tRNA Modification by S -Adenosylmethionine:tRNA Ribosyltransferase-Isomerase ASSAY DEVELOPMENT AND CHARACTERIZATION OF THE RECOMBINANT ENZYME*

The enzyme S -adenosylmethionine:tRNA ribosyl-transferase-isomerase catalyzes the penultimate step in the biosynthesis of the hypermodified tRNA nucleoside queuosine (Q), an unprecedented ribosyl transfer from the cofactor S -adenosylmethionine (AdoMet) to a modi-fied-tRNA precursor to generate epoxyqueuosine (oQ). The complexity of the reaction makes it an especially interesting mechanistic problem, and as a foundation for detailed kinetic and mechanistic studies we have carried out the basic characterization of the enzyme. Importantly, to allow for the direct measurement of oQ formation, we have developed protocols for the preparation of homogeneous substrates; specifically, an overexpression system was constructed for tRNA Tyr in an E. coli queA deletion mutant to allow for the isolation of large quantities of substrate tRNA, and [U- ribosyl - 14 C]AdoMet was synthesized. The enzyme shows optimal activity at pH 8.7 in buffers containing various oxyanions, including acetate, carbonate, EDTA, and phosphate. Unexpectedly, the enzyme was inhibited by Mg 2 (cid:1) and Mn 2 (cid:1) in millimolar concentrations. The steady-state kinetic parameters were determined to be K m AdoMet (cid:2) 101.4 (cid:3) M , K m tRNA (cid:2) 1.5 (cid:3) M , and k I/ Bam HI sites of pBluescript II SK( to give the plasmid pBlue- TGT. structure of the tgt confirmed by sequencing and subcloned as an Nde I/ Bam HI fragment into the Nde I/ Bam HI sites of the expression vector pET-11a to give the expression plasmid pET-TGT.

The post-transcriptional processing of tRNA involves a number of functionally distinct events essential for tRNA maturation (1)(2)(3)(4). The phenomenon of nucleoside modification is perhaps the most remarkable of these events, imparting a rich structural diversity to tRNA not observed in other RNAs (4). Over 80 modified nucleosides have been characterized in tRNA (4), many of which are conserved across broad phylogenetic boundaries. In the main, the biochemical function of nucleoside modification in tRNA remains poorly understood, although roles from structural stabilization to modulation of translational fidelity have been revealed in a few cases (4).
The most structurally complex of these modified nucleosides is the hypermodified nucleoside queuosine (Q) 1 (Fig. 1), which contains a cyclopentenediol group appended to a (7-aminomethyl)-7-deazaguanine core structure and, in some mammalian tRNA, can be further modified by glycosylation with galactose or mannose (5). Queuosine and its derivatives occur exclusively at position 34 (the wobble position) in the anticodons of tRNAs coding for the amino acids asparagine, aspartic acid, histidine, and tyrosine (6). Each of these tRNAs possess the genetically encoded anticodon sequence GUN (where N represents any nucleotide). Although otherwise ubiquitous throughout eukarya and bacteria, queuosine is inexplicably absent from the tRNA of yeast and mycoplasma and is not found in the tRNA of archaea.
A definitive picture of the biochemical function or functions of queuosine has yet to emerge, but it has been correlated with eukaryotic cell development and proliferation (7)(8)(9)(10), neoplastic transformation (8,(11)(12)(13), tyrosine biosynthesis in animals (14), translational frameshifts essential to retroviral protein biosynthesis (15)(16)(17), and the ability of pathogenic bacteria to invade and proliferate in human tissue (18). Underlying most, if not all, of these phenomena is a role in modulating translational fidelity, consistent with the location of queuosine in the anticodon. However, evidence also exists that implicates queuosine (or the free base queuine) in signal transduction (19 -26), potentially by tRNA-independent mechanisms.
Although queuosine is present in both eukarya and bacteria, only bacteria are capable of de novo queuosine biosynthesis (Fig. 2). Eukaryotes utilize a salvage system and acquire queuosine as a nutrient factor and from intestinal flora (27) and insert queuine directly into the appropriate tRNA by the enzyme tRNA-guanine transglycosylase (TGT) (28). A related TGT is present in bacteria (29), but in this case the substrate is the queuosine precursor 7-aminomethyl-7-deazaguanine (preQ 1 ) (Fig. 2), which appears to be derived from GTP (30). Considerable effort has been directed over the past decade toward elucidating the catalytic mechanism (31,32) and RNA specificity of the bacterial TGT enzymes (33,34), and high resolution x-ray crystal structures have been solved for the enzyme from Zymomonas mobilis (35,36). The only other enzyme identified in the pathway is S-adenosylmethionine:tRNA ribosyltransferase-isomerase (QueA), which catalyzes the penultimate step in the de novo biosynthesis of queuosine, the formation of epoxyqueuosine (oQ) via the addition of an epoxycyclopentanediol ring to preQ 1 -modified tRNAs (Fig. 3). Remarkably, the epoxycyclopentanediol moiety of oQ originates from the ribosyl portion of S-adenosylmethionine (AdoMet) (37,38), the first example of the stoichiometric use of AdoMet as a "ribosyl" donor in an enzymatic reaction. The overall reaction involves the elimination of both methionine and adenine from AdoMet, the transfer of the ribosyl moiety to tRNA, and its rearrangement to form an epoxycarbocycle.
QueA has garnered considerable interest for its central role in the biosynthesis of queuosine, the unique utilization of AdoMet, and as a fascinating problem in fundamental mechanistic enzymology. The reaction catalyzed by QueA is unprecedented in biological systems, and the results of preliminary mechanistic studies are consistent with a catalytic mechanism involving sulfonium ylide and vinyl sulfonium intermediates (39), species never before implicated in an enzymatic reaction. However, in contrast to TGT, the lack of an easily accessible activity assay has stymied the basic characterization and quantitative analysis of this remarkable enzyme and precluded detailed kinetic studies. We report here a solution to this problem with protocols for the preparation of appropriate substrates, the development of a quantitative assay for the reaction with short RNA substrates, the elucidation of optimal reaction conditions, the measurement of the steadystate kinetic parameters, and a preliminary evaluation of the importance of RNA structural information for recognition and catalysis by QueA.

EXPERIMENTAL PROCEDURES
General-Buffers and salts of highest quality grade were purchased from Sigma unless otherwise noted. DTT, isopropyl ␤-D-thiogalactopyranoside, and ampicillin were from U.S. Biologicals. [U-14 C]glucose was from ICN (46 mCi/mmol), and [8-14 C]guanine was from Sigma (45.3 mCi/mmol). Bulk yeast tRNA type X and Escherichia coli tRNA tyrosine-specific I (HPLC standard) were from Sigma. Glutathione-Sepharose 4B, Sephadex G-25 (superfine, DNA grade), and nucleotides were purchased from Amersham Biosciences. POROS chromatography resins were from PerSeptive Biosystems. Benzoylated-napthoylated DEAE-cellulose was from Sigma, and NACS TM resin was from Invitrogen. Whatman GF-C disks were from Fisher. Plasmid Midi Kits and Ni 2ϩ -nitrilotriacetic acid-agarose were from Qiagen, and PUREscript RNA isolation kits were from Gentra Systems, Inc. Centriprep YM-30, Centricon YM-10, and Centricon YM-3 units were from Amicon. Dialysis was performed in Slide-A-Lyzer cassettes from Pierce. Oligonucleotides were from Operon Technologies and IDT. The plasmid pET-11a was from Novagen, pBluescript II SK(ϩ) was from Stratagene, and pTrc99B was from Amersham Biosciences. DNA from restriction digests and PCR was purified with GeneClean III (Bio 101) kits following the manufacturer's instructions. Buffers made for RNA work were prepared with diethyl pyrocarbonate-treated water and filtered through Cameo 25ES nitrocellulose filters from Osmonics, Inc. Protein concentrations were based on the Bradford dye-binding procedure (Bio-Rad). Cytoscint ES liquid scintillation mixture was from ICN.
Instrumentation-Centrifugation was performed with an Avanti J-20 XP centrifuge using JA-25.50 and JA-10 rotors, an LE-80K ultracentrifuge using a Ti-60 rotor, and a model TJ-6 swinging bucket centrifuge, all from Beckman. UV-visible spectrophotometry was performed with the Varian Cary 100 Bio. HPLC was carried out with a Hitachi system consisting of the LACHROM software, the D-7000 System Manager, an L-7100 pump, and an L-4500A diode array detector. Continuous elution PAGE was carried out with a Bio-Rad 491 Prep Cell. PCR and enzyme assays were carried out on the GeneAmp PCR System 2400 from PerkinElmer Life Sciences. Sonication was done with a model W-375 sonicator from Heat Systems-Ultrasonics, Inc. Protein and nucleic acid PAGE was carried out with the Mini Protean III from Bio-Rad. Radioactivity from enzyme assays was quantitated with a Beckman LS 6500 liquid scintillation counter. An Amersham Biosciences Typhoon 9200 variable mode imager with ImageQuant 5.2 software was used to measure and analyze radioactivity in gels and for fluorescence detection of ethidium bromide-or SybrGreen-stained nucleic acids in PAGE and agarose gels.
Bacterial Strains, Plasmids, and Enzymes-E. coli strain DH5␣ Ϫ ,m B Ϫ ) gal) was from Stratagene. The E. coli queA deletion mutant K12QueA (37) and plasmids pHH1 (40) and pGEX-QA (37) were generous gifts from Helga Kersten (Universitä t Erlangen). Restriction enzymes were from MBI Fermentas and New England Biolabs, Inc. T 4 kinase and T 4 ligase were from New England Biolabs. Pfu and Pfu Turbo polymerase were from Stratagene. Factor Xa was from Promega. 6-Phosphogluconate dehydrogenase type V from torula yeast was purchased from ICN as an ammonium sulfate suspension. L-Glutamate dehydrogenase type I from bovine liver, pyruvate kinase type II from rabbit muscle, glucose-6phosphate dehydrogenase type VII from bakers' yeast, and myokinase from rabbit muscle were from Sigma, also as ammonium sulfate suspensions. Phosphoriboisomerase from torula yeast and hexokinase type F-300 from bakers' yeast were from Sigma as lyophilized powders and were stored in 50% glycerol and the appropriate buffer at Ϫ20°C. Recombinant adenine phosphoribosyltransferase was overproduced from pQEAPT1 in E. coli strain B25 (B25/pQEAPT1; a generous gift from Dr. Milton Taylor, Indiana University). Expression and purification of adenine phosphoribosyltransferase was essentially as described (41,42). The enzyme was judged ϳ95% pure by SDS-PAGE analysis. Activity was measured as described (43). Recombinant S-adenosylmethionine synthetase from E. coli was overproduced from pK8 (44) in E. coli DH5␣ (pK8, a generous gift of Dr. G. Markham, Fox Chase Cancer Center). Expression and streptomycin sulfate and ammonium sulfate purification steps were as described (45). The enzyme was enriched to ϳ80% as judged by SDS-PAGE; activity was measured as described (45). Recombinant 5-phosphoribosyl-1-pyrophosphate synthetase from Salmonella typhimurium was overproduced from pBRS11R in E. coli DH5␣ (pBRS11R; a generous gift from Dr. Vern L. Schramm, Albert Einstein University, New York) and purified by ammonium sulfate fractionation. The enzyme was enriched to ϳ80% as judged by SDS-PAGE. Activity was assayed as described (46). Recombinant T 7 RNA polymerase was overproduced from pT7-911Q (a generous gift from Dr. Thomas E. Shrader, Albert Einstein University) as a His 6 fusion protein in E. coli DH5␣ and purified on Ni 2ϩ -nitrilotriacetic acid agarose (Qiagen). All recombinant enzymes were overproduced using standard procedures, and cell lysis was achieved with sonication to give cell-free extracts containing the relevant recombinant proteins.
Cloning and Construction of an Homologous Overexpression System for E. coli tgt-Template DNA for the PCR-based amplification of the tgt structural gene from E. coli was prepared by isolation of the 1.8-kb SalI/BamHI fragment from the E. coli genomic clone pHH1 (40). Amplification of tgt was carried out with Pfu polymerase directed by the primers tgtSTART (5Ј-CGGTCGACCATATGAAATTTGAACTGGA-CACCACCG-3Ј) and tgtSTOP (5Ј-GCGCAGGATCCTTATTAATATTA-ATCAACGTTCAAAGG-3Ј). Designed into tgtSTART were the restriction enzyme sites SalI (italics) and NdeI (underlined), the later as part of the translation initiation codon (boldface type) for the tgt gene, whereas tgtSTOP contained a BamHI site (underlined) downstream from the translation termination codon. The PCR program included an initial hold for 5 min at 95°C, followed by 10 cycles of 95°C for 2 min, 55°C for 2 min, and 75°C for 2.5 min, followed by a 7-min hold at 75°C. After amplification, the PCR product was gel-purified (0.8% agarose), subjected to restriction digest with SalI/BamHI, and ligated into the Purification of [U-ribosyl-14 C]ATP was carried out by HPLC as described by Lim et al. (48) with a semipreparative Hypersil 5 C18 column (250 ϫ 10 mm, 5 m). The fractions containing the purified ATP were pooled, lyophilized to dryness, and dissolved in water. Concentration was determined by UV spectroscopy (⑀ 260 ϭ 15.4 ϫ 10 3 M Ϫ1 cm Ϫ1 ). After correcting for the loss of 14 CO 2 , the radiochemical yield was typically 85-95%.
Synthesis of [U-ribosyl- 14 C]S-Adenosylmethionine-The purified [U-ribosyl- 14 C]ATP was used in an enzymatic reaction to produce [U-ribosyl- 14 C]AdoMet essentially as described by Park et al. (49). [U-ribosyl- 14 C]AdoMet was purified by HPLC using a semipreparative Hypersil 5 C18 column (250 ϫ 10 mm, 5 m). The column was developed under isocratic conditions of 20 mM ammonium acetate (pH 6.0) and 2% methanol at a flow rate of 3 ml/min, and the fractions containing [U-ribosyl- 14 C]AdoMet were pooled, lyophilized to dryness, and dissolved in 10 mM H 2 SO 4 . The radiochemical yields were typically Ͼ90%. Concentration was determined by UV spectroscopy (⑀ 260 ϭ 14.9 ϫ 10 3 M Ϫ1 cm Ϫ1 ). The purity of the [U-ribosyl-14 C]AdoMet was checked by comparing with authentic material (Sigma) by TLC and HPLC.
Construction of a Homologous Overexpression System for E. coli tRNA Tyr -A synthetic gene for E. coli tRNA Tyr was constructed from four oligonucleotides using standard procedures (50) such that the 5Јand 3Ј-ends of the gene could be ligated into the EcoRI and BamHI sites, respectively, of pTrc99B. The two oligonucleotides comprising the sense strand of the E. coli tRNA Tyr gene were 5Ј-AATTCGGTGGTGG-GGTTCCCGAGCGGCCAAAGGGAGCAGACTGT-3Ј (Tyr-S1) and 5Ј-A-AATCTGCCGTCATCGACTTCGAAGGTTCGAATCCTTCCCCCACCA-CCACTTATTAAG-3Ј (Tyr-S2). The two oligonucleotides comprising the antisense strand were 5Ј-GATCCTTAATAAGTGGTGGTGGGGGAAG-GATTCGAACCTTCGAAG-3Ј (Tyr-AS1) and 5Ј-TCGATCACGGCAGA-TTTACAGTCTGCTCCCTTTGGCCGCTCGGGAACCCCACCACCG-3Ј (Tyr-AS2). The structure of the tRNA Tyr gene of the resultant plasmid pTrc-Tyr was confirmed by sequencing.
In Vivo Overexpression of preQ 1 -tRNA Tyr -Production cultures of E. coli K12QueA/pTrc-Tyr were grown in LB/amp until the A 600 reached 0.9 -1.0, when overexpression of the tRNA Tyr gene was induced by the addition of isopropyl ␤-D-thiogalactopyranoside to a final concentration of 0.5 mM. To determine the appropriate harvesting times for optimal production of preQ 1 -tRNA Tyr , 3-ml aliquots were removed immediately before and 1, 2, 4, 8, and 13 h postinduction with isopropyl ␤-D-thiogalactopyranoside. The RNA was extracted from the samples using PUREscript RNA isolation kits (Gentra Systems, Inc.) according to the manufacturer's directions. The RNA samples and commercial E. coli tRNA Tyr were then analyzed by HPLC on W-POREX 5 C4 column (250 ϫ 4.6 mm; Phenomenex) using a modification of the mobile phase developed by Pearson et al. (51). For the large scale isolation of preQ 1 -tRNA Tyr , 500-ml cultures were allowed to grow 13 h postinduction when the cells were collected by centrifugation at 3000 ϫ g for 10 min and flash-frozen in liquid nitrogen. The cells were stored at Ϫ90°C until further use.
The acid guanidinium thiocyanate/phenol/chloroform method was used to lyse cells and extract ribonucleic acids (52). After washing the crude RNA pellet with 70% ethanol, the RNA was dissolved in diethyl pyrocarbonate-treated water, and an equal volume of 8 M lithium chloride was added. The solution was placed at Ϫ20°C for 2 h, followed by centrifugation at 20,000 ϫ g for 20 min to remove large RNA. To the supernatant, containing preQ 1 -tRNA Tyr , was added an equal volume of isopropyl alcohol, and the solution cooled at Ϫ20°C for 2 h. The precipitate was collected by centrifugation at 20,000 ϫ g for 20 min, washed with 70% ethanol, and dissolved in 3 mM sodium citrate (pH 6.3). The concentration of tRNA was determined by UVvisible spectroscopy using ⑀ 260 ϭ 15 units/mg/ml. Approximately 25% of the total tRNA was determined to be preQ 1 -tRNA Tyr by tRNAtyrosine synthetase activity assays (53).
Purification of preQ 1 -tRNA Tyr -Crude tRNA was initially fractionated by chromatography on benzoylated-napthoylated DEAE-cellulose according to Gillam et al. (54). The preQ 1 -tRNA Tyr -containing fractions, identified by QueA and tRNA-tyrosine synthetase activity assays, were pooled, and the tRNA was precipitated by the addition of isopropyl alcohol and cooling. The tRNA precipitate was washed twice with 70% ethanol and dialyzed against 10 mM NaOAc (pH 4.5), 10 mM MgCl 2 , and 0.2 M NaCl. The enriched preQ 1 -tRNA Tyr was further purified by chromatography on NACS TM resin according to Thompson et al. (55). The preQ 1 -tRNA Tyr -containing fractions were pooled, and the tRNA was precipitated by the addition of isopropyl alcohol and cooling. After washing with 70% ethanol, the purified preQ 1 -tRNA Tyr was dissolved in 3 mM citrate (pH 6.3).
To remove bound metal ions from the preQ 1 -tRNA Tyr , EDTA was added to a concentration of 5 mM, and the solution was dialyzed at 80°C for 10 min against 3000 volumes of 3 mM sodium citrate (pH 6.3), followed by dialysis at 6°C over 5 h with two buffer changes. The preQ 1 -tRNA Tyr was then precipitated by the addition of isopropyl alcohol and cooling. After washing with 70% ethanol, the purified preQ 1 -tRNA Tyr was dissolved in 3 mM citrate, pH 6.3. The concentration of preQ 1 -tRNA Tyr was determined by UV-visible spectroscopy using ⑀ 260 ϭ 816,000 M Ϫ1 cm Ϫ1 (determined on the World Wide Web at paris.chem. yale.edu/extinct.html). Based on tyrosine-tRNA synthetase and QueA activity assays the tRNA was determined to be Ն85% preQ 1 -tRNA Tyr (900 pmol/A 260 ).
Synthesis and Purification of Minihelix RNA-An RNA 17-mer corresponding to the E. coli tRNA Asn anticodon stem loop (5Ј-GCGGACU-GUUAAUCCGC-3Ј) was synthesized by in vitro transcription with recombinant T 7 RNA polymerase (56). After elimination of the DNA template with RNase-free DNase and phenol/chloroform/isoamyl alcohol extraction, the minihelix RNA was purified by continuous elution PAGE essentially as described by Cunningham et al. (57). The fractions containing the minihelix RNA (determined by urea-PAGE and TGT activity assays) were pooled and concentrated using a Centricon YM-3. RNA concentration was determined by absorbance at 260 nm with ⑀ ϭ 162,000 M Ϫ1 cm Ϫ1 and mass ϭ 5442.4 Da (determined at paris.chem. yale.edu/extinct.html). The minihelix RNA was judged Ն95% pure by urea-PAGE.
Insertion of preQ 1 into Minihelix RNA-An aliquot of TGT (75 g) was added to a 1-ml solution containing 20 mM HEPES (pH 7.5), 10 mM MgCl 2 , 50 mM KCl, 2 mM DTT, 37 M minihelix RNA, and 50 M [8-14 C]guanine. After 3 h at 37°C, the reaction was terminated by the addition of one-tenth volume of 2 M NaOAc (pH 4.0) and phenol/ chloroform/isoamyl alcohol. The aqueous phase was recovered and applied to a Quik-Sep column (Isolab Inc.) containing 2 ml of Sephadex G-25. The sample was centrifuged (1.5 min at 700 ϫ g) to separate the unreacted guanine from the minihelix RNA, the resin was washed with 3 mM citrate (pH 6.3), and the fractions containing the minihelix RNA were combined and concentrated using a Centricon YM-3 unit to give minihelix RNA with a specific activity of 2.8 mCi/mmol.
The base preQ 1 , synthesized as described (58,59), was inserted into the minihelix RNA by a TGT-catalyzed reaction under the above conditions with preQ 1 in a 12-fold excess over minihelix RNA. After work-up as above, the concentrated minihelix RNA contained essentially no radioactivity (specific activity Ͻ 0.005 mCi/mmol), indicating that Ͼ99.8% of the minihelix RNA contained preQ 1 .
Expression and Purification of GST-QueA Fusion Protein-Recombinant GST-QueA was overproduced in E. coli DH5␣/pGEX-QA using standard expression conditions. Affinity chromatography on glutathione-Sepharose 4B was carried out essentially as described previously (37).
Following affinity chromatography, the protein was loaded (1 mg/ run) onto a POROS HQ column (4.6 ϫ 100 mm) equilibrated with 20 mM Tris-HCl (pH 8.0), 10 mM MgCl 2 , 2 mM DTT, and 10% glycerol (buffer E). A series of linear gradients were developed from buffer E to 1 M KCl in buffer E (buffer F) at a flow rate of 4 ml/min, and fractions containing GST-QueA were pooled and dialyzed against 50 mM Tris-HCl (pH 8.0), 100 mM KCl, 1.0 mM DTT, and 10% glycerol. From SDS-PAGE analysis, the protein was judged to be Ͼ95% pure.
Cleavage of GST-QueA with Factor Xa Protease-A sample of the fusion protein was cleaved in a solution containing 50 mM Tris-HCl (pH 8.0), 100 mM KCl, 1 mM CaCl 2 , 4 mM DTT, Factor Xa (1.0 g/1.0 mg of fusion protein), and GST-QueA. The reaction was carried out overnight at room temperature, and QueA was purified by HPLC under the conditions listed above.
QueA Activity Assays-Routine assays of QueA activity were initially carried out in 100 mM Tris-HCl (pH 8.7), 50 mM KCl, 10 mM MgCl 2 , 0.5 mM DTT, 25 g of crude preQ 1 -tRNA Tyr , and 50 M [U-ribosyl- 14 C]AdoMet in a final volume of 50 l at 37°C. Reactions were terminated at the appropriate times by the addition of 3 volumes of cold 10% trichloroacetic acid, and the precipitated tRNA was collected on Whatman GF-B filters by vacuum filtration, rinsed with cold 5% trichloroacetic acid followed by ethanol, and dried, and the radioactivity was quantitated by liquid scintillation counting. After optimum conditions for enzyme activity were elucidated, routine assays were carried out in 100 mM glygly (pH 8.7), 100 mM EDTA, 100 mM KCl, 0.5 mM DTT, 3.75 M preQ 1 -tRNA Tyr , and 100 M [U-ribosyl-14 C]AdoMet in a final volume of 50 l at 37°C. Activity assays were initiated and terminated as described above.
QueA Activity Assays with Minihelix RNA-Reactions with preQ 1modified minihelix RNA were carried out in 100 mM glygly (pH 8.7), 100 mM EDTA (pH 8.7), 100 mM KCl, 0.5 mM DTT, with variable minihelix RNA, [U-ribosyl- 14 C]AdoMet and GST-QueA in a final volume of 50 l at 37°C. Reactions were terminated at the appropriate time by decreasing the pH to 6.5 with the addition of acetic acid and heating briefly at 95°C. The samples were loaded onto a Quik-Sep column containing 250 l of DEAE-cellulose equilibrated in 50 mM imidazole (pH 6.5). The columns were washed with 15 volumes of 50 mM imidazole (pH 6.5), and the minihelix RNA was eluted with 7 column volumes of 50 mM imidazole (pH 6.5) and 1 M NaCl. The unreacted AdoMet (wash) and minihelix RNA were both collected and counted.
Ionic Strength and pH Optima-The pH profile for QueA was carried out in 100 mM buffer, 50 mM KCl, 20 mM MgCl 2 , 0.5 mM DTT, 1.5 M preQ 1 -tRNA Tyr , 100 M [U-ribosyl-14 C]AdoMet, and 100 nM GST-QueA under initial velocity conditions. The pH profile was also carried out under the above conditions with 100 mM EDTA and no MgCl 2 . The buffers used were MES (pH 5.6 -6.6), MOPS (pH 6.7-7.7), Tris-HCl (pH 7.5-8.7), CHES (pH 8.7-9.3), and CAPS (pH 9.8 -11.0). Optimal ionic conditions were tested in 100 mM glygly (pH 8.7), 0.5 mM DTT, 1.5 M preQ 1 -tRNA Tyr , 100 M [U-ribosyl-14 C]AdoMet, and 100 nM GST-QueA under initial velocity conditions. Steady-state Kinetic Assays-The steady-state kinetic parameters for GST-QueA were determined from activity assays carried out in 100 mM glygly (pH 8.7), 100 mM EDTA (pH 8.7), 100 mM KCl, 0.5 mM DTT, with variable preQ 1 -tRNA Tyr (or minihelix RNA) and [U-ribosyl- 14 C]AdoMet in a final volume of 50 l at 37°C. GST-QueA had a final concentration of 200 nM with variable AdoMet and saturating preQ 1 -tRNA Tyr (15 M) and 50 nM with variable preQ 1 -tRNA Tyr and saturating AdoMet (1 mM). Reactions with preQ 1 -tRNA Tyr were terminated by the stepwise addition of carrier yeast tRNA Type X (to a final amount of 20 g/assay) and 3 volumes of cold 10% trichloroacetic acid and worked up as described above. Reactions with preQ 1modified minihelix RNA were terminated and worked up as described above for assaying minihelix RNA substrates. Time course assays (up to 20 min) were performed in order to determine initial velocity conditions using nonlinear regression of dpm versus time. Michaelis-Menten parameters were calculated from the average of minimally

Characterization of Recombinant QueA
four replicates by nonlinear regression analysis of the initial velocity versus substrate concentration using Kaleidagraph 3.0 (Synergy Software, Reading, PA).

RESULTS
Preparation of Substrates-To prepare AdoMet with specific radiochemical labeling in the ribosyl moiety, ATP was first synthesized from glucose using the methodology of Schramm and co-workers (47), followed by the enzymatic conversion of ATP to AdoMet by S-adenosylmethionine synthetase as described by Park et al. (49). Of the nine enzymes required for the conversion of glucose to ATP, seven are commercially available. 5-Phosphoribosyl-1-pyrophosphate synthetase and adenine phosphoribosyltransferase, the two enzymes not commercially available, were overproduced in E. coli and purified as described (41,42) for use in the synthesis. Thus, starting with [U-14 C]glucose, [U-ribosyl-14 C]ATP was synthesized enzymatically; purification of the ATP by reversed-phase HPLC (Fig. 4) provided [U-ribosyl- 14 C]ATP in radiochemical yields of Ͼ85% after accounting for the loss of 14 CO 2 from C1 of glucose.
Recombinant E. coli S-adenosylmethionine synthetase was overproduced in E. coli from pK8 and purified through the ammonium sulfate step as previously described (45). S-adeno-sylmethionine synthetase-catalyzed formation of [U-ribosyl- 14 C]AdoMet from [U-ribosyl-14 C]ATP and methionine was carried out in reactions containing 20% acetonitrile, conditions that relieve the severe product inhibition observed with Sadenosylmethionine synthetase (49,60), and the crude [U-ribosyl- 14 C]AdoMet was subjected to reversed-phase HPLC (Fig.  4) to provide pure [U-ribosyl-14 C]AdoMet in radiochemical yields of Ͼ90%. The purified AdoMet was subsequently stored at Ϫ90°C with sulfate as the counter ion (pH Ͻ1.5) to maximize chemical stability and minimize epimerization at the sulfur atom (61,62).
In order to prepare significant quantities of substrate tRNA, an overexpression system for tRNA Tyr was constructed by synthesizing the gene for E. coli tRNA Tyr (I) and ligating it into the pTrc99B vector, which had previously been shown to be an efficient vehicle for the overexpression of E. coli tRNA Asp (63), to give the expression plasmid pTrc-Tyr. After confirming the structure of the gene through sequencing, pTrc-Tyr was expressed in K12QueA, an E. coli queA deletion mutant (37) that produces tRNA containing preQ 1 at position 34 instead of queuosine. HPLC analysis of crude tRNA isolated from K12QueA/ pTrc-Tyr cells at various postinduction times indicated that maximal tRNA Tyr was present ϳ13 h postinduction (data not shown).
Crude tRNA was isolated using the acid guanidinium thiocyanate/phenol/chloroform protocol (52), and based on tyrosine-tRNA synthetase assays, ϳ25% of the crude tRNA was determined to be tRNA Tyr , whereas tRNA from the untransformed strain comprised only ϳ3% tRNA Tyr (data not shown). The crude tRNA was fractionated first by mixed mode chromatography on benzoylated-napthoylated DEAE-cellulose as previously described (54,64), with tRNA Tyr eluting at 1 M NaCl and 10% ethanol (data not shown). The tRNA Tyrenriched fractions were further purified on NACS TM resin (65), with tRNA Tyr eluting at 800 mM NaCl (data not shown). Because we anticipated screening various metal ions for their effect on QueA activity, the purified preQ 1 -tRNA Tyr was dialyzed at high temperature (80°C) in citrate buffer containing EDTA to ensure complete removal of all bound metal ions.
A minihelix RNA corresponding to the anticodon stem-loop of E. coli tRNA Asn was synthesized via in vitro transcription using recombinant T 7 RNA polymerase, and the minihelix RNA was purified by continuous flow electrophoresis (data not shown) (57,66). Purification in this way provided the desired 17-mer in large amounts free of all other transcription products as determined from analytical PAGE (data not shown).
To quantify the amount of preQ 1 incorporated into the minihelix RNA by recombinant E. coli TGT, [8-14 C]guanine was first incorporated into the RNA using TGT (67). After careful measurement of the specific radioactivity of the labeled RNA, the [8-14 C]guanine was subsequently eliminated by the TGTcatalyzed incorporation of preQ 1 . Based on comparisons of specific radioactivity, greater than 99.8% of the [8-14 C]guanine present in the minihelix RNA was replaced by preQ 1 .
Effect of GST on the Catalytic Activity of QueA-Previous investigations of QueA activity had employed only the GST-QueA fusion protein (37-39), but it was not known whether the GST domain compromised catalytic activity. To resolve this issue, the activities of GST-QueA and the cleaved QueA were measured and compared under identical conditions; both QueA and GST-QueA had identical molar specific activities, 46.9 Ϯ 7.0 and 46.7 Ϯ 6.3 mmol min Ϫ1 mol Ϫ1 enzyme, respectively.
General Properties of Recombinant GST-QueA-The pH profile of GST-QueA exhibited a bell curve with activity between FIG. 4. A, purification of [U-ribosyl-14 C]ATP. Purification was carried out by mixed mode HPLC using a semipreparative Hypersil 5 C18 column (250 ϫ 10 mm, 5 m; Phenomenex). The ATP eluted under isocratic conditions of 83.3 mM triethylammonium acetate (pH 6.0) and 6% methanol at 4 ml/min. The peak at ϳ10 min is due to adenine, the peak at ϳ12 min is NADPH, and ATP is at ϳ18 min. B, purification of [U-ribosyl-14 C]AdoMet. Purification was carried out by reversed-phase HPLC using the above column with elution under isocratic conditions of 20 mM ammonium acetate (pH 6.0) and 2% methanol. [ 14 C]AdoMet elutes at ϳ13 min, and the small amount of residual [ 14 C]ATP elutes at ϳ5 min. pH 7.5 and 10.5 (Fig. 5), suggesting the involvement of acidbase catalysis, where a minimum of two ionizable residues are required for protonation/deprotonation steps. Optimal activity was observed at pH 8.7 and thus was the pH used for subsequent characterization studies.
Monovalent ions gave only a slight change in activity, with the optimal concentration at 100 mM of KCl or NaCl (Table I).
Interestingly the presence of Mg 2ϩ or Mn 2ϩ in millimolar concentrations inhibited QueA activity but had little or no effect in the submillimolar range (Fig. 6). In contrast, the presence of Ca 2ϩ in millimolar concentrations resulted in a modest stimulation of activity. Unexpectedly, the presence of several anions significantly increased activity (Table I), with optimal activity occurring in the presence of the oxyanions EDTA (100 mM), acetate (750 mM), carbonate (25 mM), or phosphate (25 mM). This phenomenon was also observed with genomic DNA, where activity was highest in the presence of 200 g/ml DNA. Furthermore, replacing Tris buffer with glygly increased activity another 2-fold (Table I). Other anionic species, including fluoride and thiocyanate, were more moderate activators, whereas sulfate exhibited comparably little activation of the enzyme. Remarkably, activation of QueA was completely eliminated in the presence of millimolar concentrations of magnesium (Fig.  6). Overall there was an approximate 10-fold difference in the specific activity of QueA when assaying activity in Tris buffer compared with glygly and one of the oxyanionic activators ( Table I).
Development of a New Activity Assay Appropriate for Small RNA Substrates-The failure of small RNAs to precipitate quantitatively with trichloroacetic acid required the development of an alternate assay for the analysis of small RNA substrates of QueA. A number of methods that utilized selective binding of either AdoMet or RNA to chromatographic matrices were investigated, and good results were obtained with RNA binding to DEAE-cellulose. Since this was also an inexpensive matrix, protocols were optimized for this assay and involved terminating reactions by decreasing the pH to 6.5 with the addition of acetic acid, briefly heating to 95°C, and loading the assay solutions into Quik-Sep columns containing 250 l of DEAE-cellulose equilibrated in 50 mM imidazole (pH 6.5). The unreacted AdoMet was collected by eluting the columns with 50 mM imidazole (pH 6.5), and the RNA was subsequently collected by eluting with 50 mM im-idazole (pH 6.5) and 1.0 M NaCl. The assay methodology was validated by carrying out replicate assays with preQ 1 -tRNA Tyr and processing one set with the trichloroacetic acid protocol and the other with the DEAE columns; initial velocity data from both series were indistinguishable (data not shown).
Steady-state Kinetics-QueA activity assays were performed under initial velocity conditions at variable substrate concentrations (with saturating co-substrate) to obtain steady-state kinetic parameters for the enzyme-catalyzed reaction. Initial velocities determined from incubations with saturating preQ 1 -tRNA Tyr (15 M) and variable AdoMet gave a K m of 104 Ϯ 8.6 M for AdoMet and a k cat of 2.3 Ϯ 0.1 min Ϫ1 (Fig. 7), whereas incubations with saturating AdoMet (1 mM) and variable preQ 1 -tRNA Tyr provided a K m of 1.5 Ϯ 0.1 M for preQ 1 -tRNA-Tyr (Fig. 7). When the preQ 1 -modified minihelix RNA was the variable substrate, the K m was determined to be 37.6 Ϯ 4.5 M, with a k cat of 14.7 Ϯ 0.7 min Ϫ1 . The kinetic parameters are collected in Table II.  a Reactions were carried out in 100 mM gly-gly (pH 8.7), 0.5 mM DTT, 100 M AdoMet, 1 M tRNA Tyr , and 100 nM GST-QueA plus added chemical.
c Reaction was carried out in 100 mM gly-gly (pH 8.

DISCUSSION
The enzyme QueA catalyzes the penultimate step in the biosynthesis of the hypermodified tRNA nucleoside queuosine (Fig. 3), a complex reaction involving the loss of adenine and methionine from the co-factor AdoMet, and the transfer and rearrangement of the ribosyl moiety to tRNA in the formation an epoxycarbocycle. Although the enzyme has been known for almost a decade, the absence of a commercial source of AdoMet with specific radiochemical labeling in the ribosyl moiety rendered quantitative assays of QueA activity via the direct meas-urement of epoxyqueuosine problematic; previous studies utilized assays in which QueA activity was observed either indirectly by the inability of TGT to insert [ 3 H]guanine into tRNA after reaction of QueA with AdoMet (37) or directly through the incorporation of tritium into tRNA from the in situ generation of [2,5Ј,8-3 H]AdoMet from [2,5Ј,8-3 H]ATP in a coupled S-adenosylmethionine synthetase/QueA reaction (38). Whereas the latter assay allowed for the confirmation of enzymatic activity from the direct observation of epoxyqueuosine formation, the assay was inefficient (ϳ84% of the 3 H is in a position not incorporated into epoxyqueuosine) and not amenable to the acquisition of accurate velocity data, since the concentration of AdoMet remains unknown when it is generated in situ. The need to acquire quantitative initial velocity data therefore necessitated the synthesis of appropriately labeled AdoMet.
With the preparation of homogeneous substrate tRNA, preQ 1 -modified minihelix RNA, and ribose-labeled AdoMet as described above, it was possible to carry out quantitative measurements of QueA activity via the direct measurement of epoxyqueuosine formation. The observation of identical molar specific activities for the GST-QueA fusion protein and the cleaved QueA suggests that the GST domain does not compromise enzyme activity. However, because the QueA isolated after Factor Xa cleavage still possesses an additional 7 Nterminal amino acids due to the structure of the fusion protein (37), it can be argued that the cleaved QueA is inappropriate for measuring native enzyme activity. To partially address this, we subcloned the E. coli queA gene into a pET expression vector and overproduced the wild-type QueA; based on the comparable levels of enzymatic activity and levels of recombinant enzyme measured in cell-free extracts containing GST-QueA or wild-type QueA, we conclude that the presence of the N-terminal fusion does not compromise enzyme activity.
The pH at which optimal QueA activity was observed (pH 8.7), while high, is similar to the optimal pH reported for TGT activity (32) and is consistent with the need to deprotonate the 5Ј-carbon during the course of the reaction. Although there is no obvious need for Mg 2ϩ or Mn 2ϩ in the QueA-catalyzed reaction, the inhibition observed with these metals was surprising, given the ubiquity and importance of Mg 2ϩ to tRNA tertiary structure (68 -70). Likewise, the activation observed in the presence of certain oxyanions was unexpected, particularly at the concentrations seen (Table I).
We have obtained preliminary evidence that an enzyme-AdoMet complex can form in the presence of MgCl 2 (detected by native PAGE), whereas in the absence of MgCl 2 no significant binding of AdoMet to free enzyme occurs (data not shown). Given that we have recently determined that the enzyme follows an ordered sequential kinetic mechanism in which tRNA binds first, followed by AdoMet, 2 these observations are consistent with Mg 2ϩ -dependent binding of AdoMet to free enzyme to form a dead end complex. Although the basis for the activation of QueA remains unclear, the inhibition pattern exhibited by Mg 2ϩ , unchanged in the presence or absence of activating anions, suggests that inhibition and 2 S. G. Van Lanen and D. Iwata-Reuyl, manuscript in preparation. activation are mechanistically distinct phenomena (Fig. 6). Further experiments will clearly be required to clarify the molecular basis of these phenomena; nevertheless, the characterization of these phenomena has allowed us to develop optimal conditions for analyzing enzyme activity.
Enzyme activity assays based on the precipitation of tRNA with trichloroacetic acid are widespread (72,73), but this protocol is not suitable when small RNAs are employed as substrates because they are refractory to trichloroacetic acid precipitation. Although other methods have been developed for small RNAs (33,34), we sought a protocol that would give us the flexibility of measuring both product formation as well as substrate consumption, so we investigated column-based ion exchange methods that allowed for the separate collection of product and unreacted substrate. The binding of RNA to DEAE-cellulose proved effective for resolving AdoMet from both small RNA and full-length tRNA and provided a simple and efficient protocol for processing activity assays.
QueA exhibits classic hyperbolic Michaelis-Menten kinetic behavior (Fig. 7), with both the tRNA and minihelix RNA serving as efficient substrates for the enzyme (k cat /K m ϭ 1.7 and 0.4 M Ϫ1 min Ϫ1 , respectively). Although the K m for minihelix RNA is 25-fold higher than for tRNA, this is almost completely offset by the larger k cat , such that k cat /K m for the two differ by ϳ4-fold. The similar values of k cat /K m indicate that all of the RNA structural elements required by QueA for recognition and catalysis are present in the anticodon stemloop. Indeed, given that the sequence of the minihelix RNA differs from the anticodon stem-loop of tRNA Tyr at four positions (including the presence of 2-methylthio-N 6 -isopentyladenosine at position 37 in tRNA Tyr ), the differences in the kinetic parameters may be due to these sequence differences, and there may be no RNA contacts outside of the anticodon region at all. This is consistent with previous footprinting experiments (71) and is perhaps not surprising, given that QueA catalysis is peripheral to the polynucleotide backbone, involving a preexisting modified base not present elsewhere in the tRNA. This can be contrasted with TGT catalysis, the prior step in queuosine biosynthesis, where replacement of the canonical base requires reaction directly with the sugar-phosphate backbone and where RNA structural information is clearly essential to ensure that reaction occurs only at the wobble position of the appropriate codons. However, even in this case the decrease in k cat /K m was only 30 -40-fold (33,34), demonstrating that the information content of the anticodon stem-loop is sufficient for achieving the requisite specificity. Thus, perhaps the more relevant question regarding QueA recognition and catalysis is not which part of the tRNA is required but whether any part is required.