The miaA Mutator Phenotype of Escherichia coli K-12 Requires Recombination Functions

Abstract

miaA mutants, which contain A-37 instead of the ms²i⁶A-37 hypermodification in their tRNA, show a moderate mutator phenotype leading to increased GC→TA transversion. We show that the miaA mutator phenotype is dependent on recombination functions similar to, but not exactly the same as, those required for translation stress-induced mutagenesis.

The miaA gene encodes a tRNA prenyltransferase that catalyzes the addition of a Δ²-isopentenyl group from dimethylallyl diphosphate to the N⁶-nitrogen of adenosine adjacent to the anticodon at position 37 of 10 of 46 Escherichia coli tRNA species that read codons starting with U residues (i⁶A-37 in Fig. 1) (3, 5, 19, 26, 35). In E. coli, the i⁶A-37 tRNA modification is further methylthiolated by the action of the miaB gene product, which is dependent on iron, and possibly another enzyme activity (MiaC) to form ms²i⁶A-37 (Fig. 1), except in tRNA^Sec (11, 12). Methylthiolation is dependent on prior formation of i⁶A-37 by the MiaA tRNA prenyltransferase (3), and miaA mutants contain fully unmodified A-37 residues in their tRNA.

FIG. 1.

Formation of the ms²i⁶A-37 modification in E. coli tRNA by the MiaA and MiaBC enzyme activities. See the text for details. Ubi, ubiquinone; IPP, isopentenyl diphosphate; DMAPP, dimethylallyl diphosphate; Met, methionine; Cys, cysteine; SAM, S-adenosylmethionine; SAH, S-adenosylhomocysteine.

The presence of A-37 instead of the ms²i⁶A-37 tRNA modification in miaA mutants of E. coli and Salmonella enterica serovar Typhimurium results in multiple defects in translation efficiency, codon context sensitivity, and fidelity (8, 20). These translation defects impart broadly pleiotropic phenotypes to miaA mutants, including decreased growth rate and yield (10), altered sensitivity to amino acid analogs (10), increased oxidation of certain amino acids and tricarboxylic acid cycle intermediates (32), altered utilization of primary carbon sources (32), and suppression of Tet(M) protein-induced tetracycline resistance (4). In contrast, the presence of i⁶A-37 instead of ms²i⁶A-37 in miaB mutants leads to milder defects in translation than those of miaA mutants (8, 11).

We demonstrated previously that miaA mutants exhibit a moderate mutator phenotype that results in GC→TA transversion mutations (5, 6). The genetic basis for this miaA mutator phenotype has not been determined, nor has it been determined whether miaB mutants show a mutator phenotype. However, it was shown that the miaA mutator phenotype was not due to polarity on the expression of the downstream hfq gene, which encodes a pleiotropic regulator, or hflA-region genes, which encode a protease (32). Recently, Humayun and coworkers described a new pathway of mutagenesis that results in increased transversion mutations in response to translation stress (1, 17, 28, 29). This translation stress-induced mutagenesis (TSM) pathway is induced in mutA mutants, which contain an anticodon mutation in the glyV tRNA gene that causes insertion of glycine instead of aspartic acid in proteins, such as the proofreading subunit of DNA polymerase (30, 31). Unexpectedly, the TSM pathway was recently shown to depend on recombination functions (1, 28, 29). In reviewing inducible mutagenic pathways in E. coli, Humayun hypothesized that the miaA mutator phenotype may be a manifestation of the TSM pathway (17). In this report, we show that the miaA mutator phenotype is dependent on recombination functions similar to, but not exactly the same as, those required for the TSM pathway. We also demonstrate a correlation between the GC→TA mutator phenotype and the severity of defects in translation in different tRNA modification-deficient mutants.

miaA mutator phenotype depends on recombination functions.

Previously, we detected the miaA mutator phenotype in the Cupples-Miller mutation tester strain CC104, which can revert to lacZ⁺ only by a specific GC→TA transversion, but not in the five other tester strains for the remaining transversion and transition mutations (6). To determine the genetic requirements of the miaA mutator phenotype, we constructed strains in which the miaA tRNA modification mutation was combined with mutations defective in recombination, the SOS response, or some pathways of DNA repair (Table 1).

TABLE 1.

Bacterial strains constructed in this study

Strain or plasmid	Genotype or descriptiona	Source or referenceb
Strains
BW17593	DE3 lacX74 Cre(wild type) phn(EcoB) recD1903::Tn10-11Tetr	B. Wanner collection
CC104	P90 C ara Δ(lac-pro)XIII (F′ lacIZ proB+); GC→TA tester strain	7
CC105	P90C ara Δ(lac-pro)XIII (F′ lacIZ proB+); AT→TA tester strain	7
CGSC6661	uvrA::Tn10 λ− IN(rrnD-rrnE)	22
GE922	MC4100 malE::Tn10 lexA3(Ind−)	G. Weinstock collection
GE1878	KM1230 strA tsl-1 recA430	G. Weinstock collection
N2101	recB268::Tn10 Tetr	21
NU611	W3110 hisT::Kanr(BstEII)	2
RW82	ΔumuDC595::cat uvrA6 trp ilv-325 thy-325 thr-1 leu6 proA his-4 argE3 thi-1 lacY1 galK2 ara-14 xyl-5 mtl-1 tsx-33 strA31 supE44	36
SH2101	S90C polBΔ1 Strr Sper	13
TX2547	JC7623 miaA::ΩCmr	32
TX2591	CC104 miaA::ΩCmr	CC104 × P1vir(TX2547)
TX2605	CC104 hisT::Kanr	CC104 × P1vir(NU611)
TX2658	CC105 miaA::ΩCmr	Lab stock
TX3081	CC104 mutY::mini-Tn10 Kanr	Lab stock
TX3128	CC104 miaA::ΩCmrmutY::mini-Tn10 Kanr	TX3081 × P1vir(TX2591)
TX3224	CC104 ΔumuDC::cat	CC104 × P1vir(RW82)
TX3289	CC104 miaA::ΩKanr ΔumuDC::cat	TX2591 × P1vir(RW82)
TX3303	CC104 uvrA::Tn10 Tetr	CC104 × P1vir(CGSC6661)
TX3327	CC104 miaA::ΩCmruvrA::Tn10Tetr	TX2591 × Plvir(CGSC6661)
TX3328	CC104 miaA::ΩCmrrecA430	TX2591 × P1vir(GE1878)
TX3346	CC104 miaB::Tn10d Cmr	12
TX3351	CC104 recA430	CC104 × P1vir(GE1878)
TX3935	CC104 polBΔl	CC104 × P1vir(SH2101)
TX3938	CC104 miaA::ΩCmrpolBΔl	TX2591 × P1vir(SH2101)
TX4183	CC104 lexA3(Ind−)	CC104 × P1vir(GE922)
TX4184	CC104 miaA::ΩCmrlexA3(Ind−)	TX2591 × P1vir(GE922)
TX4201	CC104 recD1903::Tn10 Tetr	CC104 × P1vir(BW17593)
TX4202	CC104 miaA::ΩCmrrecD1903::Tn10 Tetr	TX2591 × P1vir(BW17593)
TX4204	CC104 recB268::Tn10 Tetr	CC104 × P1vir(N2101)
TX4206	CC104 miaA::ΩCmrrecB268::Tn10 Tetr	TX2591 × P1vir(N2101)
Plasmids
PControl	pLS4 (pBR322 with Kanr)	16
PMutL	MutL+ cloned in pLS4; overexpresses MutL	16
PMutS	MutS+ cloned in pLS4; overexpresses MutS	16

^aCm^r, Kan^r, Str^r, Spe^r, and Tet^r indicate resistance to chloramphenicol, kanamycin, streptomycin, spectinomycin, and tetracycline, respectively. 

^bStrains were constructed by generalized transduction by phage P1vir (25) propagated on the indicated donor strains. 

Mutation frequencies were determined essentially as described by Cupples and Miller (see Fig. 2) (7). Briefly, strains were grown overnight to saturation at 37°C with shaking in 5 ml of Luria-Bertani broth (10 g of NaCl per liter) supplemented with 30 mg of l-cysteine per liter. Bacterial cells were washed twice by collection by low-spread centrifugation (≈4,000 × g) and resuspension in 5 ml of minimal (E) salts (9) lacking a carbon source. Bacterial pellets were resuspended in 0.5 ml of minimal (E) salts. Resuspended cells (0.1 ml) were spread onto plates containing minimal (E) salts, 0.4% (wt/vol) lactose, 0.01 M FeSO₄, and 1.5% (wt/vol) Bacto agar or serially diluted in minimal (E) salts and spread onto plates containing 0.4% (wt/vol) glucose instead of lactose. Plates containing lactose or glucose were incubated at 37°C for 3 days or 1 day before colonies were counted to determine the number of lacZ⁺ revertants or viable F′ episome-containing bacteria, respectively. Reconstitution experiments in which fixed numbers of lacZ⁺ revertants of CC104 miaA⁺ and CC104 miaA were purposefully added to cultures before plating indicated that CC104 miaA⁺ lacZ⁺ colonies began to appear after 1 day and reached a maximum number that remained constant on days 2 and 3 (data not shown). In contrast, CC104 miaA lacZ⁺ colonies did not appear until after 2 days and reached a maximum number by 3 days (data not shown); therefore, mutation frequencies were determined after 3 days of incubation at 37°C.

FIG. 2.

Mutation frequencies of Cupples-Miller tester strains CC104 and CC105, which can revert to lacZ⁺ only by GC→TA and AT→TA transversions, respectively (7), containing a miaA mutation combined with mutations defective in recombination, the SOS pathway, or DNA repair. Mutation frequencies were determined as described in the text. Experiments were performed independently at least three times, and standard errors of the means are shown.

We could not detect the miaA mutator phenotype in CC104 miaA containing mutations in recA, recB, or recD, which are defective in homologous recombination functions (1, 17, 29) or in CC105 (AT→TA tester) as reported previously (6) (Fig. 2). These patterns were confirmed in qualitative experiments in which blue lacZ⁺ papillae were observed in lawns of bacteria on plates containing minimal medium and 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal) (data not shown). In these experiments, we used the recA430 allele, which is partially defective in homologous recombination and SOS induction (18), because strains containing this mutation grow faster than recA null mutants. Still, this leaky recA allele prevented the miaA mutator phenotype. The recB or recD transposon insertion mutations abolish homologous recombination (18) or result in hyper-recombination in some cases (18) and increased F′ episome replication (15), respectively. In contrast to recA, recB, and recD mutants, the miaA mutator phenotype was still observed to varying extents in CC104 miaA (GC→TA tester) containing lexA(Ind⁻), polB, uvrA, umuDC, and mutY mutations, which confer defects in SOS induction, DNA polymerase II, nucleotide excision repair, SOS mutagenesis, and GO repair, respectively (Fig. 2) (1, 13, 17, 24, 29, 37). Although the miaA mutator phenotype was evident, its extent was somewhat reduced in the lexA(Ind⁻) (2.3-fold) and mutY (1.8-fold) mutants compared to that of the CC104 miaA⁺ and CC104 miaA pair (3.7-fold) (Fig. 2).

The absence of the miaA mutator phenotype in both recA and recB mutants is consistent with the interpretation that this phenomenon depends on recombination functions, similar to TSM (1, 17, 28, 29). Likewise, the miaA mutator phenotype and TSM were not dependent on polB and umuDC functions (Fig. 2) (1, 27). However, there are noteworthy differences between the miaA mutator phenotype and TSM. Only GC→TA transversion was increased in miaA mutants, whereas the mutA mutator phenotype, which has been hypothesized to induce TSM (17), led primarily to an increase in AT→TA transversions in tester strain CC105 (7.8-fold) as well as AT→CG and GC→TA transversions in tester strains CC101 and CC104 (5.1-fold and 3.7-fold, respectively) (23). The recD mutation abolished the miaA mutator phenotype, whereas a recD mutation failed to affect induction of TSM in mutA mutants (29). Finally, the miaA mutator phenotype was reduced, but not abolished, by the lexA(Ind⁻) mutation, whereas TSM was not reduced in cells unable to induce the SOS response (28). Nonetheless, the striking dependence of both the miaA mutator phenotype and TSM on recombination functions suggests that miaA mutations may partly induce the TSM response, albeit more weakly than mutA mutations. Testing of this hypothesis will require a much deeper understanding of the mechanism of the TSM response and of the reasons why it depends on recombination functions, especially in cells lacking preformed DNA lesions (28).

miaB and hisT tRNA modification mutants lack a mutator phenotype.

We tested whether other tRNA modification mutations induced GC→TA transversion mutations (Fig. 3). We found that the hisT mutant, which lacks pseudouridine residues at positions 38, 39, and 40 in the anticodon stem-loop of tRNAs (3, 8, 33, 35), has a barely detectable mutator phenotype. For reasons that are not understood, there was more variation in lacZ⁺ reversion of hisT mutants than in that of miaA mutants (Fig. 3). Mutation of miaB did not cause a mutator phenotype.

FIG. 3.

Mutation frequencies of strain CC104 (GC→TA tester) containing a miaA or miaB mutation or a hisT mutation and thus defective in ms²i⁶A-37 or pseudouridine tRNA modification, respectively. Experiments were performed as described for Fig. 2.

These results have three implications. First, induction of GC→TA transversion, and possibly TSM, seems to be correlated with the severity of translation defects and pleiotropic phenotypes caused by tRNA undermodification, where miaA > hisT > miaB (3, 8, 33). Second, undermodification of ms²i⁶A-37 to i⁶A-37 in miaB mutants does not induce mutagenesis. The i⁶A-37 tRNA undermodification also occurs in miaA⁺ bacteria that are subjected to mild limitation for iron, and it has been suggested that this undermodification acts as a physiological switch in response to this stress (3). Our results show that this undermodification likely does not act as a switch to increase GC→TA transversion. Third, tRNA modifications in response to other stress conditions have been reported (3). With the exception of oxygen-dependent hydroxylation of ms²i⁶A-37 to form ms²io⁶A-37 in Salmonella (3), the mechanisms and physiological implications of these tRNA undermodifications remain largely unexplored (3). It is possible that other stress conditions may trigger GC→TA transversion, and possibly the TSM pathway, through tRNA undermodification in wild-type bacteria.

Overexpression of MutS mismatch repair protein suppresses the miaA mutator phenotype.

Recently we reported that overexpression of the MutS DNA mismatch binding protein leads to a recA-independent decrease in GC→TA transversion mutations in wild-type E. coli and mutY mutants (37). Therefore, we tested whether overexpression of the MutS and MutL mismatch repair proteins decreases GC→TA transversion in miaA mutants (Fig. 4). We found that overexpression of MutS or MutL suppressed the miaA mutator phenotype, with MutS eliciting the stronger effect (Fig. 4). This result suggests that the miaA mutator phenotype, and possibly part of the TSM response, may involve a mismatch recognizable by the MutS protein, which is present at just-sufficient levels in exponentially growing E. coli cells and is strongly down-regulated in bacterial cells that are in stationary phase, such as those used in these experiments (14, 34).

FIG. 4.

Mutation frequencies of strain CC104 miaA (GC→TA tester) containing an empty vector plasmid (pControl) or plasmids that overexpress the MutL (pMutL) or MutS (pMutS) mismatch repair proteins by about 20- and 60-fold, respectively (16). Experiments were performed as described for Fig. 2, except that 50 μg of kanamycin per ml was added to growth media and plates to maintain the plasmids.