCR using Pfu DNA polymerase with aprimer (provided by the manufacturer) annealed to the 5�-endof the cDNA and a nested CaALA1-specific primer annealed600 bp downstream of ATG1. After PCR-driven amplification,the double-stranded cDNA products were cloned andsequenced.Sequencing of the Mitochondrial Form of CaAlaRS—Deter-

mination of the amino terminus of the processed mitochon-drial form of CaAlaRS was carried out by the Edman degrada-tion method. First, mitochondria were isolated fromtransformants carrying the wild-type (pIVY97) and ATG9mutant (pIVY118) constructs (14), and the His6-tagged pro-teins expressed were purified by nickel-nitrilotriacetic acid col-umn chromatography. After SDS-polyacrylamide gel electro-phoresis, the proteins were transferred to a nitrocellulosemembrane and stainedwithAmido Black, and the protein bandof the correct size was removed and sequenced.Complementation Assays for the Cytoplasmic Function of

ALA1—The yeast ALA1 knock-out strain TRY11 was asdescribed (15). This strain is maintained by a plasmid encodingAlaRS and the URA3marker. Complementation assays for thecytoplasmic function of plasmid-borne ALA1 and derivativeswere carried out by introducing a test plasmid into TRY11 anddetermining the ability of transformants to grow in the pres-ence of 5-fluoroorotic acid (5-FOA). The cultures were incu-bated at 30 °C for 3–5 days or until colonies appeared (photosfor the complementation assays were taken at day 3 after incu-bation). The transformants evicted the maintenance plasmidwith theURA3marker in the presence of 5-FOA. Thus, only anenzyme with the cytoplasmic AlaRS activity encoded by thesecond plasmid (with the LEU2 marker) could rescue thegrowth defect.Complementation Assays for the Mitochondrial Function of

ALA1—Complementation assays for the mitochondrial func-tion of plasmid-borneALA1 and derivativeswere carried out byintroducing a test plasmid (carrying a LEU2marker) and a sec-ondmaintenance plasmid (carrying aHIS3marker) intoTRY11and selecting on a plate containing 5-FOA. The second main-tenance plasmid used in this assay contained ALA1(I(�1)stop),which expresses a functional cytoplasmicAlaRS but is defectivein mitochondrial AlaRS activity. In the presence of 5-FOA, thefirst maintenance plasmid (containing a URA3 marker) wasevicted from the co-transformants, whereas the second main-tenance plasmid was retained. Thus, all co-transformants sur-vived 5-FOA selections due to the presence of the cytoplasmicAlaRS derived from the second maintenance plasmid. The co-transformants were further tested onYPGplates for theirmito-chondrial phenotypes at 30 °C, with results documented at day3 after plating. Because a yeast cell cannot survive on glycerolwithout functional mitochondria, the co-transformants do notgrow on YPG plates unless a functional mitochondrial AlaRS ispresent. Complementation assays for various VAS1 constructsand their derivatives were conducted essentially the same wayas those for ALA1 constructs, except that a VAS1 knock-outstrain, CW1, was used as the test strain (16).

A Bifunctional tRNA Synthetase

OCTOBER 20, 2006 • VOLUME 281 • NUMBER 42 JOURNAL OF BIOLOGICAL CHEMISTRY 31431

by guest on February 4, 2020http://w

ww

.jbc.org/D

ownloaded from

Western Blot—The protein expression patterns of variousCaALA1 constructs were determined by a chemiluminescence-based Western blot analysis following standard protocols. TheCaALA1 constructs were first transformed into INVSc1 (Nova-gen), and the resultant transformants were subsequently grownin a selection medium lacking leucine. The total, cytoplasmic,and mitochondrial fractions were prepared from each of thetransformants according to the protocols described by Daumet al. (14). 45 �g of the protein extracts were loaded onto a gel(size, 8 � 10 cm) containing 8% polyacrylamide and electro-phoresed at 130 volts for 1�2 h. The resolved proteins weretransferred onto a nitrocellulose membrane using a semidryblotting device. The membrane was hybridized with a horse-radish peroxidase-conjugated anti-His6 tag (Invitrogen) oranti-FLAG antibody (Sigma) and then exposed to x-ray filmafter the addition of the appropriate substrates.

Growth Curve Assay—Growthcurve assays for the plasmid-bornemitochondrial AlaRS activities werecarried out in YPG broth. TRY11was first co-transformed with a sec-ond maintenance plasmid (carryinga HIS3 marker) and a test plasmid(carrying a LEU2 marker), and theresultant co-transformants wereplated on a FOA plate. After FOAselections, one colony of the survi-vors was picked and inoculated into3 ml of SD broth lacking histidineand leucine and grown to stationaryphase. The cells were washed threetimes with YPG broth, and appropri-ate amounts were transferred to aflask containing 10 ml of YPG brothto a final cell density of A600 � 0.1.The cell culture was shaken in a 30 °Cincubator, and the cell density of theculture was checked every 4 h for aperiod of 48 h.

RESULTS

Cross-species Complementationof an S. cerevisiae ala1� Strain by aHomologue of C. albicans—Unlikemost yeast tRNA synthetases thathave two distinct nuclear genes (onecoding for the cytoplasmic enzymeand the other for its mitochondrialcounterpart), CaALA1 appears tobe the only ALA1 homologue in theyeast C. albicans. We wonderedwhether this gene actually providesAlaRS function in vivo and whetherit could code for both cytoplasmicandmitochondrial activities, aswiththe case of ALA1 in S. cerevisiae. Tofurther our understanding on thisgene, we first scanned the 5�-termi-

nal nucleotide sequences of this gene for potential translationstart codons that might be involved in the synthesis of proteinisoforms. As shown in Fig. 1A, there are two nearby in-frameATG codons, i.e. ATG1 and ATG9, close to the 5�-end of itsopen reading frame. In addition, four potential non-ATG initi-ators, i.e. non-ATG codons that differ from ATG by a singlenucleotide, are present in the sequence between ATG1 andTGA(-47), the closest termination codon.We wondered whichof these triplets is the authentic start sites for CaALA1. Beforeproceeding, the transcription profiles of this gene in vivo wereelucidated using 5�-RACE. Fig. 1B showed that a single tran-script, with its 5�-end mapped to nucleotide position �24 rela-tive to ATG1, was amplified by RT-PCR using total RNAextracts ofC. albicans as the templates (Fig. 1B). This transcriptwas, thus, considered to be the template for protein translation.In addition, sequences of the cDNAs (�600 bp determined)

FIGURE 1. Determining the transcription initiation sites of CaALA1. A, the 5� sequences of CaALA1, extend-ing from 141-bp upstream to 102-bp downstream of ATG1. The first two ATG triplets, ATG1 and ATG9, areboxed, and the upstream non-ATG triplets that differ from ATG by a nucleotide are underlined. Labeled on topof the nucleotide sequence is the transcription initiation site at bp �24 relative to ATG1. The amino acidresidues deduced from the CaALA1 open reading frame (between Met-1 and Leu-34) are shaded. The cleavagesite for mitochondrial matrix processing peptidase is marked with Œ. B, the cDNA product of 5�-RACE. Lane 1,DNA markers; lane 2, 5�-RACE product. The migrating positions of the DNA markers on a 2% agarose gel and the5�-RACE product are labeled on the left and right, respectively. C, comparison of the sequence identities amongAlaRS proteins of different origins. Ca, C. albicans; Sc, S. cerevisiae; Sp, S. pombe; Ec, E. coli. D, alignment of theamino-terminal sequences of AlaRS proteins of different origins. Amino acid residues conserved among thesesequences are shaded.

A Bifunctional tRNA Synthetase

31432 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 281 • NUMBER 42 • OCTOBER 20, 2006

by guest on February 4, 2020http://w

ww

.jbc.org/D

ownloaded from

obtained from RT-PCR were identical to those of the genomicDNA (data not shown), suggesting that the possibilities of alter-native splicing at the 5�-end ofCaALA1mRNAs and translationinitiation from an AUG codon spliced from afar could be ruledout. Comparison of the protein sequences among AlaRSs ofdifferent origins showed that CaAlaRS (deduced from its puta-tive open reading frame starting at ATG1) shares a significantlyhigher sequence identity to those from S. cerevisiae (cytoplas-mic form) (�68%) and Schizosaccharomyces pombe (�61%)than to the Escherichia coli enzyme (�39%) (Fig. 1C). Mostintriguingly, this protein appears to have an amino-terminal15-residue appendage that is absent from the other yeast cyto-plasmic enzymes compared (Fig. 1D). It should be noted thatthe cytoplasmic form of AlaRS of S. cerevisiae starts from thesecond Met on the sequence shown in Fig. 1D. We surmisedthat if this protein does have mitochondrial function, thisappendage could serve as at least part of its mitochondrial tar-geting signal.Because the genetic system has not been well developed for

C. albicans, we wondered whether we could test the biologicalfunctions of CaALA1 in a closely related and well developedyeast system such as S. cerevisiae. To this end, CaALA1 clonedin various vectors was transformed into a S. cerevisiae ala1�

yeast strain, TRY11, and tested for its complementing activity.As shown in Fig. 2, the wild-type CaALA1 gene cloned inpRS315 (a low copy number vector), pRS425 (a high copy num-ber vector), or pADH (a high copy number vector with a con-stitutive ADH promoter) efficiently rescued both the cytoplas-mic and mitochondrial defects of TRY11, i.e. transformantscarrying the plasmid-borne CaALA1 gene formed coloniesafter 2�3 days of growth on FOA (Fig. 2B) and YPG (Fig. 2C)

plates, respectively. These results indicated that the homolo-gous gene from C. albicans could overcome the species barrierand encodes both cytoplasmic and mitochondrial AlaRSactivities.Generation of Two Functionally Overlapping Protein Iso-

forms through Alternative Translation Initiation—The ques-tion arose as to howmany protein isoforms are generated fromCaALA1 and whether the upstream non-ATG triplets (relativeto ATG1) are involved in the translation initiation. To shedlight on this matter, various CaALA1 constructs were clonedusing pRS425 as the vector and tested for their complementa-tion activities. First, the codon at position �1 was mutated to astop codon TAA to block all the possible translational eventsinitiated upstream of ATG1 and then tested for its effect on thecytoplasmic and mitochondrial functions of this gene. Fig. 3shows that the newly introduced stop codon affected neitherthe mitochondrial nor cytoplasmic function of this gene (seepIVY102), suggesting that the upstream potential non-ATGinitiators are not involved in the synthesis of the alanineenzyme(s). We next aimed at the two nearby ATG triplets, i.e.ATG1 andATG9, for their possible participation in translation.Mutation of ATG1 (see pIVY100 or pSAM35) or insertion oftwo nucleotides between ATG1 and ATG9 (causing ATG1 tobe out-of-frame with respect to the rest of the open readingframe) (see pSAM25) specifically impaired the mitochondrialfunction of this gene (Fig. 3, B and C), suggesting that ATG1was the sole initiator responsible for the translation of themito-chondrial form and the cytoplasmic form was initiated else-where, possibly from ATG9. Further mutation of ATG9 inpIVY100 (resulting in pIVY104) abolished the remaining cy-toplasmic activity, indicating that the cytoplasmic function ofpIVY100 was indeed provided by protein product initiated atATG9. However, much to our surprise, mutation of ATG9alone did not impair the cytoplasmic function as expected (seepSAM33); instead, the ATG9 mutant still retained both activi-ties. These results suggested that the long form that is initiatedfrom ATG1 provided both the cytoplasmic and mitochondrialfunctions, whereas the shorter form that is initiated fromATG9provided only the cytoplasmic activity. Given that the out-of-frame mutant (see pSAM25) still retained the cytoplasmicfunction, it is likely that the second ATG triplet can be recog-nized by scanning ribosomes as a remedial translation start siteeven in the presence of the first ATG triplet. Therefore, thecytoplasmic function of this gene is probably contributed byboth isoforms under normal conditions. It is noteworthy thatmutation of ATG1 to GCG (resulting in pSAM35) also createdan out-of-frame ATG triplet (between the nucleotides �2 and�1) andhad a phenotype similar to that of pSAM25.Toprovidea more quantitative data on the mitochondrial complementa-tion activity of these mutants, a growth curve assay was subse-quently carried out in YPG broth. Consistent with the observa-tions made from the complementation assay (Fig. 3C),transformants carrying pSAM33 (with ATG9 inactivated) orpSAM20 (with the wild-type gene) grew well in YPG broth,whereas growth of those carrying pIVY100 (with ATG1 inacti-vated), pSAM25 (with an out-of-frame mutation between thetwo ATG codons), or pIVY104 (with both ATG triplets inacti-vated) was severely impaired (Fig. 3D). Analysis of the relative

FIGURE 2. Cross-species complementation of a S. cerevisiae ala1� strainby CaALA1. TRY11 was transformed with the wild-type CaALA1 gene clonedin various vectors and then tested for its growth phenotypes. Complementa-tion of the cytoplasmic and mitochondrial defects of the ala1� strain wasshown by its ability to lose the maintenance plasmid and grow on a FOA anda YPG plate, respectively. A, summary of the ALA1 constructs and theircomplementation activities. B, complementation assays for the cytoplasmicfunction on a 5-FOA plate. C, complementation assays for the mitochondrialfunction on a YPG plate. In B and C, the numbers 1–5 denote constructs shownin A. Cyt, cytoplasmic; Mit, mitochondrial.

A Bifunctional tRNA Synthetase

OCTOBER 20, 2006 • VOLUME 281 • NUMBER 42 JOURNAL OF BIOLOGICAL CHEMISTRY 31433

by guest on February 4, 2020http://w

ww

.jbc.org/D

ownloaded from

FIGURE 3. Identifying the translation initiation sites of CaALA1. Various mutations were individually introduced into CaALA1 cloned in pRS425 and testedfor their effect on the complementing activities. A, summary of the CaALA1 constructs and their complementation functions. TRY11 was transformed with thewild-type and mutant CaALA1 constructs and then tested for its growth phenotypes. Nucleotide sequences shown include nucleotides �3��27, i.e. codons�1��9, relative to ATG1. Codons that have been mutated are shaded. B, complementation assays for the cytoplasmic function on a 5-FOA plate. C, comple-mentation assays for the mitochondrial function on a YPG plate. D, growth curves of the TRY11 transformants containing various plasmid-borne CaALA1constructs in YPG broth. E, RT-PCR. The relative amounts of specific CaALA1 mRNA generated from some of the constructs shown in A were determined byRT-PCR. As an internal control, the relative amounts of actin-specific mRNA in each sample were also determined. In B--E, the numbers 1– 8 denote constructsshown in A. Cyt, cytoplasmic; Mit, mitochondrial.

A Bifunctional tRNA Synthetase

31434 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 281 • NUMBER 42 • OCTOBER 20, 2006

by guest on February 4, 2020http://w

ww

.jbc.org/D

ownloaded from

levels of specific mRNAs generated from each of these con-structs indicated that similar levels ofCaALA1 transcripts weregenerated from or maintained in the transformants carryingthewild-type ormutantCaALA1 constructs as determined by asemiquantitative RT-PCR experiment (Fig. 3E). This observa-tion suggested that these mutations had little effect on the sta-bility of the specificmRNAs generated from the constructs and,therefore, most likely modulated only the translation initiationactivity of the individual initiators.Because the observations made above were a result of AlaRS

proteins produced from a high copy number plasmid, we wereafraid that they might not accurately reflect protein activity atphysiological levels. Therefore, wild-type and mutant CaALA1genes were cloned into a low copy number shuttle vector (Fig.4A) and tested for their cytoplasmic and mitochondrial func-tions. As shown in Fig. 4, B and C, these constructs had com-plementing activities similar to the corresponding constructscloned in a high copy number vector.Partition Pattern of ATG1- and ATG9-initiated CaAlaRS

Isoforms—To investigate whether theCaALA1 constructs con-tain similar activities when highly expressed from a constitutiveADH promoter, some of the representative constructs shown inFig. 3Awere subcloned into pADH and tested for their comple-mentation functions. As shown in Fig. 5,A–C, these constructscontained similar complementation functions to those clonedin pRS425, except for the ATG1 mutant, which contained onlycytoplasmic function when cloned in pRS425 but containedboth cytoplasmic and mitochondrial functions when cloned inpADH (compare pIVY100 and pIVY98). One likely possibilityleading to this outcome is that the ATG9-initiated form con-

tains a cryptic mitochondrial targeting signal that normallydoes not play a role in mitochondrial localization but can berecruited to function when the protein is highly expressed.To directly look at the protein expression levels and elucidate

the correlations between complementation functions and par-tition patterns of the isoforms within the cell, the total, mito-chondrial, and cytoplasmic fractions were isolated from each ofthe transformants harboring various CaALA1 constructs. Asshown in Fig. 5D, the proteins expressed from the wild-typeCaALA1 construct were partitioned between cytoplasm andmitochondria (lane 1, pIVY97), but mutations that inactivatedboth of the ATG initiators completely abolished the synthesisof the isoforms (lane 4, pIVY149). Interestingly, when the firstATG initiator was inactivated (lane 2, pIVY98), the protein lev-

FIGURE 4. Complementation by CaALA1 constructs cloned in a low copynumber vector. Various mutations were individually introduced into CaALA1cloned in pRS315 and tested for their effect on the complementing activities.A, summary of the CaALA1 constructs and their complementation functions.B, complementation assays for the cytoplasmic function on a 5-FOA plate. C,complementation assays for the mitochondrial function on a YPG plate. In Band C the numbers 1– 6 denote constructs shown in A. Cyt, cytoplasmic; Mit,mitochondrial.

FIGURE 5. Localization of CaAlaRS isoforms expressed from a high copynumber vector. Various mutations were individually introduced into CaALA1cloned in pADH and tested for their effect on the complementing activities. A,summary of the CaALA1 constructs and their complementation functions.Nucleotide sequences shown include nucleotides �3��27, i.e. codons�1��9, relative to ATG1. Codons that have been mutated are shaded. B,complementation assays for the cytoplasmic function on a 5-FOA plate. C,complementation assays for the mitochondrial function on a YPG plate.Cyt, cytoplasmic; Mit, mitochondrial. D, fractionation and Western blots. Thetotal, cytoplasmic, and mitochondrial fractions of the transformants harbor-ing various CaALA1 constructs were isolated and analyzed by Western blotsusing anti-His6 tag antibody (upper three panels). As internal controls, thecytoplasmic and mitochondrial fractions extracted from these transformantswere probed with an antibody mixture containing both anti-phosphoglycer-ate kinase (PGK) and anti-Hsp60 (lower two panels). The rightmost lane in thelower two panels shows the hybridization patterns of the total extracts oftransformants containing pIVY97 (shown as Cont.). Indicated on the right arethe relative migrating positions of CaAlaRS, Hsp60, and phosphoglyceratekinase (PGK), respectively. E, relative levels of CaAlaRS isoforms. The totalfractions of the transformants were serially diluted and analyzed by Westernblots. In B–E, the numbers 1–5 denote constructs shown in A.

A Bifunctional tRNA Synthetase

OCTOBER 20, 2006 • VOLUME 281 • NUMBER 42 JOURNAL OF BIOLOGICAL CHEMISTRY 31435

by guest on February 4, 2020http://w

ww

.jbc.org/D

ownloaded from

els in the total fraction remained almost unchanged, whereasthe protein band in the mitochondrial fraction drasticallydecreased (compare lanes 1 and 2), suggesting that the mito-chondrial proteins came largely from initiation at ATG1. Inaddition, it came as a surprise to us at first to find that theprotein levels in the cytoplasmic fraction of this mutant appre-ciably increased. We surmised that the basis underlying thisunexpected observation could probably be attributed to the factthat ATG9 served only as a remedial initiation site in the wild-type construct (lane 1) but became the first available ATG ini-tiator in the ATG1 mutant (lane 2), leading to the higherexpression of the “cytoplasmic” form. To assess the initiatingactivity of the remedial initiation site, i.e. ATG9, more accu-rately, ATG1was left unaltered, and two extra nucleotides wereinserted into the sequence between the two ATG initiators,causing the firstATG to be out-of-framewith respect to the restof the open reading frame (lane 5, pIVY152). Under such con-ditions the protein band in the total fraction only slightlydecreased due to loss of the ATG1-initiated protein, whereasthe protein level in the cytoplasmic fraction was almostunchanged (compare lanes 1 and 5). But most significant of all,no protein was seen in the mitochondrial fraction of this out-of-frame mutant (lane 5), suggesting that the ATG9-initiatedprotein form was exclusively confined to the cytoplasm whenATG9 serves only as a remedial initiation site (compare lanes 1and 5) and could be forced into mitochondria, possibly due tothe presence of a cryptic mitochondrial targeting signal, whenATG9 serves as the first available initiator, resulting in higherexpression of the short form (compare lanes 2 and 5). By con-trast, when the second ATG initiator was inactivated (lane 3;pIVY118), the levels of the proteins in the total or cytoplasmicextracts drastically decreased, suggesting that the cytoplasmicproteins came largely from initiation at ATG9 in the wild-typeconstruct (compare lanes 1 and 3), and theATG1-initiated pro-tein form can be partitioned in both compartments (lane 3),with the major portion targeted to the mitochondria. It is note-worthy that the protein level in the mitochondrial fraction ofthis mutant appreciably increased as compared with the wild-type construct (compare lanes 1 and 3). We surmised that per-haps this unexpected increase was due to alterations of thepotential MPP cleavage site, which might affect its processingand in turn its distribution. As a control, themitochondrial andcytoplasmic fractions were also probed with a mixture of anti-phosphoglycerate kinase (a cytoplasmic marker protein) andanti-Hsp60 (a mitochondrial marker protein) to check forcross-contamination. As shown in Fig. 5D, no serious cross-contamination was seen in these preparations (lower two pan-els). To further quantify the initiating activity of ATG1 andATG9, the relative protein levels in the total fractions ofpIVY152 (an out-of-frame mutant) and pIVY118 (an ATG9mutant) were compared. Fig. 5E showed that the initiatingactivity of ATG9 is around 4-fold as high as that of ATG1 underthe conditions used. As a control, the relative protein levelsgenerated frompIVY97 (wild type)were also shown. This resultsuggested that the long and short forms account for �20 and�80% of the total proteins generated from CaALA1.We next checked the protein partition patterns for CaAlaRS

isoforms expressed from a low copy number plasmid. As shown

in Fig. 6, the partitioning patterns obtained were very similar tothose shown in Fig. 5. However, the protein levels expressedfrom ATG1 under native conditions were only about 4% rela-tive to those initiated fromATG9 (Fig. 6B). Because the relativeamount of protein produced from ATG1 under native condi-tions is much lower than that produced under conditions ofoverexpression, the faint band seen in lane 6 (and possibly aportion of the protein seen in lane 4) is likely a result of slightcontamination from the cytoplasmic fraction. Such contamina-tion becomes significant under native conditions but is tooslight to be seen under conditions of overexpression.To determine whether the ATG1-initiated protein form is

indeed processed in mitochondria, we subsequently sequencedthe His6-tagged CaAlaRS proteins purified from themitochon-drial fraction of transformants harboring pIVY97 (wild-type)and pIVY118 (Met-9 mutation to Ala-9). The results showedthat the purified proteins have an amino-terminal sequence ofMSSNTTI and ASSNTTI, respectively, suggesting that theATG1-initiated protein can be targeted to mitochondria andprocessed by matrix-processing peptidase between residueseight and nine (Fig. 1A). To rule out the possibility that proteinsoverexpressed from theADH promotermay overload the proc-essing system and lead to aberrant cleavage, the His6-taggedCaAlaRS proteinwas also purified from themitochondrial frac-tion of transformants harboring a ATG9 mutant constructcloned in a low copy number vector and sequenced.As it turnedout, the processed mitochondrial form contained an expectedamino-terminal sequence of ASSNTTI.Demonstration of Two Distinct Protein Isoforms Initiated

from ATG1 and ATG9—Because the ATG1- and ATG9-initi-ated protein isoforms are very similar in size (�106 kDa), it isimpractical to directly distinguish them by Western blot. Fur-thermore, the long protein formmay be processed in the mito-chondria to a size similar to the short form (Fig. 5). To ascertain

FIGURE 6. Localization of CaAlaRS isoforms expressed from a low copynumber vector. Transformants harboring various CaALA1 constructs clonedin pRS315 were subjected to fractionation, and the relative levels of cytoplas-mic and mitochondrial CaALA1 proteins were probed by Western blot. A, frac-tionation and Western blots. The total, cytoplasmic, and mitochondrial frac-tions of the transformants harboring various CaALA1 constructs were isolatedand analyzed by Western blots using anti-FLAG antibody (upper panel). Asinternal controls, the total, cytoplasmic, and mitochondrial fractionsextracted from these transformants were probed with an antibody mixturecontaining both anti-phosphoglycerate kinase (PGK) and anti-Hsp60 (lowerpanel). Indicated on the right are the relative migrating positions of CaAlaRS,Hsp60, and phosphoglycerate kinase, respectively. Lanes 1, 4, and 7, wild-type; lanes 2, 5, and 8, ATG9 mutant; lanes 3, 6, and 9, out-of-frame mutant. B,relative levels of CaAlaRS isoforms. The total fractions of the transformantswere serially diluted and analyzed by Western blots. In B the numbers 2 and 3(circled) denote constructs shown in A. Cyt, cytoplasmic; Mit, mitochondrial.

A Bifunctional tRNA Synthetase

31436 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 281 • NUMBER 42 • OCTOBER 20, 2006

by guest on February 4, 2020http://w

ww

.jbc.org/D

ownloaded from

whether two distinct proteins are independently initiated fromthese two nearby initiators, we next fused the 5�-terminalsequence of CaALA1 (base pairs �40��30) to a smaller gene,lexA (thereby both avoiding mitochondrial localization andincreasing relative size difference of the resulting fusions) andtested for its protein expression profile. These fusions wereexpressed under the control of anADHpromoter and separatedon an 18% SDS-PAGE (size, 16 � 20 cm). As shown in Fig. 7, aminor upper band and amajor lower bandwere simultaneouslygenerated from pIVY155 (containing both ATG1 and ATG9),whereas mutation of ATG1 or introduction of an out-of-framemutation between ATG1 and ATG9 impaired the productionof the upper band (see pIVY156 and pIVY159), suggesting thatthe upper band is initiated from ATG1 and the lower bandshown in pIVY156 and pIVY159 is initiated from ATG9. Con-sistently, further mutation of ATG9 in pIVY156, resulting inpIVY161, abolished both protein bands (lane 4). However,when only ATG9 was mutated (pIVY160, lane 3), two distinctprotein bands of similar amounts could be seen, suggesting thatthe ATG1-initiated protein was partially processed to a mole-cule with size similar to the ATG9-initiated protein. Therefore,the lower band shown in pIVY155 (lane 1) is likely composed ofproteins initiated from ATG1 (processed) and ATG9. Itappears that mutation of ATG9 to GCG (Met-9 to Ala-9) inpIVY160 reduces the processing rate of the protein by matrixprocessing peptidase (compare lanes 1 and 3 and Fig. 5D, lanes1 and 3). Thus, ATG9 functions as a remedial translation initi-ator even in the presence of ATG1 (pIVY159) and is about 4times as efficient as ATG1 (compare lanes 3 and 5), similar toresults obtained in Fig. 5. Similar results were observed when

these fusions were expressed from the native CaALA1 pro-moter (data not shown).Mapping the Mitochondrial Targeting Signal of CaAlaRS—

As shown in Fig. 1D, the protein form initiated from ATG1appears to have an amino-terminal 15-residue appendage thatis absent from the other yeast cytoplasmic AlaRSs compared;therefore, we wondered whether this leader peptide actuallyparticipates in protein localization into mitochondria. To mapthemitochondrial targeting sequence ofCaAlaRS, the cytoplas-mic form of ValRS (designated here as ValRSc) was chosen asthe reporter protein, which is by itself confined exclusively inthe cytoplasm even when overexpressed (15). Three CaALA1fragments with distinct 3�-ends were independently amplifiedby PCR and fused in-frame to the 5�-end of VAS1c (coding forValRSc), resulting in various CaALA1-VAS1c constructs (Fig.8A). These constructs were transformed into CW1, a vas1�

strain, and tested for their ability to rescue the growth defects ofthe knock-out strain on FOA and YPG plates, respectively.Contrary to our anticipation, fusion of the peptide containingresidues 1–34 of CaAlaRS to ValRSc (resulting in pIVY117) didnot confer a mitochondrial phenotype to the fusion, i.e. trans-formants harboring pIVY117 could not grow on the YPG plate(Fig. 8, B–C). The passenger protein could be successfully tar-geted to mitochondria only when the CaAlaRS portion wasextended to residue 42 or 53 (see pIVY105 and pSAM37,

FIGURE 7. Demonstration of two distinct protein isoforms initiatedfrom ATG1 and ATG9. A, summary of the CaALA1-lexA* constructs. Awild-type or mutant CaALA1 sequence (bp �40��30) was fused in-frameto the 5�-end of lexA* (where its ATG initiator has been mutated), resultingin various CaALA1-lexA* fusions. The ATG initiators and their mutants(shaded) were labeled on top of the sequences. The open and striped boxesrepresent CaALA1 and lexA sequences, respectively. B, Western blot. Theconstructs were transformed into INVSc1, and their protein expressionwas analyzed by Western blot using anti-LexA antibody. In B, the numbers1–5 correspond to constructs shown in A.

FIGURE 8. Mapping the mitochondrial targeting sequence of CaAlaRS.The vas1� strain, CW1, was transformed with various CaALA1-VAS1c con-structs and then tested for its growth phenotypes. A, summary of the CaALA1-VAS1c constructs and their complementation activities. Various CaALA1 frag-ments (�370��102 bp, �370��126 bp, or �370��159 bp) were fusedin-frame to the 5�-end of VAS1c, resulting in pIVY117, pIVY105, and pSAM37,respectively. The ATG initiators were labeled on top of the sequences. Theopen and solid boxes represent CaALA1 and VAS1c sequences, respectively.B, complementation assays for the cytoplasmic function on a 5-FOA plate.C, complementation assays for the mitochondrial function on a YPG plate.Cyt, cytoplasmic; Mit, mitochondrial. D, RT-PCR. The relative amounts ofspecific CaALA1-VAS1c mRNA generated from the constructs were deter-mined by RT-PCR. As an internal control, the relative amounts of actin-specific mRNA in each sample were also determined. In B–D, the numbers1–3 denote constructs shown in A.

A Bifunctional tRNA Synthetase

OCTOBER 20, 2006 • VOLUME 281 • NUMBER 42 JOURNAL OF BIOLOGICAL CHEMISTRY 31437

by guest on February 4, 2020http://w

ww

.jbc.org/D

ownloaded from

respectively). This result suggested that the mitochondrial tar-geting sequence of CaAlaRS extends from residue 1 to 42. Tocheck if these fusions express similar levels of mRNAs, the rel-ative levels of specific CaALA1-VAS1cmRNAs generated fromeach construct were determined by a semiquantitative RT-PCRexperiment using a set of primers, with the forward primerannealed to base pairs �1��25 of CaALA1 and the reverseprimer to base pairs�400��425 ofVAS1. Fig. 8D showed thatsimilar levels of the cDNA products were generated from thesefusions, suggesting that lack of the mitochondrial function inpIVY117 was not caused by different levels of RNA expression(or degradation) but, rather, by the absence of an efficientmito-chondrial transit signal.Given that the long form of CaAlaRS can be dual-targeted,

wewonderedwhether this unique feature could be attributed toa poor mitochondrial targeting signal such that only a portionof the preprotein is targeted to the mitochondria and whetherthis feature could be passed on to the CaAlaRS-ValRSc fusionproteins. As shown in Fig. 9, mutation of ATG1 to TCT (result-ing in pIVY106) selectively impaired the mitochondrial func-tion of the CaALA1-VAS1c fusion (compare pIVY105 andpIVY106), whereas mutation of ATG9 to GCG (resulting inpIVY107) specifically eliminated the cytoplasmic function ofthe fusion (compare pIVY105 and pIVY107). Additionally,double mutations that inactivated both of the initiators led to aconstruct (see pIVY108) that was defective in both functions.This observation suggested that the ATG1- and ATG9-initi-ated fusions were targeted exclusively to mitochondria andcytoplasm, respectively. Thus, the ATG1-initiated signal pep-tide (containing residues 1–42) is in effect a strong mitochon-drial targeting signal under the conditions used.

DISCUSSION

In the present work we discovered that CaALA1, the onlyalanyl-tRNA synthetase gene in C. albicans, could overcomethe species barrier and substitute for both the cytoplasmic andmitochondrial functions ofALA1 in S. cerevisiae (Figs. 2 and 3).As with the case in S. cerevisiae (10), these functions are pro-vided by two distinct protein isoforms that are synthesizedthrough alternative initiation from two nearby in-frame startcodons (Fig. 3). However, several characteristic features regard-ing the mechanism of translation initiation and partition of theprotein isoforms appear to be idiosyncratic to the gene and areworthy of further attention. First, although two distinct proteinforms are generated from this gene, the short form appears tobe redundant anddispensable for the cytoplasmic function (Fig.3). Second, different fromALA1 of S. cerevisiae, no non-canon-ical initiators are involved in the translation of a minor, mito-chondrial form in this case; instead, both of the isoforms ofCaAlaRS are initiated from AUG triplets (Fig. 3). Last but notleast, whereas the bifunctional phenotype of ScALA1 is contrib-uted by two functionally exclusive protein isoforms; the longform of CaAlaRS per se is a dual-targeted (and, thus, bifunc-tional) protein (Fig. 3). To our knowledge, this appears to be thefirst example in yeast, wherein a naturally occurring form of atRNA synthetase can play roles in both compartments.Despite the fact that the mitochondrial targeting signal of

CaAlaRS extends from amino acid residue 1 to 42 (Fig. 8), thecleavage site of matrix-processing peptidase was mappedbetween residue positions 8 and 9 (Fig. 1). As a result, the “pro-cessed” mitochondrial form has an amino terminus identical tothat of the short form. Analysis of the CaAlaRS isoforms withthe PSORTII program (17) showed a 65% likelihood of mito-chondrial import for the long formbut only a 26% likelihood forthe short form. Furthermore, as with many classical mitochon-drial targeting signals (18), this 42-residue peptide is rich inpositively charged (17%) and hydroxylated residues (26%) but isalmost devoid of acidic residues (2%). Strangely enough,although the long form could be distributed and, thus, wasfunctional in both compartments, fusion of the full-length sig-nal peptide to a cytoplasmic passenger did not confer a bifunc-tional phenotype to the fusion (Fig. 9). These results sug-gested that the mechanism of protein localization forCaAlaRS is different from what we have observed for a trun-cated version of the mitochondrial form of ValRS, where amitochondrial preprotein can be made bifunctional simplyby weakening its mitochondrial targeting signal (15). Moreexperiments are currently under way to elucidate the mech-anism that contributes to the dual-targeting nature of thelong form. In this aspect the yeast FUM1 gene (coding forfumarase) represents an interesting example. A single spe-cies of primary translation product is generated from FUM1and is responsible for both the cytoplasmic and mitochon-drial fumarase activities in vivo (19). As it turns out all FUM1gene products are first targeted to the mitochondrial matrixand then a significant fraction of the processed proteinsarrives back in the cytoplasm (20). Thus, the mature forms ofthe cytoplasmic and mitochondrial fumarases have the sameamino termini (21).

FIGURE 9. Testing the efficiency of the mitochondrial targeting signal ofCaAlaRS. The vas1� strain, CW1, was transformed with various CaALA1-VAS1cconstructs and then tested for its growth phenotypes. A, summary of theCaALA1-VAS1c constructs and their complementation activities. A wild-typeor mutant CaALA1 sequence (�370��126 bp) was fused in-frame to the5�-end of VAS1c, resulting in various CaALA1-VAS1c constructs. The ATG initi-ators and their mutants (shaded) were labeled on top of the sequences. Theopen and solid boxes represent CaALA1 and VAS1c sequences, respectively. B,complementation assays for the cytoplasmic function on a 5-FOA plate. C,complementation assays for the mitochondrial function on a YPG plate. In Band C, the numbers 1– 4 denote constructs shown in A. Cyt, cytoplasmic; Mit,mitochondrial.

A Bifunctional tRNA Synthetase

31438 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 281 • NUMBER 42 • OCTOBER 20, 2006

by guest on February 4, 2020http://w

ww

.jbc.org/D

ownloaded from

Inmammalian cells the small ribosomal subunit often skips aweak translation initiator, such as a non-AUG codon or anAUG codonwithin a suboptimal sequence context, and contin-ues scanning downstreamon themessage until it encounters anAUG triplet within a more favorable sequence context. Thisprocess is referred to as “leaky scanning” (22–23). Although thismechanismhas been observed frequently inmammals (24–27),there are only a few known examples in yeast. Examples includeMOD5 (coding for isopentenyl pyrophosphate:tRNA isopente-nyl transferase) (28) and CCA1 (coding for ATP (CTP):tRNAnucleotidyltransferase) (29). In these two instances leaky scan-ning occurs probably because the first AUG codon is locatedtoo close to the 5�-end of themRNA,making it less accessible tothe initiating ribosome. In the case of CaALA1, because onlyone transcript with its 5�-end located at nucleotide position�24 relative to ATG1 is available (Fig. 1), it is, therefore, likelythat recognition of the second AUG triplet as well as produc-tion of the short isoform is also mediated by leaky scanning.Evidence supporting this argument came from the observationthat expression of the AUG9-initiated short form drasticallyincreased when AUG1 was mutated (Fig. 5 and 7).The specificity of an aminoacylation reaction is accomplished

by direct recognition of the cognate tRNA by the specific synthe-tase. In some instances recognition depends mainly on the antic-odon, whereas in others it depends more on the sequences/struc-tures in the acceptor stem (right next to the amino acid couplingsite) (30). For example, Drosophila melanogaster cytoplasmictRNAAla has a G3-U70 base pair in the acceptor stem as its majoridentity determinant, whereas the GU pair has been translocatedto position 2:71 in its mitochondrial isoacceptor. Consequently,D. melanogaster mitochondrial AlaRS can only aminoacylate itsmitochondrial tRNAsbutnot its cytoplasmic equivalents (31).Weshould mention that the G3-U70 base pair is conserved in allknown tRNAAla sequences from prokaryotes, Archaea, eukaryotecytoplasm,andchloroplasts. In this sense loweukaryotes suchasS.cerevisiae andC. albicans appear to represent a divergent point inthe course of coevolution of tRNAAla and its cognate synthetase,where a single nuclear AlaRS gene is capable of aminoacylatingalanyl-tRNAs in both compartments of the same cell in these or-ganisms (10). Furthermore, theC. albicans enzymes are capable ofcross-species complementation, adding further emphasis to theimportance of the G3-U70 element in recognition specificity.

Acknowledgment—We are grateful to Grace Lin of National CentralUniversity for critical reading of the manuscript.

REFERENCES1. Carter, C. W., Jr. (1993) Annu. Rev. Biochem. 62, 715–7482. Martinis, S. A., and Schimmel, P. (1996) inEscherichia coli and Salmonella

Cellular and Molecular Biology (Neidhardt, F. C., ed) 2nd Ed., pp.887–901, American Society for Microbiology, Washington, D. C.,

3. Giege, R., Sissler, M., and Florentz, C. (1998) Nucleic Acids Res. 26,5017–5035

4. Pelchat, M., and Lapointe, J. (1999) Biochem. Cell Biol. 77, 343–3475. Marechal-Drouard, L., Weil, J. H., and Dietrich, A. (1993) Annu. Rev. Cell

Biol. 8, 115–1316. Steinmetz, A., and Weil, J. H. (1986)Methods Enzymol. 118, 212–2317. Natsoulis, G., Hilger, F., and Fink, G. R. (1986) Cell 46, 235–2438. Chatton, B., Walter, P., Ebel, J.-P., Lacroute, F., and Fasiolo, F. (1988)

J. Biol. Chem. 263, 52–579. Souciet, G., Menand, B., Ovesna, J., Cosset, A., Dietrich, A., andWintz, H.

(1999) Eur. J. Biochem. 266, 848–85410. Tang,H. L., Yeh, L. S., Chen,N.K., Ripmaster, T., Schimmel, P., andWang,

C. C. (2004) J. Biol. Chem. 279, 49656–4966311. Chang, K. J., Lin, G.,Men, L. C., andWang, C. C. (2006) J. Biol. Chem. 281,

7775–778312. Chang, K. J., and Wang, C. C. (2004) J. Biol. Chem. 279, 13778–1378513. Wang, W., and Malcolm, B. A. (1999) Biotechniques 26, 680–68214. Daum, G., Bohni, P. C., and Schatz, G. (1982) J. Biol. Chem. 257,

13028–1303315. Ripmaster, T. L., Shiba, K., and Schimmel, P. (1995) Proc. Natl. Acad. Sci.

U. S. A. 92, 4932–493616. Wang, C. C., Chang, K. J., Tang, H. L., Hsieh, C. J., and Schimmel, P. (2003)

Biochemistry 42, 1646–165117. Nakai, K., and Horton, P. (1999) Trends Biochem. Sci. 24, 34–3618. Neupert, W. (1997) Annu. Rev. Biochem. 66, 863–91719. Wu, M., and Tzagoloff, A. (1987) J. Biol. Chem. 262, 12275–1228220. Stein, I., Peleg, Y., Even-Ram, S., and Pines O. (1994) Mol. Cell. Biol. 14,

4770–477821. Sass, E., Blachinsky, E., Karniely, S., and Pines, O. (2001) J. Biol. Chem. 276,

46111–4611722. Kozak, M. (1991) J. Biol. Chem. 266, 19867–1987023. Kozak, M. (1999) Gene (Amst.) 234, 187–20824. Acland, P., Dixon, M., Peters, G., and Dickson, C. (1990) Nature 343,

662–66525. Saris, C. J., Domen, J., and Berns, A. (1991) EMBO J. 10, 655–66426. Hann, S. R., Sloan-Brown, K., and Spotts, G. D. (1992) Genes Dev. 6,

1229–124027. Packham, G., Brimmell, M., and Cleveland, J. L. (1997) Biochem. J. 328,

807–81328. Slusher, L. B., Gillman, E. C., Martin, N. C., andHopper, A. K. (1991) Proc.

Natl. Acad. Sci. U. S. A. 88, 9789–979329. Wolfe, C. L., Lou, Y. C., Hopper, A. K., and Martin, N. C. (1994) J. Biol.

Chem. 269, 13361–1336630. Schimmel, P., Giege, R., Moras, D., and Yokoyama, S. (1993) Proc. Natl.

Acad. Sci. U. S. A. 90, 8763–876831. Lovato, M. A., Chihade, J. W., and Schimmel, P. (2001) EMBO J. 20,

4846–4853

A Bifunctional tRNA Synthetase

OCTOBER 20, 2006 • VOLUME 281 • NUMBER 42 JOURNAL OF BIOLOGICAL CHEMISTRY 31439

by guest on February 4, 2020http://w

ww

.jbc.org/D

ownloaded from

WangHsiao-Yun Huang, Yu Kuei, Hen-Yi Chao, Shun-Jia Chen, Lu-Shu Yeh and Chien-Chia

a Class II tRNA SynthetaseCross-species and Cross-compartmental Aminoacylation of Isoaccepting tRNAs by

doi: 10.1074/jbc.M601869200 originally published online August 23, 20062006, 281:31430-31439.J. Biol. Chem. 

  10.1074/jbc.M601869200Access the most updated version of this article at doi:

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

  http://www.jbc.org/content/281/42/31430.full.html#ref-list-1

This article cites 30 references, 16 of which can be accessed free at

by guest on February 4, 2020http://w

ww

.jbc.org/D

ownloaded from