Structure and Function of the c-myc DNA-unwinding Element-binding Protein DUE-B*

Local zones of easily unwound DNA are characteristic of prokaryotic and eukaryotic replication origins. The DNA-unwinding element of the human c-myc replication origin is essential for replicator activity and is a target of the DNA-unwinding element-binding protein DUE-B in vivo. We present here the 2.0Å crystal structure of DUE-B and complementary biochemical characterization of its biological activity. The structure corresponds to a dimer of the N-terminal domain of the full-length protein and contains many of the structural elements of the nucleotide binding fold. A single magnesium ion resides in the putative active site cavity, which could serve to facilitate ATP hydrolytic activity of this protein. The structure also demonstrates a notable similarity to those of tRNA-editing enzymes. Consistent with this structural homology, the N-terminal core of DUE-B is shown to display both d-aminoacyl-tRNA deacylase activity and ATPase activity. We further demonstrate that the C-terminal portion of the enzyme is disordered and not essential for dimerization. However, this region is essential for DNA binding in vitro and becomes ordered in the presence of DNA.

During each division of somatic cells DNA replication is regulated so that the genome is copied in its entirety only once (1,2). The classical replicon hypothesis states that the initiation of DNA replication at replication origins is controlled by initiator protein binding at regulatory sites called replicators and has served as a useful description for prokaryotic replication (3). In contrast, a concise description of eukaryotic replication is complicated by the fact that eukaryotic chromosomes contain multiple origins of replication and the replicator sequences controlling them do not share easily recognizable conserved sequences (4,5). However, a structure common to yeast and mammalian origins of repli-cation is a region of easily unwound DNA termed a DNAunwinding element (DUE) 3 (6 -13).
Replication initiates within a 3.5-kb region upstream of the human c-myc gene (14 -16). Semi-conservative DNA synthesis was shown to originate in this region using nascent DNA abundance assays in chromosomal and plasmid-based systems in vitro and in vivo (13, 16 -26). Transposition of a 2.4-kb c-myc origin fragment to an ectopic site in the HeLa genome promotes bidirectional DNA replication initiation at the site of insertion, confirming its function as a chromosomal replicator (13,22). This 2.4-kb region contains an AT-rich region comprising three matches to the Saccharomyces cerevisiae ARS consensus sequence and a ϳ40-bp DUE zone of predicted helical instability, also termed the far upstream element or FUSE (27), which has been shown to be sensitive to single stranddirected reagents in vitro and in vivo (19,27). Deletion of the DUE/ARS region eliminates c-myc origin activity (13), and a heterologous DUE restores origin activity, 4 implying that a DUE is essential for chromosomal replication origin activity.
In addition to a region of easily unwound DNA, replication origins frequently contain binding sites for auxiliary factors that recruit constituents of the replication complex. Yeast one-hybrid studies utilizing the DUE from the c-myc origin as bait identified a 24-kDa polypeptide capable of specifically engaging the DUE in vivo (28) and in vitro in the presence of other nuclear proteins (29). This protein, termed DUE-B, is conserved in bacteria, yeasts, and eukaryotes with homologs identified in rodents, amphibians, and fish. Recombinant DUE-B purified from baculovirus-infected insect cells is a homodimer that co-purifies with ATPase activity (29). In HeLa cells, short interfering RNA-mediated knock down of DUE-B delays entry into S phase and promotes cell death. Moreover, immunodepletion of DUE-B from Xenopus egg extracts inhibits DNA replication and addition of purified, recombinant DUE-B expressed in HeLa cells restores replication activity to these extracts, emphasizing the interspecies conservation of DUE-B function (29). These observations provide strong evidence that DUE-B is involved in the initiation of DNA replication.
To gain further insights into the function of DUE-B in DNA replication, we have solved the three-dimensional crystal structure of the recombinant protein to a resolution of 2.0 Å and, guided by our crystal structure, have carried out further bio-chemical characterization of DUE-B. Homodimerization of DUE-B is mediated by an extensive set of interactions that result in the formation of a continuous ␤-sheet structure between the two monomers. The C-terminal residues that are involved in mediating interactions with target DNA are largely disordered in the crystal, in agreement with results from limited proteolysis experiments. The three-dimensional structure of DUE-B reveals notable similarity to the overall domain structures of both D-aminoacyl-tRNA deacylases and the archaeaspecific editing domain of threonyl-tRNA synthetase (30 -32).
In vitro analyses confirm that DUE-B possesses both D-aminoacyl-tRNA deacylase and ATPase activities.

EXPERIMENTAL PROCEDURES
Purification and Crystallization-Recombinant human DUE-B containing a C-terminal His 6 tag (rDUE-B) was purified from baculovirus-infected insect cells (29). Forty-eight hours after infection the cells were lysed and DUE-B purified by Ni ϩ2 -NTA resin affinity chromatography (Qiagen) according to the manufacturer's instructions. Fractions containing Ͼ95% pure rDUE-B (200 mM imidazole eluate) were identified by SDS-PAGE and silver staining. Protein was concentrated and buffer exchanged into 10 mM Tris-Cl, pH 7.6, 100 mM NaCl, 10% glycerol by centrifugal filtration (Millipore). Immediately prior to crystallization, protein samples were exchanged into a buffer composed of 100 mM NaCl and 10 mM HEPES, pH 7.5, by ultrafiltration.
Crystals suitable for diffraction studies were obtained by the hanging drop method by mixing 2 l of protein (6 mg/ml) with 2 l of a well solution (200 mM KCl, 50 mM sodium cacodylate, pH 6.5, and 10% (w/v) polyethylene glycol 8000) and equilibrating against the latter solution at 12°C. For cryo-crystallography, crystals were serially transferred into the precipitant solution containing incremental concentrations of glycerol up to a final concentration of 30% and flash-cooled by plunging directly into liquid nitrogen. Native diffraction data were collected at an insertion device synchrotron source using an ADSC CCD detector (Beamline 17-ID, Advanced Photon Source, Argonne National Laboratory, Argonne, IL) at 100 K. Data were indexed and scaled using HKL2000 (33) to a limiting resolution of 2.0 Å. A derivative data set was collected at 100 K from crystals soaked in 1 mM ethylmercuric phosphate using a RAXIS IVϩϩ image plate detector system to a limiting resolution of 3.0 Å. The data collection and processing statistics are summarized in Table 1.
Phasing, Model Building, and Refinement-Initial crystallographic phases were obtained from a mercurial derivative (34,35). Although the resultant electron density was not readily interpretable, solvent flattening and non-crystallographic symmetry averaging (36) between the multiple copies of DUE-B in the crystallographic asymmetric unit allowed for an initial trace of the protein main chain atoms. Multiple rounds of 2-and 4-fold non-crystallographic symmetry averaging (36), followed by cycles of manual rebuilding and crystallographic refinement using REFMAC5 (37), allowed assignment of the first 151 amino acids of each of the four monomers in the crystallographic asymmetric unit. Refinement against the 2.0 Å synchrotron data yielded a final model with a working R factor of 20.4% and a free R factor of 23.7%. The final model contains the first 151 amino acid residues for each DUE-B monomer, 379 water molecules, and 4 magnesium ions. The refined coordinates have been deposited in the Protein Data bank with accession number 2OKV.
Enzyme Assays-Protease digestions of DUE-B were performed with trypsin, chymotrypsin, or V8 protease at 1:20 -1: 400 w:w protease:DUE-B. Aminoacylation of Escherichia coli tRNA was performed as described by Soutourina et al. (38) to minimize contamination by L-aspartic acid. A 500-l reaction containing 100 mM Tris, pH 7.6, 5 mM MgCl 2 , 50 mM KCl, 0.5 mM EDTA, 2.5 mM ATP, 1% glycerol, 0.6 mM ␤-mercaptoethanol, 5 M D-[ 3 H]aspartic acid (PerkinElmer), 25 M E. coli tRNA (Sigma), and 1250 units of mixed E. coli tRNA synthetases (Sigma) was incubated for 10 min at 37°C. After overnight precipitation with ethanol at Ϫ20°C, the supernatant was lyophilized, resuspended in water, and used again in an aminoacylation reaction at 37°C for 90 min with an additional 12.5 nmol E. coli tRNA and 1250 units of E. coli tRNA synthetase. The reaction was extracted with phenol-chloroform, precipitated with ethanol, and used fresh in deacylation reactions. E. coli tRNA (Sigma) aminoacylated with D-[ 3 H]aspartate was incubated in a 50-l reaction containing 20 mM Tris-HCl, pH 7.8, 5 mM MgCl 2 , and 500 fmol of the indicated form of recombinant DUE-B. The reaction was incubated at 37°C for 10 min before addition of glycogen and precipitation with ethanol at Ϫ20°C overnight. After centrifugation the supernatant was analyzed by liquid scintillation counting. To assay for ATPase activity, reactions containing 25 mM Tris, pH 7.6, 1 mM dithiothreitol, 0.2 mg/ml bovine serum albumin, 5 mM MgCl 2 , 100 M unlabeled ATP, 100 nM [␥-32 P]ATP (PerkinElmer), and the indicated amount of recombinant DUE-B were incubated at 37°C for 60 min. Reactions were stopped by addition of EDTA to 50 mM and frozen by dropping into liquid N 2 . An aliquot of each reaction was spotted on a polyethyleneimine (PEI)-cellulose TLC plate (Selecto Scientific) that was developed in 0.8 M acetic acid/0.8 M LiCl and analyzed using a PhosphorImager.
To assay for DUE-B dimer stability, rDUE-B was mixed with Xenopus egg extract for 40 min at room temperature. The mixture was subsequently incubated with Ni ϩ2 -NTA-agarose beads for 4 h at 4°C.
Electrophoretic Mobility Shift Assays-Recombinant forms of DUE-B containing or lacking the C terminus were used in electrophoretic mobility shift assays to determine DNA binding. The DNA probe was generated using PCR in the presence of [␣-32 P]dCTP with primers amplifying a 123-bp region of the c-myc replicator. Preparation of the protease-resistant DUE-B core by trypsin digestion (1:100 w:w trypsin:DUE-B) was as described, except the reaction volume was 10 l and contained 25 pmol DUE-B. After 1 h at 37°C a 0.5-l aliquot was removed for SDS-PAGE analysis, and the remaining reaction mixture was brought up to 15 l with TBE (Tris borate-EDTA) (0.5ϫ), NaCl (50 mM), loading dye (1ϫ final concentration; Promega), and 25 fmol 32 P-labeled PCR product. Samples were then loaded on a 4% native polyacrylamide (0.5ϫ TBE) gel that had been pre-run for 30 min at room temperature. The gel was run at 100 V for 50 min at room temperature, dried, and analyzed by autoradiography.

RESULTS AND DISCUSSION
Details of Overall Topology-We have solved the crystal structure of human DUE-B to 2.0 Å resolution using non-crystallographic symmetry averaging of phases determined from a mercurial derivative (Table 1). Although mass spectrometric analysis of dissolved crystals of DUE-B documents that crystals contain the full-length polypeptide encompassing all 209 residues, interpretable electron density is only visible for the first 151 amino acids. Predictive analysis and limited proteolysis studies (see below) are consistent with the view that the final 58 amino acid residues are unstructured in the absence of additional ligands.
The N-terminal 151 residues of DUE-B fold into a compact domain of the ␣/␤ class of proteins (Fig. 1a). Residues Lys-2 through Val-15 form a long, extended ␤ strand (␤1) that is sharply bent at residue Thr-9. Two ␤ strands, encompassing residues Glu-18 through Ile-23 (␤2) and residues Gly-26 through Gly-32 (␤3), form the core of the ␤-sheet structure. Residues Gln-39 through Asn-51 form helix ␣1, followed by the next strand (␤4) of the sheet structure created by Glu-73 through Ser-78. This is followed by a second long helix (␣2) encompassing residues Thr-99 through Thr-116. Residues Ile-122 through Asp-124 form a short ␤ strand (␤5). The long, final ␤ strand (␤6), created by residues Met-131 through Glu-146, with a kink at residue Gly-139, completes the structure of this domain of DUE-B (see Fig. 2 for a secondary structure annotation of the DUE-B monomer).
Solution studies show that DUE-B is a homodimer (29), and this dimeric interaction is observed in the crystal structure. Within the asymmetric unit, there are four copies of the DUE-B monomer. Pairs of monomers form tight interacting dimer, yielding two copies of the DUE-B homodimer in the asymmetric unit. Each homodimer buries a total of 2295 Å 2 of surface area. The dimer interface is created mainly by intermolecular interactions between strand ␤6 residues Met-131 through Gly-139 from each monomer to create an antiparallel ␤-sheet structure that extends across the homodimer interface (Fig. 1b).
The primary sequence of human DUE-B shows strong evolutionary conservation of sequences found in higher and lower eukaryotes and in bacteria (Fig. 2). The topological arrangement of the DUE-B monomer bears many of the structural elements of polypeptides that contain nucleotide binding motifs (Fig. 3). Structure-based comparison of DUE-B with the ATPbinding protein MJ0577 (1MJH) (39) yields a root mean square deviation of 3.7 Å over 79 residues (Fig. 3b). A structural comparison with liver alcohol dehydrogenase demonstrates that the DUE-B N-terminal domain structure is similar to half of the canonical Rossman fold (Fig. 3c). A superposition of 62 ␣ carbon atoms common to DUE-B and alcohol dehydrogenase (1A4U) (40) yields a root mean square value of 3.7 Å. Given the observation that recombinant DUE-B possesses the ability to hydrolyze ATP (29), it is quite likely that the structural similarity between the N-terminal domain of DUE-B and these proteins extends to its function as a nucleotide binding motif, as illustrated below.
Stability of the DUE-B Dimer-DUE-B isolated from Xenopus egg extract (xDUE-B), HeLa nuclei, or human recombinant DUE-B expressed in insect cells (rDUE-B) is in the form of a homodimer (29). When rDUE-B was mixed with Xenopus egg extract and reisolated by binding to Ni ϩ2 -NTA-agarose, rDUE-B did not co-isolate with xDUE-B (Fig. 4a), implying that the rDUE-B and xDUE-B dimers do not disproportionate in vitro. Similarly, rDUE-B did not dimerize with HeLa DUE-B when mixed in vitro with HeLa nuclear extracts (not shown). In contrast, when His 6 -tagged human DUE-B was expressed in HeLa cells and purified by Ni ϩ2 -NTA-agarose, the ectopic  and endogenous forms of DUE-B co-isolated in extracts from cells arrested in G 1 , S, or M phase of the cell cycle (Fig. 4b). These results suggest that the DUE-B heterodimers and, by extension, DUE-B homodimers formed in vivo are stable over the cell cycle.
The crystal structure of DUE-B suggests that the C-terminal portion of the protein is not involved in dimerization. To test this directly, a C-terminal truncation mutant missing amino acids 148 -209 was expressed in HeLa cells. As shown in Fig. 4c, the truncated form of DUE-B efficiently dimerized with the endogenous wild type protein. Taken together, these results show that DUE-B dimers are stable in vivo and in vitro and that the dynamic C-terminal region is not essential for dimerization.
Protease Sensitivity of the DUE-B C Terminus and DNA Binding-The C-terminal portion of DUE-B (amino acids 152-209) is not visible in the crystal structure and is predicted to be dynamically disordered (41)(42)(43). Trypsin or chymotrypsin digestion of His 6 -tagged human DUE-B (rDUE-B) purified from baculovirus-infected Sf9 insect cells produced a resistant core of ϳ17 kDa, close to the predicted size of the monomeric N-terminal domain visible in the DUE-B crystal (Fig.  5a). Protease digestion removed the C-terminal His 6 tag during production of the resistant core (Fig. 5b), and formic acid cleavage of fulllength DUE-B between Asp-156 and Pro-157, or formic acid treatment of trypsin-digested DUE-B, resulted in a product of closely similar size to that of the protease-resistant core (Fig. 5c), demonstrating that it is the C terminus that is removed by proteolysis.
Disordered regions of DNAbinding proteins often become ordered upon DNA binding (42). Incubation of rDUE-B with a 54-or 123-bp DNA fragment containing the DUE (29) resulted in the resistance of the full-length protein to protease digestion (Fig. 6a). Comparative time course digestions of DUE-B in the absence and presence of DNA (Fig. 6b) revealed a significant (4-to 5-fold) delay in susceptibility to digestion in the presence of the polynucleotide ligand. Consistent with the lack of sequence specificity of purified DUE-B binding (29), the DUE was not essential for resistance to trypsin digestion, as poly(dI-dC)⅐poly(dI-dC) fully protected DUE-B and mixed E. coli tRNAs moderately protected DUE-B against trypsin digestion (not shown). By contrast, a 54-nt singlestranded oligonucleotide containing one strand of the DUE or a 21-nt oligonucleotide from the DUE/ARS segment did not afford significant protection against digestion (Fig. 6c). Because DUE-B was isolated by virtue of its binding to the c-myc DUE in vivo and selectively binds a double-stranded DNA fragment containing the DUE in vitro in the presence of HeLa nuclear extract (29), the observation that double-stranded DNA preferentially protects the C terminus of DUE-B against protease suggests that in the presence of in vivo binding partners DUE-B may recognize some feature of incipiently unwound DNA prior to actual strand separation.
Purified full-length rDUE-B is able to bind double-stranded DNA without sequence specificity in an electrophoretic mobility shift assay (29). As shown in Fig. 6d, to test whether the C terminus of DUE-B is necessary for DNA binding, the electrophoretic mobility shift assay was repeated using the protease- resistant DUE-B core, the C-terminal truncation mutant, or the trypsin-treated C-terminal truncation mutant (which is resistant to trypsin digestion, as expected from the results of Fig. 5). Full-length rDUE-B was able to form a stable complex with the double-stranded DNA probe whereas neither the protease-resistant core nor the C-terminal truncation mutant bound to the probe, implying that the C terminus is necessary for binding of purified DUE-B to DNA.
Structural Similarities to tRNA-editing Enzymes-Human DUE-B shows strong evolutionary conservation with tRNAbinding proteins in yeasts and metazoans. The overall conformation of the N-terminal 151-amino acid domain of DUE-B bears significant similarity to the nucleotide binding Rossman fold found in aminoacyl-tRNA synthetases (44) and demonstrates significant homology to the structures of bacterial tRNA-editing enzymes, specifically to those of the E. coli D-Tyr-tRNA Tyr deacylase (30,31) and the archaea-specific editing domain of threonyl-tRNA synthetase (32) (Fig. 7, a-c). A superposition of the ␣ carbon atoms of DUE-B with those of the tRNA-editing enzymes yields a root mean square deviation value in the range of 1.8 to 2.2 Å. Lim et al. (31) have identified putative active site residues based on modeling studies and mechanistic inferences. All of the residues suggested to participate in the hydrolytic mechanism of the tRNA-editing enzymes are conserved within the structure of human DUE-B (Fig. 2).
This structural similarity to tRNA-editing enzymes is intriguing, given the biological evidence that DUE-B is involved in the initiation of DNA replication (28,29). Consideration of the proposed catalytic reactions carried out by the DUE-B-and tRNA-editing enzymes suggests that the structural homologies may likely extend to functional similarities. tRNA deacylases preserve protein homochirality by hydrolyzing the aminoacyl linkage between a misincorporated D-amino acid and the terminal adenosine in the CCA-3Ј tail of tRNA (45). The reaction involves direct nucleophilic attack on the carbonyl carbon of the D-amino acid (Fig. 8A). Hydrolysis of the aminoacyl-tRNA   (29) and reisolated using Ni ϩ2 -NTA beads. An aliquot of the input mixture and the bound and unbound fractions were electrophoresed by SDS-PAGE and Western-blotted with anti-DUE-B antiserum. b, His 6 -tagged rDUE-B was expressed in HeLa cells and reisolated using Ni ϩ2 -NTA beads from cells in asynchronous growth (ASYN) or cells arrested in G 1 with mimosine (MIM), S phase with hydroxyurea (HU) or aphidicolin (APH), or in mitosis with nocodazole (NOC). Protein bound to the Ni ϩ2 -NTA beads was electrophoresed by SDS-PAGE and analyzed by Western blotting with anti-DUE-B antiserum. c, the C-terminal DUE-B truncation mutant was expressed in HeLa cells and reisolated using Ni ϩ2 -NTA beads from cells in asynchronous growth. The bound protein was electrophoresed by SDS-PAGE in parallel with whole cell extract (wce) and trypsin-digested DUE-B as a molecular weight marker. FIGURE 5. Disordered DUE-B C terminus. a, rDUE-B was digested with trypsin or chymotrypsin for the indicated times. SDS-PAGE gels were silverstained after electrophoresis. b, rDUE-B was digested with trypsin or V8 protease and Western blotted with anti-DUE-B or anti-His 6 antibodies. c, rDUE-B was digested with trypsin or cleaved between Asp-156 and Pro-157 with formic acid as indicated. Duplicate SDS-PAGE gels were silver-stained or Western-blotted with anti-DUE-B antibodies.
ester results in the formation of an acyl-enzyme intermediate, which is then hydrolyzed by an activated water molecule to complete the deacylation reaction (31). During the ATPase catalytic cycle of DUE-B, the adenine base of ATP is probably housed in a location similar to that in the tRNA-editing enzymes. The hydrolytic reaction likely proceeds with the nucleophilic attack of the Thr-81 hydroxyl on the ␥ phosphate of ATP (Fig. 8b).
A Magnesium Ion near the Putative Active Site-Enzymes that catalyze phosphoryl transfer reactions often contain divalent metal ions at the active site (46 -48). These metals are thought to facilitate the hydrolytic reaction in a number of different ways, including stabilization of the developing charge on the ␥-phosphoryl oxygens during the transition state, lowering the pK a of a nucleophilic water molecule, or selective stabilization of the substrate for in-line chemistry. There is a single magnesium ion located adjacent to the putative active site in all four molecules of DUE-B in the crystallographic asymmetric unit (Fig. 1). Enzyme residues including the O⑀1 of Gln-6 and the backbone carbonyls of Val-4 and Cys-28 (which are conserved in the bacterial forms of the protein) coordinate the metal ion. The coordination of this metal ion is weak, with coordination distances ranging from 2.6 to 2.9 Å. This weak coordination scheme is reminiscent of that observed in other enzymes that catalyze phosphoryl transfer reactions, such as the RNA-dependent RNA polymerases from poliovirus (49), rabbit hemorrhagic disease virus (50), and Norwalk virus (51). In each of these structures, the metal ions are weakly bound at the active site in the absence of nucleotide and/or substrate but associate tightly with octahedral coordination chemistry in the presence of nucleotide and/or substrate. We note that the presence of such divalent ions has not been observed in the structures of the homologous tRNA-editing enzymes (30 -32), suggesting that this may be a feature unique to DUE-B and its vertebrate homologs. Interestingly, the PPM family of eukaryotic protein serine/ threonine phosphatases requires divalent metal ions for catalytic activity (52). The 7-motif arrangement of invariant amino acids essential for catalysis in these phosphatases is reproduced in human DUE-B, suggesting that DUE-B may be a mixed function phosphohydrolase. The structure of the N-terminal domain of DUE-B reveals a notable similarity to that of tRNA-editing enzymes (30 -32). However, the presence of a divalent magnesium ion near the putative active site may also facilitate ATP hydrolysis. To test the enzymatic activities of human DUE-B and the role of the putative active site Thr-81 in catalysis, rDUE-B expressed in Sf9 cells was incubated with bacterial tRNA aminoacylated with D-aspartate. As shown in Fig. 9a, both the full-length and the C-terminal truncation mutant displayed D-aminoacyl-tRNA deacylase activity. By contrast, a T81A point mutant did not display this activity, confirming the intrinsic deacylase activity of the wild type protein and demonstrating the conservation of the active site threonine function (cf. Figs. 1, 7). rDUE-B was also able to hydrolyze the terminal phosphate from [␥-32 P]ATP in vitro (Fig. 9b), suggesting that the acquisition of Mg ϩ2 binding endowed the protein with ATPase activity. The T81A mutation, but not the C-terminal truncation, also eliminated the ATPase activity, indicating that both D-aminoacyl-tRNA deacylase and ATPase activities are properties of the N-terminalordered domain of the DUE-B. Based on co-crystal structures of the archaeal tRNA-editing domain of threonyl-tRNA synthetase bound to substrate analogs, Hussain et al. (53) propose an alternative mechanism for mis-acylated tRNA hydrolysis in the posttransfer complex where the terminal adenosine 2Ј OH acts as a base to promote nucleophilic attack on the carbonyl carbon of the amino acid. However, our proposed mechanism for aminoacyl-tRNA hydrolysis and ATP hydrolysis is consistent with our mutational studies that demonstrate a loss of each activity in the T81A mutant. These differences in the proposed mechanisms are not surprising given the low sequence identity between the two enzymes.
The presence of D-aminoacyl-tRNA deacylase activity in a protein involved in DNA replication is intriguing in view of the tight binding and stimulation of DNA polymerase ␣ enzymes from wheat embryos and HeLa cells by threonyl-tRNA synthetases (54,55) and the strong structural resemblance of accessory ␤ subunits of animal mitochondrial DNA polymerases to tRNA synthetase enzymes (56,57). Indeed, constituents of the aminoacyl-tRNA synthetase complex containing nine synthetases and three interacting multifunctional proteins (AIMPs) carry out diverse functions as cell cycle-related signal-  ing molecules. Of particular interest are p38/AIMP2, which is transported to the nucleus upon transforming growth factor ␤ treatment, where it binds and promotes ubiquitination of the FUSE-binding protein and down-regulates c-myc expression (58), and p18/AIMP3, which is translocated to the nucleus in response to DNA damage to activate the ATM/ATR (ataxia telangiectasis-mutated/ATM and Rad3-related) kinases and increase p53 levels (44). Thus, DUE-B appears to represent yet another homolog of an ancient tRNA-metabolizing enzyme that has evolved additional roles in DNA replication and cell cycle progression (29).