Advertisement
Research Article

Prophage Induction and Differential RecA and UmuDAb Transcriptome Regulation in the DNA Damage Responses of Acinetobacter baumannii and Acinetobacter baylyi

  • Janelle M. Hare mail,

    jm.hare@morehead-st.edu

    Affiliation: Department of Biology and Chemistry, Morehead State University, Morehead, Kentucky, United States of America

    X
  • Joshua C. Ferrell,

    Affiliation: Department of Biology and Chemistry, Morehead State University, Morehead, Kentucky, United States of America

    X
  • Travis A. Witkowski,

    Affiliation: Department of Biology and Chemistry, Morehead State University, Morehead, Kentucky, United States of America

    X
  • Alison N. Grice

    Affiliation: Department of Biology and Chemistry, Morehead State University, Morehead, Kentucky, United States of America

    X
  • Published: April 07, 2014
  • DOI: 10.1371/journal.pone.0093861

Abstract

The SOS response to DNA damage that induces up to 10% of the prokaryotic genome requires RecA action to relieve LexA transcriptional repression. In Acinetobacter species, which lack LexA, the error-prone polymerase accessory UmuDAb is instead required for ddrR induction after DNA damage, suggesting it might be a LexA analog. RNA-Seq experiments defined the DNA damage transcriptome (mitomycin C-induced) of wild type, recA and umuDAb mutant strains of both A. baylyi ADP1 and A. baumannii ATCC 17978. Of the typical SOS response genes, few were differentially regulated in these species; many were repressed or absent. A striking 38.4% of all ADP1 genes, and 11.4% of all 17978 genes, were repressed under these conditions. In A. baylyi ADP1, 66 genes (2.0% of the genome), including a CRISPR/Cas system, were DNA damage-induced, and belonged to four regulons defined by differential use of recA and umuDAb. In A. baumannii ATCC 17978, however, induction of 99% of the 152 mitomycin C-induced genes depended on recA, and only 28 of these genes required umuDAb for their induction. 90% of the induced A. baumannii genes were clustered in three prophage regions, and bacteriophage particles were observed after mitomycin C treatment. These prophages encoded esvI, esvK1, and esvK2, ethanol-stimulated virulence genes previously identified in a Caenorhabditis elegans model, as well as error-prone polymerase alleles. The induction of all 17978 error-prone polymerase alleles, whether prophage-encoded or not, was recA dependent, but only these DNA polymerase V-related genes were de-repressed in the umuDAb mutant in the absence of DNA damage. These results suggest that both species possess a robust and complex DNA damage response involving both recA-dependent and recA-independent regulons, and further demonstrates that although umuDAb has a specialized role in repressing error-prone polymerases, additional regulators likely participate in these species' transcriptional response to DNA damage.

Introduction

Cells that experience damage to their DNA have evolved mechanisms of sensing, repairing, and replicating this damaged DNA. In most bacteria, DNA damage from various sources such as UV radiation, alkylating chemicals (e.g. mitomycin C (MMC)), and antibiotics can induce up to 10% of the genome in this SOS response [1]. Induced SOS genes encode proteins that sense damage, control cell division, and repair, replicate and recombine DNA for continued cellular survival [2][4]. These processes are often carried out in an error-free manner, using conserved SOS genes such as ssb, recA, recN, ruvA, ruvB, uvrA, uvrB, and uvrD in repair and recombination processes [3] and sulA in controlling the bacterial cell cycle [5], [6]. However, DNA damage left unrepaired can also lead to the induction of SOS gene products that carry out error-prone replication of this damaged DNA. These error-prone polymerases, formed by the homodimerization of UmuC and two molecules of self-cleaving UmuD (DNA polymerase V, [7]), or DinB/DinP (DNA polymerase IV [8]) are responsible for SOS mutagenesis.

The mechanism by which these SOS genes are specifically transcribed when the cell experiences DNA damage is through relief of LexA repression [9]. This de-repression occurs after RecA binds ssDNA, an indicator of DNA damage [10], and induces LexA self-cleavage [11]. The normal state of repression in the absence of DNA damage thus prevents constitutive production of the entire SOS regulon, and SOS mutagenesis.

This general model of SOS gene induction and function, which has been developed to a significant extent in Escherichia coli [2], [4], is conserved throughout proteobacterial classes, albeit imperfectly. Gammaproteobacteria in the order Enterobacteriales often possess one LexA protein that recognizes a conserved SOS box in SOS gene promoters [2]. However, in the Pseudomonadales (containing the opportunistic, often multidrug-resistant pathogens Acinetobacter baumannii and Pseudomonas aeruginosa) and Xanthomonadales orders, more divergent responses to DNA damage exist. For example, Pseudomonas putida possesses two different LexA proteins, each controlling separate regulons [12], and Geobacter sulfurreducens also has two LexA proteins, which do not bind the recA promoter [13]. This diversity highlights the need for additional examination of not only the mechanisms of SOS gene control but also SOS gene identity in this order.

Further components of the SOS model of gene regulation are absent in Acinetobacter species of this order. None of its fellow members of the family Moraxellaceae possess umuD homologs [14], which may have implications for the ability of these organisms to undergo SOS mutagenesis after DNA damage. Additionally, no lexA homolog has been identified in this genus [14]. Nevertheless, in the non-pathogenic genetic model organism Acinetobacter baylyi ADP1 [15], previous investigations of the DNA damage response demonstrated that two genes are induced by mitomycin C and UV exposure in this strain. These two induced genes are recA (which unlike in other bacteria, does not require recA for its own induction [16], nor contains an SOS box in its promoter [16]), and ddrR, a gene of unknown function found only in the genus Acinetobacter [14]. ddrR is transcribed divergently from umuDAb [17], which is itself an unusual component of the DNA damage response of this species. UmuDAb is a UmuD homolog that is required for full induction of ddrR [17], but it is not known whether ADP1 uses it to induce other genes that are, in other bacteria, part of the SOS response. UmuDAb carries out self-cleavage in a RecA-dependent manner after cells experience diverse forms of DNA damage, and thus shares features with both the DNA polymerase V component UmuD and the LexA repressor [18]. Recent work demonstrates that in A. baumannii ATCC 17978, UmuDAb binds to and represses the promoters of umuDC homologs [19] and so might serve as a LexA analog for this genus.

Multiple umuD and umuC homologs co-exist in A. baumannii strains, and at least some of these strains (ATCC 17978, AB0057) display DNA damage-induced mutagenesis [14]. These observations suggest that these strains possess a mechanism of sensing DNA damage and inducing at least error-prone polymerase production under these conditions, and suggest that specialized UmuD function has evolved in this species. Whether this mechanism is a global response to DNA damage, induces SOS genes found in other species, or requires the action of RecA and/or other repressors is unknown.

These unusual features in the DNA damage responses of Acinetobacter species prompted us to use RNAseq experiments to define the transcriptome of A. baylyi ADP1 after DNA damage, and compare its response to that of the opportunistic pathogen, A. baumannii ATCC 17978. Our aims in these experiments were to determine both the existence and identity of any global DNA damage-induced transcriptome in these species, and the possible requirements for RecA and UmuDAb in regulating such a response. Although UmuDAb has been shown to regulate some DNA damage-induced genes [17], [19], the limited similarity between umuDAb and lexA [14] suggests that it may not directly substitute for all LexA function, and allows for the possibility that additional regulators might exist. This stress response of this pathogen is also relevant, as environmental stresses such as dessication and exposure to UV radiation used for decontamination [20] are encountered in health care settings where nosocomial Acinetobacter pathogens abound. Examination of DNA damage and stress responses have been specifically identified as areas in which our knowledge of Acinetobacter virulence is lacking [21].

We observed that the organization, gene content, and regulation of the induced and repressed genes in the mitomycin C-induced DNA damage transcriptome differed significantly between A. baylyi ADP1 and A. baumannii ATCC 17978. These experiments also established different uses for RecA in these two species' DNA damage responses, and suggested that UmuDAb is only one of multiple repressors of the DNA damage response in both species, serving a specialized role in regulating the transcription of error prone polymerases throughout the genome. These error-prone polymerase genes, as well as known virulence-associated genes, were found in bacteriophage particles in A. baumannii ATCC 17978 after DNA damage, which could facilitate the spread of mutation-inducing and other virulence genes to other bacteria.

Materials and Methods

Bacterial strains and growth conditions

A. baylyi strains ADP1, ACIAD1385 (ΔrecA::KanR), and ACIAD2729 (ΔumuDAb::KanR) [22] as well as A. baumannii strains ATCC 17978, its isogenic recA insertion mutant [23], and a ΔumuDAb::KanR null mutant were grown at 37°C in minimal media plus succinate [17] for transcriptome and RT-qPCR analyses, and in Luria-Bertani broth for the production of phage particles. For both RNASeq transcriptome and RT-qPCR analyses, a 3 ml overnight culture, grown at 37°C at 250 rpm, was diluted 1:25 into 5 ml fresh media and grown with shaking for two hours, at which time the culture was split in two and 2 μg/mL mitomycin C (MMC) was added to one culture. Further incubation for three hours served to induce gene expression. DNA damage-induced mutagenesis after UV-C exposure was conducted as described previously [14].

Mutant strain construction

Null mutations of umuDAb (A1S_1389), umuD (A1S_0636) and rumB (A1S_1173) in A. baumannii ATCC 17978 were constructed by replacing the coding sequence of each gene with either the kanamycin resistance gene from the Invitrogen pCRII vector (for umuDAb and umuD), or the streptomycin/spectinomycin resistance cassette from pUI1638 (for rumB), as described previously [22]. Primer sequences are listed in Table S1; “up” primers amplified DNA upstream of each coding sequence, and “dw” primers amplified DNA downstream of each coding sequence. The kanamycin resistance gene was amplified with primers Kmup and Kmdw [24] and the streptomycin/spectinomycin resistance cassette was amplified from pUI1638 [25] with primers StrepSpecFor and StrepSpecRev. Splicing overlap extension PCR was used to construct linear DNA fragments from these three pieces, 300 ng of which was electroporated into A. baumannii ATCC 17978 cells. (In the rumB replacement, the linear fragment was first cloned into the suicide vector pEX18Gm before electroporation [26]). Transformants were selected on LB plates containing 30 μg/mL kanamycin or 10 μg/mL each of streptomycin and spectinomycin. Mutants were confirmed with PCR analyses to contain allelic replacements of the wild type allele and were not merodiploids.

RNASeq experiments and analyses

RNA was purified from one milliliter samples (biological triplicates for 17978; duplicates for ADP1) and processed through the Epicentre MasterPure RNA Purification kit. Further removal of contaminating DNA was performed using the Ambion DNA-free rigorous DNase treatment. RNASeq experiments were conducted with the assistance of Cofactor Genomics (St. Louis). RNA quality was assessed on a BioRad Experion instrument to have a quality corresponding to an RNA Integrity Number equal or greater than 9. Whole transcriptome RNA was extracted from total RNA by removing large and small ribosomal RNA (rRNA) using RiboMinus Bacterial Kit (Invitrogen). Five ug of total RNA was hybridized to rRNA-specific biotin-labeled probes at 70°C for 5 minutes. The rRNA-probe complexes were then removed by streptavidin-coated magnetic beads, and rRNA free transcriptome RNA was concentrated using ethanol precipitation.

In cDNA synthesis, 1 μg of transcriptome RNA was incubated with fragmentation buffer (Illumina RNA-seq kit) for 5 minutes at 94°C. Fragmented RNA was purified with ethanol precipitation. First-strand cDNA was prepared by priming the fragmented RNA using random hexamers and followed by reverse transcription using Superscript II (Invitrogen). The second-strand of cDNA was synthesized by incubation with second-stranded buffer, RNase Out and dNTP (Illumina RNA-seq kit) on ice for 5 minutes. The reaction mix was then treated with DNA Pol I and RNase H (Invitrogen) at 16°C for 2.5 hours.

In constructing cDNA libraries, double-stranded cDNA was treated with a mix of T4 DNA polymerase, Klenow large fragment and T4 polynucleotide kinase to create blunt-ended DNA, to which a single 3′ A base was added using Klenow fragment (3′ to 5′ exo-) provided by an Illumina RNA-seq kit. Size selection of adaptor-ligated DNA was performed by cutting the target fragment out of a 4–12% acrylamide gel. The amplified DNA library was obtained by in-gel PCR using a Phusion High-Fidelity system (New England Biolabs).

Sequencing and cluster generation was performed according to the sequencing and cluster generation manuals from Illumina (Cluster Station User Guide and Genome Analyzer Operations Guide). Primary data were generated using the Illumina Pipeline version SCS 2.8.0 paired with OLB 1.8.0. NovoAlign version 2.07.05 was used for all sequence alignment; aligner algorithm specifics can be obtained from novocraft.com. The coverage depth for sequencing of the A. baylyi ADP1 libraries was an average of 67.5-fold for the wild type strain, 69.7-fold for ACIAD1385, and 148-fold for ACIAD2729, with an average percent coverage of reference bases of 95.8%, 98.2%, and 99.8%, respectively. Coverage depth for the sequenced A. baumannii ATCC 17978 libraries was 235-fold for the wild type strain (99.3% of reference bases covered), 238-fold for the recA strain (98.7% of reference based covered), and 203-fold for the umuDAb strain (97.3% of reference bases covered). An average of 878 million to 1.1 billion total bases were generated for each A. baumannii ATCC 17978 library and over 300 million total bases were generated for each A. baylyi ADP1 library. Clusters were linearly normalized by multiplying each sample's coverage by the total reads of the lower read-count sample divided by the respective sample's total reads, and the induction ratio of reads between MMC-treated and untreated samples was then calculated. Genes were considered induced if this ratio was greater than or equal to 2.0, and considered repressed if this ratio was less than or equal to 0.5 and if the expression levels in at least two wild type uninduced samples were above the detection threshold. The detection threshold for each sample corresponded to one Illumina read in a million that aligned to the reference genome sequences (CP000521, CP000522, and CP000523 for A. baumannii ATCC 17978 [27] and its two plasmids, respectively; CR543861.1 for A. baylyi ADP1 [28]). These sequence datasets have been submitted to the NCBI Short Read Archive (http://www.ncbi.nlm.nih.gov/sra) under accession number SRP036862.

RT-qPCR analysis

RNA samples for RT-qPCR were produced from 1 mL of triplicate biological samples with the Epicentre MasterPure RNA Purification Kit, after which additional removal of DNA was performed using Ambion DNA-free rigorous DNaseI treatment. Removal of contaminating DNA was verified by the absence of PCR products amplified when PCR was performed with primers listed in Table S2 and S3: 17978umuDCRTFor and 17978umuDCRTRev for A. baumannii strains 17978, 17978 recA, and 17978 ΔumuDAb. For A. baylyi strains ADP1 and ACIAD1385, umuDAb#2RTRev and umuDAb#RTFor were used, and for ACIAD2729, dnaNRTFor and dnaNRTRev were used. Genomic DNA was used as a positive control. RNA sample quality was confirmed on an E-Gel EX 2% agarose gel (Invitrogen) before use.

cDNA was synthesized from 1 μg of RNA with oligo(dT) and random hexamers by a modified Moloney murine leukemia virus reverse transcriptase in a 25 μL reaction (Bio-Rad iScript cDNA Synthesis kit). Five μL of a 1:100 dilution of this cDNA was used to perform RT-qPCR using BioRad iTaq SYBR Green Supermix on an Applied Biosystems 7300 Real-Time PCR system in a 15 μL reaction. Technical triplicates were run for each of the three biological replicates in ABI MicroAmp Optical 96-well clear reaction plates, with the following cycling conditions: 95°C for 5 minutes, followed by 40 cycles of 95°C for 15 seconds and 60°C for 1 minute. A dissociation step of 95°C for 15 seconds, followed by 60°C for 30 seconds and 95°C for 15 seconds was used to check product integrity. No template controls for each primer set confirmed absence of product formation. Each RT-qPCR plate contained wild type and one mutant strain sample, comparing reference primers and test primers for the gene of interest.

RT-qPCR primers were designed using PrimerBlast (NCBI) and are listed in Tables S2 and S3. PCR efficiency was evaluated for every primer set by dilution of genomic DNA over 5 logs of template concentration and was between 94% and 105% for all primer sets. Efficiency was calculated using the formula E = 10(−1/slope) of the standard curve generated with the primer set, where efficiency = (E-1)×100%. Primer concentration used was 400 μM. All test gene primer sets were compared to the reference gene primer set 16SrRNA#RTFor and 16SrRNA#2RTRev for A. baylyi ADP1 (Table S2) and 1797816rRNARTFor and 1797816rRNARTRev for A. baumannii ATCC 17978 (Table S3). Validation of these reference primers was performed by observing no significant difference between MMC-treated and untreated samples in either A. baumannii ATCC 17978 or A. baylyi ADP1 in six independent experiments (p>0.05 in a paired t-test). Transcriptional changes were calculated using the 2−ΔΔCT method [29] and GraphPad InStat was used to conduct all statistical analyses.

Bacteriophage purification, electron microscopy and analyses

Phages were produced in cultures grown in LB broth. Overnight cultures were diluted 1:25 into fresh medium and grown at 37°C with shaking for 1 hour before MMC at 2 μg/mL was added. The cultures' optical density at 600 nm was measured each hour for six additional hours after induction with MMC. At either three or six hours, 1 mL of culture was centifuged at 13,000 rpm for two minutes, and the supernatant was filtered through a 0.22 μm filter. This filtrate was centrifuged at 13,200 rpm for one hour at room temperature and the pellet was resuspended in phage buffer (10 mM Tris, pH 7.5, 10 mM MgSO4, 68.5 mM NaCl, 1 mM CaCl2). These samples were processed through the Ambion DNA-free DNase Treatment & Removal kit if PCR analyses were performed.

The resulting phage suspension was processed for transmission electron microscopy. Phage samples were placed on freshly made carbon coated formvar grids for 5 minutes, rinsed with phage buffer and deionized water for ten seconds each, and stained twice with 1% uranyl acetate for one minute each. Uranyl acetate was wicked off and the grid was air dried. Micrographs were taken using 80kV accelerating voltage on a JEOL 100CX transmission electron microscopy onto Kodak 4489 film, then scanned with a Minolta Dimage Scan Multi Pro film scanner at 2400 dpi. The capsid diameter, tail width, and tail length of twelve phage particles observed in micrographs was measured with Image J software [30]. The arithmetic mean of the measurements was reported. Analysis of the genome structure and content of the three cryptic prophage regions was performed using the web server Phage Search Tool (PHAST) [31].

Results

Previous reports had indicated that ddrR [17] and recA [16] were induced by DNA damage in A. baylyi ADP1, and recent observations indicated that multiple error prone polymerases were induced by various forms of DNA damage in A. baumannii ATCC 17978 [19], [32]. However, in the absence of a LexA homolog encoded by these species [14], it was not known whether multiple genes were induced in these species, nor how this response might be regulated. RNA-Seq experiments were performed to test whether A. baylyi ADP1 and A.baumannii ATCC 17978 (henceforth abbreviated as ADP1 and 17978, respectively) possessed a genome-wide transcriptional response to mitomycin C exposure. Genes were considered induced (or repressed) if their expression increased (or decreased) by 2.0-fold or more, relative to their expression in untreated cultures.

A. baylyi ADP1 possess a DNA damage transcriptomes of SOS response genes, a CRISPR/Cas system, and other genes

Sixty-six genes, or 2.0%, of all ADP1 genes were induced (Table 1), indicating a global system of regulating gene expression in response to this form of DNA damage. These 66 induced genes were widely dispersed throughout the chromosome, and included 8 putative operons of two genes each. In addition to these induced genes, an astonishing 38.4% of all ADP1 genes were repressed upon DNA damage.

thumbnail

Table 1. Genes induced in A. baylyi ADP1 after MMC-induced DNA damage and their regulation by UmuDAb and RecA.

doi:10.1371/journal.pone.0093861.t001

A core set of 6 SOS genes for gamma proteobacteria such as Acinetobacter includes recA, ssb, ruvA, ruvB, recN, and uvrA [33], with a larger set of 36 genes induced in Escherichia coli [2] and Pseudomonas aeruginosa [34], the best-studied organism in the order to which Acinetobacter belongs (Pseudomonadales). Surprisingly, only 6 of these 36 genes were induced, and 7 genes were repressed, while 9 genes were neither induced nor repressed (Table 2). ~40% of all SOS genes (dinI, dinG, hokE, lexA, molR, pcsA, polB, sbmC, sulA, ybfE, ydjM, ydjQ, yebG, yigN; all of which are present in E. coli and some of which are present in P. aeruginosa) are not encoded in the ADP1 genome, although none of these were ‘core’ SOS genes. RT-qPCR experiments confirmed that recA, ssb, umuDAb, and ddrR were each induced >2-fold.

thumbnail

Table 2. Regulation and presence of canonical SOS genes in Acinetobacter species.

doi:10.1371/journal.pone.0093861.t002

In addition to the SOS response genes involved in recombination and repair, the induced genes included nucleotide metabolism-related nrdAB (ACIAD0722/0724) encoding ribonucleotide reductase, dgt (ACIAD2613), involved in breakdown of ssDNA (a trigger of the SOS response, [10]), and endonuclease G (ACIAD3408) (Table 1). Numerous genes encoded chaperones (htpG, ACIAD0316; acuD/papD, ACIAD0388) or other proteins involved in oxidative or other stresses, notably dnaK (ACIAD3654), hemO (heme oxygenase, ACIAD1478), dps (ACIAD1205), and the ribosomal inhibitor raiA (ACIAD3535). Detoxification enzymes encoded by aphC (ACIAD2103) and gst (ACIAD0445) were also induced. gst encodes a member of the glutathione S-transferase family, which is involved in oxidative stress and xenobiotic/antimicrobial agent detoxification [35] and was the most highly induced gene in both species. Additionally, acuA (ACIAD0387), which encodes the thin pilus subunit mediating ADP1 adhesion [36] was induced, as were the adaptive immunity Cascade proteins associated with a Type I-F CRISPR/Cas locus (ACIAD2479-2484) identified by the CRISPRFinder web server [37]. ACIAD2481 and 2482 were further confirmed to be induced with RT-qPCR experiments (data not shown).

No genes were induced that are involved in natural competence, or were in either of the two putative prophage regions identified for ADP1 [28]. It is not known whether these putative prophages are functional [28]. One of these two prophages regions may be too small (9 kb) to encode a functional prophage, while the other, although possessing an appropriate genome size (53 kb), seems not to encode several proteins (e.g. in DNA replication and capsid and tail structural elements) normally required to produce functional bacteriophage particles, and only one third of its genes are of phage origin (Figure 1A).

thumbnail

Figure 1. Comparison of the ADP1 and 17978 chromosomal regions indicated as putative prophages.

This to-scale diagram indicates the size and predicted gene functions from PHAST analysis within the chromosomal regions designated as putative prophages. Locations in the genome are represented by the ACIAD and A1S gene designation boundaries for ADP1 and 17978, respectively. (A) All genes represented in the ADP1 prophage region were not induced after DNA damage, and approximately one-third of the genes in this region were either hypothetical conserved phage genes, or genes of predicted phage function (with only 2 each of DNA metabolism/replication, capsid/packaging and tail genes). (B) Three chromosomal regions of induced genes in A. baumannii ATCC 17978 overlapped with three regions of the genome designated as putative prophages by our PHAST analysis [31] and as cryptic prophages CP5, CP9 and CP14 [38]. Genes that were not induced in this study are marked with asterisks. Previously described virulence-associated esv genes [27] are named below each gene, and error-prone polymerase alleles are shown in bright cyan color (see legend). The specific gene function of each gene in order of its placement in each prophage is described in Table 4.

doi:10.1371/journal.pone.0093861.g001

The DNA damage transcriptome of A. baumannii ATCC 17978 includes three prophages

In A. baumannii ATCC 17978, 152 genes, or 3.7% of the genome, was induced (Table 3), indicating that the expression of multiple genes was regulated in response to this form of DNA damage. In 17978, as for ADP1, few of the canonical SOS genes responded to MMC-induced DNA damage as in the E. coli model (Table 2). Only 4 of these 36 genes were induced, with core SOS genes recA, umuD and ssb induced, similar to ADP1. Two genes were repressed (holB and ruvC, also repressed in ADP1), while 16 genes were neither induced nor repressed, and the same 14 genes as were absent from the ADP1 genome were absent in the 17978 genome. There was no significant difference between ADP1 and 17978 in the number of genes in these four classes (induced, repressed, unaffected, absent) as tested with Chi-square analysis (p>0.05).

thumbnail

Table 3. Genes induced in A. baumannii 17978 after MMC-induced DNA damage and their regulation by UmuDAb and RecA.

doi:10.1371/journal.pone.0093861.t003

In contrast to the dispersal of induced genes throughout the chromosome in ADP1, the location of 90% of all 17978 induced genes nearly perfectly overlapped with three regions predicted to contain prophages by our analysis with the Phage Search Tool (PHAST), a web server designed to rapidly and accurately identify, analyze, and annotate prophages [31] (Figure 1B). These three regions were also the only regions to be identified as cryptic prophages (CP; CP5, CP9, and CP14) in 17978, based on their presence in some but not all epidemic-associated A. baumannii strains [38]. Ninety-nine percent of all genes within these prophage regions were induced.

These cryptic prophages encode some of the error-prone polymerase components that were induced in 17978: umuDrumB (A1S_1173-1174) in CP5, and A1S_2014-2015 in CP9. (A1S_2014, putatively transcribed in an operon with A1S_2015, belongs to a newly described family of SOS response associated thiol autopeptidases (SRAP; [39]) and was therefore included in the category of error-prone polymerase components.) Non-phage associated polymerases or polymerase components included umuDAb (A1S_1389) and a umuDC operon (A1S_0636-0637) that is located in a genomic island that may have been horizontally acquired from a Yersinia plasmid [38]. Notably, all error prone polymerases or polymerase components, as well as the conserved ddrR gene adjacent to umuDAb, were induced in both the RNASeq and additional RT-qPCR experiments. Both umuD and umuC genes in each operon were induced, although the level of induction in each case was greater for umuD than umuC, consistent with its position at the beginning of the operon and with its use in a 2:1 ratio to the umuC gene product in DNA polymerase V activity [7].

Virulence-associated genes were also induced in 17978. Previous studies in a Caenorhabditis elegans model of A. baumannii ATCC 17978 infection demonstrated that ethanol-stimulation of virulence was dependent upon 12 esv genes [27]. Two of these, esvK1 and esvK2, which were encoded in CP14, were induced, with the induction of esvK1 further tested and confirmed in RT-qPCR experiments. While the induction of esvI, encoded by CP9, fell just below the RNASeq induced cutoff ratio of 2.0, it was induced ~5-fold after MMC treatment in RT-qPCR experiments. No CRISPR-Cas system is present in 17978, although some isolates of the EU clone lineage I possess a CRISPR-Cas system [40].

Although none of the genes on the 17978 plasmids pAB1 and pAB2 were induced, 6 of 11 pAB1 genes, and 2 of 6 pAB2 genes were repressed. Overall, 11.4% of all 17978 genes were repressed.

The A. baylyi DNA damage transcriptome includes four different regulons of MMC-induced genes

In the SOS response of gammaproteobacteria, RecA action is typically required to relieve SOS genes from repression by either LexA [4] or a prophage repressor [41]. We conducted RNASeq analysis on both recA and umuDAb mutant strains of ADP1 and 17978 to test whether recA regulated these transcriptomes, and whether umuDAb was a global regulator of DNA damage-induced (and/or repressed) genes in a LexA-analogous manner.

These experiments demonstrated a complex picture of regulation, with the ADP1 transcriptome possessing four regulons of induced genes that differentially required umuDAb and recA (Table 1). Figure 2A shows these four regulons, which were supported by statistical testing (repeated measures analysis of variance within each regulon; p<0.05). Twelve genes were regulated by both umuDAb and recA; 13 genes required recA only. Unexpectedly, we found a regulon of 22 genes that were induced after DNA damage but required neither recA nor umuDAb for this induction. Additionally, 17 genes were regulated only by umuDAb, but all of these were still moderately induced in the umuDAb mutant, having an average induction ratio of 1.70-fold (only slightly below the cutoff for being considered induced), and were not investigated further. These categories were validated by RT-qPCR experiments: ddrR required both recA and umuDAb, dnaN required only recA, and ssb and nrdA required neither recA nor umuDAb. In the eight induced operons, the regulation was the same throughout the operon, supporting the categorization and physiological relevance of the regulation method.

thumbnail

Figure 2. Distribution of regulation mechanisms for mitomycin C-induced and repressed transcriptome in ADP1 and 17978.

The absolute number of genes induced (A) or repressed (panel B) by MMC in the transcriptome of ADP1 and 17978 is shown. The designation of regulon is represented by the following terms: Neither (genes requiring neither umuDAb nor recA for regulation), Both (genes requiring both umuDAb and recA for regulation), RecA (genes requiring only recA for regulation), or UmuDAb (genes requiring only umuDAb for regulation). (A) Many more repressed genes were observed in ADP1 than 17978, with UmuDAb sufficing for this repression in most genes; 17978 repressed genes required either UmuDAb or both UmuDAb and RecA. (B) A greater number of induced genes was observed in 17978 than ADP1, and these genes required either RecA or both RecA and UmuDAb. In comparison, ADP1 induced genes belong to four regulons (Neither, Both, RecA, or UmuDAb).

doi:10.1371/journal.pone.0093861.g002

This variety in regulatory requirements also extended to the induced SOS genes (Figure 3A). None of the five canonical SOS genes that were induced (recA, dnaN, uvrA, ssb, and ruvA) depended upon umuDAb. Only three of the five SOS genes were recA-regulated, none of the five were umuDAb-regulated, and strikingly, ssb was regulated by neither recA nor umuDAb (Figure 3A). This regulation was confirmed in RT-qPCR experiments.

thumbnail

Figure 3. Mitomycin-C induced and repressed SOS genes in ADP1 and 17978 differentially require RecA and UmuDAb.

Gene induction ratios obtained from RNASeq transcriptome experiments are shown, with the induction (panel A) and repression (panel B) of canonical SOS genes. Gene names prefaced by “AD” indicate ADP1 genes; “AB” indicates 17978 genes, with A1S numbers of 17978 genes listed for umuDC alleles. The placement of the horizontal axis in each panel represents the cutoff level for a gene to be considered induced (A) or repressed (B). Bars above the horizontal axis indicate induced genes (panel A), and bars below the horizontal axis indicate repressed genes (panel B), with bars rising either below (panel A) or above (panel B) this axis to have lost their induction or repression, respectively, in the umuDAb and/or recA mutant strains. (A) Induced genes did not require umuDAb in either ADP1 or 17978, except for the category of umuDC alleles, and recA was required for all induced 17978 genes but only some induced ADP1 genes (ruvA and uvrA). (B) Repressed genes required only umuDAb in ADP1 but required both umuDAb and recA in 17978.

doi:10.1371/journal.pone.0093861.g003

In contrast, throughout the ADP1 genome, including all repressed SOS genes, 87% of the repressed genes required only umuDAb to be repressed, with just 6% requiring both umuDAb and recA, and 7% requiring neither of these genes for repression (Figure 2B).

The A. baumannii DNA damage transcriptome requires RecA regulation and displays a specialized regulatory role for the UmuDAb repressor

In contrast to ADP1, 17978 exhibited only a recA-dependent path of inducing genes—with the exception of recA itself and A1S_2020, which were induced 2.0 to 2.2 –fold, respectively, in the recA mutant. However, the 17978 induced transcriptome contained two DNA-damage induced regulons: i) 123 genes regulated by recA (i.e. umuDAb-independent), and ii) 27 genes regulated by recA and umuDAb (i.e. umuDAb-dependent) (Table 3, Figure 2A). Within the umuDAb-independent regulon, there was a significant difference between the induction of the wild type vs. the umuDAb samples (p<0.05 in a Wilcoxon matched-pairs signed-ranks test), suggesting a possibly different role of umuDAb from simple repression. Consistent with the proportions of genes in these two regulons, 85% of the induced genes in the three prophages CP5, CP9, and CP14 required recA only, and this regulation was not significantly different for conserved hypothetical genes vs. genes typically found in bacteriophages (p>0.05, Fisher's exact test.). These observations were consistent with the possibility of gene repression by a prophage-encoded repressor. Of the 8 induced canonical SOS genes (which includes 6 alleles of umuDC), only the umuDC alleles were dependent on umuDAb for induction (Figure 3A). The recA and ssb genes' induction were umuDAb independent (Figure 3A). This regulation of ssb, umuDAb, and recA was confirmed in RT-qPCR experiments.

In the DNA damage-repressed transcriptome, this pattern was reversed: umuDAb was required for 99% of the genes' repression, with recA also required in ~49% of the cases. This was also observed in the repressed SOS genes, where umuDAb was required for repression after DNA damage of holB and ruvC, but recA was required for repression as well (Figure 3B). However, repression of 7 of the 8 genes located on the plasmids pAB1 and pAB2 required both umuDAb and recA.

We further tested whether all of the prophage-encoded error-prone polymerase alleles (CP5 (A1S_1173/1174, umuDrumB) and CP9 (A1S_2014-15)) were regulated similarly to their chromosomal counterparts umuDC (A1S_0636-0637) and the regulatory umuDAb gene (A1S_1389). All were regulated by recA and de-repressed in the umuDAb mutant (i.e. had high expression in the absence of MMC exposure), with this regulation confirmed by RT-qPCR experiments (Figure 4). This was not observed for non- umuDC-related alleles, either prophage- or chromosomally- encoded (recA, esvK1, and ssb): although recA-dependent, they were not regulated by umuDAb or de-repressed in the umuDAb mutant (Figure 4).

thumbnail

Figure 4. RT-qPCR experiments indicate that umuDAb is required for repression of error-prone polymerase components, not all DNA damage-induced genes.

Delta Cq values from RT-qPCR experiments measuring expression of selected A. baumannii ATCC 17978 genes demonstrates the repressing activity of UmuDAb only for error prone polymerase components. The expression of each gene in both wild type and umuDAb null mutant is shown, with gene identity and A1S number listed on the x axis. Each gene was assayed in one RT-qPCR experiment (plate), with error bars indicating standard error of the mean from technical triplicates of biological triplicates.

doi:10.1371/journal.pone.0093861.g004

DNA damage in A. baumannii induces bacteriophage particle production that contain virulence genes

PHAST characterization of the prophages apparently encoded by the three induced regions of the 17978 chromosome indicated that the majority of each prophage's genes (65% in CP14, 70% in CP5, and 81% in CP9) were either conserved hypothetical phage genes or phage genes of a function identified by homology (Figure 1B, Table 4). Of these genes typically found in bacteriophages, 65 - 78% (depending on the CP) were most similar to phage genes from the viral family Siphoviridae. In CP9, 68% of the phage genes were most similar (identity ranging from 60 – 100%) to genes in the Acinetobacter siphovirus BΦ-B1251, which was found in a sewage sample and lysed a carbapenem-resistant A. baumannii clinical isolate [42]. Both CP5 and CP14 were composed of a variety of phages' genes, with no one species being in the majority. This difference was statistically significant (p<0.05, Fisher's exact test).

thumbnail

Table 4. Description of gene functions in order of appearance in each prophage in 17978.

doi:10.1371/journal.pone.0093861.t004

All three prophage regions were within the size range for non-Bacillus siphovirus genomes (14–56 kb [43], [44]) and were organized into modules of (in this order): lysogeny/regulation, DNA metabolism, DNA packaging and head, tail, and lysis genes (Figure 1B), which is the same organization as in genomes from the family Siphoviridae [43], [45]. This analysis and annotation by the PHAST software, as well as manual characterization and genome size of the three prophage regions, suggested that the 45 kb CP5 was an intact prophage, encoding the requisite morphological (capsid, packaging, tail), DNA replication and lysogeny regulation (including repressors; Table 4) gene products indicative of a functional prophage. However, the 49 kb CP9 and the 22 kb CP14 also contained these genes and may be intact prophages as well. Thus the composition as well as the induction of these prophages differed from the (uninduced) prophage loci present in ADP1, the larger (~53 kb) of which contains only roughly one third of its genes as phage genes (either as conserved hypotheticals or of known function) (Figure 1A), [28], most of which resembled genes from the Family Myoviridae.

We hypothesized that because ~99% of all the genes in these prophages were induced after DNA damage, bacteriophage particles might be produced under these conditions. When 17978 cells were grown in LB medium in the presence of MMC, a decrease in culture turbidity was observed beginning around two hours post-exposure, relative to untreated cells (Figure 5A). Transmission electron microscopy was used to visualize intact phage particles of uniform morphology from filtered supernatants of these cultures in three independent experiments. Morphological analyses of these phages showed them to have a non-enveloped capsid of approximately 57 nm in diameter, and a long, thin (11 nm), flexible tail of approximately 167 nm that possessed tail fibers (Figure 5B). These morphological features, together with the size, content and organization of the three prophage regions, suggest that the phage particles may belong to the viral family Siphoviridae. Bacteriophages in the Myoviridae family visually resemble siphoviruses but possess a wider and inflexible tail. Furthermore, viruses in the Myoviridae family are lytic, and thus not consistent with the temperate nature of the phages that we observed.

thumbnail

Figure 5. Mitomycin C treatment induces production of bacteriophage particles in 17978.

(A) Overnight LB cultures of 17978 cells were diluted into fresh LB medium and grown for 0.75 hours before addition of 2 μg/mL MMC. After approximately two hours of MMC treatment, the optical density leveled off and decreased slightly but continued to increase in the absence of MMC treatment. Error bars represent standard error of the mean from three independent experiments. (B) Electron micrograph of bacteriophage particles at 100,000× magnification, showing polyhedral capsid, long, flexible tail and tail fibers. Results shown are representative of three independent experiments producing and imaging bacteriophage particles.

doi:10.1371/journal.pone.0093861.g005

We tested whether these similar-looking bacteriophages represented the products of CP5, CP9, CP14, or a mixture of all three of these prophages, as was suggested by the induction of genes in all three prophage regions. Phage particles were purified away from bacterial fragments and chromosomal DNA in both MMC-treated and untreated cultures by DNAse treatment of supernatant that had been 0.22 μm filtered and precipitated. PCR amplification experiments were performed on these DNAse-treated, purified samples to determine whether genes from each prophage region were present in the particles we observed. Primers that amplified portions of rumB (from CP5), esvI and umuC A1S_2015 (from CP9), and esvK1 (from CP14) all yielded PCR products only from 3 hour-MMC-treated, purified culture supernatants, but not from untreated, purified culture supernatants, in three independent experiments. This suggests that all three putative prophage regions might produce phage particles when induced by MMC, although it was not unambiguously determined that these particles are encoded by each of the three prophage regions independently, or whether one of these phage, e.g. CP5, might have served as a helper virus for the production of particles containing CP9 or CP14 prophage DNA.

We next tested the hypothesis that the umuD-rumB (A1S_1173-1174) operon that we observed in the phage lysate is responsible for the DNA damage-induced mutagenesis previously observed in this strain [14], [32]. Compared to the frequency of rifampin-resistant mutants observed in wild type 17978 cells after UV exposure, a rumB null mutant displayed only approximately ~40% of the rifampin mutation frequency after DNA damage (in four independent experiments). This suggests that if CP5 produced phage particles, these could transduce these genes into a new host and so allow error-prone replication of DNA in this host. However, a similar, partial (~65%) reduction of rifampin resistance frequency in a non-phage encoded umuD (A1S_0636) null mutant was also observed (in six independent experiments). The apparent redundancy of these error-prone polymerases in the DNA damage-inducible mutagenesis occurring in the 17978 strain is likely a reflection of these polymerases being of bacteriophage as well as bacterial origin in this species.

Discussion

These transcriptome studies of A. baumannii ATCC 17978 and A. baylyi ADP1 indicated that a genome-wide system of inducing and repressing genes after DNA damage exists in these species. Between 2% and 4% of these species' genes were induced after mitomycin C treatment, but their distribution throughout the chromosome differed greatly, with localization of most (~90%) of the 17978 genes into three prophages but wide dispersal of ADP1 induced genes throughout the chromosome. There was little overlap in the DNA damage-induced transcriptomes of these organisms with either canonical SOS genes (only 11–17% of which were induced) or each other (only recA, ssb, umuDAb, ddrR, and gst were induced in both species). recA, ssb, and umuD are core SOS genes, whereas gst encodes a member of the glutathione S-transferase (GST) family, which protects against oxidative stress and detoxifies endogenous, xenobiotic and antimicrobial compounds [35]. A. baumannii ATCC 17978, like many proteobacteria, possesses multiple (9) gst genes [46], as does A. baylyi ADP1 [28]. The induced gst genes (A1S_0408 and ACIAD0445) share 69% amino acid identity, and are present in a highly syntenic chromosomal location in these two species, allowing for the possibility that A1S_0408 and ACIAD0445 may be members of the GST family that participate in the DNA damage response.

Besides a subset of the canonical SOS genes, stress proteins and chaperones, ADP1 induced the genes of a CRISPR/Cas system, which are bacterial adaptive immunity/defense modules. The cell processes foreign, e.g. bacteriophage, DNA molecules and forms a CRISPR array locus in the chromosome composed of short segments of these DNA sequences [47]. The next time similar DNA molecules enter the cell, Cascade proteins (cas gene-encoded, typically adjacent to the CRISPR repeat locus) and transcribed CRISPR sequences bind to and cleave the incoming foreign DNA. The A. baylyi ADP1 Type I-F CRISPR/Cas locus consists of the Cascade proteins encoded by cas3/cas2 (ACIAD2477), as well as two conserved hypothetical genes, csy2, csy3, cas6, and cas1 (ACIAD2479-2484, which were induced by MMC). It is intriguing that in A. baylyi ADP1 cells, which are naturally competent for the uptake of and transformation with DNA [15], this CRISPR/Cas defense against foreign DNA appears to be functional. Short transcribed CRISPR RNA molecules (crRNA), identical to those comprising the CRISPR repeats adjacent to the induced ACIAD2479-2484 genes, accumulate in ADP1 cells after treatment with nalidixic acid [48], a well-known inducer of the SOS response. The dependence of these crRNA molecules' formation on new protein synthesis [48] is consistent with induction of the cas genes that we observed. A link between a CRISPR/Cas system and DNA repair has been observed in E. coli, where the Cas1 nuclease YgbT acts on both branched DNAs and in antiviral immunity [49]. However, to our knowledge, this is the first evidence of transcriptional induction of a CRISPR/Cas system gene by DNA damage. To the limited extent that CRISPR/Cas genes' expression has been studied, a constitutive level has been assumed [48], but the uninduced level of A1S_2479-2484 expression is modest, being below the average uninduced level of the 66 induced genes, but approximately four times the detection threshold of the RNASeq experiments.

Our data are largely consistent with those observed in recent microarray studies of A. baumannii ATCC 17978 in which 39 genes were induced more than 1.5-fold after MMC treatment [19], with 77% of that study's genes also induced in our experiments. The greater number of induced genes observed in our study (152), as well as the variation in the specific identity of the induced genes may be because of the different methodologies used (RNA-Seq vs microarray), and also because Aranda et al. used a rich medium source (LB broth), a shorter induction time of two hours, and one-quarter the amount of MMC as in this study. The invariant conservation of the induction of all error-prone polymerases and polymerase components in this and other studies [19], [32], however, supports the centrality of these genes to the DNA damage response of this species.

Further transcriptional profiling of umuDAb and recA mutant strains of both species after MMC treatment allowed determination of the roles of these putative regulators in the DNA damage responses. In the DNA damage-induced transcriptomes, recA was required for the induction of only 38% of the ADP1 induced genes, but virtually all of the 17978 induced genes, which is consistent with both the known SOS response mechanism [4] and the involvement of recA in antimicrobial resistance, general stress responses, and virulence in 17978 [23]. This recA dependence is also consistent with the repression of the prophage genes by a prophage-encoded repressor [41] as opposed to a LexA-like, UmuDAb-mediated repression of these genes. However, we observed that 9–19% of each of the three prophage genomes required umuDAb for gene induction in addition to recA, which argues against a solitary action of RecA-facilitated autocleavage of a prophage repressor in the response we observed. The action of UmuDAb, a potential LexA homolog, was complex in both species, playing a role in only 44% of ADP1 induced genes, and in 16% of 17978 induced genes, including both those encoded in prophages and in the chromosome. The large number of repressed genes in the DNA damage transcriptomes, especially of ADP1 (Figure 2B), was unexpected, with the repressor action of UmuDAb being consistent with its involvement in the repression of the vast majority of these genes in ADP1, although its action may be indirect rather than direct.

The de-repression that we observed of ddrR and all umuDC alleles in a null umuDAb mutant is consistent with recent observations that UmuDAb binds to, and regulates, the promoters of these genes in A. baumannii ATCC 17978 [19], although those studies used a umuDAb insertion mutant and not a null mutant. However, our genome-wide profiling of umuDAb regulation of induced genes found that unlike for the umuDC alleles, the induction of the majority (83.5%) of all genes in 17978 was umuDAb-independent. Either UmuDAb is not the sole LexA-like repressor in this species, or has a mechanism of action unlike LexA, because a LexA-regulated regulon of DNA damage-induced genes would have become de-repressed in the absence of DNA damage, which was not observed (except for the umuDC and ddrR genes). Furthermore, umuDAb was required for the induction of genes that are not error-prone polymerases and which were encoded in prophage regions (Table 3). These data suggest that UmuDAb does not serve as a direct replacement of LexA for the entire DNA damage regulon in this genus, instead serving a more specialized role in repressing error-prone polymerases. This specialized UmuDAb role invokes an additional DNA damage-related repressor to regulate gene expression after DNA damage, which is consistent with the failure of RecA to regulate its own induction, seen both in this study and previously for A. baylyi ADP1 [16] and A. baumannii [32].

In having multiple umuDAb-dependent and –independent regulons, the behavior of Acinetobacter in regulating their genes after DNA damage is more like its closer pseudomonad relatives, which contain multiple regulons of DNA damage-induced genes involving different (LexA) repressor proteins [12], than it is to enteric bacteria such as E. coli. These Acinetobacter species, like P. aeruginosa, also repressed many more genes than they induced in response to DNA damage, and both genera repressed multiple canonical SOS genes in a lexA-independent manner (recG in ADP1, and holB and ruvC in both ADP1 and 17978), and induced nrdAB and prophage genes [34].

Our observation of the 17978 strain possessing DNA damage-inducible bacteriophages that encode mutation-inducing (error prone) polymerase genes may hold significant implications for the evolution of virulence and antibiotic resistance in related strains. CP5 encodes the umuDrumB operon, which this study found to be responsible for at least half of the DNA damage-induced mutagenesis, while CP9 encodes A1S_2015, annotated as an “error-prone lesion bypass DNA polymerase V” that might also contribute to mutagenesis after DNA damage [32]. Multiple DNA damage-inducing agents–UV-C exposure [14], [32] as well as methyl methanesulfonate, dessication, and ciprofloxacin [32]–are capable of inducing mutagenesis (as measured by rifampin resistance) in A. baumannii ATCC 17978 and AB0057 [14]. A. baumannii strains AB0057 and 3909 also contain CP5 that encodes the umuDrumB genes [38], while A. baumannii ATCC 19606 and D1279779, strains not investigated by DiNocera et al., also possess a very similar CP5-like prophage region that encodes umuDrumB (Figure S1). This indicates the possibility of a widespread mechanism in this species for spread of these error-prone polymerase genes in response to multiple stimuli. Virulence-associated genes such as esvK1 and esvK2 (encoded in CP14), and esvI (encoded in CP9) that contributed to ethanol-stimulated virulence in a model of C. elegans infection by the 17978 strain [27] are also encoded by these prophages and could contribute to the evolution of strains through transduction by bacteriophages that may be produced from, or encapsidate, the genomes of CP5, CP9, or CP14, although these phages have not yet been shown to infect other hosts.

The overall patterns of UmuDAb and RecA usage in these species suggests that diverse mechanisms exist in A. baylyi ADP1 for the repression and induction of genes, which include a regulon induced by neither UmuDAb nor RecA. In contrast, A. baumannii ATCC 17978 almost universally depends on RecA (as well as UmuDAb) but also uses additional, unknown repressors and/or regulators, possibly of prophage origin, in addition to UmuDAb. These species therefore offer robust model systems in which to study the processes of gene regulation after DNA damage, with A. baumannii additionally posing a relevant biological problem in its possible dissemination of error-prone polymerases.

Supporting Information

Figure S1.

CP5-like prophage regions present in A. baumannii strains. The three to-scale diagrams indicate CP-like prophage regions present in A. baumannii strains ATCC 19606 and D1279779. Analysis and image production was performed using the PHAST webserver, with the color-coding indicating the likely function assigned to each coding sequence. The numbered bar indicates the nucleotide number in the genome, with coding regions in the three forward frames shown above the bar and coding regions in the three reverse frames shown below the bar for each strain.

doi:10.1371/journal.pone.0093861.s001

(TIF)

Table S1.

PCR primers used in constructing umuDAb, umuD, and rumB mutants of A. baumannii ATCC 17978.

doi:10.1371/journal.pone.0093861.s002

(DOCX)

Table S2.

Primers used in RT-qPCR experiments in A. baylyi ADP1.

doi:10.1371/journal.pone.0093861.s003

(DOCX)

Table S3.

Primers used in RT-qPCR experiments in A. baumannii ATCC 17978.

doi:10.1371/journal.pone.0093861.s004

(DOCX)

Acknowledgments

We gratefully acknowledge Veronique de Berardinis for A. baylyi ADP1 strains ACIAD1385 and ACIAD2729, and thank the Germán Bou lab for the recA strain of A. baumannii ATCC 17978. We also thank John Andersland, Rodney King, Kurt Gibbs, and James Bradley for technical assistance and helpful discussions.

Author Contributions

Conceived and designed the experiments: JMH JCF. Performed the experiments: ANG JMH JCF. Analyzed the data: JMH TAW ANG. Wrote the paper: JMH ANG.

References

  1. 1. Khil PP, Camerini-Otero RD (2002) Over 1000 genes are involved in the DNA damage response of Escherichia coli. Mol Microbiol 44: 89–105. doi: 10.1046/j.1365-2958.2002.02878.x
  2. 2. Walker GC (1996) The SOS Response of Escherichia coli. Washington, D.C.: ASM Press.
  3. 3. Friedberg EC (1995) DNA repair and mutagenesis. Washington, D.C.: ASM Press.
  4. 4. Little JW, Mount DW (1982) The SOS regulatory system of Escherichia coli. Cell 29: 11–22. doi: 10.1016/0092-8674(82)90085-x
  5. 5. Huisman O, D'Ari R (1981) An inducible DNA replication-cell division coupling mechanism in E. coli. Nature 290: 797–799. doi: 10.1038/290797a0
  6. 6. Schoemaker JM, Gayda RC, Markovitz A (1984) Regulation of cell division in Escherichia coli: SOS induction and cellular location of the sulA protein, a key to lon-associated filamentation and death. J Bacteriol 158: 551–561.
  7. 7. Tang M, Shen X, Frank EG, O'Donnell M, Woodgate R, et al. (1999) UmuD' (2)C is an error-prone DNA polymerase, Escherichia coli pol V. Proc Natl Acad Sci U A 96: 8919–8924. doi: 10.1073/pnas.96.16.8919
  8. 8. Kim SR, Maenhaut-Michel G, Yamada M, Yamamoto Y, Matsui K, et al. (1997) Multiple pathways for SOS-induced mutagenesis in Escherichia coli: an overexpression of dinB/dinP results in strongly enhancing mutagenesis in the absence of any exogenous treatment to damage DNA. Proc Natl Acad Sci U A 94: 13792–13797. doi: 10.1073/pnas.94.25.13792
  9. 9. Brent R, Ptashne M (1981) Mechanism of action of the lexA gene product. Proc Natl Acad Sci U A 78: 4204–4208. doi: 10.1073/pnas.78.7.4204
  10. 10. Horii T, Ogawa T, Nakatani T, Hase H, Matsubara H, et al. (1981) Regulation of SOS functions: purification of E. coli LexA protein and determination of its specific site cleaved by the RecA protein. Cell 27: 515–522. doi: 10.1016/0092-8674(81)90393-7
  11. 11. Little JW, Edmiston SH, Pacelli LZ, Mount DW (1980) Cleavage of the Escherichia coli lexA protein by the recA protease. Proc Natl Acad Sci U A 77: 3225–3229. doi: 10.1073/pnas.77.6.3225
  12. 12. Abella M, Campoy S, Erill I, Rojo F, Barbé J (2007) Cohabitation of two different lexA regulons in Pseudomonas putida. J Bacteriol 189: 8855–8862 doi:10.1128/JB.01213-07.
  13. 13. Jara M, Nunez C, Campoy S, Fernandez de Henestrosa AR, Lovley DR, et al. (2003) Geobacter sulfurreducens has two autoregulated lexA genes whose products do not bind the recA promoter: differing responses of lexA and recA to DNA damage. J Bacteriol 185: 2493–2502. doi: 10.1128/jb.185.8.2493-2502.2003
  14. 14. Hare JM, Bradley JA, Lin CL, Elam TJ (2012) Diverse DNA damage responses in Acinetobacter include the capacity for DNA damage-induced mutagenesis in the opportunistic pathogens Acinetobacter baumannii and Acinetobacter ursingii. Microbiology 158: 601–611. doi: 10.1099/mic.0.054668-0
  15. 15. Young DM, Parke D, Ornston LN (2005) Opportunities for Genetic Investigation Afforded by Acinetobacter baylyi, a nutritionally versatile bacterial species that is highly competent for natural transformation. Annu Rev Microbiol 59: 519–551. doi: 10.1146/annurev.micro.59.051905.105823
  16. 16. Rauch PJ, Palmen R, Burds AA, Gregg-Jolly LA, van der ZeeJR, et al. (1996) The expression of the Acinetobacter calcoaceticus recA gene increases in response to DNA damage independently of RecA and of development of competence for natural transformation. Microbiology 142: 1025–1032. doi: 10.1099/00221287-142-4-1025
  17. 17. Hare JM, Perkins SN, Gregg-Jolly LA (2006) A Constitutively Expressed, Truncated umuDC Operon Regulates the recA-Dependent DNA Damage Induction of a Gene in Acinetobacter baylyi Strain ADP1. Appl Env Microbiol 72: 4036–4043. doi: 10.1128/aem.02774-05
  18. 18. Hare JM, Adhikari S, Lambert KV, Hare AE, Grice A (2012) The Acinetobacter regulatory UmuDAb protein cleaves in response to DNA damage with chimeric LexA/UmuD characteristics. FEMS Microbiol Lett 334: 57–65. doi: 10.1111/j.1574-6968.2012.02618.x
  19. 19. Aranda J, Poza M, Shingu-Vázquez M, Cortés P, Boyce JD, et al. (2013) Identification of a DNA-Damage-Inducible Regulon in Acinetobacter baumannii. J Bacteriol 195: 5577–5582 doi:10.1128/JB.00853-13.
  20. 20. Rutala WA, Gergen MF, Weber DJ (2010) Room decontamination with UV radiation. Infect Control Hosp Epidemiol 31: 1025–1029. doi: 10.1086/656244
  21. 21. Mortensen BL, Skaar EP (2012) Host-microbe interactions that shape the pathogenesis of Acinetobacter baumannii infection. Cell Microbiol 14: 1336–1344 doi:10.1111/j.1462-5822.2012.01817.x.
  22. 22. De Berardinis V, Vallenet D, Castelli V, Besnard M, Pinet A, et al.. (2008) A complete collection of single-gene deletion mutants of Acinetobacter baylyi ADP1. Mol Syst Biol.
  23. 23. Aranda J, Bardina C, Beceiro A, Rumbo S, Cabral MP, et al. (2011) Acinetobacter baumannii RecA Protein in Repair of DNA Damage, Antimicrobial Resistance, General Stress Response, and Virulence. J Bacteriol 193: 3740–3747. doi: 10.1128/jb.00389-11
  24. 24. Aranda J, Poza M, Pardo BG, Rumbo S, Rumbo C, et al. (2010) A rapid and simple method for constructing stable mutants of Acinetobacter baumannii. BMC Microbiol 10: 279–290. doi: 10.1186/1471-2180-10-279
  25. 25. Eraso JM, Kaplan S (1994) prrA, a putative response regulator involved in oxygen regulation of photosynthesis gene expression in Rhodobacter sphaeroides. J Bacteriol 176: 32–43.
  26. 26. Hoang TT, Karkhoff-Schweizer RR, Kutchma AJ, Schweizer HP (1998) A broad-host-range Flp-FRT recombination system for site-specific excision of chromosomally-located DNA sequences: application for isolation of unmarked Pseudomonas aeruginosa mutants. Gene 212: 77–86. doi: 10.1016/s0378-1119(98)00130-9
  27. 27. Smith MG, Gianoulis TA, Pukatzki S, Mekalanos JJ, Ornston LN, et al. (2007) New insights into Acinetobacter baumannii pathogenesis revealed by high-density pyrosequencing and transposon mutagenesis. Genes Dev 21: 601–614. doi: 10.1101/gad.1510307
  28. 28. Barbe V, Vallenet D, Fonknechten N, Kreimeyer A, Oztas S, et al. (2004) Unique features revealed by the genome sequence of Acinetobacter sp. ADP1, a versatile and naturally transformation competent bacterium. Nucleic Acids Res 32: 5766–5779. doi: 10.1093/nar/gkh910
  29. 29. Livak KJ, Schmittgen TD (2001) Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods San Diego Calif 25: 402–408 doi:10.1006/meth.2001.1262.
  30. 30. Abramoff MD, Magalhaes PJ, Ram SJ (2004) Image Processing with ImageJ. Biophotonics Int 11: 36–42.
  31. 31. Zhou Y, Liang Y, Lynch KH, Dennis JJ, Wishart DS (2011) PHAST: a fast phage search tool. Nucleic Acids Res 39: W347–352 doi:10.1093/nar/gkr485.
  32. 32. Norton MD, Spilkia AJ, Godoy VG (2013) Antibiotic resistance acquired through a DNA damage-inducible response in Acinetobacter baumannii. J Bacteriol 195: 1335–1345 doi:10.1128/JB.02176-12.
  33. 33. Erill I, Escribano M, Campoy S, Barbé J (2003) In silico analysis reveals substantial variability in the gene contents of the gamma proteobacteria LexA-regulon. Bioinforma Oxf Engl 19: 2225–2236. doi: 10.1093/bioinformatics/btg303
  34. 34. Cirz RT, O'Neill BM, Hammond JA, Head SR, Romesberg FE (2006) Defining the Pseudomonas aeruginosa SOS response and its role in the global response to the antibiotic ciprofloxacin. J Bacteriol 188: 7101–7110 doi:10.1128/JB.00807-06.
  35. 35. Allocati N, Federici L, Masulli M, Di Ilio C (2009) Glutathione transferases in bacteria. FEBS J 276: 58–75 doi:10.1111/j.1742-4658.2008.06743.x.
  36. 36. Gohl O, Friedrich A, Hoppert M, Averhoff B (2006) The thin pili of Acinetobacter sp. strain BD413 mediate adhesion to biotic and abiotic surfaces. Appl Environ Microbiol 72: 1394–1401 doi:10.1128/AEM.72.2.1394-1401.2006.
  37. 37. Grissa I, Vergnaud G, Pourcel C (2007) CRISPRFinder: a web tool to identify clustered regularly interspaced short palindromic repeats. Nucleic Acids Res 35: W52–57 doi:10.1093/nar/gkm360.
  38. 38. Di Nocera PP, Rocco F, Giannouli M, Triassi M, Zarrilli R (2011) Genome organization of epidemic Acinetobacter baumannii strains. BMC Microbiol 11: 224 doi:10.1186/1471-2180-11-224.
  39. 39. Aravind L, Anand S, Iyer LM (2013) Novel autoproteolytic and DNA-damage sensing components in the bacterial SOS response and oxidized methylcytosine-induced eukaryotic DNA demethylation systems. Biol Direct 8: 20 doi:10.1186/1745-6150-8-20.
  40. 40. Hauck Y, Soler C, Jault P, Mérens A, Gérome P, et al. (2012) Diversity of Acinetobacter baumannii in four French military hospitals, as assessed by multiple locus variable number of tandem repeats analysis. PloS One 7: e44597 doi:10.1371/journal.pone.0044597.
  41. 41. Roberts JW, Roberts CW (1975) Proteolytic cleavage of bacteriophage lambda repressor in induction. Proc Natl Acad Sci U S A 72: 147–151. doi: 10.1073/pnas.72.1.147
  42. 42. Jeon J, Kim J, Yong D, Lee K, Chong Y (2012) Complete genome sequence of the podoviral bacteriophage YMC/09/02/B1251 ABA BP, which causes the lysis of an OXA-23-producing carbapenem-resistant Acinetobacter baumannii isolate from a septic patient. J Virol 86: 12437–12438 doi:10.1128/JVI.02132-12.
  43. 43. Brüssow H, Desiere F (2001) Comparative phage genomics and the evolution of Siphoviridae: insights from dairy phages. Mol Microbiol 39: 213–222. doi: 10.1046/j.1365-2958.2001.02228.x
  44. 44. Petrovski S, Dyson ZA, Seviour RJ, Tillett D (2012) Small but sufficient: the Rhodococcus phage RRH1 has the smallest known Siphoviridae genome at 14.2 kilobases. J Virol 86: 358–363 doi:10.1128/JVI.05460-11.
  45. 45. Deghorain M, Van Melderen L (2012) The Staphylococci phages family: an overview. Viruses 4: 3316–3335. doi: 10.3390/v4123316
  46. 46. Longkumer T, Parthasarathy S, Vemuri SG, Siddavattam D (2014) OxyR-dependent expression of a novel glutathione S-transferase (Abgst01) gene in Acinetobacter baumannii DS002 and its role in biotransformation of organophosphate insecticides. Microbiol Read Engl 160: 102–112 doi:10.1099/mic.0.070664-0.
  47. 47. Bhaya D, Davison M, Barrangou R (2011) CRISPR-Cas systems in bacteria and archaea: versatile small RNAs for adaptive defense and regulation. Annu Rev Genet 45: 273–297 doi:10.1146/annurev-genet-110410-132430.
  48. 48. Klaiman D, Steinfels-Kohn E, Kaufmann G (2013) A DNA break inducer activates the anticodon nuclease RloC and the adaptive immunity in Acinetobacter baylyi ADP1. Nucleic Acids Res. doi:10.1093/nar/gkt851.
  49. 49. Babu M, Beloglazova N, Flick R, Graham C, Skarina T, et al. (2011) A dual function of the CRISPR-Cas system in bacterial antivirus immunity and DNA repair. Mol Microbiol 79: 484–502. doi: 10.1111/j.1365-2958.2010.07465.x