INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Isopentenylated adenine is frequently found at position 37 of eubacterial and some eukaryotic tRNAs that read codons with U in the first position. In Escherichia coli, 2-methylthio-N⁶-(isopentenyl)adenosine (ms²i⁶A) is found, whereas Salmonella typhimurium contains the hydroxylated derivative (ms²io⁶A) (9). The majority of the modified nucleosides (including ms²io⁶A) are synthesized after transcription of the tRNA has been completed (23, 53), and the formation of ms²io⁶A is thought to proceed according to the pathway depicted in Fig. 1. The MiaA enzyme catalyzes the first step, transfer of a dimethylallyl group from dimethylallyl diphosphate onto the adenine at position 37 of the tRNA (22, 35, 41). An iron-requiring methylthiolation then follows (8, 46, 61). Since a miaA mutant contains mainly A37 and not ms²A37 (18, 59), the enzyme(s) involved in the methylthiolation appears to recognize the isopentenyl group. This second step is believed to involve at least two distinct reactions, thiolation of i⁶A37 to s²i⁶A37 (1) and methyl transfer from S-adenosylmethionine (24), giving ms²i⁶A37. In S. typhimurium, but not in E. coli, the isopentenyl group is further modified by the addition of a hydroxyl group, resulting in ms²io⁶A37. This last step is catalyzed by the MiaE protein (42) and occurs only aerobically (6). The hydroxylating enzyme seems to recognize the ms² group, since only small amounts of io⁶A are produced when methylthiolation is physiologically (6) or genetically (20) disturbed.

View larger version (14K): [in this window] [in a new window]   FIG. 1.   Postulated biosynthetic pathway for ms2i6A in E. coli and ms2io6A in S. typhimurium. The gene designations correspond to identified genetic loci or postulated (not yet identified) functions. Known cofactors and substrates are indicated. DMAPP, dimethylallyl diphosphate; SAM, S-adenosyl-L-methionine; R, the ribose moiety. The dashed arrow indicates that the reaction in question is used to a lesser extent. The hydroxylation reaction (mediated by MiaE, giving io6A and ms2io6A) does not occur in E. coli.

We previously described an S. typhimurium mutant (miaB2508::Tn10dCm) whose tRNA lacks the methylthiolated species and instead contains i⁶A37 (20). To gain further insight into the mechanism of the methylthiolation reaction, we have now identified the affected gene. Complementation analysis established that the f474 (yleA) open reading frames (ORFs) of E. coli and S. typhimurium encode a protein involved in the methylthiolation step of the biosynthesis of ms²i(o)⁶A [ms²i(o)⁶A denotes either ms²i⁶A or ms²io⁶A].

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, plasmids, and growth conditions. The bacterial strains and plasmids used in this study are listed in Table 1. Bacteria were grown in Luria-Bertani (LB) medium or glucose minimal medium (58). For growth of S. typhimurium cultures to be used in P22 transduction, NAA medium was used (20). Chloramphenicol, tetracycline, kanamycin, ampicillin, and carbenicillin were added to 12.5, 25, 50, 50, and 50 µg per ml, respectively, in LB medium. In minimal medium, 12.5 µg of chloramphenicol per ml or 6.25 µg of tetracycline per ml was used.

Plasmid constructions. DNA inserts in plasmids pTX637 to pTX640, which were used in mRNA mapping (below), were amplified by PCR with the primers listed in Table 2. HindIII sites were introduced into primers MiaBP1, MiaBT2, and MiaBT3, and BamHI sites were introduced into primers MiaBP2, MiaBP3, and MiaBT1. PCR fragments amplified from strain CC104 chromosomal DNA with MiaBP1-MiaBP2, MiaBP1-MiaBP3, MiaBT1-MiaBT2, and MiaBT1-MiaBT3 were first cut with HindIII and BamHI and then cloned into pGEM3Z (Promega) (cut with HindIII and BamHI) to make plasmids pTX637, pTX638, pTX639, and pTX640, respectively.

Genetic techniques. Transductions with phage P22 HT105/1 (int-201) (48) were performed as described elsewhere (15). Generalized transduction with phage P1vir was performed as described previously (49). To avoid the degradation of plasmids originally grown in E. coli, plasmids were first introduced into the restriction-deficient and modification-proficient S. typhimurium GT907 by transformation (39). The plasmids were thereafter moved into the relevant S. typhimurium genetic background by P22 transduction.

DNA manipulations. Procedures for DNA restriction digestions, agarose gel electrophoresis, DNA ligation, and transformation of competent E. coli cells were performed essentially as described by Sambrook et al. (47). Restriction enzymes and DNA ligase was purchased from Boehringer (Mannheim, Germany), Promega (Madison, Wis.), and New England Biolabs (Beverly, Mass.). Medium-scale plasmid preparations were done with Bio-Rad (Hercules, Calif.) Quantum or Qiagen (Hilden, Germany) midiprep kits. Small preparations of plasmid DNA were made with the Qiagen or Promega miniprep kits. Chromosomal DNA was isolated with the Genome kit from Bio 101 Inc, Vista, Calif.

Amplification of DNA by PCR. PCR amplification was performed with Taq DNA polymerase (Boehringer) or Vent (exo⁺) DNA polymerase (New England Biolabs) by using the buffers supplied with the enzymes. For amplification of the S. typhimurium miaB mutant alleles, a high-salt buffer containing 0.6 M KCl and 20 mM MgCl₂ was used. Routinely, 15 pmol of the appropriate primers (Table 2) and about 100 ng of template DNA were added to the reaction mixture. In some cases, cell suspensions were used instead of pure DNA. The cell suspension was obtained by resuspending one colony in 50 µl of distilled water, incubating it at room temperature for 30 min, and then using 1 to 2 µl of the suspension in the PCR mixture. Cycling was performed in a PC-960G microplate gradient thermal cycler (Corbett Research, Sydney, Australia) or in a PTC-100 programmable thermal controller (MJ Research, Waterstown, Mass.). The first cycle was preceded by a 3- to 5-min 95°C denaturation step, and the reaction temperature cycle was 95°C for 1 min, 50 to 68°C (depending on the primers used) for 1 min, and 72°C for 3 min. The cycle was repeated 25 to 30 times and was followed by a 5-min extension at 72°C. The PCR products were visualized on 0.9 to 1.5% agarose gels. The fragments were either purified directly from the PCR mixture by using a Promega PCR clean-up kit or GenElute PCR DNA Purification kit (Supelco, Bellefonte, PA) or isolated from the agarose gel and then purified with the GeneClean III kit from Bio 101.

Southern hybridization. For colony hybridization, bacterial colonies were lifted onto a Hybond N⁺ membrane (Amersham, Little Chalfont, England) by replica plating, and the cells were lysed in situ with 0.5 M NaOH (28). The released DNA was immobilized on the membrane by UV illumination. For screening of plasmids, 1 to 2 µg of DNA was digested with suitable restriction enzymes. The digests were separated on 0.7% agarose gels and blotted onto the membrane by capillary transfer (51). Hybridizations were carried out as specified by the manufacturer (Amersham). The probe used was a ~350-bp [-³²P]dATP-labeled miaB⁺ fragment generated by PCR from plasmid pUST135 by using primers T-I and MotH.

DNA sequencing and sequence analysis. Double-stranded plasmid DNA containing the miaB⁺ region of S. typhimurium was sequenced with an ABI Prism cycle-sequencing kit (Perkin-Elmer, Norwalk, Conn.) in an ABI Prism 377 DNA sequencer. Gel-purified PCR fragments were used as templates for sequencing mutant alleles of S. typhimurium miaB. Two or three PCR products from separate PCR runs of the same strain were mixed before being sequenced. Sequencing of the E. coli miaB mutant was performed at the Core facility of the Department of Microbiology and Molecular Genetics, University of Texas Houston Medical School. The sequences were analyzed by using the University of Wisconsin Genetics Computer Group programs and the BLAST server at the National Center for Biotechnology Information (2).

Localized mutagenesis of miaB with hydroxylamine. A P22 phage stock (0.5 ml; titer of 4.7 × 10¹⁰ PFU per ml) propagated on strain TT2342 was treated with 0.4 M hydroxylamine (32). The mutagenesis was performed at 37°C and was interrupted after 39 h of incubation. The mutagenized phage lysate (10⁸ PFU per ml) was thereafter used to transduce strain GT1825. Transductants, which had received a portion of the zbf-99::Tn10dTc region, which is 83% linked to miaB (20), were selected by plating the transduction mixture on TYS plates (42) containing 15 µg of tetracycline per ml and 10 mM EGTA. The level of amber suppression in the lacI-lacZ gene of the F' in the Tc^r transductants was then screened by printing on minimal plates containing histidine, leucine, and 40 µg of 5-bromo-4-chloro-3-indolyl--D-galactopyranoside (X-Gal) per ml. Strain GT1825 gives blue colonies on such a plate at 41°C because of efficient function of the supF30 amber suppressor tRNA. In contrast, strain GT1975, containing the miaB2508::Tn10dCm allele, is light blue at this temperature (20). Colonies which exhibited a reduced level of LacI-LacZ fusion protein (as judged by the color of the colony) compared to GT1825 were chosen for further studies.

HPLC analysis of the contents of modified nucleosides in tRNA. tRNA was isolated either as described by Connolly and Winkler (11) or by the procedure described by Buck et al. (7). tRNA was digested to nucleosides with nuclease P1 and alkaline phosphatase (25). The resulting hydrolysate was analyzed by high-performance liquid chromatography (HPLC) by the method of Gehrke and Kuo (26) on a Supelcosil LC-18S column (Supelco) with a Waters or Shimadzu HPLC system.

Determination of the transcriptional start point by primer extension. Total cellular RNA from S. typhimurium cells grown in LB medium at 37°C to a cell density of about 7 × 10⁸ cells per ml was isolated by extraction with hot phenol (60). The oligonucleotide primers Sekv14 and PCR2 were end labeled with [-³²P]ATP by using T4 polynucleotide kinase. The labeled primers were purified from unincorporated nucleotides by chromatography on a Quick Spin Sephadex G-25 column (Boehringer Mannheim). Approximately 0.1 pmol of labeled primer was mixed with 20 µg of RNA. The mixture was then denatured at 95°C and subjected to annealing and primer extension with avian myeloblastosis virus reverse transcriptase (Pharmacia Biotech) as described previously (27). To estimate the sizes of the reaction products, dideoxy sequencing reaction products obtained with plasmid pUST137 and primers Sekv14 and PCR2, respectively, were electrophoresed along with the primer extension reaction products on a standard 6% urea-polyacrylamide sequencing gel.

RNase T2 protection assays of chromosomal transcripts. RNA from E. coli CC104 was purified from cells grown in LB medium to mid-exponential phase (approximately 5 × 10⁸ cells per ml). RNA was prepared by adding portions of bacterial cultures directly to lysis solutions without intervening steps as described previously (56). RNase T2 protection assays of transcripts from the bacterial chromosome were completed as described previously (56). RNA probes 1 to 4 and the corresponding complementary probes 1C to 3C for mapping of the miaB region (see Fig. 4) were synthesized by using the following phage RNA polymerase and linearized plasmid templates: probe 1 (T7; pTX638 with HindIII); probe 1C (SP6; pTX638 with BamHI); probe 2 (T7; pTX637 with HindIII); probe 2C (SP6; pTX637 with BamHI); probe 3 (T7; pTX640 with HindIII); probe 3C (SP6; pTX640 with BamHI); probe 4 (T7; pTX639 with HindIII). A series of labeled, undigested RNA molecules of known lengths were used as size standards to determine the lengths of the protected fragments (standard errors  5 to 10%) (56). Each hybridization reaction mixture contained 40 µg of total RNA.

Nucleotide sequence accession number. The S. typhimurium DNA sequence described in this work was deposited in the EMBL database under accession no. AJ249116.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Identification of the insertion point of the Tn10dCm transposon of the miaB2508::Tn10dCm mutant. The miaB2508::Tn10dCm mutation, which gives rise to undermodification of ms²io⁶A tRNA in S. typhimurium, is located close to the nag locus at 16 min on the S. typhimurium chromosome (20). To establish the exact location of Tn10dCm in this mutant, we took advantage of the fact that there is another transposon (zbf-99::Tn10dTc) located 2 to 3 kb from the miaB2508::Tn10dCm insertion (20). We intended to amplify the region between these two transposons by PCR by using as template a lysate from strain GT3171 (miaB2508::Tn10dCm zbf-99::Tn10dTc). However, we did not detect a PCR product of the expected size (2 to 3 kb). Instead, we found a fragment of approximately 1.3 kb (data not shown) when the Tn10XbaI and T-L primers were used. The obtained fragment was ligated into the pT7Blue T vector (Novagen), yielding plasmid pUST135. Sequencing of the insert revealed that one end of the fragment consisted of Tn10dCm DNA while the other end was the result of the T-L primer binding to a nontransposon sequence. A search for sequence homologies showed that the amplified fragment was homologous to the f474 (yleA) ORF of E. coli identified by Blattner et al. (4). This ORF lies (together with four other ORFs) between the cutE (lnt) gene and a tRNA cluster located close to the nag operon of E. coli (Fig. 2). This demonstrates that the Tn10dCm transposon of the S. typhimurium miaB mutant had been inserted in the f474 (yleA) gene, which is hereafter called miaB.

View larger version (9K): [in this window] [in a new window]   FIG. 2.   Genetic organization of the region between the cutE (lnt) gene and the nag operon at 14.8 to 15.1 min of the E. coli chromosome. Previously identified genes and their directions of transcription are indicated by solid arrows. The tRNA cluster is represented by a solid box. Shaded arrows indicate putative ORFs. (There is also an alternative nomenclature for the E. coli ORFs, where f474 is denoted yleA, o391 is yleB, f359a is ybeZ, and f155b is ybeY.) The vertical arrow indicates the point where the miaB::Tn10dCm transposon had been inserted. Horizontal bars represent the inserts of the S. typhimurium plasmids. Plasmids pUST138 and pUST190 are high-copy number and low-copy number versions, respectively, carrying the same insert.

Construction of a miaB::Tn10dCm mutant of E. coli. An E. coli miaB::Tn10dCm mutant is useful for further studies of the effects of ms²i⁶A tRNA undermodification on spontaneous mutagenesis (11, 12) and bacterial physiology (20). Since galE mutants of S. typhimurium can be infected by the E. coli-specific phage P1 (40), we first moved the miaB2508::Tn10dCm mutation of S. typhimurium GT2176 into a galE background (strain MST1934) by P22 transduction. Integration of S. typhimurium DNA into the E. coli chromosome can be achieved by using a mutS or mutL E. coli recipient (44). The miaB2508::Tn10dCm allele was therefore moved first into an E. coli mutS strain by P1vir transduction (resulting in strain TX3325) and thereafter into a Su⁺3 lacZ(UGA) strain of E. coli to verify the antisuppressor activity of the miaB mutation (20). We also moved the miaB mutation into strain CC104 and showed that in such a background, the mutation did not influence the growth rate. This was in accordance with earlier observations of the S. typhimurium miaB2508::Tn10dCm mutant (20). Sequencing of a plasmid (pTX641) carrying chromosomal DNA from strain TX3346 verified that the DNA flanking the transposon was homologous to the f474 ORF, thereby confirming the location of miaB. The tRNA from the E. coli miaB strains showed an accumulation of i⁶A, whereas ms²i⁶A was absent (data not shown). These results confirmed the construction of the miaB::Tn10dCm mutant of E. coli.

Construction and properties of minimal miaB⁺ plasmids. We isolated a plasmid containing the miaB⁺ wild-type gene of S. typhimurium from an S. typhimurium genomic library (29), which was introduced into E. coli DH5. Colony hybridization with a miaB⁺ (f474⁺) probe (see Materials and Methods) followed by Southern hybridization identified one plasmid, pUST136, which gave a positive signal. Subcloning from pUST136 resulted in a smaller construct (pUST137), which complemented the ms² deficiency of the miaB2508::Tn10dCm mutant (data not shown). The insert in this plasmid was sequenced and was shown to be homologous to the E. coli ORFs f155b, f359a, f474, and o391 (Fig. 2). Thus, this region of the S. typhimurium chromosome is very similar to its counterpart in E. coli.

Analysis of MiaB sequences. A search for amino acid sequence homologies in the Swiss-Prot and GenBank databases was performed by using the BLAST utility at the National Center for Biotechnology Information server. The polypeptides showing homology to MiaB from S. typhimurium and E. coli could be divided into three groups. One consists of polypeptides highly homologous to MiaB, in organisms presumably containing the ms²i(o)⁶A modification. The second comprises polypeptides of high homology but from organisms presumably not containing the ms²i(o)⁶A modification. The polypeptides in the first two groups have more than 35% similarity to the S. typhimurium MiaB peptide. The entries in the third group showed less overall similarity (<35%) but had high homology within two of the conserved motifs in MiaB. These two motifs, which were present in almost all of the MiaB homologues, are indicated in Fig. 3. A few proteins of known function were found within this group, but no conclusion about the roles of these two motifs could be drawn based on these enzymes, since they constitute a seemingly heterogeneous group. It is likely that these two conserved motifs are involved in some fundamental type of interaction exhibited by a variety of enzymes.

View larger version (34K): [in this window] [in a new window]   FIG. 3.   Amino acid sequences of the MiaB polypeptides of S. typhimurium and E. coli as deduced from the DNA sequences. The E. coli sequence differs from the S. typhimurium sequence at only eight positions, which are shown in italics in the figure. The solid boxes indicate amino acids that are conserved in at least 50% of the MiaB homologues (that is, polypeptides giving E values of 109 or less in BLAST). Shaded boxes represent amino acids conserved in 50% of the MiaB homologues if amino acids with similar properties also were considered.

Characterization of point mutations in the miaB gene of S. typhimurium. Point mutations in miaB were identified to provide information about domain function and to further rule out polarity effects. The S. typhimurium chromosome was mutagenized with hydroxylamine (see Materials and Methods) to obtain missense mutations. Since the miaB2508::Tn10dCm null mutation reduces the efficiency of the suppressor tRNA_CUA^Tyr at least twofold (20), it was possible to detect other mutations in miaB by monitoring amber suppression of the lacI-lacZ hybrid gene. Of about 5,500 colonies, 26 were picked that exhibited a reduced level of suppression. An analysis of the modified nucleosides in their tRNA showed that four of these 26 isolates contained i⁶A instead of ms²io⁶A when grown at 41°C (Table 3). The miaB26 mutant (GT3828) contained ms²io⁶A when grown at 37°C, while the other three were deficient in the ms² group at both temperatures.

Transcriptional organization of the miaB gene. To investigate whether transcription of miaB is linked to the adjacent genes (Fig. 4A), the E. coli miaB transcript was mapped by RNase T2 protection assays with RNA probes 1 to 4 and the complementary probes 1C to 3C (Fig. 4B).

View larger version (43K): [in this window] [in a new window]   FIG. 4.   (A) Structure and transcription of the region surrounding the miaB gene at 15 min in the chromosome of E. coli K-12. The figure is drawn to scale. The orientations of the ORFs of o391, miaB, and f359a are indicated by arrows. Locations of the putative Po391, P1miaB, P2miaB, and Pf359a promoters and the Rho-independent TermiaB terminator were determined as described in the text. Horizontal lines represent RNA probes 1 to 4 for mapping in vivo transcripts by the RNase T2 protection assay (shown in panel B). (B) RNase T2 protection assays of transcripts from the o391-miaB-f359a region of the E. coli K-12 chromosome. The positions of RNA probes 1 to 4 used for this assay are indicated in panel A. Probes 1 to 4 correspond to the miaB and f359a noncoding strand and hybridized with miaB and f359a transcript. Probes 1C and 2C were complementary to probe 1 and 2, respectively, and hybridized to o391 transcripts. Lanes: a, RNA probes hybridized with 40 µg of total cellular RNA; b, RNA probes hybridized with 50 µg of tRNA (self-hybridization control); S, RNA size standards. Po391, P1miaB, and P2miaB are transcripts initiated from the putative promoters Po391, P1miaB, and P2miaB, respectively; TermiaB is the transcript terminated at the TermiaB terminator; f359a indicates f359a transcripts.

View larger version (70K): [in this window] [in a new window]   FIG. 5.   Mapping of the 5' ends of the miaB (A) and the o391 (B) mRNA by primer extension analyses with RNA isolated from S. typhimurium grown at 37°C. The Sekv14 primer was used in both sequencing and extension reactions for mapping of the miaB transcripts, while the PCR2 primer was used in mapping of the o391 transcriptional start point. Lanes A, C, G, and T represent the DNA-sequencing ladder. Lanes: 1, pUST138 plasmid in GT522 background; 2, pUST137 plasmid in GT522 background; 3, GT522 without any plasmid.

View larger version (17K): [in this window] [in a new window]   FIG. 6.   (A) Nucleotide sequences of the E. coli and S. typhimurium miaB promoter region. Included are the sequences encoding the N-terminal parts of the MiaB and o391 peptides. Putative 35 and 10 promoter sequences and potential Shine-Dalgarno sequences (SD) are indicated. The solid triangles represent the mRNA 5' ends identified in primer extension analyses, and the open triangles indicate mRNA 5' ends detected in RNase T2 protection assays. Ends detected by RNase T2 assays are accurate to only about 10 nt. Shaded boxes indicate positions in the E. coli sequence (accession no. AE000170), which differs from the S. typhimurium sequence (accession no. AJ249116). (B) Nucleotide sequence of the terminator regions of the miaB gene of E. coli and S. typhimurium. Included are the sequences encoding the C-terminal part of the MiaB peptide and the N-terminal part of the F359a peptide. The inverted repeats of the putative terminator stem are indicated by arrows.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The genetic organization of tRNA modification genes does not seem to conform to a common theme (62). Among the tRNA modification genes whose genetic organizations have been resolved, only one, the trmA gene (36), has hitherto been shown to be a monocistronic operon. The queA and tgt genes, which are both required for biosynthesis of Q34, are grouped in a polycistronic operon (43, 45). tRNA modification genes can also be found in complex operons, either with seemingly unrelated genes, as in the case with the miaA gene (12, 55, 57), or, like the trmD gene (10), together with genes encoding other components involved in translation. In this study, we identified the miaB gene of S. typhimurium and E. coli, which is involved in the methylthiolation step in the biosynthesis of the modified nucleoside ms²i(o)⁶A37 in tRNA. Our experiments revealed that the S. typhimurium homologue to the f474 (yleA) ORF of E. coli had been truncated by the Tn10dCm insertion in the miaB2508::Tn10dCm mutant. The f474 ORF is located between the cutE (lnt) gene and the nag operon at 14.9 min on the E. coli chromosome. Sequencing of a 3.7-kb chromosomal fragment from S. typhimurium covering the miaB (f474) gene showed that the genetic organization in this area is highly homologous to that in the E. coli counterpart (Fig. 2). The only major difference is a 51-bp insertion almost immediately downstream of the miaB UAA stop codon in S. typhimurium (Fig. 6B). Putative Rho-independent transcription termination signals were identified in this area between the miaB gene and the downstream ORF f359a (Fig. 6B). In E. coli, the 3' end of the miaB transcript coincides with this possible secondary RNA structure. This, together with the fact that more than 95% of the E. coli miaB transcripts ended at this terminator and that the downstream f359a ORF was transcribed from its own promoter, suggests that the miaB gene is transcribed as a monocistronic unit in E. coli (and most probably in S. typhimurium as well). Thus, we present the second case of a tRNA modification gene being transcribed as a monocistronic unit.

Putative FIS-binding sites have been observed in the promoter regions of several tRNA modification genes (trmA [29], queA-tgt [50], and miaA [57]; reviewed in reference 62). However, no sequence similar to a FIS consensus binding site (33) is present in the miaB promoter. Neither did the miaB promoters contain the discriminator sequence associated with stringent control (54). These two observations indicate that the expression of the MiaB protein is probably not regulated in the same manner as that of its tRNA substrate. It is known that the MiaB-mediated methylthiolation reaction requires iron (8, 46, 61) and that the concentration of iron in the cell is tightly regulated by Fur, which is a repressor for many iron-regulated operons (13). However, no putative Fur-binding site (16, 21) is present in the promoter of miaB.

miaB seems to be transcribed from two upstream promoters in E. coli and S. typhimurium cells grown in LB medium (Fig. 4 and 5). However, in E. coli, transcripts from the putative promoter proximal to the miaB translation start codon are present in excess over those from the upstream promoter, whereas in S. typhimurium, the numbers of transcripts from the two putative promoters are about equal. Further work is needed to determine whether this expression pattern represents differential transcription initiation or transcript processing in these two bacterial species and whether miaB transcription is differently regulated. Whether the overlap of the o391 and upstream miaB promoters has regulatory significance also remains to be determined.

HPLC analysis of the modified nucleosides in the tRNA of the miaB2508::Tn10dCm mutant (20) and of miaB point mutants (see Results) show that they contain i⁶A instead of ms²io⁶A. This suggest that the gene disrupted in these mutants encodes a protein involved in the addition of the thio part of the modification, by encoding either a thiotransferase or a bifunctional enzyme catalyzing both the thiolation and the methylation reaction. However, when Agris et al. (1) showed that the methylthiolation reaction occurs in two steps, they noticed that the s²i⁶A derivative was unstable. Therefore, we cannot rule out that we failed to detect s²i⁶A under the conditions used in our experiments. To address this issue, we extracted tRNA from a rel met strain grown in rich medium lacking methionine, analogous to the way Agris et al. (1) obtained methyl-deficient tRNA. In the HPLC analysis of isopentenylated nucleosides from this tRNA, we did not detect any s²i⁶A derivative as monitored by mass spectrometry; only i⁶A was present (data not shown). Thus, at present, we may be unable to detect s²i⁶A derivatives by these methods.

Two conserved motifs were found in almost all of the polypeptides homologous to MiaB. One is an array of three cysteine residues spaced by 3 and 2 amino acids (CxxxCxxC; positions 157 to 164 in MiaBSt [Fig. 3]), and the other is a highly conserved motif (IVGFPGET; positions 307 to 314 in MiaBSt [Fig. 3]) located in the C-terminal part of the proteins. The cysteine motif has previously been noticed and is denoted "uncharacterized protein family UPF0004 signature" (PS01278) in the PROSITE database. This stretch of amino acids in MiaB was also recognized when searching the BLOCKS 11.0 database (31). The protein block matching this cysteine motif is found in a group of enzymes known as radical activating enzymes, and their signature motif is designated PS01087 in PROSITE. The MiaB cysteines match the second half of the PS01087 motif, which is presumed to bind iron (52). The fact that the MiaB cysteine pattern only partly matches the radical activating enzyme signature probably indicates that both types of enzyme have active sites where an iron molecule is coordinated by the cysteines but that the rest of the active site is different due to differences in catalytic specificity. Since iron has been implicated in the methylthiolation of i⁶A (8, 46, 61), it is likely that the MiaB protein is a thiotransferase, assuming that the cysteine motif of MiaB really binds iron.

A potential binding site for pyridoxal phosphate (motif PS00595 in PROSITE) was also identified in the E. coli and S. typhimurium MiaB sequences (LVxxTSRKxxxxxxGxTxN; positions 388 to 402 in MiaBSt [Fig. 3]) and in several MiaB homologues. Lipsett and Peterkofsky (37) observed that incorporation of sulfur into tRNA was enhanced if pyridoxal phosphate was included in the reaction mixture. However, contradictory results have also been obtained (30), claiming that pyridoxal phosphate is not a cofactor for tRNA sulfur transferase activity in E. coli. Still, studies of the enzymatic reaction catalyzed by two purified enzymes involved in the biosynthesis of s⁴U showed that this reaction was dependent on pyridoxal phosphate (38). Thus, the presence of a putative pyridoxal phosphate-binding site in MiaB supports the suggestion that the MiaB enzyme is involved in a thiolation reaction.

The MiaB polypeptide contained only a weak similarity to the glycine-rich motif (D/E)GxGxGxxxx (where  is a hydrophobic residue and  is a charged or polar amino acid), which has been proposed to be involved in binding of S-adenosylmethionine (AdoMet) (63). Interestingly, two of the point mutations obtained in this study are located within this region (G79D and C83Y [Fig. 3]), indicating that the function of MiaB may be decreased by mutations in this region. However, the two glycine residues present in MiaB from S. typhimurium and E. coli are conserved in only two of the homologues from other organisms (YleA from Haemophilus influenzae and Yhe2 from Pseudomonas aeruginosa). Thus, the weak similarity to an AdoMet-binding site and the lack of conservation of these amino acids suggest that the miaB gene does not encode a methyltransferase. However, this possibility cannot be excluded, because Djordjevic and Stock (17) found that the AdoMet-binding site of different types of methyltransferases have common sizes and shapes, even though there is a lack of sequence identity. It is also possible that the structure of the MiaB enzyme is unique and that the structure of the AdoMet-binding site deviates from the ones observed in proteins which exhibit the consensus pattern mentioned above.

Finally, the MiaB polypeptide was compared with the amino acid sequence of the MiaA protein. This alignment revealed that these two peptides have only 16% identity and that there are no obvious shared motifs. If there is some common structure, for instance for recognition of the adenine moiety or the tRNA, it is thus probably reflected in overall structure and not in sequence similarity. Alternatively, the MiaB enzyme may recognize just the i⁶A modified base and not the tRNA molecule. The homology between MiaB and the MiaE protein was even lower (13% identity).

In summary, the presence of putative iron- and pyridoxal phosphate-binding sites favors the conclusion that MiaB is the thiotransferase required in ms²i(o)⁶A37 tRNA modification. The absence of an apparent AdoMet-binding site suggests that MiaB may not be bifunctional and carry out the methyltransferase reaction as well. However, we cannot rule out that MiaB carries out the methyltransfer by using a nonconsensus AdoMet-binding site. These hypotheses will be tested by future biochemical characterization of purified MiaB.