Previous Article |
Table of Contents
| Next Article 
BIOLOGICAL SCIENCES / GENETICS
An editing-defective aminoacyl-tRNA synthetase is mutagenic in aging bacteria via the SOS response
The Skaggs Institute for Chemical Biology, and Departments of Molecular Biology and Chemistry, The Scripps Research Institute, 10550 North Torrey Pines Road, BCC-379, La Jolla, CA 92037
Contributed by Paul Schimmel, December 6, 2006 (received for review November 30, 2006)
 |
Abstract |
|---|
 Top Abstract ResultsDiscussionMethodsAcknowledgementsReferences |
|---|
aminoacylation errors | error-prone DNA polymerases | amino acid misincorporation | genetic code ambiguity
For example, mice that carry a mild mutation in the editing domain of alanyl-tRNA synthetase suffer from ataxia (1). The mutant alanyl-tRNA synthetase mischarges tRNAAla with serine and, as a consequence, mistranslation occurs that leads to induction of the unfolded protein response. That, in turn, leads to the degeneration of Purkinje cells in the cerebellum, beginning within 3 weeks of age. Similarly, in human cell culture, an inducible, editing-defective valyl-tRNA synthetase caused cellular degradation and apoptosis (2).
With these recent results in mind, we were motivated to investigate the possibility that mistranslation could also lead to heritable genetic changes. Previously, we generated a knock-in mutant allele of ileS that encodes no functional editing domain (3). This ileSAla allele has a critical section of the coding sequence of the editing domain replaced with 11 codons for alanine. Relative to the isogenic wild-type strain, this strain had a diminished growth rate, was highly sensitive to a structural analogue of isoleucine, and had an increased sensitivity to antibiotics (4). Surprisingly, the editing-defective mutant was no more sensitive to mutagens and had no greater mutation frequency than the isogenic wild-type strain (4). Given the gradual cellular degeneration observed in Purkinje cells in response to an editing-defective alanyl-tRNA synthetase (1), we wondered whether aging might also lead to accumulated genetic degeneration in a bacterial model system. If so, then a connection of mistranslation might be made to the long-standing observations of the accumulation of genetic mutations in aging individuals that lead to disease.
|
Results |
|---|
|
TopAbstractResultsDiscussionMethodsAcknowledgementsReferences |
|---|
Following this line of reasoning, the susceptibility to mutation of wild-type and ileSAla (editing-defective) Escherichia coli was explored under conditions of aging. Such studies are referred to as MAC. These require plating bacteria on solid media for either 1 or 7 days, followed by determining the frequency of spontaneous rifampicin (Rif)-resistant (RifR) mutants in the population after each time period (5–7). This phenotype is known to at least partially depend on the SOS response. The latter response includes, among others, the LexA-dependent induction of error-prone DNA polymerase (pol) II (encoded by polB), pol IV (encoded by dinB), and pol V (encoded by umuDC). [DNA pol II is a high-fidelity DNA pol when restarting replication, but error-prone when correcting certain types of DNA damage (8).]
We determined the MAC phenotype for wild-type and ileSAla E. coli. Bacteria were allowed to grow as colonies for either 1 or 7 days. At the end of each period, all of the colonies on each assay plate were collectively resuspended in liquid media and titered for viable cells and the number of RifR cells in the population. Little difference in the spontaneous appearance of RifR colonies was observed between strains after 1 day (Mann–Whitney U test; n = 67 and 68 for wild-type and ileSAla strains, respectively; P = 0.8) (9). After 7 days incubation on LB agar, the frequency of spontaneous mutants increased in both strains, but to a significantly greater extent in the editing-defective strain (Mann–Whitney U test; n = 69 for both strains; P < 0.05).
Wild-type E. coli (strain MG1655) is reported to have a 5.5-fold increased frequency of RifR clones after MAC (comparing median of day 7 to median of day 1) (5). Similarly, the wild-type strain used here, which is derived from strain MG1655, had an increased mutation frequency of 
6-fold (Fig. 1A). By comparison, the median value of RifR frequency increased 15-fold in the editing-defective strain. The change in mutation frequency at day 7 is highly significant relative to day 1, for both editing-defective and wild-type strains (Mann–Whitney U test; P << 0.0005 for each).
|
The putative high- and low-mutation frequency subpopulations were separately analyzed for statistical significance. A Mann–Whitney U test revealed a significant difference between the highly mutated ileSAla and wild-type subpopulations (P < 0.005), whereas the ileSAla and wild-type subpopulations that experienced lower levels of mutations were not significantly different (P = 0.14). (The same test revealed that the subpopulations with lower mutation rates after 7 days still had a significantly greater mutation frequency than after 1 day; P << 0.0005 for both wild-type and ileSAla strains.) The bimodal distribution is indicative of a subpopulation that is experiencing a higher mutation rate, and this effect is greatly exacerbated in the editing-defective strain. Median mutation frequencies for the more frequently mutating subpopulations were much greater than for the population as a whole. The editing-defective strain had increased in mutation frequency 330-fold, whereas the wild-type strain had an increased mutation frequency of 95-fold.
DNA Sequence Analysis of Mutated rpoB.
The target of Rif is the 
-subunit of RNA pol, so that mutations that confer RifR localize to the rpoB gene (10). In particular, RifR mutations typically fall in one of three clusters of rpoB. These regions have been extensively sequenced and characterized in earlier work (10–12). To examine the mutations in detail, 48 clonal DNA isolates were sequenced for each genotype (wild-type and ileSAla) after each time period (1 and 7 days). Of course, clones isolated from the same plate (LB plus Rif) could not be definitively characterized as independent mutations if the genetic change was identical. As a consequence of this consideration, we were limited to 39 and 29 unique mutants for ileSAla strains and 38 and 30 unique mutants for the wild-type strains on days 1 and 7, respectively.
The frequencies of each type of transition and transversion and the total frequency of insertions were determined (Fig. 2A). The GC 
AT transition was most common and was identified in >50% of all mutations for both the wild-type and ileSAla strains. The frequency of GC TA transversions increased by 2-fold on day 7 relative to day 1. This distribution of base-pair changes is similar to previous RifR mutations that were caused by MAC in other tested strains (5). Thus, the enhanced mutation rate of the ileSAla strain yields the same distribution of specific transitions and transversions as seen with the wild-type organism (
2 test, P = 0.82 for day 1; 2 test, P = 0.47 for day 7).
|
AT mutation is a signature for DNA-repair-defective mutants (including ung, nth, nei, ada, and ogt) (5, 13, 14). Thus, the products of one or more of these genes are failing to adequately function under conditions of MAC. In addition, mutS can be implicated in GC AT mutations, as MutS is known to be strongly down-regulated in stationary phase (15) and may therefore be failing to function at adequate levels under MAC. Significantly, increased expression of wild-type mutS in cells subjected to MAC is reported to diminish the MAC phenotype (5). However, mutS is associated with both GC AT and AT GC mutations (12). The paucity of AT GC mutations suggests that a loss of MutS activity is not a significant contributor to the MAC phenotype in this case. Finally, two known mechanisms (mutYM or miaA) may be responsible for the increased frequency of GC TA mutations (12, 16–18). In addition, these two mechanisms may be functioning to differing extents in the editing-defective and wild-type strains. We also analyzed the positions altered within the RNA pol coding sequence. The frequency with which any given residue is found changed was mapped onto a linear representation of the protein sequence (Fig. 2B). Few of these resulted in mutations that were unique or significantly more numerous in either the wild-type or ileSAla strain. We therefore chose a different approach to understanding the source of different mutation frequencies in the two strains when exposed to MAC.
Role of SOS Response in MAC. Some of the genes associated with the SOS response have been shown to be involved in MAC (5–7). The SOS response is activated by the presence of ssDNA (8, 19). RecA protein binds to ssDNA, forming a nucleoprotein filament that is then able to activate the self-cleavage activity of LexA. Once cleaved, LexA is no longer capable of repressing the SOS regulon. Among the genes activated by the SOS response are polB, dinB, and umuDC, which encode DNA pols II, IV, and V, respectively. These proteins function to repair DNA damage and are known to be low-fidelity DNA pols.
Exposure to ciprofloxacin (Ci) (20, 21) and related quinolone antibiotics (22) are known to induce the SOS response. These antibiotics target DNA gyrase and toposiomerase IV (23), which relax supercoiling during DNA replication. Ci binds to these proteins and prevents rejoining of DNA ends. RecA protein can then bind to these free DNA ends and ultimately induce the SOS response through LexA. When E. coli with an active SOS response is exposed to Ci, resistant mutants continue to arise over the course of several days (20, 21). Thorough genetic studies showed that the only mutants that arise in E. coli lacking the SOS response are those that were acquired before exposure to the drug (21, 24). For this reason, the accumulation of CiR mutants after exposure to Ci is diagnostic for the SOS response.
A dramatic difference was observed between editing-defective and wild-type strains in the accumulation of Ci-resistant (CiR) mutants over time (Fig. 3A). After 2 days of growth on minimal media with Ci, the editing-defective strain acquired a much greater mutation rate than did the wild-type strain. This observation further supports the idea that ileSAla strains accumulate CiR mutations at an increased rate, because of exposure to the antibiotic and concomitant induction of the SOS response.
|
To test directly for the involvement of the SOS response, the editing-defective strain was combined with either a lexA derivative that was incapable of self-cleavage [lexA(S119A), which is a constitutive repressor of the SOS regulon] or an isogenic wild-type construct. These strains were tested for the accumulation of CiR mutants. The constitutive repressor lexA mutants resulted in 10-fold diminished spontaneous CiR mutants in overnight cultures of both the wild-type and editing-deficient strain. This result is consistent with the SOS response being active even in culture conditions. The differences in mutation rate between the ileSAla/lexA(S119A) and the ileSAla/lexA+ strains was especially striking (Fig. 3C). In particular, the lexA-deficient, editing-defective strain had a far lower mutation rate. These results demonstrated that the inability to induce the SOS response resulted in a dramatically diminished accumulation of CiR mutants.
|
Discussion |
|---|
|
TopAbstractResultsDiscussionMethodsAcknowledgementsReferences |
|---|
There are several candidate enzymes that may induce the SOS response by increasing the frequency of breaks in DNA (25). These include topoisomerases and gyrases (the same class of enzymes that are the targets of Ci), which relax supercoiling. Similarly, recombinases such as RecG or RuvC promote strand migration to begin the process that ultimately relieves lesion blockage of pols. PriA and pol II are candidates as well because they reestablish strand polymerization after lesion repair. If any of these enzymes were to malfunction because of translational errors, free DNA ends may result. [In addition to the effect of mistranslation on SOS-induced enzymes, we considered more general effects of mistranslation that could induce the SOS response. For example, based on the types of rpoB mutations (especially GC AT; Fig. 2A), one or both of oxidation and alkylation are occurring at elevated levels in aging bacteria (5, 13, 14). Lesions that cause DNA replication to stall or arrest may occur if the enzymes normally responsible for repair of DNA oxidation or alkylation are impaired because of mistranslation in the editing-defective strain (25). These lesions would result in DNA breaks.] In either case, the free DNA ends would initiate the formation of RecA nucleoprotein filaments that can then induce LexA self-cleavage and induce the SOS response. Of course, the SOS response itself is likely to be mutagenic.
It was a formal possibility that LexA with translational errors have a diminished ability to regulate the SOS response, which results in SOS response induction. However, the SOS response remains inhibited in the lexA(S119A) mutant (Fig. 3C), showing that the regulatory function of LexA is retained.
Historically, the major assay for MAC has been the spontaneous accumulation of mutations that confer Rif resistance (5–7). We followed this precedent, recognizing that the detection of mutations caused by MAC can vary depending on the antibiotic resistance being tested (5). This variance is doubtless caused by intrinsic differences in the mechanism of action of each antibiotic, making some better than others for mutation detection. (We confirmed this kind of variance by testing for resistance to fosfomycin, which, although weaker with regard to the frequency of spontaneous mutants that result in resistance to Rif, has previously been shown to have a significant MAC phenotype. The results with the wild type and the editing-defective strain were inconclusive.)
In our work with Rif, the dramatic increase after 7 days in the frequency of rpoB mutations at R529 suggested a selective advantage conferred by this particular mutation. This particular mutation may result in similar effects to mutations known to affect the interaction between ppGpp, the alarmone signal of the stringent response, and RpoB. These mutations are known to limit RNA polymerization (26), resulting in fewer ribosomes to synthesize proteins (27). The mutants, therefore, suffer less from amino acid starvation, the cause of the stringent response (27). Following this line of reasoning, all combinations of relA/relA+ and ileSAla/ileS+ were constructed in E. coli and tested for the MAC phenotype. However, no differences were observed for relA/relA+ strains (data not shown).
We also considered the possibility that the enhanced mutation rate coming from mistranslation in aging bacteria may be caused, in particular, by the variant protein structures resulting from misincorporation of amino acid metabolites in the cell that are not among the canonical 20 amino acids. Examples include the naturally occurring metabolites 
-aminobutyrate, homocysteine, and norvaline (28). [Owing to its large hydrophobic binding pocket that must accommodate the isopropyl side chain of isoleucine, isoleucyl-tRNA synthetase weakly activates a broad spectrum of amino acids (28).] Thus, the inclusion in the media of an amino acid analog such as norvaline might increase the misfolded proteins within the cells, thereby increasing the SOS or stress response. Pursuing this possibility, sublethal levels of norvaline were tested for their effect on the SOS response in the editing-defective versus the wild-type strain. Each strain accumulated CiR mutants in the presence of norvaline at a comparable rate to when the analogue was absent (data not shown). Thus, the enhanced mutation rate caused by mistranslation in aging bacteria most likely occurs through mistranslation that inserts one or more of the standard, canonical amino acids and thereby sets up the system for induction of error-prone DNA pols.
Translational errors have been shown to result in an increased mutation frequency. In particular, mutA and mutC are alleles of glyV and glyW, respectively, which encode tRNAs specific for glycine (29). These mutants replace the tRNA anticodon for glycine with that for asparagine, resulting in statistical proteins that incur Asp Gly substitutions. The source of the increased mutation rate is DnaQ, which provides an error-correcting exonuclease function to the replicative DNA pol III (30). This cause of increased mutation rate is known as translational stress-induced mutagenesis (TSM). The mechanism of TSM is not completely understood, however, as the phenotype appears to require several recombination-associated enzymes (31). By comparison, we have shown here the unambiguous involvement of a known pathway for mutagenesis, namely the SOS response. Other studies have shown that mutations that result from the SOS response are specifically caused by the activity of the error-prone DNA pols (21).
A number of human homologues of error-prone DNA pols have been identified (8). Several of these have been associated with various types of cancer. For example, DNA pol 
, an error-prone DNA pol, is expressed at elevated levels in breast cancer cells (32). Pol causes an increased level of mutagenesis, and immunodepleted nuclear extracts exhibited a diminished mutation frequency. Similarly, DNA pol 
, the human homolog of the E. coli DNA pol IV (dinB), is capable of translesion synthesis that can result in error-free, error-prone and –1 frameshifting, depending on the lesion (33–35). Pol can be up-regulated in cultured lung cancer cells and leads to loss of heterozygosity (LOH), the most common alteration found in tumors (36). LOH is likely induced by replication slippage, which is especially likely to be caused by the frameshifting effect of pol slippage, indicating that up-regulation of pol predisposes a cell to cancer. These specific cases are further supported by a more general study in which the expression of specialized DNA pols were examined in a number of tumor samples (37). In >45% of the samples, at least one specialized DNA pol was expressed at least 2-fold higher relative to the replicative DNA pols.
These examples serve to illustrate that the acquisition of a mutator phenotype may play a crucial role in susceptibility to cancer (38–40). Our bacterial system has shown that mutations in the editing site of a tRNA synthetase can stimulate induction of error-prone DNA pols. Thus, while recent previous work demonstrated that a heritable mutation in the editing domain of a tRNA synthetase leads to cellular degeneration of Purkinje cells because of misfolded proteins (1), this work showed that such mutations also cause heritable genetic degeneration from the induction of error-prone DNA pols.
|
Methods |
|---|
|
TopAbstractResultsDiscussionMethodsAcknowledgementsReferences |
|---|
MAC. An increased propensity for mutation caused by aging was assayed essentially as described (7). In brief, 100–1,000 colonies were allowed to grow on solid LB agar for either 1 or 7 days. After this period, bacteria were scraped off of the plate in 2 ml of LB, and the titer of viable colonies was determined by serial dilution and spotting a 5-µl aliquot on LB plates. In addition, the titer of spontaneous RifR mutants was determined. Growth of RifR mutants was rapid, and colonies were counted after overnight growth to ensure that Rif (which is moderately unstable) had not degraded. In each of six separate experiments, either three or four separate plates were assayed at each time point for each strain, PS8078, PS8079, PS8080, PS8229, PS8231, and PS8233, thereby resulting in >60 separate data points (RifR frequency determinations) for each genotype after each period. The Mann–Whitney U test was used to determine the statistical significance of differences in distributions (9). One additional experiment testing the acquisition of fosfomycin resistance was conducted.
Sequencing of rpoB Mutations. Two clones from each plate of two independent MAC experiments were isolated and grown in microplates such that a total of 48 clones of each genotype were isolated after 1 and 7 days. Clones were stored in LB plus 15% glycerol at –80°C. Clones were revived in 250 µl of LB and grown overnight at 37°C. A 2-µl aliquot of these overnight cultures was diluted to 10 µl with water, lysed by boiling, and used as a template for PCR of the region of the rpoB gene that is known to harbor the locus for the majority of RifR-conferring mutations. As described, primers 1525-up (5'-ggcgatctggataccctgatgc) and 2198-down (5'-cggagtcaacggcaacagcac) were used to PCR-amplify this region (5) by using Platinum Taq HiFi (Invitrogen, Carlsbad, CA) and the thermocycling program, 35x (94°C, 30 s/56°C, 30 s/68°C, 2 min, 30 s), 68°C 10 min, 4°C. PCR products were purified (Qiagen, Valencia, CA) and sequenced with the primer 1525-up.
Mutation Rate Determination. Overnight MSglc cultures of strains were diluted 1:10 in PBS, and a 50-µl sample was plated on MSglc plus Ci or MSglc plus Ci plus norvaline for lexA+ and lexA(wt)-KanR strains, whereas a 50-µl sample of undiluted cultures was plated for lexA(S119)-KanR strains. The starting titer of each culture was determined by serial dilution and subsequent spotting of 5 µl on LB. The number of new colonies on minimal media plates was counted each day after the first, and the positions of colonies were marked (21). Six replicate plates were counted for each strain. However, to determine the mutation rate, the number of viable cells that had not formed colonies needed to be determined (21). Therefore, on days 2 and 5, all of the visible colonies were excised from one plate per strain (equating to three plates per genotype). The remaining agar was mixed with 10 ml of PBS to elute the surviving bacteria. The number of viable cells in solution was then determined by spotting serial dilutions (5 µl) on LB. The mass of agar used for elution of viable, CiS cells was compared with the total mass of agar (before excision of visible colonies) to calculate the total number of viable, CiS cells that had not formed colonies.
CiR clones were saved by storing overnight cultures in LB plus 15% glycerol. To determine whether there was a functional difference between preexposure and postexposure mutations, the MIC of CiR clones was determined. Clones isolated from growth on MSglc plus Ci were revived in LB from day 2 (representing preexposure CiR mutants) and days 5–7 (representing postexposure CiR mutants). LB cultures were diluted 1:100 in MSglc and again grown overnight. These cultures were diluted 1:500 in 2 ml of PBS and allowed to coat the surface of MSglc plates, and excess liquid was then removed. Plates were dried, and then a Ci Etest strip (AB Biodisk, Piscataway, NJ) was placed on the plate. Etest strips allow for the precise determination of the MIC.
|
Acknowledgements |
|---|
|
TopAbstractResultsDiscussionMethodsAcknowledgementsReferences |
|---|
|
Footnotes |
|---|
Abbreviations: MAC, mutagenesis in aging colonies; Ci, ciprofloxacin; CiR, Ci-resistant; Rif, rifampicin; RifR, Rif-resistant; pol, polymerase; MIC, minimum inhibitory concentration; Kan, kanamycin; KanR, Kan-resistant.
*To whom correspondence should be addressed. E-mail: schimmel{at}scripps.edu
Author contributions: J.M.B. designed research; J.M.B. performed research; J.M.B. and P.S. analyzed data; and J.M.B. and P.S. wrote the paper.
The authors declare no conflict of interest.
© 2007 by The National Academy of Sciences of the USA
|
References |
|---|
|
TopAbstractResultsDiscussionMethodsAcknowledgementsReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg What's this?
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
