The Human Lysyl-tRNA Synthetase Gene Encodes Both the Cytoplasmic and Mitochondrial Enzymes by Means of an Unusual Alternative Splicing of the Primary Transcript*

Two cDNAs encoding human lysyl-tRNA synthetase have been identified. One encodes the cytoplasmic form of the enzyme identified previously. The second cDNA contains the same sequence but with a 180-bp insertion at the 5′-end of the mRNA. This results in a predicted protein whose carboxyl 576 amino acids are identical to those of the cytoplasmic enzyme but with a different amino terminus of 49 amino acids that contains a putative mitochondrial targeting sequence. Expression of the two lysyl-tRNA synthetase-green fluorescent protein gene fusions in a human cell line confirmed that the cytoplasmic form was targeted to the cytoplasm and the mitochondrial form to mitochondria. The genomic lysyl-tRNA synthetase gene consisted of 15 exons. The two isoforms were created by alternative splicing of the first three exons of the gene. The cytoplasmic form was created by splicing exon 1 to exon 3. The inclusion of exon 2 between exons 1 and 3 produced an mRNA encoding the mitochondrial isoform with an additional upstream small open reading frame, consisting mainly of a portion of the 5′ coding region of the cytoplasmic isoform. This is the first example of mitochondrial targeting sequence being encoded on the second exon of a gene. Ribonuclease protection analysis showed that the mRNA encoding the cytoplasmic isoform makes up approximately 70%, and the mitochondrial isoform approximately 30%, of the mature transcripts from the lysyl-tRNA synthetase gene. The mitochondrial form of the enzyme, purified after expression in Escherichia coli, aminoacylated in vitro transcripts corresponding to both the cytoplasmic and mitochondrial tRNALys, despite the difference in the discriminator base sequence in the acceptor stems of these tRNAs.

The tRNA synthetases are a group of enzymes responsible for aminoacylating tRNAs with appropriate amino acids. In recent years, a considerable amount of information has been obtained about their functions, but there are still a number of uncharacterized members of this diverse group of enzymes.
Genes encoding lysyl-tRNA synthetases have been cloned from a number of organisms, and crystal structures have been obtained for one of the two Escherichia coli lysyl-tRNA synthetases, LysU (1), and the Thermus thermophilus lysyl-tRNA synthetase (2). The majority of the lysyl-tRNA synthetases are class II enzymes; however, some archaebacteria have a class I lysyl-tRNA synthetase (3). The cDNA encoding the human cytoplasmic lysyl-tRNA synthetase has been described previously, and the protein purified after expression in E. coli has been characterized (4). In mammals, the cytoplasmic form is a component of a synthetase complex that contains an additional seven tRNA synthetase activities (5,6). The amino-terminal region of the protein has been implicated in the interactions with the multienzyme complex (5,6). The gene encoding human lysyl-tRNA synthetase has been localized to chromosome 16q23-q24 (7).
Since protein translation occurs in both the cytosol and mitochondria of eukaryotic cells, the cell generally requires tRNA synthetase activities for each amino acid in both subcellular locations. In some cases, two distinct genes encode the cytoplasmic and mitochondrial isoforms, e.g. human histidyl-tRNA synthetase (8). Alternatively, one gene may encode both forms of the protein. In the case of glycyl-tRNA synthetase, this is achieved by using alternative transcription initiation sites, with the furthest upstream transcript start site resulting in the addition of a mitochondrial targeting sequence to the amino terminus of the protein (9,10). A number of the human mitochondrial tRNA synthetases have yet to be identified and characterized. In yeast, the cytoplasmic and mitochondrial isoforms of the lysyl-tRNA synthetase are encoded by two distinct genes (11,12). To identify and clone the gene encoding the mitochondrial form of human lysyl-tRNA synthetase, we searched the NCBI human EST 1 data base to identify potential ESTs that were homologous, but not identical, to the human lysyl-tRNA synthetase. We identified a human EST cDNA sequence that appeared to be an altered form, or alternative splicing product, of the cytoplasmic form of the gene (AA356156). We investigated the possibility that the alternative form of the gene encoded the mitochondrial lysyl-tRNA synthetase.

EXPERIMENTAL PROCEDURES
Overexpression and Purification of Lysyl-tRNA Synthetase-Plasmid pM368 (4), containing the full-length cytoplasmic lysyl-tRNA synthetase cloned into a derivative of E. coli expression vector pET19b, was kindly given to us by Dr. Kiyotaka Shiba (Japanese Foundation for Cancer Research, Tokyo, Japan). The cDNA encoding the mitochondrial lysyl-tRNA synthetase was cloned into E. coli expression vector pET24d (Novagen). The second codon in the cDNA, encoding leucine, was changed to one encoding alanine to improve protein stability following expression in E. coli, in accordance with the N-end rule (13). These constructs were expressed in E. coli BL21(DE3) Codon Plus (RIL) * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  (Stratagene) at 15°C overnight in the presence of 1 mM isopropyl-1thio-␤-D-galactopyranoside to produce lysyl-tRNA synthetase fusion proteins, with amino-or carboxyl-terminal hexahistidine regions, which were subsequently purified using Talon metal affinity resin (CLON-TECH). The cytoplasmic form constituted approximately 90% of the total purified protein; the mitochondrial form constituted approximately 80% of the total purified protein (data not shown), as determined after gel electrophoresis.
Isolation of cDNAs Encoding the Mitochondrial Lysyl-tRNA Synthetase and the 5Ј-Region of the Full-length Mitochondrial Transcript-The 5Ј-portion of the lysyl-tRNA synthetase gene encoding the mitochondrial isoform was obtained by 5Ј-rapid amplification of cDNA ends (14) using total RNA isolated from human osteosarcoma cell line 143B. cDNA, synthesized by an avian myeloblastosis virus reverse transcription system (Promega), was polyadenylated at the 5Ј-end by terminal transferase (Roche Molecular Biochemicals). The 5Ј region of the mitochondrial transcript was cloned by nested PCR using nested adapter primers (14) and nested primers corresponding to a sequence predicted to encode the mitochondrial targeting peptide. PCR products were cloned into the pGEMT-easy vector (Promega) and sequenced by the Kimmel Cancer Center DNA Sequencing Facility (Thomas Jefferson University).
Intracellular Location of Lysyl-tRNA Synthetase Isoforms-To investigate the localization of the proteins expressed from the two forms of the lysyl-tRNA synthetase mRNA, the full-length mRNAs were cloned into the mammalian expression vector pEF/myc/cyto/GFP (Invitrogen). DNA fragments encoding the full-length proteins were cloned into NcoI and PstI sites of the vector so that fusion proteins would be obtained with GFP added to the carboxyl termini. The human osteosarcoma cell line 143B was transfected with these constructs using Effectene (Qiagen). Cells on coverslips were fixed 24 h after transfection. In some cases, the cells were immunostained with a monoclonal antibody directed against the human cytochrome oxidase subunit I (mouse monoclonal 1D6-E1-A8; Molecular Probes, Inc.) and appropriate fluorescent secondary antibodies (Chemicon).
In Vitro Aminoacylation Studies-Cytoplasmic tRNA Lys was synthesized by in vitro run off transcription (15), using pLysF119 (4) as the template. Since the T7 RNA polymerase initiates with G and since the mitochondrial tRNA Lys begins with C, the mitochondrial tRNA Lys gene was cloned so that tRNA Lys was preceded by a hammerhead ribozyme (16). This plasmid was used as a template to make a run-off transcript that was then incubated to release a tRNA transcript starting with the correct nucleotide. The resulting tRNAs were purified by electrophoresis through polyacrylamide-urea gels. Aminoacylation was performed at 37°C as described previously (4) using 1 M tRNA, 0.1 Ci/l [ 3 H]lysine (89 Ci/mmol; PerkinElmer Life Sciences), and 0.9 -88 nM mitochondrial lysyl-tRNA synthetase or 5-500 nM cytoplasmic lysyl-tRNA synthetase in a total reaction volume of 20 l. At 3-min intervals, 4-l aliquots were spotted on 3MM filter paper and soaked in cold 5% trichloroacetic acid to terminate the reaction. The amount of chargeable tRNA in the purified transcript was calculated following a time course study to determine the plateau level for aminoacylation. These were carried out using 1.8 M mitochondrial lysyl-tRNA synthetase with 0.2 M cytoplasmic tRNA Lys transcript or 0.7 M mitochondrial tRNA Lys transcript and using cytoplasmic lysyl-tRNA synthetase with 0.4 M cytoplasmic tRNA Lys transcript or 0.8 M mitochondrial tRNA Lys transcript.
Quantitation of the Relative Amounts of mRNA Transcripts Encoding the Cytoplasmic or Mitochondrial Forms of Lysyl-tRNA Synthetase-The relative amounts of each transcript in total cellular RNA extracts from 143B cells were quantitated using the Ribonuclease Protection Assay (RPA) system RPA II (Ambion) according to the manufacturer's suggested protocol. A gene fragment corresponding to exon 2 (121 bp, encoding the mitochondrial targeting sequence) and part of exon 3 (78 bp, encoding a region common to both cytoplasmic and mitochondrial proteins) was subcloned into pGEM-Teasy. The resulting plasmid was used as a template for in vitro transcription in the presence of [␣-32 P]UTP to give a labeled fragment of 363 nt. Different amounts of total RNA were tested in incubations with a large excess of the 363-nt transcript. After nuclease digestion, the protected fragments were separated by electrophoresis through a 6% polyacrylamide gel. Signals corresponding to the mitochondrial and cytoplasmic transcripts were quantitated by a PhosphorImager (Molecular Dynamics, Inc., Sunnyvale, CA).
Isolation of the Genomic Region Containing the Lysyl-tRNA Synthetase Gene-A human blood genomic library (Novagen) was screened with probes corresponding to different regions of the lysyl-tRNA synthetase gene. To isolate the 5Ј region of the lysyl-tRNA synthetase locus, we identified BAC clones that had been described as containing this gene and the GEF-2 gene that, like lysyl-tRNA synthetase, has been mapped to chromosome 16q23-q24. We hypothesized that such clones were likely to contain the lysyl-tRNA synthetase gene and not lysyl-tRNA synthetase pseudogenes. Using this criterion and the NCBI Unigene identifier Hs.3100 for lysyl-tRNA synthetase and Hs.6518 for GEF-2, with the Caltech human Unigene information navigator, we identified BAC clones that potentially contained the lysyl-tRNA synthetase gene. Comparison of Lysyl-tRNA Synthetase Genes among Eukaryotic Species-We used the following GenBank TM EST and genomic sequences to compare the genomic or mRNA 5Ј sequences encoding lysyl-tRNA synthetase: Caenorhabditis elegans mitochondrial isoform mRNA CO9136.1, cytoplasmic isoform mRNA AV188647.1, genomic sequence U41105.1; Drosophila melanogaster mitochondrial isoform mRNA AI516192.1, cytoplasmic isoform mRNA AE003447.1, genomic sequence AE003447.1; mouse mitochondrial isoform mRNA AW258396, cytoplasmic isoform mRNA W53766; zebrafish (Danio rerio) mitochondrial isoform mRNA AW421393, and cytoplasmic isoform mRNA, AW421565.1.

Identification of Two Forms of mRNA Encoding Lysyl-tRNA
Synthetase-BLAST searches of the NCBI/GenBank TM human EST data base were used to identify cDNAs, homologous to cytoplasmic lysyl-tRNA synthetase, that might potentially encode the mitochondrial lysyl-tRNA synthetase. We identified a human EST (GenBank TM accession number AA356156), containing the 5Ј region of the cytoplasmic lysyl-tRNA synthetase gene, that appeared to be an altered form, or represent an alternative splicing, of the cytoplasmic form of the gene. BLAST searches of the human EST and HTGS data bases with both bacterial and eukaryotic lysyl-tRNA sequences failed to identify other sequences related to lysyl-tRNA synthetase. This suggested that a single gene gave rise to two forms of mRNA that encode a cytoplasmic and a mitochondrial lysyl-tRNA synthetase.
Isolation of a cDNA Encoding the Mitochondrial Isoform of Human Mitochondrial Lysyl-tRNA Synthetase-The cDNA sequence identified by searches of the EST data base was similar to that of the cytoplasmic form of mitochondrial lysyl-tRNA synthetase, although it contained an extra 180 bp inserted into the protein-coding region. This sequence altered the reading frame so that the amino terminus of the cytoplasmic mitochondrial lysyl-tRNA synthetase was replaced with an amino terminus predicted by the program Mitoprot (17) to encode a mitochondrial targeting sequence.
On the basis of the sequence of the cDNA identified by a search of the human EST data base, we designed primers to isolate and characterize the full-length coding region of the putative mitochondrial lysyl-tRNA synthetase and also to examine the noncoding 5Ј region of the mRNA. We amplified and isolated a cDNA that encoded a full-length (625-aa) putative mitochondrial isoform of mitochondrial lysyl-tRNA synthetase, using total RNA isolated from human osteosarcoma cell line 143B. With the exception of the 5Ј region of the mRNA, the sequence was identical to that of the cytoplasmic lysyl-tRNA synthetase isoform, identified in GenBank TM as gene KIAA0070 (GenBank TM accession number D31890).
To investigate the 5Ј sequence of the putative mitochondrial lysyl-tRNA synthetase transcript, we performed 5Ј-rapid amplification of cDNA ends using total RNA isolated from cell line 143B. The sequence obtained confirmed that the cDNA encoding the mitochondrial isoform contained an upstream region corresponding to the 5Ј sequence of the cytoplasmic form of the lysyl-tRNA synthetase (Fig. 1, A and B).
Using primers homologous to the 5Ј-and 3Ј-ends of the cDNA encoding cytoplasmic lysyl-tRNA synthetase, two full-length cDNAs encoding both the cytoplasmic and mitochondrial isoforms of mitochondrial lysyl-tRNA synthetase were amplified by reverse transcriptase-PCR. The sequences were identical except for a 180-bp insertion close to the 5Ј-end to produce the putative mitochondrial isoform (see Fig. 1A). This insertion introduces a sequence encoding Lys-Trp-Trp-STOP into the cytoplasmic isoform reading frame after the codon for Lys 20 . The next ATG is 24 bp downstream from the stop codon and presumably is the initiation codon of the mitochondrial isoform. The reading frame of the mitochondrial isoform encodes 48 amino acids before resuming the coding sequence of the cytoplasmic isoform at amino acid Glu 22 . The 48 amino acids have the characteristics of a mitochondrial targeting sequence. The program Mitoprot (17) predicts a mitochondrial location for this protein and also identifies a potential mitochondrial signal cleavage site between Lys 16 and Thr 17 . The predicted translation products of the cDNAs differ at their amino termini but share a common carboxyl-terminal region of 576 amino acids. The full-length cDNA encoding the mitochondrial lysyl-tRNA synthetase has been deposited in GenBank TM (accession number AF285758).
Genomic Structure of the Human Lysyl-tRNA Synthetase Gene-A portion of the genomic lysyl-tRNA synthetase gene was obtained from a human genomic DNA library using probes corresponding to the full-length cytoplasmic cDNA sequence (4). A clone (p47/4) was identified. Southern and DNA sequence analysis showed that this contained the 3Ј region of the gene (corresponding to exons 4 -15 of the final sequence) but lacked the 5Ј region of the gene. Repeated screening attempts failed to find the 5Ј region of the gene in this library.
To obtain the 5Ј portion of the gene, we identified BAC clones containing the gene encoding lysyl-tRNA synthetase, as de-scribed under "Experimental Procedures." BAC clones were screened by Southern analysis, using a probe corresponding to the region of the cDNA encoding the putative mitochondrial targeting sequence. Eight clones were identified that contained the 5Ј portion of the lysyl-tRNA synthetase gene on a 13-kb BamHI fragment. One, BAC clone 2256 J17, was selected for more detailed analyses. A combination of Southern and DNA sequence analysis confirmed that the 13-kb BamHI fragment contained the first four exons of the lysyl-tRNA synthetase gene. A partial genomic sequence of the lysyl-tRNA synthetase locus was determined by direct sequencing of the BAC clone 2256 J17 and the genomic library clone. The introns 1, 2, 3, 4, 7, and 8 were not fully sequenced, but their sizes were estimated by restriction mapping, Southern analysis, and PCR amplification of the BAC clones and genomic DNA.
The complete gene contained 15 exons and extended over approximately 20 kb of the genome (Fig. 2). The intron-exon borders are shown in Table I. A comparison of the genomic sequence with those of the two forms of cDNA obtained showed that the two forms of lysyl-tRNA synthetase were obtained by alternative splicing of the first three exons of the gene (Fig. 3). The first exon encodes the amino-terminal 20 amino acids of the cytoplasmic form of the enzyme. The second exon encodes an open reading frame with an initiating methionine and 48 amino acids including a region with the characteristics of a mitochondrial targeting sequence. The cytoplasmic form of the enzyme is encoded by an mRNA containing the first exon spliced to the third exon, creating the cDNA for the cytoplasmic form of the enzyme described previously (4). The mitochondrial form is encoded by an mRNA whose 5Ј region consists of the first three exons spliced consecutively. The reading frame created by the first exon is terminated by a stop codon in the second exon. The initiating methionine of the mitochondrial isoform is 24 bp downstream from this stop codon (Fig. 1).
Upstream from the initiation ATG of the human cytoplasmic lysyl-tRNA synthetase gene, on the opposite strand, is a reading frame of 1200 bp starting at position Ϫ243. Investigation of the corresponding cDNA in the NCBI EST data base allowed us to construct a putative full-length cDNA. We used this information to design primers that were used to clone the mRNA by PCR from cDNA.
The protein encoded by the upstream reading frame, which we designated KRU (lysyl-tRNA synthetase reading frame upstream) in Fig. 2 had no strong homologies to known proteins. A ProfileScan analysis for protein motifs identified a BRCT domain (residues 78 -101) and a Myb domain (residues 128 -188). This protein has since been identified as a human ortholog of Rap1 (Ref. 18; GenBank TM accession number AF262988), a protein, localized at telomeres, that affects telomere length.
The region of chromosome 16 containing the lysyl-tRNA synthetase and Rap1 is represented by "working draft" sequences in GenBank TM (accession numbers AC025287 and AC011934).
There is a region of 243 bp between the two initiation codons for lysyl-tRNA synthetase and KRU/Rap1. This region lacks a conventional TATA sequence, as is characteristic of housekeeping genes, but presumably contains a bidirectional promoter and is predicted to contain several Sp1 binding domains in both orientations. A ribonuclease protection assay analysis of mRNA isolated from human 143B cells showed that the major initiation site for the lysyl-tRNA synthetase transcript occurs at the position corresponding to Ϫ31 (data not shown).
Analysis of Other Eukaryotic Lysyl-tRNA Synthetases-Analysis of the GenBank TM EST and genomic sequence data bases (see "Experimental Procedures" for the appropriate ac-cession numbers) showed that it is likely that one gene provides both cytoplasmic and mitochondrial isoforms of lysyl-tRNA synthetase in mouse, C. elegans, D. melanogaster, and zebrafish (D. rerio) (Fig. 1B). In each case, the cytoplasmic form would be created by the exclusion of the second exon, containing a mitochondrial targeting sequence during the splicing of the primary transcript.
Both the C. elegans and D. melanogaster genomic lysyl-tRNA synthetase sequences are predicted to contain four exons. In C. elegans, two of the three positions of the introns in the coding sequence are absolutely conserved with an equivalent human intron; the third is within 3 aa (10 bp) of the equivalent site in human DNA. Exon 1 contains the coding regions of the amino termini of both the cytoplasmic and mitochondrial forms, since the 5Ј cytoplasmic/mitochondrial boundary region of the mRNA is not interrupted by an intron in the genome. To form the mRNA encoding the cytoplasmic form, splicing proceeds from a point midway in exon 1 to exon 2. In Drosophila, the locations of the introns are conserved with the human gene only for introns 1 and 2. The sequences in the EST data base suggest that, in Drosophila, the incorporation of the mitochondrial amino terminus might be accomplished by initiation of transcription at exon 2.
Subcellular Localization of the Isoforms of Human Lysyl-tRNA Synthetase-To confirm the identities of the mRNA species predicted to encode the two isoforms of lysyl-tRNA synthetase, they were cloned into mammalian expression vector pEFmyc/cyto/GFP so that a gene encoding GFP was added, in frame, to the 3Ј-ends of both mRNAs. These constructs were then transfected into the human osteosarcoma cell line 143B for transient expression studies. Cells were fixed 24 h after transfection and examined for GFP fluorescence or immunostained. As expected, the cytoplasmic lysyl-tRNA synthetase-GFP fusion construct produced a diffuse, cellwide fluorescence pattern (Fig. 4A). The putative mitochondrial lysyl-tRNA synthetase-GFP fusion construct resulted in a punctate pattern of fluorescence characteristic of a mitochondrial distribution (Fig.  4C). To confirm that this localization was mitochondrial, cells expressing the lysyl-tRNA synthetase-GFP fusion protein were immunostained with an antibody directed against the inner mitochondrial membrane protein COX I and a rhodamine-labeled secondary antibody (Fig. 4B). The GFP and rhodamine staining co-localized. Thus, the two lysyl-tRNA synthetase mRNAs encoded enzymes that were destined for two distinct subcellular locations, the mitochondria and the cytosol.

TABLE I Exon-intron organization of the human lysyl-tRNA synthetase gene
The size of those introns not fully sequenced were estimated by PCR and Southern analyses (estimate). Where possible these were compared to available unfinished sequence data in the NCBI HTGS database (HTGS). The consensus splice sites are shown above the 5Ј and 3Ј sequences. Quantitation of the Relative Amounts of the Cytoplasmic and Mitochondrial Lysyl-tRNA Synthetase Transcripts-The relative amounts of the two lysyl-tRNA synthetase transcripts were quantitated by a ribonuclease protection assay using radiolabeled sense and antisense RNA probes designed using a cDNA sequence corresponding to a portion of the mitochondrial lysyl-tRNA synthetase isoform (Fig. 5A). The mitochondrial form of lysyl-tRNA synthetase was predicted to protect a fragment of 199 nt, while the cytoplasmic form was predicted to protect a fragment of 78 nt. The 363-nt antisense probe (Fig.   5B, lane 1) was hybridized with total RNA isolated from human osteosarcoma cell line 143B, followed by digestion with RNases A and T1. (Fig. 5B, lanes 2 and 3). The two fragments obtained after RNase digestion were estimated to be 202 and 67 nt, consistent with the predicted lengths. The sense probe was not protected by incubation with total RNA (data not shown). The ratio of the two transcripts remained constant over a 10-fold range of the amount of total RNA used (not shown). After correction for the number of radiolabeled UTPs incorporated into each probe, it was calculated that the mRNA encoding the cytoplasmic lysyl-tRNA synthetase accounted for approximately 69 Ϯ 3% of the total lysyl-tRNA synthetase mRNA, while the transcript for the mitochondrial isoform was 31 Ϯ 3% of the total.
In Vitro Aminoacylation-The full-length mitochondrial and cytoplasmic lysyl-tRNA synthetases were expressed in E. coli and purified. These enzymes were used for aminoacylation assays with gel-purified in vitro expressed transcripts for cytoplasmic and mitochondrial tRNA Lys . Prior to determining the specific activity of the enzymes, the maximum level of aminoacylation was determined as a percentage for each transcript, using both forms of the enzyme. As calculated by this plateau analysis, approximately 75% of cytoplasmic tRNA Lys transcript and 15% of the mitochondrial tRNA Lys transcript were aminoacylated by the cytoplasmic lysyl-tRNA synthetase. The average maximum levels for aminoacylation obtained using the mitochondrial lysyl-tRNA synthetase were approximately 95% of the cytoplasmic tRNA Lys transcript and 30% of the mitochondrial tRNA Lys transcript. Although the two forms of the enzyme aminoacylate both the cytoplasmic and the mitochondrial Human cell line 143B was transiently transfected with constructs containing the two full-length lysyl-tRNA synthetase coding regions fused with GFP. A, human cell line 143B was transiently transfected with a pEF/ myc/cyto/GFP construct containing the full-length human cytoplasmic lysyl-tRNA synthetase coding region fused with GFP. Diffuse GFP fluorescence was observed throughout the cytoplasm. Note the exclusion of the majority of the fluorescence from the nucleus. B and C, 143B cells were transfected with a pEF/myc/cyto/GFP construct containing the full-length human mitochondrial lysyl-tRNA synthetase coding region fused with GFP. Shown in B is the immunofluorescence pattern of staining using a primary antibody directed against the mitochondrial COX I protein and a rhodamine-labeled secondary antibody. The immunofluorescence exhibits a punctate distribution in the cytoplasm characteristic of a location in mitochondria. The direct fluorescence of GFP in the same field of view is shown in C. In cells expressing the mitochondrial lysyl-tRNA synthetase-GFP construct, the GFP fluorescence co-localizes with the immunofluorescence of COX I. This indicates that the mitochondrial lysyl-tRNA synthetase is localized in mitochondria.

FIG. 5. Quantitation of transcripts encoding isoforms of human lysyl-tRNA synthetase.
A, a schematic diagram indicating the location of the 363-nt antisense RNA probe used for the ribonuclease protection assay is shown. The region of the cDNA representing exon 1 is white, and the region of the cDNA representing exon 2 is black. The remainder of the cDNA is striped. The probe contained sequences unrelated to lysyl-tRNA synthetase at both its 5Ј-and 3Ј-ends as indicated by the stippled boxes. B, after RNase treatment the protected RNA probes were separated by electrophoresis through a 6% polyacrylamide gel and quantitated by a PhosphorImager. The undigested probe is shown in lane 1. 10% of the amount of probe used in the assay was loaded in this lane; the exposure shown for lane 1 is shorter than that for lanes 2 and 3. Lanes 2 and 3 show the result of hybridizing the probe with total RNA isolated from 143B cells (lane 2, 32 g; lane 3, 16 g) followed by RNase digestion. The regions protected by the mitochondrial transcript (mtKRS; 199 nt) and by the cytoplasmic transcript (cytKRS; 78 nt) are indicated. Sizes were calculated by comparison with a 32 P-end-labeled 100-bp ladder (data not shown).
tRNA Lys transcripts, the specific aminoacylation activity with the mitochondrial tRNA Lys was less than 1% of that with the cytoplasmic tRNA Lys . For mitochondrial lysyl-tRNA synthetase, the specific aminoacylation activities were 13.3 mol/min/ mol of protein with cytoplasmic tRNA Lys and 0.1 mol/min/ mol of protein with mitochondrial tRNA Lys . For cytoplasmic lysyl-tRNA synthetase, the specific aminoacylation activities were 3.02 mol/min/mol of protein with cytoplasmic tRNA Lys and 0.02 mol/min/mol of protein with mitochondrial tRNA Lys .
The low aminoacylation activity of the mitochondrial tRNA Lys transcript was expected, since the mitochondrial tRNA Lys in vitro transcript has been shown to have a modified hairpin structure due to the lack of a methyl group at position 1 of adenosine 9 (19). The reason for the higher plateau levels of aminoacylation and the 4-fold higher activity of the mitochondrial isoform over the cytoplasmic isoform is not known. This may reflect the difference between the cytoplasmic enzyme expressed with an amino-terminal hexahistidine tag (4) and the mitochondrial form with a carboxyl-terminal tag. DISCUSSION A variety of strategies are employed that allow a single gene to encode proteins destined for localization in two or more eukaryotic cellular compartments (reviewed in Refs. 20 -22). Single genes may give rise to multiple transcripts or single transcripts with multiple translation initiation sites, or the gene transcript may undergo alternative splicings. A single protein may also be localized to different compartments by inefficient targeting (21) or by a chimeric targeting sequence whose activity is modulated by post-translational modification (23).
Where one gene encodes proteins located in both the cytoplasm and mitochondria, the cell frequently generates two types of transcripts from the gene. This can be accomplished by alternate transcription start sites or by alternate splicing of the transcript. In each case, the result is the same. The shorter transcript encodes a nonmitochondrial form of the protein, and the longer transcript encodes a mitochondrial targeting sequence in frame with the protein produced from the initiating Met on the shorter transcript. The translated proteins are identical except for the upstream targeting sequence.
In the human lysyl-tRNA synthetase gene, exon 1 encodes the amino-terminal region of the cytoplasmic form of the enzyme. Exon 2 encodes the amino terminus of the mitochondrial protein, containing the mitochondrial targeting sequence that is presumably cleaved upon import of the protein into a mitochondrion. The third and subsequent exons encode that part of the enzyme that is common to both forms of the synthetase. The first three exons are alternatively spliced solely to incorporate or exclude the second exon that encodes the mitochondrial targeting sequence. This form of alternative splicing, to our knowledge, is thus far unique among enzymes destined for more than one cellular location. This pattern of splicing appears to be conserved among the lysyl-tRNA synthetase genes of humans, mice, zebrafish, and C. elegans but not S. cereviseae (Fig. 1B), based on our surveys of the EST and genomic data bases. An exception is Drosophila, where although the overall genomic structure is maintained, the mitochondrial transcript is predicted to commence at exon 2. A more detailed analysis of RNA transcripts from these species is needed to confirm these predictions.
As a consequence of the splicing of exon 1 to exon 2, the mRNA transcript encoding the mitochondrial isoform contains a 23-aa upstream open reading frame (uORF), derived largely from exon 1, that terminates 24 nt upstream from the mitochondrial lysyl-tRNA synthetase initiating methionine. The first 20 aa of this uORF are identical to the amino-terminal 20 aa of the cytoplasmic isoform. Such a uORF may have a role in the regulation of the translation of the downstream reading frame (24,25). The uORF in the mRNA encoding mitochondrial lysyl-tRNA synthetase may be important, since its presence has been conserved among the higher eukaryotes (Fig. 1B). In Drosophila, which appears to lack the first exon in the mitochondrial transcript, the second exon encodes a short uORF that is unrelated to the cytoplasmic gene.
According to the scanning model of translation, the ribosome scans the mRNA from the 5Ј-end and generally initiates translation at the first ATG codon it encounters (26). The translation of the second open reading frame on a transcript can occur by a "leaky scanning" mechanism, where the ribosome does not always recognize the initial ATG if the surrounding sequence is not suitable for efficient initiation. Alternatively, reinitiation of translation can occur after the termination of translation of the uORF (26). The latter mechanism is more likely in the case of mitochondrial lysyl-tRNA synthetase expression, since the sequence around the first ATG (ggaagATGgc) contains the A at the Ϫ3-position and G at the ϩ4-position that characterize a strong initiating sequence (26). Further studies are necessary to determine the mechanism for initiating translation of human mitochondrial lysyl-tRNA synthetase.
A number of human tRNA synthetases have been described in both their cytoplasmic and mitochondrial forms. In most cases, two genes give rise to the separate forms of each enzyme, e.g. histidyl tRNA synthetase (8), phenylalanyl-tRNA synthetase (27,28), and tryptophanyl-tRNA synthetase (29,30). However, in the case of glycyl-tRNA synthetase, there are two proteins derived from a single gene (10). This is due to translation from alternate initiation codons resulting in inclusion or exclusion of a mitochondrial targeting sequence (10). Although other organisms contain numerous examples of cytoplasmic and mitochondrial proteins generated by alternate splicing, human lysyl-tRNA synthetase appears to be unique, or at least unusual, in that the mitochondrial targeting sequence is encoded on the second exon. More commonly, alternate transcription initiation provides alternative 5Ј exons, with the mitochondrial targeting sequence being on the first exon e.g. human dUTPase (31). An examination of GenBank TM data bases leads us to conclude that only human glycyl-and lysyl-tRNA synthetases are likely to have both cytoplasmic and mitochondrial isoforms encoded by single genes. 2 Although between 15 and 30% of the mitochondrial tRNA Lys transcript could be aminoacylated by high levels of enzyme, the specific activities for aminoacylation of the mitochondrial tRNA Lys with both mitochondrial and cytoplasmic enzymes were less than 1% of those obtained with the cytoplasmic tRNA Lys . This was probably due to the incorrect folding of the in vitro mitochondrial tRNA Lys transcript into a modified hairpin structure because of the lack of a methyl group, at position 1 of adenosine 9, that is found in the native tRNA (19). The m 1 A9 modification appears to be necessary for the proper folding of the wild-type tRNA Lys (19,32).
In a tRNA, the "discriminator base" at position 73 in the acceptor stem is often important in ensuring the specific aminoacylation of the tRNA (33). In most cases, synthetases do not efficiently aminoacylate cognate tRNAs with substitutions at position 73. However, the cytoplasmic isoform of the lysyl-tRNA synthetase is tolerant of variations at the G 73 discriminator base position of cytoplasmic tRNA Lys and can consequently aminoacylate the E. coli tRNA Lys that has A at position 73 (4). The data presented here show that essentially the same protein is normally required to aminoacylate mitochondrial tRNA Lys with an A at the discriminator position (34) and cytoplasmic tRNA Lys that has a G at the discriminator position. In contrast, the human glycyl-tRNA synthetase, which is encoded by a single gene and functions in both cytoplasm and mitochondria, is unable to aminoacylate E. coli tRNA Gly (9) and shows specific requirements for the nucleotide at the discriminator position (35). This reflects the fact that both cytoplasmic and mitochondrial tRNA Gly have the same nucleotide at position 73.
One gene appears to encode both cytoplasmic and mitochondrial forms of S. cereviseae cysteinyl-tRNA, 2 histidyl-tRNA (36,37), and valyl-tRNA synthetases. Of those cases where one gene provides both cytoplasmic and mitochondrial tRNA synthetases, S. cereviseae cytoplasmic and mitochondrial tRNA Cys also conserve the discriminator base (U 73 ), while S. cereviseae cytoplasmic and mitochondrial tRNA His and tRNA Val differ at the discriminator position of these tRNAs. In the case of yeast tRNA Val , there are several cytoplasmic tRNAs, one of which has a different discriminator nucleotide from that common to the mitochondrial tRNA Val and other cytoplasmic tRNA Val .
The human lysyl-tRNA synthetase also interacts with the different secondary and tertiary structures of cytoplasmic and mitochondrial tRNA Lys . The mitochondrial tRNA Lys has a relatively small D loop, 5 nucleotides shorter than that of cytoplasmic tRNA Lys , and has also been suggested to have a different tertiary conformation (38). The mitochondrial tRNA Lys has a much larger interstem angle (ϳ140°) between the anticodon and acceptor stems than that of the cytoplasmic tRNA Lys (80 -90°) such that the overall shape of the tRNA does not conform to the canonical L shape (38). This suggests that the human lysyl-tRNA synthetase has considerable flexibility in regard to substrate tRNAs.
A number of mutations in the human mitochondrial tRNA Lys are associated with human diseases (39,40). Human cell lines containing the A8344G mutation have decreased levels of aminoacylated mitochondrial tRNA Lys (41). The identification of human mitochondrial lysyl-tRNA synthetase will allow investigation into whether pathogenic mutations in mitochondrial tRNA Lys affect interactions with lysyl-tRNA synthetase and show whether defects in these interactions contribute to the mechanism of pathogenesis.