The Conserved Cys-X₁-X₂-Cys Motif Present in the TtcA Protein Is Required for the Thiolation of Cytidine in Position 32 of tRNA from Salmonella enterica serovar Typhimurium

Abstract

The modified nucleoside 2-thiocytidine (s²C) has so far been found in tRNA from organisms belonging to the phylogenetic domains Archaea and Bacteria. In the bacteria Escherichia coli and Salmonella enterica serovar Typhimurium, s²C is present in position 32 of only four tRNA species—, , , and . An in-frame deletion of an S. enterica gene (designated ttcA, for “two-thio-cytidine”) was constructed, and such a mutant has no detectable s²C in its tRNA. The TtcA protein family is characterized by the existence of both a PP-loop and a Cys-X₁-X₂-Cys motif in the central region of the protein but can be divided into two distinct groups based on the presence and location of additional Cys-X₁-X₂-Cys motifs in terminal regions of the sequence. Mutant analysis showed that both cysteines in this central conserved Cys-X₁-X₂-Cys motif are required for the formation of s²C. The ΔttcA1 mutant grows at the same rate as the congenic wild-type strain, and no growth disadvantage caused by the lack of s²C was observed in a mixed-population experiment. Lack of s²C32 did not reduce the selection rate at the ribosomal aminoacyl-tRNA site (A-site) for at any of its cognate CGN codons, whereas A-site selection at AGG by was dependent on the presence of s²C32. The presence of s²C32 in peptidyl- or in peptidyl- interfered with decoding in the A-site. The presence of s²C32 in decreased the rate of translation of the CGA codon but not that of the CGU codon.

The source of sulfur for the synthesis of the thiolated nucleosides found in tRNA from Escherichia coli or Salmonella enterica serovar Typhimurium is cysteine (1). About 30 different modified nucleosides are present in tRNA from E. coli or S. enterica, and of these, five [2-thiocytidine (s²C), 4-thiouridine (s⁴U), 5-methylaminomethyl-2-thiouridine (mnm⁵s²U34), 5-carboxymehylaminomethyl-2-thiouridine (cmnm⁵s²U34), and 2-methylthio-N⁶-(cis-4-hydroxy-isopentenyl)adenosine (ms²io⁶A, in S. enterica) or 2-methylthio-N⁶-isopentenyl-adenosine (ms²i⁶A, in E. coli)] are thiolated. Although the biochemical source of sulfur is known, the mechanism by which sulfur is incorporated into these various molecules has been an enigma for many years. However, recently we and others showed that the IscS cysteine desulfurase, which is not an RNA binding protein, is involved in the synthesis of all five thiolated nucleosides present in tRNA of S. enterica and E. coli (29, 37). It is likely that the synthesis of each thiolated nucleoside requires at least one enzyme that recognizes a specific subset of tRNAs. One characterized example of this is the enzyme that catalyzes the formation of s⁴U8 at position 8 of tRNA (26, 38). The sulfur of cysteine is transferred first to IscS, thereby forming an IscS-SSH persulfid, which in turn transfers a sulfane sulfur to the protein ThiI, forming a ThiI-SSH persulfid. In the presence of ATP-Mg, this protein transfers the sulfur to the particular tRNA substrates, thereby forming s⁴U8. A mutation in the mnmA (asuE, trmU) gene inhibits the thiolation step in the formation of mnm⁵s²U34 (37, 53). Similar to the synthesis of s⁴U, the protein MnmA obtains a persulfid from IscS and, in turn, transfers the sulfur to a subset of tRNAs, thereby forming the s²U moiety of mnm⁵s²U34 (27). A defective miaB gene results in a lack of the ms² moiety of ms²io⁶A37 in S. enterica (22), and the gene encodes an iron-sulfur protein (39) belonging to the recently identified Radical SAM superfamily (50). The thiI, mnmA, and miaB genes are specific for the syntheses of s⁴U, mnm⁵s²U, and ms²io⁶A in tRNA, respectively. Apart from iscS, no gene has been identified that influences the synthesis of s²C. This nucleoside is found in tRNA from organisms belonging to the phylogenetic domains Archaea and Bacteria and so far not in tRNA from organisms belonging to the domain Eucarya (2, 32). It is present at position 32 of only four tRNA species, , , , and in E. coli and S. enterica (52). As a first step to elucidate the mechanism of thiolation of C32 in tRNA, we have identified a gene, ttcA (for “two-thiocytidine”), whose product is required for the formation of s²C32. Interestingly, there exist genes that are very similar to the ttcA gene in organisms from all three domains, including eukaryotic organisms, in whose tRNAs no s²C has so far been found.

MATERIALS AND METHODS

Bacteria, media, and growth conditions.

The bacterial strains used in this study are listed in Table 1. Cultures were grown in either Luria-Bertani (LB) broth by the method of Bertani (6) or 0.8% nutrient broth (Difco Laboratories, Detroit, Mich.) supplemented with 0.5% NaCl, adenine, tryptophan, tyrosine, phenylalanine, p-hydroxybensoate, dihydroxybensoate, and p-aminobenzoate at concentrations given in reference 17. As rich solid medium, TYS agar plates (10 g of Trypticase peptone, 5 g of yeast extract, 5 g of NaCl, and 15 g of agar per liter) were used. Minimal agar plates contained medium E (55) and appropriate metabolites and carbon sources as described previously (17). As defined liquid medium, morpholinepropanesulfonic acid (MOPS) medium (36) supplemented with the relevant carbon source at 0.4% was used. Rich-MOPS medium was described by Neidhardt et al. (35). Antibiotics were used at concentrations of 50 μg/ml for carbenicillin, 12.5 μg/ml for chloramphenicol, 50 μg/ml for kanamycin, and 100 μg/ml for spectinomycin in E. coli and 800 μg/ml for spectinomycin in S. enterica.

TABLE 1.

E. coli and Salmonella strains and plasmids used in this study

Strain or plasmid	Genotype	Origin
E. coli
GRB105	strL aroE tsaA1 trmG1 ttcA1	Laboratory collection
S. enterica
LT2	Wild type	J. Roth
GT5969	Congenic wild type to GT5970	This work
GT5970	ΔttcA1	This work
GT6443	ttcA2 (C122A, C125A)	This work
GT6635	ttcA3 (C122S) zdf-2528::Kanr	This work
GT6636	ttcA4 (C125S) zdf-2528::Kanr	This work
Plasmids and phages
DD301	λ, Charon 30 containing BamHI fragment from E. coli	M. Rydéen-Aulin
pGJ8	pUC19 containing 3.5-kb BamHI-BamHI fragment from DD301	This work
pGJ13	ttcA gene from E. coli cloned into vector pCL1921	This work
pTH34	+1 frameshifting site CCC-AGA-A inserted in the beginning of lacZ as described previously (25)	Tord Hagervall
pTH35	As for pTH34, but the +1 frameshifting site was CCC-AGG-A	Tord Hagervall

Genetic techniques.

The Hfr mapping procedure was that of Singer et al. (48), and phage P1 transduction was done by the method of Miller (33). Transduction with phage P22 HT105/I (int201) (47) was done as described previously (17).

tRNA preparation, analysis of modified nucleosides, and mass spectrometry analysis.

Bacterial strains were grown in LB medium at 37°C to about 4 × 10⁸ to 6 × 10⁸ cells/ml (100 to 150 Klett units). The cells were lysed, and total RNA was prepared (20). tRNA was separated from rRNA by LiCl extraction (4). The tRNA was precipitated with 2.5 volumes of cold ethanol twice, washed with 70% ethanol, and dried. It was then dissolved in water, and a 100-μg sample was degraded to nucleosides with nuclease P1 followed by treatment with alkaline phosphatase (24). The resulting hydrolysate was analyzed by high-performance liquid chromatography (HPLC) by the method of Gehrke and Kuo (23) on a Supelcosil LC₁₈ column (Supelco) with a Waters HPLC system. For mass spectrometry analysis, the nucleosides were eluted using the gradient of Pomerantz and McCloskey (40). The outlet from the UV detector was directly interfaced to a VG platform mass spectrometer equipped with an electrospray ionization source (Fison Instruments, Altrincham, United Kingdom). The UV chromatograms were recorded continuously and the mass spectra were recorded every 1.0 s during the 60-min chromatography. The procedure and interpretation of data for qualitative liquid chromatography-mass spectrometry analysis of nucleosides in RNA hydrolysate by a soft ionization technique have been described previously (40).

DNA sequencing and bioinformatics.

DNA sequencing of PCR fragments purified by the PCR Kleen spin kit (Bio-Rad) or of plasmid DNA purified by the Nucleospin kit (Clontech) was performed in an ABI Prism 377 DNA sequencer. DyeEnamic ETerminator cycle-sequencing premix (Amersham-Biotech) was used for sequencing. The sequences were analyzed using the University of Wisconsin Genetics Computer Group programs for sequences from S. enterica serovar Typhimurium (http://genome.wustl.edu) and the BLAST server at the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/).

Multiple sequence alignments were conducted using global progressive algorithms (MAP program, available at http://searchlauncher.bcm.tmc.edu/multi-align/multi-align.html). Phylogenetic analysis was carried out, and neighbor-joining trees were constructed using the PHYLIP software package with the Fitch method (on-line server at the Pasteur Institute: http://bioweb.pasteur.fr/seqanal/phylogeny/phylip-uk.html).

In vivo labeling with ³⁵SO₄²⁻.

Cells were grown in 15 ml of MOPS minimal medium containing 0.081 mM K₂SO₄ supplemented with 0.4% glucose. Radioactive sulfate (final concentration, 150 μCi/ml) was added to the culture, and the cells were grown to 150 Klett units. tRNA was prepared as described above. The labeled tRNA was digested to nucleosides, and the hydrolysate was subjected to HPLC analysis. The radioactivity of the eluate was monitored with a 500TR flow scintillation analyzer (Packard) coupled to the UV detector of the HPLC system.

Construction of mutants.

An in-frame deletion in the ttcA gene was constructed by the method of Link et al. (31). Flanking primers A (5′-CGC GTC GAC CGC AGG GAT GGT TCG CAT TG-3′) and D (5′-CGC GAG CTC GCT CCA CGG CGA TCT GGA AC-3′) complementary to sequences located about 600 bp outside the gene were used together with primers B (5′-CCC ATC CAC TAA ACT TAA ACA GTT ACG GCG CAG GCG CTT TTG-3′) and C (5′-TGT TTA AGTT TAG TGG ATG GGC CGG AAG AAG ATG ACA CCG CC-3′), which were complementary for DNA 36 bp internal from the beginning and 54 bp internal from the end of the ttcA gene, respectively. Primers B and C contained an overlapping region at their 5′ ends (underlined bases), with an additional 21 bases coding for an amino acid sequence not present in bacteria. PCR fragments generated by primer pairs A-B and C-D were purified from an agarose gel and mixed in a 1:1 ratio. A new PCR was performed with the mixture using primers A and D at an annealing temperature of 50°C. A fragment of 1,234 bp was purified and cloned into the pGEM(T/A) vector (Promega). The construct was sequenced, and the entire fragment was cloned into the SalI-SphI site of the vector pDM4 (34). Using this suicide vector, which harbors the sacB gene that makes the bacteria unable to grow on 5% sucrose, the constructed in-frame deletion between codons 25 and 292 was transferred to the chromosome.

The two cysteines at positions 122 and 125 in the TtcA protein were altered by the Quikchange protocol (Stratagene). The wild-type allele of the ttcA gene from S. enterica was amplified by PCR using primers s²C-st1 (5′-CGT CGC CGT ACG TCA AGC AG-3′) and s²C-st2 (5′-GTA GCG ATG TTG AAT GCG CG-3′), which were complementary to sequences 120 bp 3′ and 140 bp 5′ of the ttcA gene, respectively, and cloned into the pGEM(T/A) vector. Primers cysmutL (5′-CCGACG CAA ACG CGA CGC CAG CGA CGC GGT GGT TTT TCC-3′) and cysmutR (5′-GGA AAA ACC ACC GCG TCG CTG GCG TCG CGT TTG CGT CGG-3′) were used to change the cysteine codons (TGC) into alanine codons (GCG; underlined). The SalI-SphI fragment was cut out and cloned into the suicide vector pDM4. This mutated fragment was later transferred to the chromosome as described previously (34).

Single-amino-acid changes of Cys 122 or Cys 125 in the TtcA protein were made as follows. First, a kanamycin resistance cassette was inserted 9 nucleotides downstream of the ttcA gene by using linear transformation (16). The chromosome of the resulting strain was then used as a template for PCR, where one of the primers bound within the Km^r cassette and the other primer bound to a site where the mutation was going to be introduced. This second primer was 60 nucleotides long and contained 41 nucleotides complementary to the sequence upstream the mutated codon, AGC, and 16 nucleotides homologous to the downstream sequence (designed to place a Ser instead of the Cys present in the wild type). The obtained PCR fragment of 2.2 kb was purified and used in linear transformation using electroporation of the wild-type S. enterica. Km^r transformants were selected for and screened for the mutated gene. Individual transformants were used to generate PCR fragments over the region of ttcA, where the mutation was constructed. Because the mutagenic change in sequence also introduced a SacI (C125S) or an AluI (C122S) restriction site, we used restriction analysis of the PCR fragments to differentiate between the wild-type and mutated ttcA forms present in the transformants.

Determination of A-site selection and P-site frameshifting activity.

The rate of aminoacyl-tRNA (aa-tRNA) selection to the A-site and the P-site frameshifting ability were determined as described previously (11, 13, 54).

Determination of sensitivity to various amino acid analogs.

Strains GT5969 (ttcA⁺) and GT5970 (ΔttcA1) were tested for the sensitivity to different amino acid analogs as described previously (10, 21). Strains were grown overnight in LB rich medium at 37°C. A sample (0.1 ml) was mixed with 2 ml of 0.5% agar in 0.9% NaCl and poured onto an agar plate containing medium E (55) and 0.2% sodium citrate. Paper disks (6 mm in diameter) were placed on the surface of the plates, and a certain amount of each analog as specified by Ericsson and Björk (21) was applied to the respective disk. The plates were scored after incubation at 37°C for 24 h. The following analogs were tested: dl-aspartic acid β-hydroxamate (Asn), 1,2,4-triazole (Cys), l-glutamic acid γ-hydrazide (Gln), dl-methionine hydroxmate (Gln), azaserine (Gln), l-methionine-dl-sulfoximine (Gln), 1,2,4-dl-triazole-3-alanine (His), S-2-aminoethyl-l-cysteine (Lys), β-(2-thienyl)-dl-alanine (Phe), m-fluoro-dl-phenylalanine (Phe), thioproline (Pro), l-thiazolidine-4-carboxlicacid (Pro), dl-5-fluorotryptophan (Trp); 5-methyl-dl-tryptophan (Trp), 3-aminotyrosine (Tyr), β-dl-hydroxynorvalin (Tyr), m-fluoro-dl-tyrosine (Tyr), fluoroacetic acid (acetate), 3,4-dehydro-dl-proline, and 3-nitrotyrosine. All analogs were obtained from Sigma Chemical Co., St. Louis, Mo.

Biolog MicroPlate assay.

ES MicroPlates were purchased from Biolog Corp. and used as recommended by the manufacturer. Log-phase cells (grown in LB broth at 37°C) were washed and then diluted in 0.9% NaCl to an optical density at 590 nm of 0.35. Diluted cells (150 μl) were added to each well, and the plates were incubated at 37°C without shaking. The optical density at 620 nm was measured by a Titertek Multiscan MCC/340 reader after 6 and 16 h of incubation.

Analysis of sulfur metabolism.

To investigate the sulfur metabolism in the deletion mutant, strains GT5969 and GT5970 were grown in MOPS minimal medium supplemented with different sulfur compounds as a sole sulfur sources. The carbon source in the experiment was succinate, and the cells were grown at 37°C. The following sulfur sources were used: S-methylcysteine, methionine sulfone, N-acetyl-l-cysteine, and isethionic acid. All chemicals were purchased from Sigma Chemical Co.

RESULTS

Isolation and characterization of ttcA1 mutants of E. coli and S. enterica.

E. coli GRB105 lacks cmo⁵U in its tRNA as a result of the aroE mutation (7) and is also deficient in m⁶t⁶A and in m²A (43). Moreover, tRNA from this strain is also deficient in a compound with migration properties similar to those of a compound suggested to be s²C on the basis of the published retention time (23). The mutation causing the lack of this unknown compound was named ttcA1, and conventional mapping placed it in the area of 28 to 33 min on the chromosomal map of E. coli. DNA was prepared from an ordered set of lambda transducing phages covering this region of the chromosome and subcloned. One of the obtained plasmids, pGJ8, complemented the ttcA1 mutation. Cloning of single open reading frames generated plasmid pGJ13, which contained an open reading frame, annotated as ydaO (b1344), and which complemented the ttcA1 mutation. We conclude that the annotated ydaO gene in E. coli is the ttcA gene.

We have for many years studied the formation of another thiolated nucleoside, ms²io⁶A37, in S. enterica. Moreover, we have isolated a mutant of S. enterica which is distinct from the ttcA1 mutant of E. coli and is defective in the synthesis of both s²C and ms²io⁶A (29a). To investigate the relationship between these two mutations, we decided to make an in-frame deletion in the ttcA (STM1654) gene of S. enterica and use this deletion mutant to study the physiological consequences of s²C32 deficiency for S. enterica. tRNA modification patterns from the two resulting congenic strains, GT5969 (ttcA⁺) and GT5970 (ΔttcA1), differing only in the allelic state of the ttcA gene, were determined by HPLC analyses. Only one compound was lacking in the tRNA from the mutant with respect to the tRNA from the congenic wild-type strain (Fig. 1). After introducing the plasmid pGJ13, which contains the E. coli ttcA⁺ gene, into the ΔttcA1 mutant, the same compound reappeared in the tRNA (data not shown). We identified this compound as s²C by the following criteria. (i) The migration in the chromatography system developed by Gehrke and Kuo (23) was similar to that observed for s²C in their analysis. (ii) The UV spectrum is the same as that published for s²C (23). (iii) By mass spectrometry, the molecular weight of the unknown compound was the same as that of the protonated form of s²C (data not shown). (iv) The ³⁵S-labeled nucleoside migrating as s²C is present in wild-type tRNA but absent in tRNA from the mutant (data not shown). Hence, we conclude that the ttcA gene product is required for the thiolation of C32 to form s²C32 in tRNA.

FIG. 1.

HPLC chromatograms of tRNA hydrolysates from wild-type ttcA⁺ (top) and the ΔttcA1 mutant (bottom). The nucleosides were monitored at 254 nm. The positions of U, s²C, mnm⁵s²U, Cm, m⁷G, and m⁵U are indicated. The arrow in the bottom panel denotes the position where s²C32 should migrate if present. Only part of the chromatogram (5- to 25-min retention time) is shown.

TtcA is part of a broadly distributed protein family.

Sequence analysis of TtcA reveals that it contains several motifs that are conserved in similar proteins from a wide variety of organisms. These elements are located in the central domain of the protein and include the SGGKDS motif (positions 47 to 52) characteristic of ATPases in the PP-loop superfamily as well as the C-X₁-X₂-C motif (positions 122 to 125), which is characteristic of proteins in the thioredoxin superfamily. Alignment of over 125 similar proteins from available sequence databases revealed that these proteins cluster into two major groups based both on elements in the intervening sequence of the central domain and on conserved motifs in the N and C termini (Fig. 2). The presence and juxtaposition of the PP-loop and C-X₁-X₂-C motifs (designated T1 in Fig. 2) is the unifying attribute for this family of proteins. The subgroup of this family that includes TtcA from S. enterica contains one additional C-X₁-X₂-C motif in the C terminus of the protein. The other major subgroup, which includes almost all family members from eukaryotic and archaeal species, contains two additional C-X₁-X₂-C motifs at the N terminus and five additional conserved cysteines at the C terminus of the protein, four of which are spaced as C-X₁-X₂-C motifs.

FIG. 2.

Domain structure of TtcA family proteins. There are two major subgroups of proteins closely related to TtcA, as exemplified by the known 74 bacterial members and 1 eukaryotic member of group 1 and by the 26 archeael, 6 bacterial, and 16 eukaryotic members of group 2. The presence of the four indicated motifs and their juxtaposition within the central domain is the unifying feature of the TtcA family as a whole. T1 is the only thioredoxin-like motif whose position is conserved in all family members. Sequences intervening between these motifs and the terminal domains of these proteins partition into two major groups, as shown. Lowercase letters represent residues that are not entirely conserved.

TtcA family members are present in all major bacterial taxa (where genomic data are available), except for the actinobacteria and cyanobacteria, but are present only sporadically in eubacteria outside of the β and γ proteobacteria (Fig. 3). In Firmicutes, homologs are found primarily in the clostridia, many of which contain several homologs in group 1. Horizontal transfer seems to account for the presence of this gene in many organisms as a phylogenetic tree representing the relationship between these proteins does not correspond in many places to a standard tree based on rRNA sequences (data not shown). Many archaeal species possess two distinct genes of the TtcA family, both in group 2. Family members from Aquifex, Thermotoga, and other “archaeon-like” eubacteria partition with the partial group of archaeal homologs (Archaea I in Fig. 3). All archaeal species represented in cluster Archaea I are also present in cluster Archaea II, which partitions more closely to the eukaryotic family members. There also exists one example of a eukaryotic group 1 homolog from Giardia lamblia.

FIG. 3.

Phylogenetic tree depicting the relationship between the various groups of TtcA family proteins. Group I homologs in the top half of the diagram cluster into several branches which constitute different eubacterial lineages, and the second of these also includes a representative from the Eucarya (Giardia lamblia). Group II homologs in the bottom half of the tree cluster into three distinct groups. The cluster labeled Archaea I + Eubacteria V includes TtcA homologs which are more similar to the group I proteins than to their counterpart archaeal homologs in the branch labeled Archaea II. The eubacterial representatives in this cluster are also predominantly from deeply branching “archaeon-like” bacteria.

The two cysteines at positions 122 and 125 in the central C-X₁-X₂-C motif of TtcA are both required for the formation of s²C.

Since the -CX₁X₂C- sequence is conserved in the TtcA protein, it was of interest to investigate its influence on the formation of s²C in tRNA. We therefore replaced these two Cys residues with Ala in strain GT6443 [ttcA2 (C122A; C125A)]. Next, we exchanged Cys122 for Ser122 [strain GT6635, ttcA3 (C122S)] and Cys125 for Ser125 [strain GT6636, ttcA4 (C125S)]. tRNA from these three strains was analyzed by HPLC, and all three strains lacked s²C (data not shown). Hence, both cysteines in the conserved T1 motif C-X₁-X₂-C are essential for the activity of TtcA in the formation of s²C in tRNA.

Lack of s²C does not affect the steady-state growth rate, nor does it show any growth disadvantage in a mixed-population experiment. Several mutants lacking a modified nucleoside in the tRNA grow more slowly than their wild-type counterparts do (8). We therefore investigated whether the s²C deficiency influences growth rate in various media. Steady-state growth cultures of the congenic strains GT5969 (ttcA⁺) and GT5970 (ΔttcA1) were grown in MOPS-glucose, MOPS-acetate, and Rich-MOPS media at 37°C. We did not observe any difference in growth rates between the wild type and the ΔttcA1 mutant in the three tested media (data not shown).

Next, we did a mixed-population experiment, which compares not only the growth in logarithmic phase but also the ability to sustain and to recover from stationary phase. The ΔttcA1 mutant and the wild type were mixed in a ratio of about 1:1 in LB medium. The culture was inoculated and grown at 37°C overnight into stationary phase and diluted 200-fold each day. On day 7 the bacteria were plated onto TYS plates. The colonies were screened for the mutant versus wild-type genotype by PCR. A total of 61 clones were tested from the samples taken on day 1, just after mixing the strains, and 99 clones were tested from the samples taken on day 7. The ratio between ΔttcA1 and ttcA⁺ cells was the same on day 1 as it was on day 7 following about 340 doublings in mass (day 1, 60.6% of the clones were ΔttcA1; day 7, 61.6% were ΔttcA1; the t test showed no statistical difference). Hence, the ttcA1 mutation did not influence the ability of the mutant to compete with the wild type in a growth competition experiment in the rich medium LB broth.

Lack of s²C does not affect the growth response to 32 amino acid analogs nor oxidation of various compounds.

It was shown previously that mutants deficient in ψ38, ψ39, and ψ40 (10), ms²io⁶A37 (21, 22), or m¹G37 (30) show an altered response to various amino acid analogs, suggesting an altered expression of the corresponding biosynthetic enzymes. We therefore tested 20 analogs, but the responses of the wild type and of the ΔttcA1 mutant were the same for all tested analogs.

Some tRNA modification mutants oxidize certain compounds differently from their congenic wild-type strains, suggesting a changed metabolism (22, 30). Therefore, we tested the ability of the ΔttcA1 mutant to oxidize 95 compounds. Indicator plates (Biolog ES MicroPlates) that contained different test compounds and the dye 2,3,5-triphenyl tetrazolium chloride were used (9). If the compound is oxidized, the redox dye is reduced and gives a purple color. However, the ΔttcA1 mutant responded similarly to the wild-type control to all compounds.

Sulfur metabolism of the ΔttcA1 mutant.

Although one phenotype induced by the deletion of the ttcA gene was the deficiency in thiolation of C32 in tRNA, this phenotype might be a secondary consequence of a defect in sulfur metabolism prior to the sulfur transfer reaction leading to the formation of s²C. We therefore tested the ability of the ΔttcA1 mutant to reduce thiosulfate or sulfite to H₂S. Compared to the wild type, no difference in H₂S production was noticed for the ΔttcA1 mutant (data not shown).

The two strains GT5969 (ttcA⁺) and GT 5970 (ΔttcA1) were grown overnight in MOPS-succinate minimal medium with 0.081 mM sulfate as the sole sulfur source. The cultures were diluted 20-fold into a different minimal medium containing 0.552 mM S-methylcysteine, methionine sulfone, N-acetyl-l-cysteine, or isethionic acid instead of sulfate as the sole sulfur source. The wild-type strain and the mutant strain grew equally well on the different sulfur sources.

The presence of s²C32 in tRNAmnm⁵UCUArg increases the rate of selection at A-site located AGG codons.

The synthesis of release factor 2 (RF2) requires a frameshift to avoid a stop codon in the zero frame. The frequency of frameshifting, which occurs in the P-site, is dependent on the rate of selection of the ternary complex (aa-tRNA/EF-Tu/GTP) at the codon in the ribosomal aa-tRNA site (A-site). This RF2-programmed frameshifting site has been placed upstream of the lacZ gene in such a way that the synthesis of β-galactosidase requires a shift into the +1 frame (13). At this RF2-programmed frameshifting site, frameshifting competes with the in-frame decoding at the A-site, which makes the frameshift-dependent β-galactosidase activity inversely proportional to the rate of aa-tRNA selection at the A-site. The , which contains s²C32, decodes CGU, CGC, and CGA. Plasmids with these codons as A-site codons in the programmed frameshifting site were introduced into the congenic strains GT5969 (ttcA⁺) and GT5970 (ΔttcA1). However, in all three cases, lack of s²C32 did not influence the rate of selection of (Data not shown).

The rare AGA codon is decoded by , whereas either or decodes AGG. The rate of selection of these tRNAs at AGA or AGG codons was determined by using a nonprogrammed +1 frameshifting site. We have previously shown that at a site CCC-N, in which the CCC codon is in the zero frame and the CC-N codon is in the +1 frame, the frameshifting event occurs by slippage at the CCC codon in the P-site (41, 42). The frequency of this slippage is inversely dependent on the rate of selection of the ternary complex at the A-site. Plasmids containing CCC AGA or CCC AGG as frameshifting sites were introduced into congenic strains differing only in the allelic state of the ttcA gene. As expected, in the wild-type strain the Arg-tRNA selection at AGG was less efficient (higher β-galactosidase) than at the AGA codon (Fig. 4). Interestingly, an s²C32-dependent enhancement of about 30% was observed for the selection at the AGG codon. We conclude that the presence of s²C32 in increases the Arg-tRNA selection at the AGG codon.

FIG. 4.

The rate of A-site selection at AGA and AGG codons in wild-type (open bars) and ΔttcA1 mutant (gray bars) strains was determined by the efficiency of programmed +1 frameshifting at the prfB gene, which replaces the regulatory UGA codon by the indicated codons (13). The +1 frameshifting by decoding CCC in the P-site is influenced by the rate of Arg-tRNA selection at the AGA or AGG codon at the A-site. The difference observed at the CCC-AGG site is significant based on a t test (P = 0.023).

The presence of s²C32 in tRNAmnm⁵UCUArg interferes with the decoding in the A-site.

The s²C32 modification is located on the 5′ side of the anticodon loop of the tRNA. When the tRNA is residing in the P-site, its 5′ side is facing the A-site tRNA. In an assay monitoring how changes in the P-site tRNA influence the aa-tRNA selection at the A-site, a nucleoside change in position 32 had a strong influence (49). Therefore, it was of interest to investigate whether the thiolation of C32 also influenced the ability of tRNA to read the A-site codon. To do this we used an assay system that monitors the influence of the P-site tRNA on readthrough/termination (level of suppression) at the UAG codon in the A-site. The assay measures the interference in vivo by the P-site-located peptidyl-tRNA with the decoding activity in the A-site. Only at one site, AGG-UAG, of the seven sites tested did we observe a difference between the mutant and the wild type (Fig. 5). In that site, at the time when the suppressor tRNA reads the UAG codon, the P-site AGG codon is decoded either by or by . A 40% increase in the readthrough was observed in the mutant, suggesting that the presence of s²C32 in the P-site tRNA decreased the decoding efficiency of the UAG suppressor tRNA or increased the ability of RF1 to decode UAG.

FIG. 5.

Influence of the s²C32-containing peptidyl-tRNA on supF30-mediated suppression of UAG. Various plasmids containing P-site codon(s) CGA CGC, CGU, AGA, AGC, AGG, or AGU were introduced into wild-type (open bars) and ΔttcA1 mutant (gray bars) strains, and readthrough was monitored by measuring β-galactosidase activity. A significant difference between the wild type and the mutant was observed only in the case of AGG as the P-site codon (P < 0.0003).

s²C32 in tRNAICGArg decreases the rate of translation of the CGA codon but not of the CGU codon.

One way to increase a small difference in the efficiency of reading a codon is to measure the rate of migration of the ribosome over a repeated sequence of the same codon. Curran (12) has constructed two plasmids (pCGA9 and pCGU9), which have nine consecutive CGA or CGU codons, respectively. These consecutive codons are translationally coupled to the lacZ mRNA in such a way that the synthesis of β-galactosidase requires that the ribosomes translate the upstream cistron with the consecutive codons, terminate, and then reinitiate at the AUG codon of lacZ (12). The frequency of reinitiating is dependent on the number of ribosomes terminating at the end of the upstream cistron. In wild-type cells the CGU codons were translated more rapidly than the CGA codons (Fig. 6), in agreement with earlier results (12). Whereas there was no difference between the wild type and the ΔttcA1 mutant in reading the CGU codons, a twofold increase was observed in the mutant in reading the CGA codons. Surprisingly, the lack of s²C32 in increased the rate of decoding of the CGA codons, revealing that the presence of this modified nucleoside may have a negative modulating effect on the activity of .

FIG. 6.

Efficiency of tRNA in translating nine consecutive CGA codons (pJC1113) (A) or nine consecutive CGU codons (pJC1119) (B) (12). β-Galactosidase activity (β-gal) was measured in wild-type (open bar) and ΔttcA1 mutant (gray bar) strains. Standard error is indicated by the error bars. A t test indicated that the difference in the wild type and in the ΔttcA1 mutant was significant in panel A (P = 0.0003)

DISCUSSION

The IscS protein is pivotal in the synthesis of the five thiolated nucleosides present in tRNA of S. enterica. Whereas the synthesis of s⁴U and the s²U moiety of (c)mnm⁵s²U has recently been elucidated (26, 27, 38), the syntheses of ms²io⁶A and s²C following the participation of IscS are much less understood. Studies of the ΔiscS mutant revealed that formation of ms²io⁶A and s²C can proceed, albeit with low efficiency, in an IscS-independent way (29, 37). Thiolation of both of those nucleosides requires the presence of IscU, which is a scaffold for [Fe-S] cluster assembly (29a). This suggests that [Fe-S] proteins are involved in the thiolation of s²C32 and ms²io⁶A37. MiaB, which catalyzes the last step in the synthesis of ms²io⁶A, is a Radical SAM enzyme containing an [Fe-S] cluster (39). Biochemical characterization of the TtcA protein, which contains several conserved cysteines, will reveal whether TtcA is the [Fe-S] protein involved in the synthesis of s²C32. Still, we conclude that the TtcA peptide appears to act in a manner distinct from the ThiI and MnmA enzymes but may share the requirement of [Fe-S] cluster assembly proteins with MiaB.

Proteins closely related to TtcA are found in a wide variety of organisms in all three domains, but s²C32 has not been identified in many of these (e.g., Saccharomyces cerevisiae contains a TtcA family member, but no s²C is known to exist in its tRNA). There are only limited data concerning the taxonomic distribution of s²C32 itself, but it is consistent to speculate that group 1 family members, as defined here (Fig. 2), play a role in mediating reactions that ultimately lead to the production of s²C32 in certain tRNAs and to assume that group 2 family members from the Eucarya and the Archaea II group (Fig. 3) are present in species where enzymatic interactions resulting from a similar type of reaction do not lead to the formation of s²C32 in tRNA. The central domains of all these proteins contain enough similar features to suggest that they are all performing reactions of a similar nature, perhaps to generate a sulfur intermediate that is involved not only in the synthesis of s²C32 in some species but also in another part of general sulfur metabolism. In species where a TtcA family member does in fact mediate interactions that lead to the formation of s²C32 in tRNA, it is likely that the enzyme associates directly with a tRNA substrate because of the high degree of similarity to other PP-loop proteins known to interact with tRNA (e.g TilS [51]).

It is generally assumed that the tRNA anticodon loop comprises 7 nucleotides. However, since a bifurcated hydrogen bond contact is present between the first base (position 32) and the last base (position 38) in the anticodon loop (44, 56), such a loop comprises 5 instead of 7 nucleotides. Although this contact is with only one hydrogen bond between the residues at positions 32 and 38 (Fig. 7), it is surprisingly strong, since it is not disrupted even if the tRNA is severely unfolded by its interaction with its cognate aa-tRNA synthetase (14, 15, 45, 46). All four s²C32-containing tRNAs have an A38, and thus an S2(s²C32)—N6(A38) bifurcated hydrogen bond is present (3). However, the N-H—S hydrogen bond may be weaker than the corresponding N-H—O hydrogen bond (18). If so, a change from an S2(s²C32)—N6(A38) hydrogen bond to an O2(C32)—N6(A38) hydrogen bond as a result of a mutation in the ttcA gene should strengthen the interaction between these bases. On the other hand, a sulfur atom stabilizes stacking interactions by its greater polarizability than oxygen as well as its contribution to dispersion forces. Since the base at position 32 is partially stacked between bases at positions 31 and 33, the change from s²C32 to C32 in the ttcA mutant should destabilize the interaction between positions 32 and 38. The final influence on the stability of the anticodon loop by these two opposing forces is not clear. In either case, the structure of the anticodon of E. coli free in solution is influenced by s²C32, since the nuclease digestion pattern is dependent on the presence of this modification (5). We have shown that selection of to the AGG codon (Fig. 4), peptidyl-tRNA^Arg interference with A-site decoding (Fig. 5), and the rate of decoding of CGA by (Fig. 6) are dependent on s²C32. These results also suggest that this postulated s²C32-A38 interaction is important to keep the anticodon loop in its proper 5-base-loop structure.

FIG. 7.

Base pairing through a bifurcated hydrogen bond between the first (C32) and the last (A38) nucleotides in the anticodon loop. A dashed line indicates a hydrogen bond. This diagram was modified from reference 3.

Arginylation of in vitro, formation of a ternary complex, and binding of to the four CGN triplets were not dependent on s²C32 (28). We also did not observe any dependence on s²C32 for the selection at the A-site-located CGN codons (data not shown). However, the rate at which read nine contiguous CGA codons was higher for s²C32-depleted than for wild-type s²C32-containing (Fig. 6). This may seem inconsistent, since the wild-type form of the tRNA is apparently less efficient that the mutant form (Fig. 6). However, the translation apparatus is optimized and reflects a trade-off between speed and cost of accuracy (19). Therefore, the role of s²C32 in this optimization appears to be as an enhancement of fidelity. We observed the s²C32 dependence of translation rate only when decoded the CGA codon, not when it decoded the CGU codon (Fig. 6). The latter codon is more efficiently decoded than the CGA codon due to the poor interaction of the I34 wobble nucleoside with the A-ending codon (Fig. 6) (12). Thus, the impact of s²C32 on the function of the tRNA appears to depend on which codon is read. Alternatively, the impact of s²C32 abolition may be more easily observed when the interaction is inefficient. Similarly, we observed an s²C32 dependence in the decoding of AGG but not of AGA. The AGA codon is read by tRNAmnm⁵UCUArg, whereas the AGG codon is read efficiently by and less well by . We observed an increased readthrough at the A-site-located UAG codon when or tRNAwas present in the P-site and interacting with AGG, but not when was present in the P-site and interacting with AGA (Fig. 5). Since it is not likely that the tRNA in the P-site directly interacts with a tRNA in the A-site, this interference may be caused by a direct interaction between the P-site tRNA and RF1 decoding the stop codon in the A-site. Alternatively, the P-site-located tRNA may induce a conformational change of the A-site that influences the decoding by either RF1 or a tRNA causing readthrough. Considering these findings together, although the lack of s²C32 does not influence the physiology of the bacterium as monitored by growth rate, several specific steps in the translational decoding process were affected, suggesting that the function of s²C32 could be to modulate the flexibility of the anticodon loop, thereby optimizing the decoding process.