Regulation of Substrate Recognition by the MiaA tRNA Prenyltransferase Modification Enzyme of Escherichia coliK-12*

We purified polyhistidine (His6)-tagged and native Escherichia coliMiaA tRNA prenyltransferase, which uses dimethylallyl diphosphate (DMAPP) to isopentenylate A residues adjacent to the anticodons of most tRNA species that read codons starting with U residues. Kinetic and binding studies of purified MiaA were performed with several substrates, including synthetic wild-type tRNAPhe, the anticodon stem-loop (ACSLPhe) of tRNAPhe, and bulk tRNA isolated from a miaA mutant. Gel filtration shift and steady-state kinetic determinations showed that affinity-purified MiaA had the same properties as native MiaA and was completely active for tRNAPhe binding. MiaA had aK m app (tRNA substrates) ≈3 nm, which is orders of magnitude lower than that of other purified tRNA modification enzymes, aK m app (DMAPP) = 632 nm, and ak cat app = 0.44 s−1. MiaA activity was minimally affected by other modifications or nonsubstrate tRNA species present in bulk tRNA isolated from a miaA mutant. MiaA modified ACSLPhe with ak cat app/K m appsubstrate specificity about 17-fold lower than that for intact tRNAPhe, mostly due to a decrease in apparent substrate binding affinity. Quantitative Western immunoblotting showed that MiaA is an abundant protein in exponentially growing bacteria (660 monomers per cell; 1.0 μm concentration) and is present in a catalytic excess. However, MiaA activity was strongly competitively inhibited for DMAPP by ATP and ADP (K i app = 0.06 μm), suggesting that MiaA activity is inhibited substantially in vivo and that DMAPP may bind to a conserved P-loop motif in this class of prenyltransferases. Band shift, filter binding, and gel filtration shift experiments support a model in which MiaA tRNA substrates are recognized by binding tightly to MiaA multimers possibly in a positively cooperative way (K d app ≈0.07 μm).

(DMAPP) ‫؍‬ 632 nM, and a k cat app ‫؍‬ 0.44 s ؊1 . MiaA activity was minimally affected by other modifications or nonsubstrate tRNA species present in bulk tRNA isolated from a miaA mutant. MiaA modified ACSL Phe with a k cat app /K m app substrate specificity about 17-fold lower than that for intact tRNA Phe , mostly due to a decrease in apparent substrate binding affinity. Quantitative Western immunoblotting showed that MiaA is an abundant protein in exponentially growing bacteria (660 monomers per cell; 1.0 M concentration) and is present in a catalytic excess. However, MiaA activity was strongly competitively inhibited for DMAPP by ATP and ADP (K i app ‫؍‬ 0.06 M), suggesting that MiaA activity is inhibited substantially in vivo and that DMAPP may bind to a conserved P-loop motif in this class of prenyltransferases. Band shift, filter binding, and gel filtration shift experiments support a model in which MiaA tRNA substrates are recognized by binding tightly to MiaA multimers possibly in a positively cooperative way (K d app Ϸ0.07 M).
The tRNA prenyltransferase (EC 2.5.1.8) encoded by the miaA gene of Escherichia coli catalyzes the addition of a ⌬ 2isopentenyl group from dimethylallyl diphosphate (DMAPP) 1 to the N 6 -nitrogen of adenosine adjacent to the anticodon at position 37 of 10 of 46 E. coli tRNA species (i 6 A37, see Fig. 1 and Refs. [1][2][3][4]. On the basis of a theoretical model of yeast tRNA Ser structure, the N 6 -nitrogen of A 37 in tRNA substrates of MiaA is recessed and points inward toward the center of the anticodon loop (5). In E. coli, the i 6 A37 tRNA modification is further methylthiolated by the action of the MiaB and MiaC enzyme activities to form ms 2 i 6 A37 (Fig. 1), except for tRNA Sec (6). ms 2 i 6 A37-modified tRNA species read codons starting with U residues and include tRNA Phe , tRNA Trp , tRNA Tyr (I and II), tRNA Cys , tRNA Leu (IV and V), and tRNA Ser (II and III) but not tRNA Ser (I and V) (7,8).
The E. coli MiaA prenyltransferase is an excellent model to study fundamental aspects of the modification process. Unlike many modifications, the function of ms 2 i 6 A37 in translation has been studied extensively in vivo and in vitro (reviewed in Refs. 3, 4, and 9). The ms 2 i 6 A37 tRNA modification is thought to stabilize tRNA-mRNA interactions by improving intrastrand stacking within tRNA anticodon loops and interstrand stacking between codons and anticodons (10,11). ms 2 i 6 A37 also seems to influence the conformation of the tRNA anticodon loop and thereby affect interstrand stacking interactions between wobble bases at position 34 in tRNA and bases immediately 3Ј to codons (4,10). Lack of ms 2 i 6 A37 in the tRNA of miaA mutants of E. coli and Salmonella typhimurium results in multiple defects in translation efficiency, codon context sensitivity, and fidelity (10 -21). These translation defects impart broadly pleiotropic phenotypes to miaA mutants that contain A37 instead of ms 2 i 6 A37 in their tRNA ( Fig. 1) (3,4), including decreased growth rate and yield (12,13), altered sensitivity to amino acid analogs (12), increased oxidation of certain amino acids, and tricarboxylic acid cycle intermediates (22), altered utilization of primary carbon sources (22), moderately increased GC 3 TA transversion frequency in nutritionally limited cells (19,23), suppression of Tet(M)-protein-induced tetracycline resistance (24), decreased DNA oxidation damage in exponentially growing cells, 2 and temperature sensitivity for colony formation at 45°C (25). Lack of ms 2 i 6 A37 does not seem to affect the aminoacylation of tRNA (21,26,27) or amino acid-tRNA-EFTu⅐GTP selection (4). Many of these miaA phenotypes may be caused by disruptions in translational control mechanisms, such as the ones that regulate expression of the tryptophan (trp) (28,29) and the tryptophanase (tna) (30,31) operons.
The E. coli miaA gene was cloned (23,32), sequenced (19), and found to be a member of a superoperon with unusual structure and complex modes of regulation (22,25,33). Homologs of E. coli miaA have also been sequenced in several other organisms (34,35). Notably, the miaA homolog of yeast, designated MOD5, has been developed into an important system to study subcellular localization of proteins (36). Over 20 years ago, Rosenbaum and Gefter (37) and Söll and co-workers (38) partially purified E. coli MiaA. These pioneering studies demonstrated the substrates and some of the conditions required for MiaA activity and established the sequential pathway for ms 2 i 6 A37 biosynthesis in tRNA ( Fig. 1; reviewed in Refs. 3 and 4). In this paper, we report rapid methods of MiaA purification and analyses of MiaA steady-state kinetics and tRNA substrate utilization and binding. We also present direct quantitation of the cellular amount of the MiaA tRNA modification enzyme. Our results show that MiaA substrate selection is complicated and likely regulated by several mechanisms.
Construction of Expression Vectors pTX439 and pTX440 -Cloning and other molecular biological procedures were done by standard methods (39), unless indicated otherwise. A his 6 -tag-miaA ϩ gene fusion under the control of the T7-phage promoter was constructed in vector pET-15b by ligating together the following three fragments: (i) a 5.7kilobase fragment of pET-15b digested with NdeI and BamHI; (ii) a 519-bp NdeI-MslI fragment generated by polymerase chain reaction with primers UMiaA and LMiaA (Table I) containing the amino-terminal segment of miaA with an NdeI site added at the miaA start codon; and (iii) a 579-bp MslI-BstYI fragment containing the carboxyl-terminal segment of miaA ϩ obtained from plasmid pTX312 (22). This strategy limited the length of polymerase chain reaction-amplified DNA fragments used in the construction. Ligation mixtures were transformed into strain JM105, and the desired construct, designated pTX439, was identified by restriction digestion patterns of purified plasmid DNA, DNA sequencing of the miaA segment generated by polymerase chain reaction amplification, and complementation of a miaA::Tn10 null mutation in E. coli strain DEV15 (14,19). pTX439 was transformed into E. coli strain HMS174 to give strain TX3371, in which the His 6 -MiaA protein was overexpressed (see Ref. 40).
The native MiaA protein was overexpressed from plasmid pTX440, which was constructed by ligating a 1.1-kilobase FspI-ScaI fragment containing the intact miaA ϩ gene from pTX312 (22) downstream of the P tac promoter in plasmid pKK223-3. Ligation mixtures were transformed into strain JM105 to give strain TX3367. The orientation of the miaA ϩ fragment in purified pTX440 plasmid was confirmed by restriction digestion patterns and DNA sequencing.
Overexpression and Purification of His 6 -MiaA-960-ml cultures of strain TX3371 were grown in LB medium containing 50 g of carbenicillin per ml and induced with 1 mM IPTG as described before (40). His 6 -MiaA protein was purified by one-step batchwise Ni 2ϩ -chelation affinity chromatography as described previously (40), except that the wash buffer contained 40 mM imidazole instead of 60 mM imidazole. Eluted His 6 -MiaA was desalted on PD-10 columns into 2 ϫ TMD buffer (60 mM Tris-HCl (pH 7.5 at 24°C), 20 mM MgCl 2 , and 2 mM dithiothreitol (DTT)). Similar MiaA concentrations were found using the Bradford and D C (Lowry) protein assays with bovine serum albumin (BSA) as the standard. After desalting, an equal volume of 72% (v/v) glycerol was added to the His 6 -MiaA preparation, and the protein solutions were divided into small volumes and stored at Ϫ70°C. His 6 -MiaA samples were thawed only once and used immediately.
The His 6 -tag was cleaved from 50 g of His 6 -MiaA by digesting at room temperature for 3 h with 1 unit of biotinylated thrombin in the buffer provided by the manufacturer. The resulting protein mixture was used immediately for enzyme assays and binding studies without further purification. His 6 -MiaA purity and the extent of thrombin cleavage were determined by SDS, 15% PAGE (5.6% stacking) as described before (41).
Purification of Native MiaA Protein by Mimetic Dye Chromatography-Six 400-ml cultures of E. coli strain TX3367 were grown in LB medium containing 50 g of ampicillin per ml at 37°C with shaking (300 rpm) until they reached a turbidity of 50 Klett (660 nm) units, upon which IPTG was added to a final concentration of 1 mM. Cultures were incubated at 37°C with shaking for 3 h longer, after which they were chilled on ice. All subsequent steps were carried out at 4°C, unless noted otherwise. Cells were collected by centrifugation (5000 ϫ g) for 15 min, and pellets were suspended in 300 ml of TMD buffer (30 mM Tris-HCl (pH 7.5 at 24°C), 10 mM MgCl 2 , and 1 mM DTT). Cells were collected again by centrifugation (5000 ϫ g) for 15 min, and pellets were suspended in 18 ml of TMD buffer. The cell suspension was passed twice through a chilled French pressure cell at 20,000 p.s.i. Cell lysates were centrifuged at 150,000 ϫ g for 60 min, and supernates were filtered through 0.22-m acetate filters (Micron Separations). 60 mg of protein extract was loaded by gravity flow onto each of the 10 different mimetic dye PIKSI-kit columns, which had been equilibrated with TMD buffer. Each column was washed by gravity flow with 5 ml of TMD buffer containing 0.05 M KCl. Proteins were eluted by two consecutive 5-ml washes of TMD buffer containing 0.2 M KCl followed by 0.8 M KCl. 100-l drops of the eluants from each column were placed on Type VS filter disks (0.025 m, Millipore), which had been floated on TMD buffer, and the eluants were dialyzed against TMD buffer at 4°C for 30 min. 15 g and 5 ng of the dialyzed eluants from each column were analyzed by SDS-PAGE and assayed for MiaA prenyltransferase activ-FIG. 1. Biosynthesis of i 6 A37 in tRNA molecules by the E. coli MiaA tRNA prenyltransferase. The MiaA substrate DMAPP is not derived in E. coli directly from mevalonic acid as it is in many other organisms (69). The i 6 A37 modified base is further methylthiolated to ms 2 i 6 A37 by the MiaB and MiaC activities, which are not yet characterized (3,6). SAM, S-adenosylmethionine; Cys, cysteine; SAH, S-adenosylhomocysteine.
ity (below), respectively. The samples eluted with 0.8 M KCl from the Mimetic Red2 A6XL and Mimetic Orange1 A6XL columns had the most MiaA activity and fewest other contaminating bands (data not shown). The eluant from the Mimetic Red column was pumped (0.3 ml per min) onto a room temperature Superose 12 HR 10/30 column, which was equilibrated and eluted with TMD buffer. Fractions were collected into chilled tubes and analyzed by SDS-PAGE and assayed for MiaA activity. The fractions that had a molecular mass of about 35 kDa and maximum MiaA enzyme activity were pooled, diluted with an equal volume of 72% (v/v) glycerol, distributed into small volumes, and stored at Ϫ70°C.
Construction of Plasmids for in Vitro Synthesis of tRNA Phe , Anticodon Stem-Loop of tRNA Phe (ACSL Phe ), and Mutant Variants-A synthetic E. coli pheU gene encoding wild-type tRNA Phe ( Fig. 2A, see "Results") was constructed by ligating together six oligonucleotides (tRNAphe1 to tRNAphe6; Table I) to form an upstream SalI restriction sequence, a T7-phage RNA polymerase promoter preceding the 76-base pair (bp) pheU gene, and downstream BstNI and BamHI restriction sites (42). Positive strand oligomers tRNAphe1, tRNAphe2, and tRNAphe3 and negative strand oligomers tRNAphe4, tRNAphe5, and tRNAphe6 were phosphorylated with T4-phage polynucleotide kinase and annealed together in T(0.1)E buffer (10 mM Tris-HCl (pH 8), 0.1 mM EDTA) by heating at 90°C for 5 min followed by slow cooling to room temperature. The annealed oligomers were ligated with an equal amount of pUC18 DNA that had been digested with SalI and BamHI, and the ligation mixture was transformed into strain JM105. The desired plasmid, designated pTX442, was identified by restriction digestion patterns, which were confirmed by DNA sequencing of the entire synthetic pheU gene.
A synthetic mutant pheU gene specifying tRNA Phe (U60C) ( Fig. 2A, see "Results") was constructed by replacing tRNAphe3 and tRNAphe4 with U60C-forward and U60C-reverse (Table I) in the above cloning strategy to give plasmid pTX476. A synthetic mutant pheU gene specifying tRNA Phe (A37G) ( Fig. 2A, see "Results") was constructed by replacing tRNAphe2 and tRNAphe5 with A37G-forward and A37G-reverse (Table I) to give plasmid pTX475. A synthetic gene specifying wild-type ACSL Phe (Fig. 2B, see "Results") was constructed by annealing together SL1 and SL2, which provided an upstream HindIII site, a T7-phage polymerase promoter abutting the 17-bp ACSL Phe gene, and downstream SmaI and BamHI sites. The two oligomers were phosphorylated and annealed as described above and ligated into pUC18 cut with HindIII and BamHI to give plasmid pTX539. A mutant variant, ACSL Phe (A11G) (Fig. 2B, see "Results"), was constructed in the same way by replacing SL1 and SL2 with SL-A37G1 and SL-A37G2 to give plasmid pTX540.
Digestion mixtures were extracted once with an equal volume of TM (30 mM Tris-HCl (pH 7.5 at 24°C) 10 mM MgCl 2 )-saturated phenol:chloroform (1:1) and four times with ethyl ether before being precipitated with ethanol (39). Linearized DNA templates were transcribed in vitro by T7-phage RNA polymerase provided in the Riboprobe or Ribomax kits according to the manufacturer's instructions. Transcripts were labeled with 32 P by adding 5 l of [␣-32 P]CTP per 100-l transcription reaction mixture. Transcription reaction mixtures were incubated at 37°C overnight. DNA templates were removed by digestion with RQ1 DNase (1 unit per 1 g of DNA) at 37°C for 1 h. Digestion mixtures were extracted once with an equal volume of TM-saturated phenol:chloroform (1:1) and four times with ethyl ether before precipitation with ethanol. Pellets were collected by centrifugation in a microcentrifuge, dried, and suspended in T(0.1)E buffer.
tRNA Phe , ACSL Phe , and mutant variants were further purified from nucleotides and aborted transcripts by DEAE high performance liquid chromatography. Resuspended mixtures were applied at a flow rate of 1 ml per min to a W-POREX DEAE column equilibrated with 20 mM sodium phosphate buffer (pH 6.5). RNA molecules were eluted with a linear 0.2 to 1 M NaCl gradient (ramp ϭ 60 min) in 20 mM sodium phosphate buffer (pH 6.5) (no urea) at room temperature. Intact tRNA Phe or ACSL Phe , which eluted at about 44 or 36 min, respectively, into the gradient, were pooled, precipitated with ethanol, and stored as dried pellets at Ϫ20°C. Before use, the purified tRNA Phe and ACSL Phe preparations were suspended in T(0.1)E buffer, heated at 90°C for 2 min, and cooled slowly to room temperature. Concentrations of the tRNA Phe and ACSL Phe molecules were determined by using A 260 extinction coefficients calculated from base compositions by the Oligo 4.0 program (National Biosciences). The yield from the Riboprobe or Ribomax kit was 7.5 g of tRNA Phe from 5 g of linearized DNA in a 100-l reaction mixture or 400 g of tRNA Phe from 20 g of linearized DNA in a 200-l reaction mixture, respectively. tRNA Phe (wt) preparations analyzed by urea-20%-PAGE lacked detectable contamination by an (n ϩ 1) tRNA Phe product, which contains an extra 3Ј-nucleotide (data not shown (43)).
Purification of Bulk tRNA from E. coli-Bulk tRNA was purified from 4 liters overnight LB cultures of strains NU398 (DEV15 miaA::Tn10) and NU394 (DEV15 miaA ϩ ) as described previously (23) with some modifications. Briefly, cultures were chilled and cells were collected by centrifugation (5000 ϫ g) for 10 min and suspended in 20 ml of cold TMD buffer. All remaining steps were performed at 4°C, unless noted otherwise. 20 ml of TM-saturated phenol was added to the suspended cells, and the mixture was agitated vigorously on a wrist shaker for 1 h. Lysed cells and phenol were removed by centrifugation (14,000 ϫ g) for 30 min. The aqueous phases were loaded by gravity flow onto separate low pressure DEAE-cellulose columns (2.5 ϫ 3 cm) equilibrated with TM buffer containing 0.02 M NaCl. The columns were washed with TM buffer ϩ 0.02 M NaCl at a flow rate of 2 ml per min for 110 min and then eluted with a linear 0.02 M to 1 M NaCl gradient (ramp ϭ 110 min) in TM buffer. Samples with A 260 Ͼ 1 were pooled, precipitated with ethanol, and stored as dry pellets at Ϫ20°C. For kinetic experiments, bulk tRNA was suspended in T(0.1)E buffer, heated at 90°C for 2 min, and cooled slowly to room temperature. Concentrations of bulk tRNA were determined from A 260 (extinction coefficient ϭ 40 g per A 260 unit). The yield of bulk tRNA was 15-25 mg per 10 g of wet cells.
Gel Filtration Shift Assays-Binding reaction mixtures (200 l) contained TMD buffer and 100 g of BSA per ml, usually 3. cold 10% (w/v) trichloroacetic acid. Precipitates were collected onto 25-mm Whatman GF/C filters, which were washed with 10 ml of cold 10% trichloroacetic acid and then 10 ml of cold 100% ethanol. The filters were dried for 10 min under an infrared heat lamp and counted in 3.5 ml of Ultima Gold XR scintillation mixture (Packard). Product formation was linear with time for at least 8 min at the highest and lowest concentrations of substrates used, and initial velocities were usually determined from 2-min reactions for intermediate substrate concentrations. ATP or ADP concentrations between 30 nM and 10 M were added to tRNA Phe excess reactions to test inhibition of MiaA for DMAPP. Kinetic parameters were calculated by using the Enzfitter nonlinear regression data analysis program (Biosoft).
As noted previously for partially purified preparations (37), the activity of purified His 6 -MiaA was strongly reduced in reaction mixtures at pH values below 6 and above 10, containing 50 mM NaPO 4 buffer (pH 7.5) instead of Tris-HCl (pH 7.5), and lacking reducing agent, such as DTT (data not shown) (37,38). Omission of BSA from reaction mixtures reduced His 6 -MiaA activity by 42%, and plots of (product formation) versus (enzyme concentration) ϫ (assay time) (44) indicated that BSA stabilized His 6 -MiaA in reaction mixtures at 37°C (data not shown). His 6 -MiaA activity was maximal in reaction mixtures containing DTT, BSA, and 10 mM MgCl 2 or 1 mM MnCl 2 , but was reduced to 53, 43, 6, or Ͻ1% when 10 mM MgCl 2 was replaced by 20 mM MgCl 2 , 10 mM MgSO 4 , 10 mM MnCl 2 , or 10 mM ZnSO 4 or 1 mM EDTA (data not shown). His 6 -MiaA (5 ng) diluted into TMD containing 100 g of BSA per ml was thermally stable at 24 to 37°C for at least 15 min but was rapidly inactivated by incubation at temperatures above 45°C (data not shown).
Quantitative Western Immunoblotting Blotting-We used purified His 6 -MiaA as an antigen for the production of anti-MiaA polyclonal antibodies in rabbits (see Ref. 41). E. coli strains TX2494 (CC104 miaA ϩ ) and TX2590 (CC104 miaA::⍀Km r ) were grown in 400 ml of Vogel-Bonner (1 ϫ E) minimal salts medium supplemented with 0.4% glucose and enriched with 0.5% acid casein hydrolysate (Difco) (25). Subsequent steps was carried out as described previously (45), except that His 6 -MiaA or thrombin-treated His 6 -MiaA were used as standards. Airdried immunoblots were scanned on a Hewlett-Packard ScanJet 4C scanner, and bands were quantitated by using SigmaScan software (Jandel).
Band Shift Assays-The binding reaction mixture (50 l) contained TMD and 100 g of BSA per ml. In protein excess titrations (46 -48), the concentration of 32 P-labeled tRNA Phe or ACSL Phe molecules was fixed at 0.8 or 3.6 nM, respectively, and the concentration of His 6 -MiaA protein was varied between 5.8 nM and 14.2 M. In ligand excess titrations (46 -48), the concentration of His 6 -MiaA protein was fixed in the range of 0.5 to 1.3 M, and the concentration of 32 P-labeled tRNA Phe or ACSL Phe molecules was varied from 0.1 to 1.6 M or 0.1 to 10 M, respectively. Binding reaction mixtures were incubated for 30 min at room temperature. RNA-protein complexes were resolved from free RNA molecules by electrophoresis through native 6% (w/v) polyacrylamide gels (29:1 acrylamide:bisacrylamide) containing 10 mM Tris acetate (pH 8.0), 0.1 mM EDTA, and 1 mM DTT. Gels were prerun at 200 V at 4°C for 2 h immediately before use. 30 l of 40% sucrose was added to samples before loading them onto gels. Gels were run at 200 V at 4°C for 2 h. Radioactive bands were visualized by autoradiography of dried gels and were quantitated by direct counting in an Instant Imager (Packard).
Nitrocellulose Filter Retention Assays-Triplicate binding reactions were carried out in 100 l of TMD containing 100 g of BSA per ml. In protein excess titrations, the concentration of 32 P-labeled tRNA Phe or ACSL Phe molecules was fixed at 20 pM or 1.8 nM, respectively, and the concentration of His 6 -MiaA protein was varied between 0.56 nM to 15.9 M. In some protein excess titrations, 100 g of BSA per ml was added to filter soaking and washing solutions (48). In ligand excess titrations, the concentration of His 6 -MiaA protein was fixed in the range from 1.8 to 2.2 M, and the concentration of 32 P-labeled tRNA Phe or ACSL Phe molecules was varied from 10 nM to 5.0 M. BSA was omitted from the filter soaking and washing solutions used for ligand excess titrations, because we found that the saturation level for MiaA⅐tRNA Phe complex formation was about 2-fold higher when BSA was omitted. Binding reaction mixtures were incubated for 30 min at room temperature and passed through soaked (TMD buffer for at least 1 h) nitrocellulose filters (13 diameter, 0.45 m pore size; Schleicher & Schuell) filters contained in a manifold (Hoefer Scientific) connected to house vacuum (15-20 in Hg). Filters were washed twice with 300 l of cold TMD buffer, dried briefly on the manifold, and counted in Ultima Gold XR scintillation mixture (Packard). The background dpm of control reactions lacking His 6 -MiaA was less than 10% of the input radioactivity and was subtracted in all cases. The retention efficiency of MiaA⅐tRNA Phe (wt) complexes ranged from about 65% to about 85% at saturation in protein excess titrations when BSA was added or omitted, respectively, from soaking and washing solutions. The retention efficiency of MiaA⅐ACSL Phe (A11G) complexes was about 40% at saturation in protein excess titrations when BSA was omitted from soaking and washing solutions. MiaA⅐ACSL Phe (wt) complexes were not detected by the filter binding assay.

RESULTS
Purification of MiaA Protein-We set up rapid methods to purify large amounts of active MiaA protein for kinetic and binding studies. We constructed an in-frame translation fusion between the His 6 -containing leader peptide encoded by vector pET15b and the presumed translation start codon of MiaA (see "Experimental Procedures" (19)). IPTG induction of the T7phage promoter driving the fusion caused 1,340-fold overexpression of His 6 -MiaA in crude extracts as measured by quantitative Western immunoblotting (Fig. 3, lanes 2 and 3; "Experimental Procedures"). Single-step, batchwise, metal ion affinity chromatography on activated Ni 2ϩ resin resulted in electrophoretically pure His 6 -MiaA (Fig. 3, lane 4). The yield of His 6 -MiaA from 960 ml of bacterial culture was 6 mg with a specific activity of 700 nmol of i 6 A formed per min per mg of protein using synthetic tRNA Phe (wt) as a substrate (see below, and see "Experimental Procedures"). The His 6 -affinity tag could be completely cleaved from the fusion protein by thrombin protease (Fig. 3, lane 5; "Experimental Procedures"). Gel filtration analyses indicated that the cleaved His 6 -affinity tag was likely released from the MiaA protein (below).
We also devised a two-step purification of small amounts of native MiaA to allow comparisons with the kinetic properties of   Fig. 4; see "Experimental Procedures") (45). As standards for these analyses, we added known amounts of purified thrombin-treated MiaA to crude extracts of a miaA null mutant (Fig. 4, lanes 4 -9). We then loaded amounts of crude extract from miaA ϩ cells so that the MiaA detected was within the linear range of the standards (Fig. 4, lanes 1-3). By this analysis, we found about 4.0 ng of MiaA protein was present in 100 g of extract of E. coli K-12 growing exponentially in enriched minimal salts/glucose medium at 37°C. This amount corresponds to about 660 monomers of MiaA per cell and a cellular MiaA concentration of about 1.0 M, where the volume of an E. coli cell was taken as 1.0 ϫ 10 Ϫ12 ml (45). The spreading of MiaA standard bands on gels (Fig. 4, lanes 5-9) was caused by salt from the thrombin cleavage reaction. Similar quantitative results were obtained when His 6 -MiaA instead of thrombintreated MiaA was used as a standard, in which case the standard bands were as tight as those from the wild-type extracts (data not shown). Finally, the cellular amount of MiaA dropped about 3-fold in cells in stationary phase (Fig. 4, lanes 1 and 2) compared with exponential phase (Fig. 4, lane 3).
Preparation of RNA Substrates-Seven RNA substrates were purified for this study of MiaA enzymology. We chose wild-type E. coli tRNA Phe (Fig. 2A) as a model synthetic substrate for MiaA, because it had been characterized previously by Uhlenbeck and co-workers (49) in their analyses of aminoacyl-tRNA Phe synthetase. We synthesized tRNA Phe (wt) in vitro by using T7-phage RNA polymerase and purified the tRNA Phe (wt) away from aborted transcripts and nucleotides by DEAE high performance liquid chromatography (see "Experimental Procedures"). The synthetic tRNA Phe (wt) differed from native tRNA Phe in that it contained a 5Ј-triphosphate instead of a 5Ј-monophosphate and was fully unmodified at all positions ( Fig. 2A).
We also constructed and synthesized four mutant variants of tRNA Phe (wt) and purified bulk tRNA from an E. coli miaA mutant and its miaA ϩ parent strain. Mutant tRNA Phe (U60C) (in which U at position 60 is replaced by C; Fig. 2A) is readily cleavable by lead ions if the tRNA molecule folds properly (49). tRNA Phe (U60C) was synthesized in case tRNA Phe (wt) was not fully available as a substrate for MiaA. Mutant tRNA Phe (A37G) ( Fig. 2A) was synthesized as a specificity control and was not expected to be isopentenylated by MiaA. The synthetic ACSL-Phe (wt) and its corresponding (A11G) mutant (Fig. 2B) were synthesized to test whether the determinants for i 6 A formation were contained in the ACSL Phe . High performance liquid chromatography analyses indicate that bulk tRNA isolated from miaA mutants seems to contain all RNA base modifications except for ms 2 i 6 A37 (23). Bulk tRNA from a miaA mutant also contains the majority of tRNA species that are not substrates for MiaA. tRNA isolated from the miaA ϩ strain should be fully modified, including with ms 2 i 6 A37, and was not expected to be a substrate for purified MiaA.
Binding Activity of His 6 -MiaA to Synthetic tRNA Phe (wt), AC-SL Phe (wt), and tRNA Phe (A37G)-We determined whether the purified His 6 -MiaA was fully active for binding to tRNA Phe (wt) and vice versa by using a gel filtration shift assay (Fig. 5). tRNA Phe (wt) and bulk tRNA from a miaA mutant ran anomalously with an apparent molecular mass of 78 kDa instead of 25 kDa on a calibrated Superose 12 HR 10/30 sizing column (Fig.  5A). Purified His 6 -MiaA and thrombin-treated MiaA (23 g) had apparent molecular masses of about 34 kDa (Fig. 5B) and 37 kDa, respectively, which approximated the 34-kDa monomer molecular mass predicted for MiaA from its amino acid sequence (19). The slightly lower mobility of His 6 -MiaA compared with thrombin-treated MiaA may have been caused by interaction between the His 6 -tag and the column matrix. A relatively large amount of crude extract (6 mg) from wild-type cells produced a single peak of MiaA activity with a molecular Addition of an equimolar amount of tRNA Phe (wt) to His 6 -MiaA caused complete disappearance of the 34-kDa His 6 -MiaA monomer peak, about a 50% reduction in the free tRNA Phe (wt) peak, and the appearance of a new 110-kDa leading peak containing the His 6 -MiaA⅐tRNA Phe (wt) complex (Fig. 5C). This result showed that the purified His 6 -MiaA was fully active for binding to tRNA Phe (wt). Similar results were obtained for thrombin-treated MiaA (data not shown). Addition of His 6 -MiaA in a 10-fold molar excess over tRNA Phe (wt) caused the free tRNA Phe (wt) peak to disappear completely with a corresponding increase in the 110-kDa peak containing the His 6 -MiaA⅐tRNA Phe (wt) complex (Fig. 5D). This result showed that the tRNA Phe (wt) substrate was fully capable of binding to the His 6 -MiaA enzyme. The above conclusions were confirmed by kinetic and quantitative binding studies (below). Gel filtration shift assays were also performed with mixtures of His 6 -MiaA and the ACSL Phe (wt) microhelix (Fig. 2B) or mutant tRNA Phe (A37G) (Fig. 2A). Free ACSL Phe (wt) ran anomalously with an apparent molecular mass of 25 kDa instead of 5 kDa on the Superose 12 HR 10/30 column (data not shown). All of the ACSL Phe (wt) could be shifted into a His 6 -MiaA⅐ACSL Phe (wt) complex with an apparent molecular mass of about 45 kDa (data not shown), which approximated the sum of the predicted monomer molecular masses of His 6 -MiaA (36 kDa) and ACSL Phe (wt) (5 kDa). The free His 6 -MiaA peak also disappeared completely from binding mixtures containing equimolar amounts of tRNA Phe (A37G) and His 6 -MiaA (Fig.  5E). However, the resulting complex had an apparent molecular mass of about 63 kDa, which approximated the sum of the predicted monomer molecular masses of His 6 -MiaA (36 kDa) and tRNA Phe (A37G) (25 kDa). Thus, on the basis of complex size, His 6 -MiaA bound to the ACSL Phe (wt) microhelix and mutant tRNA Phe (A37G) in an apparent 1:1 molar ratio, whereas His 6 -MiaA and thrombin-treated MiaA seemed to form a larger, comparatively stable 110-kDa complex with tRNA Phe (wt). The stoichiometry of tRNA Phe (wt) binding is considered below.
MiaA Enzyme Kinetics-We optimized an assay for determining initial rates of [ 3 H]DMA transfer to unlabeled RNA substrates at 24 and 37°C (see "Experimental Procedures"). [ 3 H]i 6 A-modified RNA was recovered by precipitation with trichloroacetic acid. Typical Lineweaver-Burk plots of His 6 -MiaA with tRNA Phe (wt) and DMAPP are shown in Fig. 6, A and B, respectively, and steady-state kinetic data for various RNA substrates and MiaA preparations are compiled in Table II. The apparent substrate inhibition of His 6 -MiaA by tRNA Phe (wt) (Fig. 6A) was also observed for bulk tRNA from a miaA mutant (data not shown). Since the synthetic tRNA Phe (wt) substrate and bulk tRNA were prepared by different methods (see "Experimental Procedures"), it seems unlikely that this substrate inhibition (Fig. 6A) was caused by a low level contaminant in the synthetic tRNA Phe (wt) preparations. tRNA Phe (A37G) and bulk tRNA from a miaA ϩ strain were not modified by MiaA, confirming the specificity of the in vitro i 6 A37 modification reaction (data not shown).
The K m app and k cat app for tRNA Phe (wt) were the same within experimental error for His 6 -MiaA, thrombin-treated MiaA, and native MiaA (Table II, lines 1, 5, and 6). Thus, the presence of the His 6 -tag did not appreciably affect the association state (above) or the kinetic properties (Table II) of the MiaA prenyltransferase. Consequently, His 6 -MiaA was used in most subsequent experiments and will be referred to simply as MiaA hereafter. tRNA Phe (wt) and tRNA Phe (U60C) showed equivalent kinetic properties (Table II,  MiaA had nearly the same K m app and k cat app for tRNA Phe (wt) and bulk tRNA isolated from a miaA mutant (Table II, lines 1 and  3). To make this comparison, the concentration of MiaA substrates was taken as 12.9% of all tRNA species in the bulk tRNA isolated from a miaA mutant grown in LB medium at a rate of 2.5 doublings per h (7).
The ACSL Phe (wt) microhelix (Fig. 2B) was also a substrate for the MiaA enzymes (Table II, line 8). These reactions were carried out at 24°C to prevent melting of the GC-rich ACSL-Phe (wt) (predicted T m Ϸ63°C from the Oligo 4.0 program (National Biosciences)). Control experiments showed that prolonged incubation with excess MiaA led to complete i 6 A modification of ACSL Phe (wt) and that mutant ACSL Phe (A11G) was not modified by MiaA (data not shown). The k cat app /K m app substrate specificity constant was about 17-fold lower for AC-SL Phe (wt) than for tRNA Phe (wt) due primarily to an 8-fold reduction in K m app (lines 7 and 8). Competitive Inhibition of MiaA Activity by Nucleotide Diand Triphosphates and by tRNA Phe (A37G)-MiaA homologs have been sequenced from several organisms and share an ATP/GTP P-loop binding motif (50). Hence, we checked whether ATP, ADP, and other nucleotide di-and triphosphates (NDPs and NTPs) affected MiaA enzyme activity in vitro. We found that MiaA activity was strongly inhibited by ADP (Fig.  7A) and ATP (Fig. 7B). The inhibition was classically compet-itive with respect to the DMAPP substrate with K i app (ATP) ϭ 0.07 M and K i app (ADP) ϭ 0.05 M. We tested other NTPs, such as GTP and CTP, and found similar inhibition as with ATP (data not shown). Thrombin-treated MiaA lacking the His 6 -tag was inhibited by ATP or ADP to the same extent as His 6 -MiaA (data not shown). Last, we found that mutant tRNA Phe (A37G) acts as a strong competitive inhibitor of i 6 A-37 modification in tRNA Phe (wt) (K i app ϭ 4.23 nM (Fig. 7C)). This finding is consistent with the results from gel filtration (Fig. 5E) and binding studies (below).
Stoichiometry of MiaA Binding to RNA Substrates-We further investigated the composition of complexes formed between MiaA and synthetic tRNA Phe or ACSL Phe molecules by performing band shift and filter binding assays (Figs. 8 -10; see "Experimental Procedures"). For both kinds of assays, we performed protein excess titrations, in which the tRNA Phe (wt) concentration was held constant far below the estimated K d app , and the MiaA concentration was varied (Figs. 8A and 10). We also performed ligand excess titrations, in which the MiaA concentration was held constant near its estimated cellular concentration (Ϸ1.0 M; see above), and the tRNA Phe and AC-SL Phe concentrations were varied (Figs. 8B and 9).
Unexpectedly, the molar ratio of tRNA Phe (wt) or tRNA Phe (A37G) bound per MiaA at saturation was 0.5 for both types of binding assays (Fig. 9, A and B). Given that the MiaA preparations were completely active for binding (Fig. 5), this result showed that a MiaA dimer, rather than a monomer,  a Calculation of kcat app and kcat app /Km app assumed that MiaA was fully active as a monomer for both tRNA and ACSL Phe substrates. The assumption of full activity was a first approximation based on the finding that MiaA was completely active for tRNA Phe and ACSL Phe binding (Fig.  5). If MiaA acts catalytically as a dimer for tRNA substrates and a monomer for ACSL Phe substrates as implied by binding studies (see "Discussion"), then kcat app and kcat app /Km app values will be doubled, except for the ACSL Phe (wt) substrate.
bound to each intact tRNA molecule at saturation. Consistent with this interpretation, the predicted molecular mass of a MiaA 2 ⅐tRNA Phe (wt) complex (100 kDa) matched the 110-kDa molecular mass observed during gel filtration (Fig. 5C). The discrepancy between the predicted molecular mass of a MiaA 2 ⅐tRNA Phe (A37G) complex (100 kDa) and the 63-kDa complex observed during gel filtration (Fig. 5E) may indicate dissociation of the MiaA 2 ⅐tRNA Phe (A37G) complex upon dilution during gel filtration. In contrast to tRNA Phe (wt) binding, AC-SL Phe (wt) or ACSL Phe (A11G) bound MiaA in a 1:1 molar ratio at saturation (Fig. 9, A and B), suggesting that MiaA bound ACSL Phe microhelices as a monomer. This result confirmed the conclusion that the MiaA enzyme preparations were completely active for binding. The predicted molecular mass of a MiaA⅐ACSL Phe complex (41 kDa) was near that of the 45-kDa complex observed during gel filtration (above).
Examination of the gels used for the band shift assays further confirmed a difference in the way tRNA Phe (wt) and tRNA Phe (A37G) bound to MiaA. In ligand excess titrations, the MiaA⅐tRNA Phe (wt) complex formed at lower tRNA Phe (wt) concentrations (Complex 1, Fig. 8B) had a lower mobility and was possibly larger than the complex formed at saturation, which contained a 2:1 molar ratio of MiaA to tRNA Phe (wt) (Complex 2; Fig. 8B and Fig. 9). In contrast, Complex 1 predominated at tRNA Phe (A37G) concentrations as high as 0.5 M in ligand excess titrations, whereas Complex 2 appeared and predominated at tRNA Phe (A37G) concentration near saturation (data not shown). Last, we used the concentration of MiaA at half-saturation in protein excess titrations of filter binding assays (Fig. 10) to estimate K d app Ϸ0.07 M for MiaA binding to tRNA Phe (wt) (46,47,51). Because of MiaA's complicated binding behavior (above), this K d app (tRNA Phe (wt)) is probably not a simple dissociation constant but rather a function of several binding constants (e.g. see Ref. 51). Protein excess titrations of band shift assays (Fig. 8A) gave a K d app (tRNA Phe (wt)) Ϸ1.0 M, which was about 10-fold greater than that obtained by filter binding. This higher K d app (tRNA Phe (wt)) may reflect dissociation of complexes during electrophoresis. Since MiaA bound ACSL Phe (A11G) by an apparently simple (R⅐P 3 R ϩ P) mechanism, we estimated K d (ACSL Phe (A11G)) ϭ 1.1 M from ligand excess titrations of filter binding assays (Fig. 9B) (46,47). The filter binding method failed to detect binding between MiaA and the ACSL-Phe (wt) microhelix, suggesting that ACSL Phe (wt) bound to MiaA with a lower affinity than 1.0 M. DISCUSSION We report here steady-state kinetic and binding studies of the E. coli MiaA tRNA prenyltransferase modification enzyme. Only a limited number of tRNA and rRNA modification enzymes have been purified and studied to date, despite the fact that many different kinds and families of RNA modification enzymes are present in all cells (3,4,9). Biochemical studies of these RNA modification enzymes are aimed at understanding the mechanisms of RNA-protein recognition and catalysis and the functions and regulation of RNA modification in cells. The properties of the MiaA enzyme show similarities and noteworthy differences compared with other purified tRNA modification enzymes, including the E. coli TrmA m 5 U54-methyltransferase (51)(52)(53), the E. coli TrmD m 1 G-methyltransferase (54,55), the E. coli HisT pseudouridine synthase I (56), and the E. coli and Zymomonas mobilis Tgt tRNA-guanine transglycosylases (57)(58)(59)(60).
Similar to the TrmA and Tgt enzymes (51,58), MiaA can modify a 17-mer synthetic microhelix stem-loop substrate, in this case corresponding to the ACSL of tRNA Phe (wt) (Table II). Thus, the minimal recognition elements for the MiaA tRNA prenyl transfer reside in this limited ACSL structure. However, the k cat app /K m app substrate specificity is reduced significantly by about 17-fold for the ACSL Phe (wt) microhelix compared with intact tRNA molecules (Table II) (or 34-fold if MiaA is catalytically active as a dimer for tRNA and as a monomer for ACS-L Phe (wt) (below; see Table II)). In this regard, MiaA resembles the Tgt transglycosylases, whose V max app /K m app is 5-10-fold lower for an ACSL Tyr microhelix compared with an intact tRNA Tyr substrate (58). In contrast, the TrmA methyltransferase uses a T-loop microhelix as a substrate almost as well as intact tRNA molecules (reduction of k cat app /K m app Ϸ2.5-fold; (51)). At the other extreme, the TrmD methyltransferase depends strongly on intact tRNA tertiary structures and does not efficiently modify an ACSL microhelix (reduction of V max app /K m app Ϸ300-fold (54)). Recently, an attempt was made to classify tRNA modification enzymes into two families (61). The TrmA methyltransferase and other enzymes that modify the amino acid-accepting minihelix, which is composed of the acceptor and T-loop microhelices, generally do not depend on overall tRNA tertiary structure for their activities (61). In contrast, TrmD methyltransferase and other enzymes that modify the anticodon minihelix, which is composed of the anticodon and D-loop microhelices, have been found to strongly depend on intact tRNA threedimensional structure (see Ref. 61). The exception to this classification scheme is bacterial Tgt transglycosylases, which modify ACSL microhelices moderately efficiently (see above) (58,62) and interact strongly, but not exclusively, with the ACSL region of substrate tRNA molecules (57,63). MiaA also presents an exception to the strictest application of this classification scheme. However, the significantly reduced substrate specificity of MiaA for the ACSL Phe (wt) microhelix compared with intact tRNA Phe (wt) ( Table II) suggests that interactions of MiaA with regions other than the ACSL Phe are important for optimal activity. This conclusion is further supported by binding studies discussed below. Compilations of the tRNA molecules that contain ms 2 i 6 A37 or i 6 A37 modifications led to a consensus ACSL that includes possible secondary structure  open triangles), respectively. Binding amounts were corrected using retention efficiencies (Ϸ85 or 40% for tRNA Phe (wt) or ACSL Phe (A11G), respectively) determined from protein excess titrations ( Fig. 10; see "Experimental Procedures") (46 -48). BSA was omitted from filter soaking and wash solutions. Averages of two or more experiments are shown with standard errors. and sequence-specific elements required for MiaA recognition and modification (1,8). Recent experiments by Y. Motorin and H. Grosjean 3 using tRNA Ser isoaccepting species confirm that optimal MiaA activity may depend on primary, secondary, and tertiary interactions within tRNA substrates. The kinetic, binding, and footprinting properties of these and other mutant RNA substrates are currently being determined.
One noteworthy difference between MiaA and other purified tRNA modification enzymes is its extremely low K m app for native and synthetic tRNA substrates (Ϸ3 nM; Table II). By comparison, the K m app of the Tgt transglycosylases, TrmD methyltransferase, and TrmA methyltransferase for synthetic tRNA substrates is 2.0, 3.3, and 2.8 M, respectively (51,54,58), which are about 3 orders of magnitude higher than that of MiaA (Table II). Likewise, the k cat app or V max app of MiaA is about 10-fold greater for synthetic tRNA substrates than that of the TrmA and Tgt enzymes (51,58). Together, these results show that MiaA is a comparatively active enzyme with a high apparent affinity for its tRNA substrates. Consistent with this interpretation, the kinetic properties of MiaA for a synthetic, purified tRNA Phe (wt) substrate were very similar to those for bulk tRNA isolated from a miaA mutant (Table II). Thus, the presence of nonsubstrate, native tRNA molecules in the bulk tRNA preparations did not appreciably inhibit or affect MiaA substrate recognition or activity. This behavior contrasts with the purified HisT pseudouridine synthase (56) and TrmD methyltransferase, 4 which are strongly inhibited by nonsubstrate tRNA species. Moreover, the constancy of kinetic properties of MiaA for synthetic, unmodified tRNA Phe (wt) and hypomodified bulk tRNA from a miaA mutant implies that the other modifications present in tRNA molecules do not significantly affect MiaA activity. Transparency to other base modifications in tRNA was also documented for the Tgt transglycosylase (62).
We used quantitative Western immunoblotting to determine that there are about 660 MiaA monomers per cell in bacteria growing exponentially in enriched minimal salts/glucose medium (Fig. 4). Previously, the cellular amounts of tRNA modification enzymes have been measured only indirectly by comparisons of specific activities (e.g. see Ref. 64), and these studies have led to the generalization that tRNA modification enzymes are present in few copies per cell (9). To the contrary, our results show that MiaA is a relatively abundant cellular protein compared with other biosynthetic enzymes (see Ref. 41). From the k cat app of MiaA for tRNA (Table II) (Table II), and K d app (tRNA) of MiaA Ϸ0.07 M in the absence of DMAPP (below). Finally, we found that the cellular amount of MiaA was regulated and decreased about 3-fold as bacterial cells entered stationary phase (Fig. 4), which leads to a drop in the rate of tRNA synthesis (66). Early work indicated that MiaA activity was strongly inhibited by unknown compounds in some substrate preparations (37). Overexpression of E. coli tRNA Phe (wt) in E. coli caused the accumulation of hypomodified tRNA species lacking the ms 2 i 6 A37 or i 6 A37 modifications (21), implying that MiaA activity can be saturated in vivo. We found that MiaA activity was strongly inhibited by ADP, ATP, and other NTPs (see "Results" ; Fig. 7, A and B). This inhibition was classically competitive with the DMAPP substrate (Fig. 7) with a K i app of about 0.06 M (Fig. 7, A and B).
The comparatively large cellular amount of MiaA and its high activity and substrate affinities are likely needed to overcome inhibition by NTPs and NDPs in vivo. We can estimate this inhibition by first recalling that the number of MiaA substrate tRNA molecules synthesized per min is about 328 (see above) Ϸ0.54 M (see "Results"). The K m app (tRNA) of MiaA Ϸ3 nM (Table II) (1 ϩ ([I]/K i app ))) and will depend on the MiaA kinetic parameters for DMAPP (Table II) and the free, but not total, intracellular concentrations of NTPs, NDPs, and DMAPP. To our knowledge, these free concentrations are not known with certainty for E. coli. Nevertheless, one estimate of [NTP] free is 50 M, based on the K m app (ATP) of many kinases (67), and the fact that ATP is by far the most abundant nucleotide compound in enterobacterial cells (68). Assuming that [NTP] free ϩ [NDP] free actually approaches 100 M, then [DMAPP] free would only have to be about 25 M to allow MiaA to fully modify newly synthesized tRNA substrate molecules.
[DMAPP] free Ϸ25 M seems reasonable, because DMAPP and isopentenyl diphosphate (IPP) are precursors to ubiquinone, which is abundant in E. coli ( Fig. 1) (69). Moreover, 25 M is near the K m app of IPP⅐DMAPP isomerases of yeast and other organisms (70). Thus, the amount and kinetic properties of MiaA and its inhibition by NTPs and NDPs seem balanced to just allow full tRNA substrate modification.
The competitive inhibition of MiaA by ATP or ADP likely indicates an important structure-function relationship for this class of prenyltransferases. MiaA homologs from several organisms lack significant amino acid similarities with other enzymes that use IPP and DMAPP as substrates, such as the Asp-Asp-Xaa-Xaa-Asp motif (71). MiaA homologs also lack conserved Cys and His residues positioned in possible metal binding motifs (e.g. Ref. 59). On the other hand, they do share an ATP/GTP P-loop binding motif, which is also present in Agrobacterium Ti-plasmid-encoded adenine isopentenyldiphosphate transferases (ipt; tzs) that synthesize the free-base plant hormone i 6 A (cytokinin) (19,34). The conservation of the P-loop motif and the competitive inhibition of DMAPP by NDPs and NTPs suggests that these families of tRNA and adenine prenyltransferases may use the P-loop motif to bind DMAPP instead of motifs used by other prenyltransferases (71,72). This hypothesis will be tested in future studies. To our knowledge, MiaA is the first example of an RNA modification enzyme whose activity is regulated by compounds other than its substrates or products.
The results from gel filtration (Fig. 5), kinetic (Figs. 6 and 7), and binding (Figs. 8 -10) experiments can be accounted for by a model in which MiaA binds to ACSL Phe structures as a monomer but binds to intact tRNA molecules as a multimer, either a dimer with half-site occupancy or a tetramer that dissociates into half-site occupied dimers. If intact tRNA molecules bind to the MiaA multimer preferentially and only bind to the monomer at higher tRNA concentrations, then apparent substrate inhibition of MiaA by tRNA Phe (wt) would result (Fig. 6A). One attractive feature of this model is that only the correct tRNA substrates may bind tightly and possibly in a positively cooperative way to MiaA multimers. Consequently, the association state of MiaA may contribute to the process of distinguishing between substrate and nonsubstrate tRNA molecules. Additional studies are needed to test features of this model directly, determine the kinetic order of the modification reaction, and learn whether DMAPP affects the MiaA association state.