The p43 Component of the Mammalian Multi-synthetase Complex Is Likely To Be the Precursor of the Endothelial Monocyte-activating Polypeptide II Cytokine*

p43 is one of the three auxiliary components invariably associated with nine aminoacyl-tRNA synthetases as a multienzyme complex ubiquitous to all eukaryotic cells from flies to humans. The cDNA encoding the hamster protein was isolated by using mixed oligonucleotides deduced from peptide sequences. The 359-amino acid protein is the hamster homologue of the recently reported murine and human EMAP II cytokine implicated in a variety of inflammatory disorders. The sequence of several proEMAP II proteins suggests that the p43 component of the complex is the precursor of the active mature cytokine after cleavage at a conserved Asp residue. The COOH-terminal moiety of p43 is also homologous to polypeptide domains found in bacterial methionyl- or phenylalanyl-tRNA synthetases and in the yeast Arc1p/G4p1 protein that associates with yeast methionyl-tRNA synthetase. Our results implicate the COOH-terminal moiety of p43 as a RNA binding domain. In the native state, as a component of the multisynthetase complex, p43 may be required for tRNA channeling and, after proteolytic processing occurring in tumor cells, would acquire inflammatory properties possibly related to apoptosis. The release of a truncated p43 from the complex could be involved in mediation of the signaling of tumor cells and stimulation of an acute inflammatory response.

though the three nonsynthetase components were invariably encountered in this complex, their structural or functional role within this multi-subunit structure was not established.
We have recently reported the cloning of the cDNA encoding the p18 component of the complex (9). It shares sequence homology with a protein domain recovered in the NH 2 -terminal polypeptide extension of human valyl-tRNA synthetase and in the NH 2 -terminal domains of the ␤ and ␥ subunits of elongation factor 1. In the valyl-tRNA synthetase⅐EF-1H complex, this domain has been involved in protein-protein interaction between the ␤ and ␥ subunits of the eukaryotic elongation factor EF-1H (10) or between valyl-tRNA synthetase and EF-1H (11). This led us to suggest that p18 is involved in anchoring the multisynthetase complex to the elongation factor EF-1H in a transient manner, thus providing the means of a vectorial transfer of aminoacylated tRNA from the synthetases to the factor.
The p38 component of the complex was tentatively identified as casein kinase I with a role in regulating synthetase activities (12), but recent cloning of its cDNA dismisses this proposed functional assignment and suggests that p38 contributes a template protein for the assembly of the complex. 1 The identity and role of the p43 component were not established.
To gain further insight into the function of the multisynthetase complex, we focused on the identification of its auxiliary components. In the present study, we proceed to the isolation and characterization of the p43 cDNA to investigate its functional role in the complex. We show here that p43 is a RNA binding protein that shares homology to some extent with the yeast nucleic acid binding protein G4p1/Arc1p (13,14), which associates with methionyl-or glutamyl-tRNA synthetase. Furthermore, p43 is identical to a gene product identified during the course of this work to an inflammatory cytokine called EMAP II, elicited by tumor cells (15).

EXPERIMENTAL PROCEDURES
General Recombinant DNA Techniques-All recombinant manipulations were carried out using standard procedures (16). Nucleotide sequences were determined by the dideoxynucleotide chain termination method (17). Restriction endonucleases and DNA modification enzymes were purchased from Boehringer Mannheim, New England Biolabs, Stratagene, or Perkin-Elmer. Radionucleotides were from Amersham.
Purification of p43 and Sequencing of Peptide Fragments-The aminoacyl-tRNA synthetase complex was isolated from sheep liver (18). After fractionation of the components of the complex by SDS-polyacrylamide gel electrophoresis on a 10% polyacrylamide gel (19), polypeptides were recovered onto ProBlott membranes (Applied Biosystems) by electroblotting. Membrane pieces carrying p43 were cut and either sequenced immediately or subjected to in situ protease cleavage essentially as described (20,21). In situ proteolytic digestion was carried out at 37°C for 4 h, with endoproteinase Asp-N (Boehringer Mannheim) at a protease to p43 ratio of 1:50. Peptides released from the membranes were separated by reverse-phase HPLC 2 on a C18 column. The amino acid sequence of the isolated peptides was determined by J.-P. Le Caer (Laboratoire de Physiologie Nerveuse, Gif-sur-Yvette) with a gas-phase sequencer (model 470A, Applied Biosystems).
Isolation and Sequencing of the cDNA Encoding p43 from CHO Cells-Total RNA was extracted from sheep liver according to (22). The first strand cDNA synthesis on poly(A) ϩ RNA (1 g), isolated on oligo(dT) cellulose, was primed by using an oligo(dT) 15 (0.5 g) and conducted at 42°C for 90 min, with Moloney murine leukemia virus reverse transcriptase (Stratagene). PCR amplification of p43 cDNA was carried out by using a p43-specific mixed oligonucleotide as a 5Ј-terminal primer (p43-d, Table I) and 5Ј-GGGATCC(T) 20 -3Ј as a 3Ј-terminal primer. Amplification was allowed to proceed for 40 cycles of 1 min at 94°C, 2 min at 50°C, and 3 min at 72°C with AmpliTaq DNA polymerase (Perkin-Elmer). PCR products were analyzed by Southern blotting using another p43-specific mixed oligonucleotide (p43-b, Table I), end-labeled with [␥-32 P]ATP and T4 polynucleotide kinase. A 215-nucleotide-long fragment was digested with EcoRI and BamHI, corresponding to restriction sites appended to the 5Ј-ends of the primers, and subcloned into pUC18. The nucleotide sequence of the insert was determined. It contained 130 nucleotides from the 3Ј coding region of p43. A longer cDNA probe was synthesized by PCR amplification between a mixed oligonucleotide (p43-c, Table I) located in the 5Ј-upstream region of the insert and exact oligonucleotide primers deduced from the cloned sequence.
A cDNA library was constructed from poly(A) ϩ mRNA isolated from exponentially growing CHO cells, in the Uni-ZAP XR vector by using the ZAP-cDNA synthesis kit (Stratagene). A total of 0.5 ϫ 10 6 recombinant phages were plated on Escherichia coli XL1-Blue and positive transformants were selected (23) by using the sheep cDNA fragment described above and radiolabeled by random oligonucleotide priming. The nucleotide sequence of the p43-cDNA was determined on both strands.
Analysis of p43 in HeLa Cells Extract-HeLa cells were grown as monolayers at 37°C in Ham's F-12 medium containing 5% fetal calf serum (Life Technologies, Inc.). Cells grown in ten 150-cm 2 flasks were lysed by the addition of 4 ml of ice-cold lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 10 mM ␤-mercaptoethanol, 0.5% Triton X-100, 2 mM diisopropyl fluorophosphate). After centrifugation at 10000 ϫ g at 4°C for 15 min to remove cell debris, the supernatant was analyzed by filtration on a BioGel A-5 m column (1.6 ϫ 83 cm) equilibrated in 100 mM potassium phosphate, pH 7.5, 10% glycerol, 10 mM ␤-mercaptoethanol, and developed at 4°C at a flow rate of 5 ml/h. Fractions of 2.7 ml were collected. Elution of the aminoacyl-tRNA synthetase complex and of the p43 protein were monitored by the tRNA aminoacylation assay (24) and by Western blotting using anti-sheep p43 antibodies (25).
Expression of the COOH-terminal Domain of p43 in Bacteria and Purification-The carboxyl-terminal domain of the human p43 (p43Ct) was expressed in E. coli with the pET-28b expression system (Novagen). The p43 cDNA was produced by PCR between oligonucleotides 5Ј-CCCCCATGGCCAAGCCAATAGATGTTTCCC-3Ј and 5Ј-CCCCTC-GAGTTTGATTCCACTGTTGCTCA-3Ј, which introduced a NcoI site with the ATG initiation codon, and a XhoI site in frame with the vector sequence encoding the His tail, respectively. The 0.5-kilobase pair NcoI-XhoI fragment was introduced into pET-28b digested with NcoI and XhoI to give pET43Ct. The nucleotide sequence of the PCR fragment was verified. The encoded polypeptide corresponds to the human p43 from Lys 148 to its COOH-terminal residue Lys 312 (Lys 195 to Lys 359 of hamster p43), with the two additional NH 2 -terminal residues MetAlaand the COOH-terminal extension -LeuGlu(His) 6 .
Purification of p43Ct was performed starting from a 1-liter culture of BL21(DE3) ϫ pET43Ct grown at 37°C in LB-kanamycin (50 g/ml) broth to a cell density corresponding to A 600 ϭ 0.5. The recombinant protein was expressed at 37°C for 4 h after addition of 1 mM isopropyl-1-thio-␤-D-galactopyranoside. All subsequent steps were conducted at 4°C. Cells were harvested by centrifugation (6000 ϫ g for 5 min), washed with ice-cold extraction buffer (20 mM Tris-HCl, pH 8.2, 500 M NaCl, 10 mM imidazole), resuspended in 20 ml of extraction buffer supplemented with 1 mM diisopropylfluorophosphate, and lysed by sonication. After centrifugation at 10000 ϫ g for 15 min, the lysate was loaded on a His⅐Bind resin (Novagen, 1.1 ϫ 10.5 cm) charged with Ni 2ϩ . After washing with extraction buffer, bound material was eluted by a 200-ml linear gradient of imidazole from 20 mM to 200 mM in 20 mM Tris-HCl, pH 8.2, 500 mM NaCl. Fractions containing p43Ct (eluted at 45 mM) were pooled (18 ml), equilibrated in 20 mM Tris-HCl, pH 7.0, 1 mM dithioerythritol by filtration on a Sephadex G-10 column (Pharmacia Biotech Inc., 2.6 ϫ 20 cm) and applied to a 1.6 ϫ 20-cm column of SOURCE 15S (Pharmacia) equilibrated in the same buffer. The column was eluted with a 800-ml linear gradient of NaCl from 0 to 500 mM. Fractions containing p43Ct (eluted at 165 mM) were pooled (36 ml), concentrated, dialyzed against 20 mM potassium phosphate, pH 7.5, 1 mM dithioerythritol, and stored at Ϫ70°C at a protein concentration of 11 mg/ml. Protein concentration was determined by using a calculated absorption coefficient of 0.41 A 280 units⅐mg Ϫ1 ⅐cm 2 .
Analytical Gel Filtration-The apparent native molecular mass of p43Ct was determined by gel filtration on a Superose 12 HR 10/30 (Pharmacia) column equilibrated in 50 mM potassium phosphate, pH 7.5, 10 mM ␤-mercaptoethanol containing 10% of glycerol where indicated and developed at room temperature at a flow rate of 0.4 ml⅐min Ϫ1 . All samples were loaded in 0.2 ml. For a particular protein, its elution was described in term of the corresponding K av value.
where V e is the elution volume of the particular molecule, V 0 is the void volume of the column, and V t is the total bed volume. V 0 and V t were determined with thyroglobulin (669 kDa) and dithioerythritol (154 Da), respectively.

Isolation of the cDNA Encoding the p43 Component of the
Multi-synthetase Complex-The strategy used to isolate the cDNA encoding the p43 component of the complex involves three major steps: 1) the identification of peptide sequences of the protein, 2) their use to design degenerate primers for PCR cloning of cDNA fragments, and 3) the screening of a cDNA library to isolate a full-length cDNA. The p43 polypeptide was isolated from the complex after denaturation by SDS and fractionation of the 11 polypeptide components by polyacrylamide gel electrophoresis. In the complexes isolated from rat or rabbit, the p43 and p38 components cannot be easily separated (4), leading to cross-contaminating polypeptides (27). Because the p43 component of the sheep complex migrates well apart from the p38 protein, this material was chosen to initiate the cloning process.
Analysis of native p43 by automated Edman degradation reveals the presence of a blocked NH 2 -terminal residue. After transfer onto a ProBlott membrane, p43 was subjected to in situ cleavage by endoproteinase Asp-N. Peptides released from the membrane were recovered and purified by reverse-phase HPLC on a C18 column. Mixed oligonucleotides were deduced from the peptide sequences (Table I) and used as primers for PCR amplification of the sheep p43 target cDNA. After two rounds of amplification with p43-specific primers (Table I) and a poly(dT) primer, a 600-nucleotide-long cDNA of sheep origin was recovered. The 3Ј coding region of this cDNA fragment was used to screen a -ZAP CHO cDNA library. Six independent clones were recovered and analyzed.
The 1254-nucleotide-long cDNA sequence displayed a single long open reading frame of 1077 nucleotides and 5Ј-and 3Јuntranslated regions of 83 and 74 nucleotides, respectively ( Fig. 1). A TGA stop codon is located at position 21 within the 5Ј noncoding sequence, in frame with the first putative ATG initiation codon at position 84, arguing for the isolation of a full-length cDNA.
Identification of the Deduced Protein as the p43 Component of the Multi-synthetase Complex-The cDNA sequence has a coding potential for a polypeptide of 359 amino acids, with a calculated molecular mass of 39.6 kDa in agreement with the value of 43 kDa determined by SDS-polyacrylamide gel electrophoresis analysis of the complex from CHO cells (6). As previously observed by electrofocusing analysis of the complex (28), the p43 polypeptide displays a markedly basic pI, evaluated to 9.4 according to its amino acid sequence. The sequences of the three peptides obtained from the p43 component of the sheep complex (Table I) as well as some peptide sequences issued from the rat p43/p38 components (27) were recovered in the protein sequence derived from the cloned cDNA (Fig. 1). As shown below, the cDNA-encoded protein reacted with antibodies raised against the p43 component of the sheep liver complex, thereby establishing its assignment to the p43 cDNA from CHO cells.
Identification of Human p43 to the Precursor of the EMAP II Cytokine-The hamster p43 amino acid sequence was used to search the protein data libraries by using the Blast network facilities (29). The human and mouse endothelial monocyteactivating polypeptides (EMAP) II (15) were found to share 86 and 85% of identical amino acids with residues 47-359 of the p43 hamster protein, respectively. The mature form of EMAP II is a tumor-derived cytokine of about 20 kDa (30) released from a larger precursor after removal of the NH 2 -terminal moiety of the polypeptide. The cDNAs isolated for the EMAP II proteins are shorter than that obtained for the p43 protein, but amino acid sequence identities are also recovered between p43 and the translated short 5Ј-upstream regions of the human and mouse cDNAs. Therefore, the EMAP II cDNAs are likely to be incomplete cDNA species. The high sequence similarity between p43 and EMAP II suggested that they are the same gene product. We recently cloned the human homologue of p43 by screening a human cDNA library by the double hybrid approach using the p38 component of the complex as a bait. 3 The sequence of human p43 is identical to that reported for EMAP II. These results clearly indicate that p43 is related to that cytokine.
As shown in Fig. 2, the p43 polypeptide sequence can be divided into six regions, according to their amino acid composition and homology to other proteins. Domain I is a basic segment, its high pI ϭ 12.0 being essentially due to the presence of many arginine residues (9 Arg and 0 Lys, residues . This domain has not been previously isolated for the EMAP II cDNAs. Domain II is also basic (pI ϭ 9.0) but is mainly composed of lysine residues (12 Lys and 3 Arg, residues 46 -119) and displays the propensity to fold into an ␣-helix conformation as judged from secondary structure predictions (31,32). This region is conserved with the human and mouse proteins, which display 94 and 96% of identical residues, respectively. By contrast, domains III and IV are especially variable regions. The hamster, human, and mouse proteins share only about 60 and 70% of identical residues, respectively for domains III and IV. Domain III has a high serine content (10 Ser, residues 120 -157), whereas domain IV (residues 158 -193) is distinctly hydrophilic, with a stretch of lysine and glutamate 3

AAR ATH TGG GAR CA
residues and the absence of aromatic residues, a feature reminiscent of a KEKE motif identified as a possible protein-protein interaction site (33). The EMAP II cytokine starts after residue Asp 193 , located within the sequence Ser-(Ala/Thr)-Asp-Ser-Lys-Pro strictly and exclusively conserved in all the sequences of p43/EMAP II proteins shown in Fig. 2. Domain V, made of 102 amino acid residues (194 -295), is conserved between the hamster, mouse, sheep, and human p43 proteins (Ϸ90% identity) and is also recovered with 60% identity in a yeast protein identified for its affinity for quadruplex nucleic acids (G4p1; Ref. 13) or for its ability to interact genetically with a pore-associated protein involved in tRNA biogenesis (Arc1p; Ref. 14). This yeast protein interacts with cytoplasmic methionyl-and glutamyl-tRNA synthetases, enzymes associated, in mammalian cells, into the multisynthetase complex. Domain V is also recovered within the carboxyl-terminal polypeptide extension of methionyl-tRNA synthetase from the nematode Caenorhabditis elegans (73% identity) or from all the eubacterial or archae methionyl-tRNA synthetases (36 and 33% identity with Thermus thermophilus and Methanococcus jannaschii methionyl-tRNA synthetase, respectively). A hypothetical 12.3-kDa protein from the ileX-ebgR intergenic region of E. coli exactly coincides with this domain and displays 39% identity. Finally, a similar domain (36% identity) is also recovered within the amino-terminal segment of the large subunit of the tetrameric (␣ 2 ␤ 2 ) bacterial phenylalanyl-tRNA synthetase.

FIG. 2. Alignment of the amino acid sequences of p43 and related proteins.
Top panel, the p43 protein from hamster (Cricetulus griseus; p43-Cg) is schematized according to the six functional or structural domains referred to in the text. Related proteins sharing one or several of these domains are represented below. When homologous proteins have been described from different species, only one member of each family is indicated. This includes the precursor of the EMAP II cytokine from Homo sapiens (proEMAPII-Hs; accession number U10117); a quadruplex nucleic acids binding protein G4p1 from S. cerevisiae (G4p1-Sc; accession number U31348); methionyl-tRNA synthetase from C. elegans (MetRS-Ce; accession number Z73427) or T. thermophilus (MetRS-Tt; accession number M64273), taken as a representative of all eubacterial and archae methionyl-tRNA synthetases; the large ␤ subunit of phenylalanyl-tRNA synthetase from Synechococcus sp. (␤PheRS-Ssp; accession number X94345); and an unknown open reading frame from E. coli (ORFX-Ec; accession number U18997). The segment of p43 identified with the EMAP II cytokine is indicated by an arrow. The numbering refers to the position of the corresponding amino acid boundaries in the sequence of the complete proteins. Bottom panel, a detailed sequence alignment is shown in the block region corresponding to the ␤-barrel domain of phenylalanyl-tRNA synthetase. Only amino acids from homology blocks are indicated. Numbers refer to insertions. Additional sequences are the EMAP II precursor from Mus musculus (proEMAPII-Mm; accession number U10118); the partial sequence of p43 from Ovis aries (p43-Oa; this study); an unknown open reading frame from the sequence of a cDNA fragment isolated from Oriza sativa (ORFX-Os; accession number D23020); methionyl-tRNA synthetase from E. coli (MetRS-Ec; accession number K02671), Hemophilus influenzae (MRS-Hi; accession number U32753), Bacillus stearothermophilus (MRS-Bst; accession number X57925) or Bacillus subtilis (MetRS-Bsu; accession number D26185); and the ␤ subunit of phenylalanyl-tRNA synthetase from E. coli (␤PheRS-Ec; accession number V00291), B. subtilis (␤PheRS-Bsu; accession number X53057), or T. thermophilus (␤PheRS-Tt; accession number X65609). For each sequence, the position of the first amino acid is indicated. Numbers in parentheses refer to partial amino acid sequences. The secondary structure elements, indicated as arrows for ␤-strands and rectangles for ␣-helices, are based on the crystal structure of the large subunit of phenylalanyl-tRNA synthetase from T. thermophilus (34).
It corresponds to the domain B2 from the crystal structure of phenylalanyl-tRNA synthetase from T. thermophilus (34). The particular feature of this domain is that it forms a discrete six-stranded ␤-barrel structure. The carboxyl-terminal domain of hamster p43 constitutes domain VI, which shares more than 95% identical residues with human and mouse EMAP II, 41% with C. elegans methionyl-tRNA synthetase, and 40% with yeast G4p1 in a more restricted segment.
Therefore, p43 is made of several distinct protein domains, one of which, domain V, is a recurrent protein module spread over a wide range of organisms, including archae, eubacterial, and eukaryote species, as a fusion protein. A common feature of these proteins is their ability to bind nucleic acids.
Structural Behavior of Human p43-Because two distinct protein species related to p43, a component of the multisynthetase complex and a precursor of the EMAP II cytokine, have been described, we considered the possibility that two discrete polypeptides expressed from the same gene coexist within the cell. The oligomeric state of the p43 protein from cultured human HeLa cell lines was investigated. A crude extract from exponentially growing HeLa cells, obtained after lysis in the presence of 150 mM NaCl and 0.5% of Triton X-100 to thoroughly solubilize the cellular proteins, was subjected to size fractionation on a BioGel A-5 M column (Fig. 3). Fractions were assayed for lysyl-and arginyl-tRNA synthetase activities to determine the elution volume of the synthetase complex. In addition, the free monomeric form of arginyl-tRNA synthetase, of M r Ϸ67,000 (35), labeled the elution volume for small proteins. Dilute protein samples from column fractions were concentrated by precipitation with trichloroacetic acid, subjected to SDS-polyacrylamide gel electrophoresis, and analyzed by Western blotting using antibodies directed to the p43 component of the complex (Fig. 3, inset). These antibodies revealed a single polypeptide of M r Ϸ43 000, coeluting with the complex.
Expression, Purification, and Structural Characterization of the COOH-terminal Domain of p43 Produced in E. coli-The carboxyl-terminal half of p43 was produced in bacterial cells, and its structural behavior was characterized. The cDNA sequence encoding the carboxyl-terminal moiety of the human p43, p43Ct, was cloned into the pET-28b expression vector. Six histidine residues are appended to the COOH terminus of the protein to facilitate its isolation on a metal-chelate resin. The encoded protein is 175 amino acid residues in length, with a theoretical M r of 19,400. A high expression level was obtained following isopropyl-1-thio-␤-D-galactopyranoside induction, and the recombinant protein proved to be soluble in E. coli extracts. Its purification was accomplished by affinity chromatography on a Ni 2ϩ chelating column followed by fractionation on a SOURCE 15S cation exchanger column. This purification procedure yields 20 mg of p43Ct/liter of culture from BL21(DE3) ϫ pET43Ct. The purified protein is homogeneous and displays an apparent mass of 20 kDa by SDS-polyacrylamide gel electrophoresis (not shown).
The native molecular mass of the p43Ct protein produced in E. coli was determined by filtration on a Superose 12 column (Fig. 4). To favor protein-protein interactions, the column was developed at room temperature in a low ionic strength buffer containing 50 mM potassium phosphate buffer, pH 7.5, and 10 mM 2-mercaptoethanol in the presence or the absence of 10% glycerol. The molecular mass standards included conalbumin (86 kDa), bovine serum albumin (67 kDa), ovalbumin (45 kDa), chymotrypsinogen A (25 kDa), and cytochrome c (12.4 kDa). The COOH-terminal domain of p43 was eluted as a single symmetrical peak between chymotrypsinogen A and cytochrome c, with an elution volume corresponding to an apparent native molecular mass of 16 Ϯ 1 kDa (Fig. 4). No trace of a dimeric p43Ct entity could be detected, even in the presence of 10% glycerol in the equilibration buffer.
Affinity of the COOH-terminal Domain of p43 for Nucleic Acids-Because domains V and VI of p43 display sequence homology with a yeast protein independently isolated as G4p1 or Arc1p for its ability to bind to quadruplex nucleic acids or tRNA, respectively, we analyzed the nucleic acid affinity of the COOH-terminal moiety of p43. Homogeneous p43Ct was used in a band shift assay to measure its propensity to form protein-RNA complexes. When in vitro transcribed tRNAs were used in this assay, a shift of the labeled tRNA was detected (Fig. 5). The same pattern was observed with several yeast tRNA transcripts: tRNA Phe and tRNA Asp (Fig. 5), tRNA Ala , and tRNA Met (not shown). The deduced K d dissociation constant was estimated to about 40 M in the presence of a high salt concentration (150 mM NaCl) in the incubation mixture. Two sorts of complexes are formed when increasing the protein concentration, suggesting that a 1:1 and a 1:2 tRNA-protein complex are produced (Fig. 5). Because identical shifts were observed with different tRNA molecules, this tRNA-protein interaction is likely to involve structural rather than sequence recognition FIG. 3. Gel filtration behavior of p43 from a HeLa cell extract. A crude extract from exponentially growing HeLa cells was subjected to fractionation on a BioGel A-5 m column as described under "Experimental Procedures." Initial rates of lysyl-tRNA (E) and arginyl-tRNA (ࡗ) synthesis were measured. Inset, fractions 24 -60 were analyzed by Western blotting using anti-p43 antibodies. Lane C corresponds to a control sample containing 1 g of homogeneous rat liver complex. elements. Binding of p43Ct to homopolymers of 5Ј 32 P-labeled poly(rA), poly(rU), poly(rC), or poly(rG) was examined by the same band shift assay. No protein-RNA complexes were observed, except in the case of poly(rG) where a faint band of slower mobility could be detected at high protein concentration (not shown).

DISCUSSION
The p43 protein corresponds to a polypeptide of 359 amino acids with a distinctly basic pI of 9.4. The NH 2 -terminal moiety of the protein of about 200 amino acid residues displays no significant homology with proteins of known function registered in the data libraries. By contrast, the COOH-terminal region of about 170 residues is related to the COOH terminus of the dimeric methionyl-tRNA synthetase from bacterial origin, a polypeptide domain not found in the sequence of the monomeric methionyl-tRNA synthetase from yeast. Methionyl-tRNA synthetase from the nematode C. elegans also displays this extension and could be a dimer. Also noteworthy is the fact that a protein isolated from the yeast Saccharomyces cerevisiae for its ability to bind to nucleic acids, the G4p1/Arc1p protein (13,14), interacts with yeast methionyl-tRNA synthetase and has a COOH-terminal domain homologous to p43. Therefore, methionyl-tRNA synthetase is always somehow associated with this type of protein domain: 1) as a natural fusion protein in the case of the bacterial or nematode synthetase; 2) as a binary ␣␤ complex in yeast; and 3) as members of a multisynthetase complex in higher eukaryotes. In bacteria, the function of the COOH-terminal polypeptide extension of methionyl-tRNA synthetase is not known. The removal of this domain from the E. coli enzyme by controlled proteolysis (36) or genetic engineering (37) has no effect on the specific activity of the enzyme. The only noticeable effect concerns the oligomeric structure of the truncated enzyme, which behaves as a monomer instead of a dimer for the native enzyme. However, it has never been shown that the removed domain is a dimerization domain, and the possibility that its removal destabilizes the actual dimer interface cannot be excluded. The finding that the isolated COOH-terminal domain of p43 (Fig. 4) behaves as a monomer is at odds with a putative role of a dimerization domain.
A protein domain related to domain V of p43 is also recovered from the NH 2 -terminal region of bacterial phenylalanyl-tRNA synthetase. The x-ray structure of the ␣ 2 ␤ 2 T. thermophilus enzyme complexed with two tRNA Phe molecules does not provide any functional role for this domain (38). However, the ␤-barrel-like structure of this domain, called the B2 domain, suggested that it could be involved in protein-RNA interactions. Several RNA binding proteins display this type of structural organization: the anticodon binding domain of aspartyl-tRNA synthetase (39), the major cold-shock protein CspB (40), the staphylococcal nuclease (41), and the ribosomal protein L14 (42). Therefore, we tested the possibility that the isolated COOH-terminal region of p43 could be a RNA binding domain. As shown in Fig. 5, gel shift experiments revealed its propensity to associate with tRNA molecules in a structure-rather than sequence-dependent manner. This result corroborates the data previously described for the yeast G4p1 and Arc1p proteins, where it was shown that the intact proteins are nucleic acids binding proteins, and attributes this property to the COOH-terminal moiety of these proteins. The yeast p43 homologues and the p43 protein described in this study differ in their NH 2 -terminal regions. Whereas Arc1p/G4p1 interacts with methionyl-tRNA synthetase to form a binary complex, mammalian p43 associates within the multisynthetase complex. This suggests that the protein domain involved in these interactions is located within the variable NH 2 -terminal domain of these proteins.
What could be the function of p43 within the multisynthetase complex? The ARC1 gene, encoding the yeast homologue of p43, was cloned by a genetic complementation assay for LOS1, a gene involved in general tRNA export, and related to the nuclear pore complex (14). ARC1 could complement LOS1 by virtue of its ability to bind tRNA, to interact with methionyl-tRNA synthetase and to stimulate its activity. A decreased nuclear export of tRNA could be compensated by an increased synthetase activity. However, because the Arc1p-tRNA interaction seems to be a feature not restricted to tRNA Met , it must be envisioned that the stimulation of synthetase activity by Arc1p is not restrained to methionyl-tRNA synthetase. Another possibility is to consider that Arc1p is involved in the direct delivery of tRNA from the nuclear pore complex to the synthetase and participates in the channeling of tRNA within the cytoplasm. According to the tRNA channeling model developed by Deutscher and co-workers (43,44), in mammalian cells, tRNA molecules would be channeled from the synthetases to elongation factor 1 and to ribosome and back to the synthetase without mixing with other components from the cellular fluid. In that connection, newly synthesized tRNAs entering the cytoplasm through the nuclear pore complex could require association with another protein to be incorporated into the tRNA cycle. If p43 is involved in the channeling of tRNA from the nucleus to the synthetases, the finding that no free p43 was revealed in a HeLa cell crude extract (Fig. 3) suggests that p43 triggers the multisynthetase complex in the vicinity of the nuclear pore complexes.
The cloning of the p43 component of the multisynthetase complex also revealed an unexpected link between protein synthesis and the cytokine network. A tumor-derived cytokine, called EMAP II, was identified for its propensity to activate endothelial cells, leukocytes and monocytes, and to potentiate the effect of tumor necrosis factor (30). The mature EMAP II is a 22-kDa polypeptide generated by cleavage of a precursor (15). Little is known about the molecular mechanisms implicated in controlling the synthesis and processing of the precursor into a mature cytokine. The murine and human EMAP II cDNAs (15) encode a protein of 34 kDa, which proves to be identical to p43. In the present study, we were unable to identify any free and/or truncated form of p43 in extracts from exponentially growing HeLa cells. Therefore, one likely hypothesis concerns a possible coupling between the breakdown of protein synthesis in tumor cells and the release of an active EMAP II cytokine. This release could involve a specific degradation pathway. It is of particular interest to notice that the site of cleavage occurs after the Asp residue in the sequence Ser-Ala-Asp-Ser conserved in p43/EMAP II from human, mouse, hamster, and sheep and is absent in the homologous proteins that have no cytokine properties (Fig. 2). This putative protease recognition site is strikingly similar to that of a family of intracellular cysteine proteases called interleukin-1␤-converting enzyme, which have an absolute requirement for an Asp residue at position Ϫ1 of the cleavage site (45). However, the presence of a KEKE motif upstream of the cleavage site also suggests that degradation of p43 could occur via the multicatalytic protease (33). Partial proteolysis of the multisynthetase complex with the removal of the COOH-terminal domain of p43 that would lead to a breakdown of the supramolecular organization of the translational machinery could prime a cascade of events leading to cell lysis and release of the truncated p43 in the cell environment, generating various inflammatory processes. One alternative possibility is to consider that EMAP II is not directly issued from the complex but is a maturation product of a de novo synthesized p43 protein.
The identification of p43 as a possible precursor of an inflammatory cytokine elicited by tumor cells opens the route to decipher its dual biological function: 1) its role in normal cells, as a component of the synthetase complex, possibly related to its RNA binding capacity and 2) its postulated relationship with inflammatory disorders and apoptotic processes when released from tumor cells.