The DdrR Coregulator of the Acinetobacter baumannii Mutagenic DNA Damage Response Potentiates UmuDAb Repression of Error-Prone Polymerases

ABSTRACT

Acinetobacter baumannii strain 17978 is an opportunistic pathogen with a unique DNA damage repair response that lacks the LexA repressor but induces ~150 genes after DNA damage. It uses the UmuD homolog UmuDAb and the small protein DdrR, unique to Acinetobacter, to repress multiple horizontally acquired umuDC error-prone polymerase genes through an unknown mechanism. We used reverse transcription-quantitative PCR and immunoblotting to elucidate UmuDAb regulatory requirements and DdrR contributions to the corepression of this specialized regulon. Mutations in the putative UmuDAb helix-turn-helix (HTH) domain could not repress the expression of the UmuDAb/DdrR regulon. A ddrR insertion mutation in these HTH mutant backgrounds produced even greater derepression of the regulon, suggesting that DdrR exerts an additional level of control over this mutagenic response. These ddrR HTH mutant A. baumannii cells overexpressed UmuDAb, cleaving it after treatment with the DNA-damaging agent mitomycin C. This showed that DdrR was not required for UmuDAb self-cleavage and that UmuDAb repression and self-cleavage actions were independent. An uncleavable umuDAb mutant with an A-to-Y change at position 83 (A83Y) could neither induce the UmuDAb/DdrR regulon nor conduct SOS mutagenesis. However, a prophage-encoded umuDrumB operon was still partially induced after DNA damage in this mutant. Surprisingly, that prophage’s putative repressor was dispensable for prophage-encoded umuDrumB induction, implying another repressor’s involvement. This study revealed that UmuDAb behaves like LexA, requiring an N-terminal HTH motif for repression and C-terminal self-cleavage for gene induction and subsequent SOS mutagenesis, and DdrR cooperates with it to exert an additional level of repressive control on this pathogen’s mutagenic response to DNA damage.

IMPORTANCE Acinetobacter baumannii is a nosocomial pathogen that acquires antibiotic resistance genes through conjugative transfer and carries out a robust mutagenic DNA damage response. After exposure to conditions typically encountered in health care settings, such as antibiotics, UV light, and desiccation, this species induces error-prone UmuD′₂C polymerases. This mutagenic capability increases A. baumannii survival and virulence and is regulated by the UmuDAb/DdrR corepressor system unique to the Acinetobacter genus. Our study has revealed that the DdrR protein provides an additional layer of control in preventing mutagenic polymerase expression by enhancing UmuDAb repression actions. Understanding these repressors could lead to new drug targets, as multidrug resistance in hospital-acquired infections has decreased treatment options, with limited new drugs being developed.

INTRODUCTION

Many multidrug-resistant Acinetobacter baumannii isolates cause nosocomial infections and deaths worldwide (1). Historically, A. baumannii strains have been considered to have low virulence (2), but they are successful pathogens, in part because they have acquired a “mobile resistome” of antibiotic resistance genes by the horizontal gene transfer of large conjugative plasmids, such as pAB3 and pAB04-1 (3, 4), and exchange DNA between strains and with other Gram-negative pathogens, such as Pseudomonas aeruginosa or Klebsiella pneumoniae (3, 5, 6). Increasing carbapenem resistance in A. baumannii, indicative of multidrug resistance, threatens our ability to effectively treat infections with the limited antimicrobials available (1, 7).

This opportunist can also upregulate its existing virulence and drug resistance pathways after exposure to DNA-damaging conditions, such as antibiotics, desiccation, or UV light (8–10). These conditions induce error-prone polymerases (EPPs), such as UmuD′₂C (polymerase V [Pol V]) (9, 11, 12), which is encoded by many Gram-positive and Gram-negative bacteria (13–16). When these polymerases, which lack 3′-to-5′ exonuclease activity (15), conduct translesion synthesis, mutations result from a process called SOS mutagenesis (17). The SOS response’s error-prone DNA repair and recombination pathways, including SOS mutagenesis, have been proposed and investigated as drug targets (18, 19).

In Escherichia coli, the SOS response involves over 1,000 genes (20), with ~30 SOS genes directly regulated by the LexA repressor (21, 22). LexA dimers repress gene expression by binding via an N-terminal winged helix-turn-helix (HTH) domain (23–26) to the SOS box, which is a DNA sequence upstream of the SOS genes (27–29). DNA damage triggers RecA to facilitate LexA self-cleavage, which reduces the affinity of LexA for SOS boxes (30–32) and results in induced SOS gene transcription, whose products carry out error-free and error-prone DNA repair (28). Although A. baumannii does not encode a lexA gene, it induces multiple genes and undergoes SOS mutagenesis after exposure to UV light, clinically used antibiotics, or mitomycin C (MMC) (8–12). In Acinetobacter species, a transcriptional repressor similar in size to LexA, called UmuDAb (33, 34), is one regulator of this response (11, 12, 34). Unlike LexA, UmuDAb represses only a subset of DNA damage-induced genes: umuDAb, several alleles of the error-prone polymerase V genes umuD and umuC, and ddrR (11, 12). In the absence of DNA damage, the UmuDAb protein of Acinetobacter baylyi strain ADP1 requires a predicted N-terminal HTH structure to repress this regulon (35). The UmuDAb binding site differs from the Escherichia coli SOS box in its specific sequence and has a longer central spacer region (11, 35). After DNA damage, induction of gene expression requires UmuDAb self-cleavage at an AG site (or CG site in UmuD) (33, 35) that is conserved in other S24 protease family members, LexA and UmuD (11, 34).

Another unusual feature of the Acinetobacter DNA damage response is that DdrR coregulates UmuDAb-repressed genes (36). Unique to Acinetobacter spp., DdrR is a small protein of ~80 amino acids that I-TASSER analysis predicted to have a disordered N terminus and two alpha-helices at its C terminus (37). In ADP1 and A. baumannii strains, ddrR and umuDAb are divergently transcribed, with their adjacent −35 promoter consensus elements controlled via a single overlapping UmuDAb binding site (36). The mechanistic role of DdrR in coregulating the UmuDAb/DdrR regulon is not yet known, as its requirement for coregulatory behavior in the absence of DNA damage is compatible with it either assisting UmuDAb in repression or preventing UmuDAb self-cleavage under these conditions (36). Our previous two-hybrid analyses (37) demonstrated UmuDAb dimerization but did not indicate direct physical interaction between DdrR and UmuDAb, possibly due to interference of the hybrid protein domains. However, a recent study using surface plasmon resonance spectrometry and electrophoretic mobility shift assays showed that DdrR interacts with UmuDAb and an A. baumannii putative prophage repressor (A1S_1144) (38). Although LexA usually does not use a cofactor to regulate gene expression, in Gram-positive Bacillus thuringiensis a small protein (gp7) encoded in the bacteriophage GIL01 linear genome directly binds to LexA to enhance its affinity for SOS boxes and hinder LexA self-cleavage (39). This may occur because unlike most phages, GIL01 does not encode a stress-sensitive repressor but instead uses gp7 to recruit the B. thuringiensis LexA to repress its lytic genes and maintain lysogeny (40). We hypothesize that DdrR may behave like gp7 in binding to the UmuDAb repressor to modify either its binding or self-cleavage activities.

Our goal in this study was to determine whether DdrR actions assist UmuDAb binding and gene repression or affect UmuDAb cleavage and gene induction. Here, we show that ddrR mutations further derepressed the UmuDAb/DdrR regulon in UmuDAb HTH mutant strains. DdrR thus provided an extra level of repression to the UmuDAb residues that we identified as critical for its action as a repressor. DdrR did not affect UmuDAb cleavage ability, which was required to induce the multiple horizontally acquired umuDC genes in this species and thus its Pol V-mediated SOS mutagenesis (10, 12). These unique repressors make this regulon a potential target for preventing mutagenic acquisition of drug resistance.

RESULTS

UmuDAb N-terminal domain helices are required for repression.

We hypothesized that UmuDAb DNA binding and repression require the function of its N-terminal alpha-helices, which have been predicted to form an HTH domain (37), leading us to substitute lysine (41) or helix-disrupting proline residues (42) into either helix 1 (an E-to-K change at position 24 [E24K]) or helix 2 (K40P N41P), respectively, with site-directed mutagenesis. These mutations (which we refer to as HTH1 and HTH2) were present in strains MF1 and MF2 as unmarked allelic replacements or linked to a Kan^R cassette in strains JH1724 and JH1740. We measured these strains’ expression levels of the UmuDAb/DdrR regulon (target genes) in untreated and DNA damage (MMC) treatment conditions. This regulon includes both the regulators A1S_1388 ddrR and A1S_1389 umuDAb as well as the multiple umuD and umuC alleles that encode the components of the error-prone polymerase UmuD′₂C pAB3-located A1S_0636-0637 umuDC, prophage-located A1S_1174-1173 umuDrumB and A1S_2015 umuC, and A1S_2008 umuC (12, 36). We previously reported the expression data of the UmuDAb regulon in a null umuDAb mutant of A. baumannii ATCC 17978 (12).

Reverse transcription-quantitative PCR (RT-qPCR) experiments indicated that HTH1 and HTH2 mutations increased all of these genes’ expression in the absence of DNA damage (untreated), compared to strain 17978 wild-type (WT) expression on the same RT-qPCR plate (Fig. 1A; see also Fig. S1A in the supplemental material). In HTH1 strain MF1, in which expression increased an average of 32-fold (9.5-fold in A1S_2008 to 56-fold for A1S_2015), this increase showed a nonsignificant trend in a one-way analysis of variance (ANOVA). However, based on a t test performed on each RT-qPCR plate, all genes showed higher untreated expression in this HTH1 strain (P < 0.05). The increased expression was significant for the HTH2 strain MF2 (P < 0.05 in a one-way ANOVA with Dunnett’s multiple-comparison testing, untreated WT expression versus HTH1 and HTH2 strain expression). Furthermore, the loss of repression was so great that for every target gene except ddrR (see Fig. 1A), this untreated expression was higher than WT expression under MMC-treated conditions.

FIG 1.

RT-qPCR results for UmuDAb-regulon genes in umuDAb HTH mutants and umuDAb HTH - ddrR double mutants. RT-qPCR experiments measured gene expression (2^−ΔCq) in wild-type (WT) and umuDAb mutant strains of A. baumannii ATCC 17978 in the absence or presence of DNA damage (0 or 2 μg/mL MMC, respectively). The expression of two representative genes (umuDAb A1S_1389 and umuD A1S_0636) in the UmuDAb regulon are shown (see Fig. S1 in the supplemental material for the other genes evaluated). (A) Mutation of either helix 1 or 2 of the putative HTH domain in UmuDAb resulted in higher expression relative to that in WT cells in the absence of DNA damage. HTH1, strain MF1; HTH2, strain MF2. *, P < 0.05 in a one-way ANOVA test with Dunnett’s multiple-comparison posttesting. (B) Gene expression in strains mutated in ddrR and the UmuDAb putative HTH domain (HTH1 ddrR, strain JH1701; HTH2 ddrR, strain JH1702). This ddrR mutation resulted in higher expression than in the single HTH mutant strains in the absence of DNA damage (one-way ANOVA test with Šidák’s multiple-comparison posttesting) (data not shown), and as in panel A, hgher than expression in WT cells in the absence of DNA damage (P < 0.05, one-way ANOVA test with Dunnett’s multiple-comparison posttesting). The SD of the means from technical triplicates of biological triplicates are shown. ****, P < 0.0001; ***, P < 0.001; **, P < 0.01; *, P < 0.05; ns, P ≥ 0.05.

The qualitative effects of the HTH1 and HTH2 mutations were not identical either. HTH1 mutants showed slightly higher gene expression after DNA damage for the regulatory genes umuDAb and ddrR, but all genes were slightly repressed after DNA damage in the HTH2 mutant strain (P > 0.05, one-way ANOVA) (Fig. 1A). Additionally, for all genes tested (except A1S_2008), untreated expression was significantly lower in the HTH1 mutant than in the HTH2 mutant (P < 0.05, one-way ANOVA with Šidák’s multiple-comparison testing). These results suggested that the helix 1 E24K mutation is not as severe in disrupting UmuDAb repression as the helix 2 K40P N41P mutation. We observed the same trends in strains carrying unmarked (MF1 and MF2 [Fig. 1A and Fig. S1A]) and kanamycin resistance-marked (JH1724 and JH1740 [data not shown]) umuDAb mutations.

ddrR mutation in UmuDAb HTH mutants further derepresses the UmuDAb-DdrR regulon.

We did not know whether DdrR enhanced the repression actions (DNA binding ability) of UmuDAb or its self-cleavage ability, as either was consistent with the higher expression seen in the ddrR mutant strain (36). To test how DdrR actions affected the repression of regulated genes in the absence of DNA damage, we constructed an insertional mutation in ddrR in the umuDAb mutant strains MF1 and MF2. We then compared gene expression in these ddrR/umuDAb double mutants (JH1701 and JH1702, respectively) to the single umuDAb mutants by using RT-qPCR. For nearly every gene in the regulon, the expression level in untreated cells was significantly higher in both of the HTH ddrR double mutant strains JH1701 and JH1702 than in the HTH mutant cells alone (P < 0.05, one-way ANOVA with Šidák’s multiple-comparison testing) (Fig. 1B and Fig. S1). The two exceptions included nonsignificant trends in A1S_2015 expression in JH1701 versus MF1 and A1S_1389 expression in JH1702 versus MF2.

Furthermore, this mutation of ddrR increased the trend of MMC induction in HTH1 mutants and repression in HTH2 mutants. In JH1701 (ddrR HTH1) cells, all UmuDAb/DdrR regulon genes were significantly induced, unlike the nonsignificant trends seen in the HTH1 single-mutant strain MF1. Likewise, in JH1702 (ddrR HTH2) cells, all genes were repressed, unlike the nonsignificant trends seen in the HTH2 single-mutant strain MF2 (P < 0.05, one-way ANOVA with Šidák’s multiple-comparison testing). Finally, consistent with the HTH2 mutant resulting in more derepression than the HTH1 mutant, expression of all genes was higher in the JH1702 strain than in JH1701 (Fig. 1B).

The conserved A83-G84 site is required for UmuDAb self-cleavage, gene induction, and SOS mutagenesis after DNA damage.

We hypothesized that UmuDAb self-cleavage was required for gene induction after DNA damage, as reported in A. baylyi ADP1 (35) and for LexA (31). To identify the self-cleavage site, we constructed umuDAb mutant strains with an A83Y mutation at the predicted self-cleavage site that it shares with LexA and UmuD proteins of the S24 protease family (43, 44). This yielded strains BC1 (unmarked) and JHAYK (umuDAb A83Y linked to a Kan^R gene). We then used immunoblotting to visualize UmuDAb cleavage in strains expressing UmuDAb or UmuDAb A83Y. Because UmuDAb was not visible in whole-cell lysates from WT or umuDAb A83Y strains, we expressed the UmuDAb from the pIX3 vector that we previously used to observe UmuDAb expression and cleavage in A. baylyi (33). After MMC treatment of plasmid-carrying 17978 cells, we observed the UmuDAb′ cleavage product (predicted size, 13.6 kDa) in WT cells expressing UmuDAb (Fig. 2, lanes 2 to 5) but not UmuDAb A83Y (Fig. 2, lanes 7 to 10). To confirm that the plasmid-based and not the chromosomal umuDAb allele was the source of the cleavage product, we also expressed UmuDAb from pIX3 in the umuDAb A83Y strain BC1 and observed the same-sized UmuDAb′ cleavage product in that strain also (see Fig. S2).

FIG 2.

UmuDAb A83Y does not cleave after DNA damage. Western blot of A. baumannii cell lysates, probed with polyclonal antibodies against UmuDAb and showing molecular weight markers (in kilodaltons [kDa]) in the last lane. The lanes contain lysates from WT cells collected beginning at T = 0 min and at 10-min intervals after adding 2 μg/mL MMC. After MMC treatment, UmuDAb′ cleavage products were seen expressed from the WT (pIXUDGm) plasmid, but no cleavage products were detected from the plasmid encoding UmuDAb A83Y (pIXUDAYGm). The ~40-kDa (red) band in each lane is recognized by antibodies for the 36-kDa RpoA protein of E. coli and was used to normalize signal and protein amounts for accurate quantification of target proteins. The ~19-kDa cross-reactive bands could correspond to UmuD (A1S_0636 and/or A1S_1174), which are 63 to 66% identical to UmuDAb and each other, or a similar-sized nonspecific band observed previously (33). Symbols: − and +, grown in the absence or presence of MMC, respectively; UmuDAb′= UmuDAb 84 to 202. The N-terminal fragment of UmuDAb is not visible because the polyclonal antibodies used were raised against UmuDAb residues 84 to 202. The blot shown is representative of 2 replicates.

We tested whether the uncleavable umuDAb A83Y gene product present in strain BC1 could support SOS mutagenesis by comparing the frequency of its rifampin resistance after DNA damage to that of WT cells (8). BC1 cells did not demonstrate increased rifampin resistance after UV exposure, and additional mutation of ddrR in strain BC1 (yielding strain DC3) did not alter this result. In multiple experiments (3 out of 4 for strain BC1 and 2 out of 4 for strain DC3), 0 rifampin-resistant colonies grew under either DNA-damaging or control conditions, rendering an accurate calculation of the mutation frequency difficult. Rifampin resistance frequencies estimated from viable cell counts ranged from 10⁻¹⁰ to 10⁻⁹ for these strains. In comparison, WT 17978 cells had an ~950-fold induced rifampin resistance frequency of 1 × 10⁻⁶ after UV exposure.

Using RT-qPCR experiments, UmuDAb-repressed target gene expression was not induced after DNA damage in the BC1 strain possessing an uncleavable UmuDAb (Fig. 3A and Fig. S3; the one exception is shown in Fig. 3C). In the absence of DNA damage, these genes' expression levels were not different than in WT cells (P < 0.05, one-way ANOVA with Dunnett’s multiple-comparison testing). These results suggested that the inability to induce gene expression in BC1 was due to its inability to cleave UmuDAb.

FIG 3.

RT-qPCR results for UmuDAb-regulon genes in umuDAb cleavage and umuDAb cleavage/ddrR double mutants. RT-qPCR experiments measured gene expression (2^−ΔCq) in WT and umuDAb mutant strains of A. baumannii ATCC 17978 in the absence or presence of DNA damage (0 or 2 μg/mL MMC, respectively). Two representative genes in the UmuDAb regulon (umuDAb A1S_1389 and umuD A1S_0636) are shown; see Fig. S3 in the supplemental material for the other genes. (A) Mutation of A83 in the A83-G84 UmuDAb self-cleavage site prevented target gene induction in the presence of DNA damage. A83Y, strain BC1. Asterisks denote statistical significance of (see the Fig. 1 legend for exact definitions) in a one-way ANOVA test with Dunnett’s multiple-comparison posttesting against the A83Y ddrR strain DC3. (B) Gene expression in the DC3 strain mutated in ddrR and the UmuDAb self-cleavage site. This ddrR mutation resulted in higher expression than the WT and BC1 mutant strains (one-way ANOVA test with Šidák’s multiple-comparison posttesting, as depicted by asterisks). (C) Expression of the umuDC operon A1S_1174-1173 was inducible after DNA damage, even when the UmuDAb self-cleavage site was mutated. P values are denoted by asterisks (see Fig. 1 legend) with the # symbol(s) similarly indicating induction of an individual gene after MMC treatment in a one-way ANOVA test with Šidák’s multiple-comparison posttesting. The SD of the means from technical triplicates of biological triplicates are shown.

The sole exception to this loss of induction was seen in the A1S_1174-1173 (umuDrumB) operon, which was still induced after DNA damage in the UmuDAb A83Y mutant strain BC1 (Fig. 3C). These genes are encoded by a prophage (12), prompting us to hypothesize that mutating that prophage’s putative cI repressor (A1S_1144) would cause derepression of umuDrumB. However, the A1S_1144:kanR strain AJ1144 still repressed A1S_1174-1173 in the absence of MMC treatment and induced the genes after DNA damage, as in WT cells (Fig. 4). The other UmuDAb-repressed target genes were also still repressed without MMC treatment in AJ1144 (one-way ANOVA after Šidák’s multiple-comparison test) (data not shown).

FIG 4.

Expression of A1S_1174-1173 was unaffected in the putative prophage repressor (A1S_1144) mutant strain. RT-qPCR experiments measured the expression of A1S_1174-1173 in WT and A1S_1144 mutant strains of A. baumannii ATCC 17978 in the absence (0 MMC) or presence of DNA damage (2 μg/mL MMC). Each gene was tested in one PCR plate and analyzed using one-way ANOVA with Šidák’s multiple-comparison tests to compare the mutant strain to wild type in untreated conditions and the induction of each strain after treatment. Significance is denoted (see the legend for Fig. 1 for definitions; # symbols indicate significant induction of an individual gene after MMC treatment); ns, P ≥ 0.05.

DdrR repression actions are independent of UmuDAb cleavage ability.

LexA and prophage repressors bind to DNA to inhibit target gene transcription. In the presence of DNA damage, repressor self-cleavage and dissociation from promoters allows transcriptional induction. We hypothesized that if DdrR acted to assist UmuDAb binding DNA but not self-cleavage inhibition, a ddrR mutant would still derepress gene expression in an uncleavable UmuDAb background. To test this, we compared target gene expression in the ddrR umuDAb A83Y mutant strain DC3 to that in the umuDAb A83Y strain BC1. All target genes were derepressed in DC3 cells grown in the absence of DNA damage, although A1S_1174-1173 and A1S_2015 were also induced after DNA damage (Fig. 3 and Fig. S3).

UmuDAb self-cleavage does not require DdrR or DNA binding ability.

These gene expression experiments indicated that target gene induction after DNA damage required the UmuDAb self-cleavage site A83 (Fig. 3) but did not require DdrR (Fig. 1B). To directly test for the possible DdrR impact on UmuDAb cleavage, we examined UmuDAb self-cleavage ability in WT and ddrR mutant strains with immunoblotting experiments. We observed that the ddrR mutant strain JH1700 produced UmuDAb′ cleavage products after MMC treatment (Fig. 5 and Fig. S4), suggesting that the absence of DdrR did not prevent UmuDAb self-cleavage.

FIG 5.

A ddrR-deficient strain cleaves UmuDAb similarly to WT cells. A Western blot shows A. baumannii ddrR cell lysates, probed with polyclonal antibodies against UmuDAb. JH1700 (pIXUDGm) cells were grown for 4 h then treated with 2 μg/mL MMC. Each lane contains lysate from cells collected beginning at T = 0 min and at 10-min intervals after adding MMC. The ~40-kDa (red) band in each lane is recognized by antibodies against the 36-kDa RpoA protein of E. coli. Symbols: − and +, grown in the absence or presence of MMC, respectively; UmuDAb′= UmuDAb positions 84 to 202. The blot shown is representative of 4 replicates.

We also evaluated mutant UmuDAb protein production to determine whether the increased expression of umuDAb in the HTH mutants and ddrR/HTH double mutants translated to increased UmuDAb levels, or whether the increased gene expression was instead due to lower production of mutant UmuDAb proteins available to carry out repression. Western blotting of lysates from WT cells did not show UmuDAb protein in either the presence or absence of DNA damage (Fig. 6, lanes 1 and 2). However, lysates from HTH1, HTH2, or ddrR/HTH2 mutant strains all showed increased UmuDAb production (Fig. 6), similar to the increased expression seen in RT-qPCR experiments (Fig. 1). Strain MF2 expressed more full-length UmuDAb before MMC treatment than MF1 (~2.2-fold more in two experiments, standard deviation [SD], 0.2) (Fig. 1A and Fig. S5). Similarly, the HTH2/ddrR mutant strain JH1702 expressed 2.7-fold more UmuDAb than did MF2 (SD, 1.4 in two experiments). Furthermore, all of the HTH mutants’ cell lysates contained UmuDAb′ cleavage products after MMC treatment, suggesting that UmuDAb self-cleavage does not require the ability to repress gene expression.

FIG 6.

UmuDAb helix mutants are overexpressed in the absence of DNA damage and can cleave after MMC treatment. A Western blot of A. baumannii cell lysates, probed with polyclonal antibodies against UmuDAb, shows molecular weight markers (in kilodaltons) in lane 5. WT, 17978 wild type; HTH2, umuDAb K40P N41P strain MF2; HTH1, umuDAb E24K strain MF1; ddrR HTH2, ddrR umuDAb K40P N41P strain JH1702. Cell cultures were grown in the presence (+) or absence (−) of 2 μg/mL MMC. The ~40-kDa (red) band in each lane is recognized by antibodies for the 36-kDa RpoA protein of E. coli and was used to normalize signal and protein amounts for accurate quantification of target proteins. The blot shown is representative of 2 replicates.

DISCUSSION

This work shows that the small protein DdrR, unique to the genus Acinetobacter, enhances UmuDAb repression of a specific subset of DNA damage-induced genes in the opportunistic pathogen A. baumannii. This represents a novel control mechanism for Pol V and its mutagenic potential, as Acinetobacter also does not encode the SOS response regulator LexA (8, 45). Instead, it has UmuDAb, an unusual SOS repressor that seems to have evolved from an ancestral umuDC locus in a recombination event (8), which is plausible due to the natural plasticity of the Acinetobacter genome (2, 6). Unlike LexA, UmuDAb also functions with a coregulator, the small protein DdrR, which is unique to the genus Acinetobacter (36). Together, they repress the DNA damage-induced Pol V genes and themselves, but by an unknown mechanism.

The known regulatory actions of DdrR could affect either UmuDAb-mediated gene repression or UmuDAb self-cleavage that relieves repression, necessitating our investigation of both of these possibilities in this study. One possibility was that DdrR might prevent UmuDAb self-cleavage, as no one has observed in Acinetobacter cells UmuDAb protein self-cleavage or investigated its requirements. Furthermore, the action of the small gp7 protein to induce a conformation of LexA that inhibits self-cleavage (39) provided a precedent for this possibility. Alternately, DdrR might contribute to DNA binding of the unconventional UmuDAb repressor. DdrR was recently observed to bind to the C terminus (amino acids 84 to 202) of UmuDAb based on surface plasmon resonance spectroscopy (38). The small gp7 protein of bacteriophage GIL01 also binds to a repressor, LexA, predominantly in its C terminus, enhancing LexA’s DNA binding activity (46). DdrR interactions with UmuDAb may also contribute to enhanced UmuDAb interactions with its operators in vivo. LexA mutations that contribute to a LexA(Def) phenotype have been all located in the N-terminal domain before the first helix, in all three N-terminal helices, or in the wing after the α3-helix (47). The novelty of the UmuDAb/DdrR regulon and its coregulators is a compelling reason to investigate how these regulators' actions interact.

This work established cleavage and repression requirements of UmuDAb in A. baumannii ATCC 17978. In this strain, site-directed mutation of the putative UmuDAb cleavage site eliminated gene induction after MMC treatment and mutagenesis after UV treatment, as shown in ADP1 (35). Similarly, site-directed mutation of UmuDAb HTH1 (equivalent to the LexA α2-helix) and HTH2 (equivalent to the LexA α3-helix) increased expression of the UmuDAb/DdrR regulon in the absence of DNA damage. This indicated that these predicted helices constitute a DNA binding domain required for transcriptional repression. These effects appeared to be due to the impaired UmuDAb structure resulting from the HTH1 and HTH2 mutations, as the mutant proteins were not only produced in cell lysates but overproduced relative to WT levels (Fig. 6). However, each helix’s contribution to UmuDAb repression was not equal, as HTH2 mutants had higher uninduced expression than HTH1 mutants (Fig. 1). This suggested that the contribution of HTH1 to repression is less than that of HTH2. Some HTH1 mutant UmuDAb molecules appeared to still bind to target promoters or bind with lowered affinity, but they were released after DNA damage, as indicated by the modest induction after DNA damage (Fig. 1A and Fig. S1A). In E. coli, the LexA α3-helix contains the recognition sequence that binds to the four nucleotides of the SOS box in the major groove (25, 48). Our results suggest that, like LexA, the UmuDAb HTH2 helix (amino acids 36 to 46 [35]) might bind to the major groove of the UmuDAb recognition sequence.

Additional support for the proposed lesser contribution of HTH1 to repression comes from observing that target gene expression in the HTH1 but not the HTH2 mutant strain was less than that of the ΔumuDAb mutant (12) (comparing each mutant to WT in the untreated condition in the same RT-qPCR plate). Possibly, the E24K mutation only slightly alters the conformation of the N terminus to hinder HTH2 placement in the major groove, as reported for LexA, for which even residues changed in the loops between helices can alter its binding efficiency (47).

We also observed increased production of UmuDAb in HTH mutants (Fig. 6), consistent with its higher expression in our RT-qPCR studies (Fig. 1 and Fig. S1), which likely reflected the inability of the mutant UmuDAb to repress its own transcription.

These results of the two helices’ contributions to repression are somewhat different than those observed for A. baylyi strain ADP1 (35), where the HTH2 mutant but not the HTH1 mutant showed significant induction of ddrR and umuDAb after MMC treatment (35). This was unexpected, given each helix's similarities (75% identity) between ADP1 and 17978 and their identical UmuDAb binding site palindromic and central core sequences (36). The same Lys (for HTH1) and Pro (for HTH2) substitutions were made, although the HTH2 mutation site was K40 N41 in 17978, versus K40 R41 in ADP1. Nevertheless, the lower identity of their N-terminal 83 amino acids (70%) compared to the entire UmuDAb protein (79%) may indicate differences responsible for the relative importance of each helix within the HTH region, which is known to be crucial for each species’ UmuDAb repression ability (Fig. 1) (35). The ADP1 regulon is much smaller than that of A. baumannii; UmuDAb only represses the genes umuDAb and ddrR, because ADP1 has no ancestral or prophage-derived umuDC operons (8, 34). However, the significance of these repression differences should be considered with caution, as the folding of LexA helices in E. coli deviates substantially from that of canonical HTH proteins, although the same α3-helix binds the major groove in both HTH protein members (47). Alternatively, even structurally conserved helices among similar proteins may have variable binding capabilities within and between helices (47).

Our finding that target genes that were merely trending inducible in the single HTH1 mutant strain MF1 were significantly induced in the ddrR HTH1 strain JH1701 demonstrated that DdrR does not prevent UmuDAb cleavage. This conclusion was consistent with immunoblotting experiments that showed UmuDAb cleavage in ddrR mutant cells after DNA damage (Fig. 5 and 6, lanes 8 and 9), contraindicating a strict requirement for DdrR in UmuDAb cleavage. In three other experiments (data not shown), the WT and ddrR mutant strains appeared to cleave UmuDAb with similar timing. Furthermore, insertional mutation of ddrR in BC1 to form the ddrR umuDAb A83Y strain DC3 did not allow cleavage or UV-induced mutagenesis in DC3.

Both the HTH1 and HTH2 mutations allowed greater gene expression in noninducing conditions. However, when we introduced a ddrR mutation into these strains, this expression increased still further, i.e., the partial relief of repression in the HTH mutants was potentiated by ddrR mutation. That the action of DdrR on repression may be at a different site than the N-terminal domain of UmuDAb is consistent with the observations by Pavlin et al., who reported that DdrR binds to the C terminus of UmuDAb (38). It is also compatible with the possibility that DdrR affects UmuDAb dimerization (37), which could, as for other repressors, be required for DNA binding. It could represent an additional level of control used by A. baumannii to modulate the expression of its mutagenic response to DNA damage, which if uncontrolled could lead to cell death.

Our past RNA sequencing data (12, 36), as well as the current study’s results, were acquired after MMC treatment. Therefore, we cannot exclude the possibility that the variety of other DNA-damaging agents that Acinetobacter encounters (e.g., UV light, fluoroquinolone antibiotics, the alkylating agent methyl methanesulfonate [MMS]) would induce different and specific damage-responsive pathways, with different participation of DdrR with other repressors than we saw in this study. Other experiments with 17978 cells support this possibility. For example, Jara et al. showed that the antibiotic tetracycline induced one umuDC operon (A1S_1174-1173) and umuDAb (A1S_1389) more than ciprofloxacin treatment (49). Similarly, others found that UV, the fluoroquinolone antibiotic ciprofloxacin, and MMS all induced the EPP genes studied here (and other specific DNA damage repair genes, such as recA and uvrA) to different extents than what we observed after MMC treatment (9, 50). Such differential induction of the expression and action of the EPPs in A. baumannii could reflect DdrR expression and/or participation with other regulators in response to specific sources of DNA damage and their needed repair.

umuDAb and ddrR are the only genes that trended toward being induced in the HTH1 mutants in strain 17978. This observation and the shared UmuDAb binding site (11) that controls their divergent transcription (36) also raised the possibility of an additional level of control on these regulators but not on the UmuD and UmuC effectors regulated by UmuDAb/DdrR.

In the UmuDAb/DdrR repressed regulon, the only genes that appear to be ancestral in the 17978 chromosome are umuDAb, ddrR, and the orphan umuC A1S_2008 (8). The GC content of the umuDAb-ddrR gene cluster, conserved in almost all Acinetobacter strains, does not indicate acquisition through horizontal gene transfer. The restricted function and number of genes controlled by UmuDAb/DdrR is unusual compared to the large SOS regulons controlled by LexA (20). This regulon is even smaller in the related A. baylyi species, which does not possess any umuDC operons (or intact prophages, such as CP9 or CP5, that might encode such operons [12]) and cannot carry out SOS mutagenesis after DNA damage (8). Their continued selection in this species is enigmatic, as they only repress their own expression (12).

Horizontal gene transfer is, however, the source of the two umuDC operons that contributed to the robust SOS mutagenesis observed in A. baumannii (8, 12). A1S_0636-0637 (umuDrumB) is located on the ~150-kb conjugative plasmid pAB3, which encodes drug resistance(s) and a type IV secretion system (3, 4). The umuDC (A1S_1174-1173) operon is encoded within the cryptic prophage CP5 (51). The prophage CP9 orphan umuC (A1S_2015)’s contribution to SOS mutagenesis is unknown. Interestingly, these two horizontally acquired operons are not regulated similarly by UmuDAb/DdrR (Fig. 3). The A1S_1174-1173 umuDC operon is repressed by UmuDAb (12) and DdrR (Fig. 3C) (36), but it is still induced by DNA damage in an uncleavable umuDAb A83Y mutant with or without DdrR. This suggested that another repressor also controls this operon, such as the A1S_1144 putative cI repressor also encoded in CP5 (51). Pavlin et al. observed that DdrR binds to A1S_1144, albeit with a slightly lower affinity than for UmuDAb, and suggested that the DdrR coregulator might work with A1S_1144 to coregulate genes in phage CP5 (38). However, we observed that A1S_1144 did not regulate A1S_1174-1173, as an A1S_1144:kan^r mutant still repressed this operon in noninducing conditions and allowed its induction after MMC treatment (Fig. 4). The same results were observed for the other umuD and umuC genes in the UmuDAb/DdrR regulon (data not shown). Furthermore, DdrR was previously shown not to repress any prophage genes except A1S_1174-1173 and A1S_2015 (36). We propose that another repressor controls these prophage umuDC genes in addition to the modest control exerted by UmuDAb/DdrR (Fig. 3C). There is a complex regulatory relationship between the UmuDAb/DdrR repressor system and the putative prophage repressors A1S_2037 (in CP9) and A1S_1144 (CP5), as both of them are uninducible in umuDAb and ddrR mutants (36).

As corepressors are not typically involved in LexA-mediated SOS regulation, the only comparison for DdrR/UmuDAb coregulation is the recently reported gp7-LexA interaction (39). The small protein gp7, encoded by bacteriophage GIL01, conscripts host LexA to bind its lysogenic gene promoters and maintain stable lysogeny. However, one key difference between the gp7-LexA and DdrR-UmuDAb coregulator pairs is that the latter are both bacterially encoded and appear to be ancestral in the 17978 chromosome (8).

DNA damage repair mechanisms in Acinetobacter are unusual and quite adaptable, given the absence of LexA from the genus (8), the presence of the DNA damage response coregulator DdrR, and the presence of multiple umuDC operons acquired through horizontal gene transfer (11, 33). The error-prone Pol V-mediated DNA repair that is part of the SOS pathway is one way in which bacteria can modify their drug resistance pathways (52, 53). Late in the SOS response, DNA repair by EPPs is prioritized over repair accuracy, and enhanced antibiotic resistance can result from such error-prone repair. While most mutations are deleterious, they can also increase the chance of surviving future challenges, creating an evolutionary advantage (54). UmuDAb and DdrR both affect the regulation of EPPs in Acinetobacter species, and this significantly increases the mutagenesis potential in strains with intact umuDC operons. Many strains now have multiple drug resistance genes on large conjugative plasmids that are freely shared between strains and even genera. This regulon is a potential target for drug discovery to combat Acinetobacter survival and mutagenesis in hospital settings.

MATERIALS AND METHODS

Media.

E. coli cells were grown in LB broth Miller (Fisher). Strains were grown on LB agar with antibiotics (kanamycin at 30 μg/mL, gentamicin at 20 μg/mL, and ampicillin at 100 μg/mL) for maintenance. Mutant A. baumannii ATCC 17978 strains were grown in defined minimal media with 10 μM sodium succinate and 0.1% Casamino Acids (CA) as a supplementary nitrogen source (or 0.2% CA for protein production from pIX3.0-based plasmids).

Mutant strain constructions.

All 17978 umuDAb and ddrR mutant strains were constructed by making site-directed mutations in pEX18GM suicide vectors containing 17978 genomic DNA; they are listed in Table 1. Electroporation and recombination of the resulting pEX17M, pEX17MK, or pEXTn14 plasmid (see below) into the 17978 chromosome was followed by selection for gentamicin and kanamycin (36). This process yielded a series of A. baumannii ATCC 17978 strains (MF1, MF2, and BC1; JH1724, JH1740, and JHAYK; or JH1701, JH1702, and DC3, respectively) carrying these chromosomal mutations (Table 1). These allelic replacement mutants were confirmed by DNA sequencing of genomic DNA PCR amplification products.

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Description or genotype	Primer(s) used in construction	Source or reference
17978	A. baumannii ATCC 17978; wild type		ATCC
JH1700	17978 ddrR::TnLK; Kmr		36
MF1	17978 umuDAb E24K	g70a_For; g70a_Rev	This study
MF2	17978 umuDAb K40P N41P	a118c_a119c_a121c_a122c_For; a118c_a119c_a121c_a122c_Rev	This study
BC1	17978 umuDAb A83Y	c26aG27Tc28a; ASc26aG27Tc28a	This study
JH1724	17978 umuDAb E24K; Kmr	g70a_For; g70a_Rev	This study
JH1740	17978 umuDAb K40P N41P; Kmr	a118c_a119c_a121c_a122c_For; a118c_a119c_a121c_a122c_Rev	This study
JHAYK	17978 umuDAb A83Y; Kmr	c26aG27Tc28a; ASc26aG27Tc28a	This study
JH1701	17978 ddrR::TnLK umuDAb E24K; Kmr	g70a_For; g70a_Rev	This study
JH1702	17978 ddrR::TnLK umuDAb K40P N41P; Kmr	a118c_a119c_a121c_a122c_For; a118c_a119c_a121c_a122c_Rev	This study
DC3	17978 ddrR::TnLK umuDAb A83Y; Kmr	c26aG27Tc28a; ASc26aG27Tc28a	This study
AJ1144	17978 A1S_1144:Kanr; Kmr		This study
pGEM-T	TA cloning vector; Ampr		Promega
pGEM-TEasy	TA cloning vector; Ampr		Promega
pCRII	Cloning vector, source of Kanr		Invitrogen
pIX3.0	Expression vector; Ampr		Qiagen
pIXGm	pIX3.0 containing Gmr of pEX18Gm		This study
pIXUDGm	pIXGm containing umuDAb	NotagNterm17978; NotagCterm17978	This study
pIXUDAYGm	pIXGm containing umuDAb A83Y	g70a_For; g70a_Rev	This study
pEX18Gm	Counterselectable suicide vector containing sacB; Gmr		55
pEXTn14	pEX18Gm suicide vector with 2.8-kb fragment of 17978 DNA with ddrR::TnLK; Gmr Kmr		36
p17M	pGEM-T with 4,022-bp 17978 genomic DNA; Ampr	MoreOutFor; MoreOutRev	This study
p17MK	p17M with Kanr of pCRII; Kmr	3662revXho; ddrRrevXho KmupXho; KmdwXho	This study
pEX17M	pEX18Gm with 4,092 bp SphI-SacI fragment of p17M; Gmr		This study
pEX17MK	pEX17M with Kanr gene; Gmr Kmr		This study
pEXTn14EK	pEXTn14 umuDAb E24K; Gmr	g70a_For; g70a_Rev	This study
pEXTn14KP	pEXTn14 umuDAb K40P N41P; Gmr	a118c_a119c_a121c_a122c_For; a118c_a119c_a121c_a122c_Rev	This study
pEXTn14AY	pEXTn14 umuDAb A83Y; Gmr	c26aG27Tc28a; ASc26aG27Tc28a	This study
pEX17M24	pEX17M umuDAb E24K; Gmr	g70a_For; g70a_Rev	This study
pEX17M40	pEX17M umuDAb K40P N41P; Gmr	a118c_a119c_a121c_a122c_For; a118c_a119c_a121c_a122c_Rev	This study
pEX17M83	pEX17M umuDAb A83Y; Gmr	c26aG27Tc28a; ASc26aG27Tc28a	This study
pEX17MK24	pEX17MK umuDAb E24K; Gmr	g70a_For; g70a_Rev	This study
pEX17MK40	pEX17MK umuDAb K40P N41P; Gmr	a118c_a119c_a121c_a122c_For; a118c_a119c_a121c_a122c_Rev	This study
pEX17MK83	pEX17MK umuDAb A83Y; Gmr	c26aG27Tc28a; ASc26aG27Tc28a	This study
pGEMTE1144	pGEM-TEasy with SphI/NdeI 17978 DNA containing A1S_1144; Ampr	A1S3606Forward; A1S1144REV	This study
pGEMTE1144K	pGEMTE1144 with pCRII KanR gene inserted into A1S_1144 NcoI site	Kmup; Kmdw	This study

Unmarked, site-directed mutations in umuDAb helix 1 (HTH1; E24K), helix 2 (HTH2; K40P N41P), or the UmuDAb self-cleavage site A83-G84 were constructed with the Agilent QuikChange II XL site-directed mutagenesis kit. Mutations were made in pEX17M, using mutagenic primers (Table 2). pEX17M contains a 4,022-bp genomic fragment from A. baumannii ATCC 17978 with A1S_3662, ddrR (A1S_1388), umuDAb (A1S_1389), and A1S_1390 cloned into SphI-SacI sites of pEX18Gm.

TABLE 2.

Primers used in this study

Primer	Purpose	Sequence (reference)
g70a_For	umuDAb E24K (HTH1) mutation	CGGCCGTAAAGCACAGTTCAATAAACCCACCAAAG
g70a_Rev	umuDAb E24K (HTH1) mutation	CTTTGGTGGGTTTATTGAACTGTGCTTTACGGCCG
a118c_a119c_a121c_a122c_For	umuDAb K40P N41P (HTH2) mutation	GACATTATTCAGTAACCAAGGTGGGATAAAGTTGACTTGAGATTCAGGAACACG
a118c_a119c_a121c_a122c_Rev	umuDAb K40P N41P (HTH2) mutation	CGTGTTCCTGAATCTCAAGTCAACTTTATCCCACCTTGGTTACTGAATAATGTC
c26aG27Tc28a	umuDAb A83Y mutation	CGCTGGTGATGGGAAACCATAAGCAACACGTTC
ASc26aG27Tc28a	umuDAb A83Y mutation	GCGACAGAACGTGTTGCTTATGGTTTCCCATCA
NotagNterm17978	umuDAb amplification for cloning into pIX3.0	AGAAGGAGATAAACAATGCCAAAGAAGAAAGAATTCGAG
NotagCterm17978	umuDAb amplification for cloning into pIX3.0	CTTGGTTAGTTAGTTATTATCTCATTCGTTTGAGGTTATAG
MoreOutFor	17978 genomic region with umuDAb and ddrR	CGTTGAGTGTGTATGAATAG
MoreOutRev	17978 genomic region with umuDAb and ddrR	CTTCTCAACCGCCTGTTTATC
Kmup	Amplify pCRII KanR cassette	CCGGAATTGCCAGCTGGG (56)
Kmdw	Amplify pCRII KanR cassette	TTCAGAAGAACTCGTCAAG (56)
KmupXho	Amplify pCRII KanR cassette on XhoI ends	AATAAGACTCGAGCCGGAATTGCCAGCTGGG
KmdwXho	Amplify pCRII KanR cassette on XhoI ends	AATAAGACTCGAGTCAGAAGAACTCGTCAAGAAG
3662revXho	XhoI restriction site construction in p17M	AATAAGACTCGAGGTTTAAAATAATAAAAATAAAGGAG
ddrRrevXho	XhoI restriction site construction in p17M	AATAAGACTCGAGTTATGAGTGGGTAAGGG
A1S3606Forward	17978 genomic region with A1S_1144	GCTCTCCTTTTAGTGATGTAAC
A1S1144REV	17978 genomic region with A1S_1144	AACTAAGCCTGATGGACCAGAT

Similar strains bearing the same E24K, K40P N41P, and A83Y mutations in umuDAb but linked to a kanamycin resistance gene (Kan^R) were constructed by making mutations in pEX17MK. pEX17MK is pEX17M bearing a Kan^R gene between the end of ddrR and the adjacent, convergently encoded A1S_3662 (between bp 1,631,751 and 1,631,752 of CP00521.1).

Strains with mutations in both ddrR and the umuDAb putative DNA-binding HTH1 or HTH2 sites (called JH1701 and JH1702, respectively) or the A83Y self-cleavage site (DC3) were constructed after mutating the plasmid pEXTn14 with mutagenic primers (Table 2).

An allelic replacement mutant of A1S_1144 was constructed in two steps. First, a PCR product containing A1S_1144 and surrounding genes was amplified from 17978 genomic DNA (Table 1) and cloned into pGEM-TEasy on SphI/NdeI ends to yield pGEMTE1144. Then, a Kan^R gene cassette amplified from pCRII was blunted and cloned into the A1S_1144 NcoI site (in pGEMTE1144) to form pGEMTE1144K. Electroporation of pGEMTE1144K into 17978 cells, followed by selection for kanamycin resistance, yielded the A1S_1144 mutant strain AJ1144.

RT-qPCR.

RT-qPCR analyses were performed as previously described (12). Biological triplicates of A. baumannii 17978 strains (see Table 1) were grown in 3-mL overnight cultures at 37°C at 250 rpm in minimal medium supplemented with 10 mM succinate and 0.1% CA. Overnight cultures were diluted 1:25 into 5 mL fresh medium and grown under the same conditions for 2 h before splitting. One culture was treated with 2 μg/mL MMC, to be consistent with a previous study’s methods and results (12), and both were incubated for 3 h to induce gene expression, as in previous studies (33). Total RNA was processed with the Epicentre MasterPure RNA purification kit. Contaminating DNA was removed with an Invitrogen Turbo DNA-free kit and was verified by the absence of PCR product when using primers specific for A1S_0636. Total RNA was measured with the Qubit RNA BR assay (Invitrogen), and all RNA samples had a value above 6 when measured with the Qubit RNA IQ assay (Invitrogen). Reverse transcription of 1 μg of total RNA was followed by qPCR, and RT-qPCR primers were verified for amplification efficiency as previously described (12). Data were analyzed in GraphPad Prism using one-way ANOVA with Šidák’s or Dunnett’s multiple-comparisons tests as described in the text and figure legends.

Western blot analyses.

UmuDAb proteins were produced from 17978 WT and mutant strains transformed with pIXUDGm or pIXUDAYGm. pIXUDGm is the EasyXpress vector pIX3.0 (Qiagen) bearing a gentamicin resistance gene (pIXGM), into which we cloned umuDAb. A blunted 866-bp ClaI-SspI fragment of pEX18GM containing the gentamicin resistance gene was cloned into a blunted pIX3.0 EcoRI site to construct pIXGm. The 17978 umuDAb (A1S_1389) gene amplified from genomic DNA using the primers listed in Table 2 and the EasyXpress linear template kit (Qiagen) was cloned into the BamHI and XhoI sites of pIXGM to create pIXUDGm. pIXUDAYGm was constructed by site-directed mutagenesis (with the primers in Table 2) of pIXUDGm.

In UmuDAb cleavage experiments, 50-mL overnight cultures were diluted to an optical density at 600 nm (OD₆₀₀) of 0.1, regrown until the OD₆₀₀ reached 0.7 to 0.9, and then treated with 2 μg/mL of the DNA-damaging agent MMC to be consistent with previous study results (33). We collected 1-mL samples every 10 min and stored cell pellets at −20°C until lysis with Bug Buster Master Mix containing HALT protease inhibitor. We mixed protein lysates with LI-COR 4× loading buffer with 1% beta-mercaptoethanol and loaded 20 μg of total protein (measured on a Qubit 4 fluorometer, using the Invitrogen Qubit protein assay kit) per lane in Nu-PAGE 12% bis-Tris gels (Invitrogen). Electrophoresis was performed at room temperature in prechilled 1× Nu-PAGE MES SDS running buffer (Invitrogen) at 125 V for 90 min. Proteins were transferred to Bio-Rad LF polyvinylidene difluoride (PVDF) membranes with the TransBlot Turbo transfer system. Membranes were first probed with rabbit anti-UmuDAb amino acids 84 to 202 (Genscript) (37) and mouse anti-Escherichia coli RNA polymerase α primary antibodies (BioLegend) and then with Li-Cor IRDye 680 RD anti-mouse and IRDye 800CW anti-rabbit secondary antibodies (33). Fluorescent images were acquired and normalized for statistical analysis with the Li-Cor Odyssey Fc imager and Image Studio version 5.2 software.

SOS mutagenesis.

Error-prone polymerase activity was observed by measuring rifampin resistance after UV exposure (200 J/m²), as described previously (8).