Acinetobacter baumannii Gastrointestinal Colonization Is Facilitated by Secretory IgA Which Is Reductively Dissociated by Bacterial Thioredoxin A

ABSTRACT Multidrug-resistant Acinetobacter baumannii is among the most common causes of infectious complications associated with combat-related trauma in military personnel serving overseas. However, little is currently known about its pathogenesis. While the gastrointestinal (GI) tract has been found to be a major reservoir for A. baumannii, as well as to potentially contribute to development of multidrug resistance, no studies have addressed the mechanisms involved in gut colonization. In this study, we address this critical gap in knowledge by first assessing the interaction between secretory IgA (SIgA), the principal humoral immune defense on mucosal surfaces, and the A. baumannii clinical isolate Ci79. Surprisingly, SIgA appeared to enhance A. baumannii GI tract colonization, in a process mediated by bacterial thioredoxin A (TrxA), as evidenced by reduction of bacterial attachment in the presence of TrxA inhibitors. Additionally, a trxA targeted deletion mutant (ΔtrxA) showed reduced bacterial burdens within the GI tract 24 h after oral challenge by in vivo live imaging, along with loss of thiol-reductase activity. Surprisingly, not only was GI tract colonization greatly reduced but the associated 50% lethal dose (LD50) of the ΔtrxA mutant was increased nearly 100-fold in an intraperitoneal sepsis model. These data suggest that TrxA not only mediates A. baumannii GI tract colonization but also may contribute to pathogenesis in A. baumannii sepsis following escape from the GI tract under conditions when the intestinal barrier is compromised, as occurs with cases of severe shock and trauma.

IMPORTANCE Acinetobacter baumannii is an emerging bacterial pathogen recently classified as a serious threat to U.S. and global health by both the Centers for Disease Control and Prevention and the World Health Organization. It also is one of the leading causes of combat-related infections associated with injured military personnel serving overseas. Little is known regarding mechanisms of gastrointestinal tract colonization despite this site being shown to serve as a reservoir for multidrugresistant (MDR) A. baumannii isolates. Here, we establish that secretory IgA, the major immunoglobulin of mucosal surfaces, promotes A. baumannii GI tract colonization via bacterial thioredoxin A as evidenced through significant reduction in colonization in IgA-deficient animals. Additionally, bacterial colonization and mortality were significantly reduced in animals challenged with a thioredoxin A-deficient A. baumannii mutant. Combined, these data suggest that thioredoxin A is a novel virulence factor, for which antithioredoxin therapies could be developed, for this important multidrug-resistant pathogen.

KEYWORDS Acinetobacter, bacterial virulence, gastrointestinal infection, SIgA, thioredoxin
A cinetobacter baumannii is an opportunistic pathogen that has become a significant concern for clinicians due to its high prevalence of multidrug resistance (1)(2)(3)(4)(5). A. baumannii isolates are intrinsically resistant to many antibiotics due to a reduced repertoire of membrane porins, naturally expressed beta-lactamases, and various efflux pumps (6). Acquired antibiotic resistance, through horizontal gene transfer, has only exacerbated the problem (1,2,4,(7)(8)(9)(10)(11)(12). Currently, the highly toxic antibiotic colistin, which targets bacterial membranes, is often used as a treatment of last resort (5,13,14). Furthermore, the gastrointestinal (GI) tract colonization has been linked to development of antibiotic resistance in A. baumannii (15), presumably due to close proximity of the organism to the enormous numbers and varieties of bacteria therein, allowing direct transfer of antibiotic resistance plasmids through bacterial conjugation (16,17). However, while the GI tract is a common site of colonization (4,15,(18)(19)(20)(21), there is currently a gap in our understanding of the mechanisms facilitating this colonization.
Secretory IgA (SIgA) contributes to GI tract homeostasis and protection against pathogens along mucosal surfaces (22)(23)(24). SIgA is transported across mucosal epithelia by polymeric immunoglobulin receptor (pIgR) and is comprised of a secretory component (SC) covalently bound to the Fc regions of dimeric IgA (25,26). Both SC and SIgA interact with various antigens in a nonspecific manner due to protein glycosylation associated with both molecules (22). Additionally, SC is thought to protect the dimeric IgA molecule from both host and bacterial proteases (22); however, bacteria have developed IgA-specific proteases (27)(28)(29)(30) as well as thiol-specific reductases produced in response to this immunoglobulin (31). Given the prevalence of A. baumannii as a mucosal pathogen, it is highly likely that it would possess similar mechanisms, although none have been described.
In this study, we set out to specifically address this critical gap in our understanding of A. baumannii pathogenesis. We found not only that SIgA contributes to A. baumannii colonization in a murine oral GI challenge model but that the organism reduces the disulfide bonds of SIgA, causing separation of SC from dimeric IgA in a process mediated by secreted bacterial thioredoxin A (TrxA).

RESULTS
Contribution of SIgA to A. baumannii GI tract colonization. Since SIgA is the primary immunoglobulin associated with mucosal surfaces, including the GI tract (22), we set out to determine if it was protective against A. baumannii GI tract colonization and infection. We first orally challenged wild-type (WT) C57BL/6 and IgA Ϫ/Ϫ mice with a PSVue-794-labeled A. baumannii clinical isolate, designated strain Ci79, that we previously found to be virulent in a mouse intraperitoneal sepsis model (32). Surprisingly, in vivo live imaging of challenged mice revealed significantly (P Ͻ 0.05) enhanced A. baumannii clearance 24 h postchallenge in the absence of IgA (Fig. 1A). We subsequently humanely euthanized and collected GI tracts 24 h postchallenge from IgA Ϫ/Ϫ and WT animals for ex vivo imaging (Fig. 1B). Again, while minimal signal was detected from GI tracts collected from IgA Ϫ/Ϫ mice, fluorescence was observed throughout WT mouse GI tracts. Loss of signal in IgA Ϫ/Ϫ mice could not be explained by bacterial dissemination from the GI tract, as no fluorescence was observed in surrounding tissues (i.e., liver, spleen, and kidney).
We next sought to determine how IgA might contribute to, rather than protect against, GI tract colonization. Utilizing an ex vivo intestinal attachment assay, sections of small intestine were obtained from humanely euthanized 6-to 10-day-old mice. While lack of IgA is reported to alter intestinal microbiota (23), suckling mice lack adult microbiota, thus minimizing this confounding influence (23,33). Additionally, although suckling WT mice do not yet produce IgA, they receive maternal IgA in milk (23). Following incubation of intestinal sections in suspensions of A. baumannii Ci79, we observed a nearly 80% reduction in bacterial attachment (P Ͻ 0.0001) in the absence of IgA (Fig. 2). We then repeated these experiments using sections of small intestine collected from humanely euthanized adult mice with similar results, suggesting that, in contrast to IgA, intestinal microbiota has little effect on A. baumannii attachment (Fig. 2).
While IgA Ϫ/Ϫ mice lack IgA production, they still produce functional pIgR (34,35). Furthermore, pIgR is constantly transcytosed from the basal membrane to the luminal surface of mucosal epithelial cells, where it is cleaved to produce free SC (working model; see Fig. S1, left, in the supplemental material) (25). Binding of immunoglobulin (IgA or IgM) simply speeds the process of transcytosis (Fig. S1, middle) (25). Therefore, IgA Ϫ/Ϫ mice still exhibit free SC in mucous linings. Interestingly, while IgA Ϫ/Ϫ mice appear to clear the fluorescent bacteria by 24 h postchallenge, clearance was even more rapid in pIgR Ϫ/Ϫ mice, with fluorescent signal becoming negligible within 8 h (Fig. 3). These data suggest that free SC, rather than intact SIgA, is the primary mediator of A. baumannii GI colonization (Fig. S1, right).
A. baumannii dissociated secretory component from SIgA. As many GI tractassociated bacteria produce IgA-specific proteases (27)(28)(29)(30), we sought to assess the direct interaction between SIgA and A. baumannii by incubating Ci79 with 50 g/ml of either human serum IgA or SIgA overnight in Dulbecco's Modified Eagle's Medium (DMEM). As expected, multiple bands were observed, indicating degradation of the immunoglobulin (Fig. 4A). However, Western blot analysis revealed clearly defined bands associated with ␣-heavy chain (serum IgA only), -light chain, and SC (SIgA only) with little smearing, indicating the absence of proteolysis (Fig. 4A). To confirm, we inoculated bacteria on LB agar supplemented with 2% skim milk to detect secreted protease (36). As expected, Ci79 exhibited no protease activity, as evidenced by lack of clearing around the bacteria (Fig. 4B). As SC is bound to dimeric IgA by disulfide bonds, release of SC from SIgA may be mediated via reductive processes. To test this possibility, we grew Ci79 (Fig. 4C), along with a variety of other A. baumannii-Acinetobacter calcoaceticus complex isolates (Fig. S2), in M9 minimal medium (M9MM) supplemented with 1 mM membrane-impermeant dithionitrobenzoic acid (DTNB) to detect secreted reductase activity (8,37). DTNB was reduced to 2-nitro-5-thiobenzoic acid (TNB; yellow color) in the presence of the bacteria in a fashion similar to the positive reduction control ␤-mercaptoethanol (␤ME; a strong thiol-reducing agent). These data suggest that A. baumannii may break down SIgA via reductive processes targeting the disulfide bonds within the immunoglobulin.
We previously genome sequenced and annotated a series of A. baumannii isolates, including strain Ci79 (38). Utilizing BLAST2GO analysis software, we assigned gene ontology terms to annotated genes and identified those encoding proteins and enzymes involved in reductive processes (GO:005514), specifically disulfide-reducing enzymes (GO:0004791 and/or GO:0015035). These criteria identified 9 genes with characteristics necessary for reduction of disulfide bonds within SIgA (Table 1). We then incubated A. baumannii Ci79 with or without 100 g/ml human SIgA for 1, 2, or 6 h and performed transcriptome RNA sequencing (RNA-seq) analysis. Although total gene expression was modulated at all intervals (Fig. 5A), significant modulation of disulfide-reducing proteins was observed only at 2 h. Furthermore, only trxA mRNA expression was significantly upregulated following SIgA exposure (Fig. 5B).
Generation of an A. baumannii thioredoxin A-null mutant. We next generated a trxA targeted gene deletion mutant (⌬trxA) in WT A. baumannii Ci79. A complemented strain, ⌬trxA c , was also generated from ⌬trxA, restoring TrxA protein expression. Generation of the ⌬trxA strain was confirmed by Southern blot analysis showing the predicted 3.4-kb fragment resulting from HindIII restriction digest of genomic DNA, compared to the 2.8-kb fragment observed with WT DNA (Fig. 6A and B). Restriction digest of WT and ⌬trxA genomic DNA with XbaI also resulted in predicted DNA fragments ( Fig.  6A and B), further confirming successful incorporation of the erythromycin resistance  A. baumannii GI Colonization Mediated by SIgA and TrxA ® gene (ermR) in place of trxA via homologous recombination. A similar homologous recombination strategy was employed to generate the trxA-complemented strain (⌬trxA c ) with trxA derived from A. baumannii Ci77 containing a synonymous small nucleotide polymorphism (SNP), resulting in loss of a SalI restriction site and yet remaining 100% identical to that of the Ci79 strain on an amino acid level. While putative trxA-complemented strains were selected based on restoration of DTNB reduction, after several transformation attempts a single stable complement clone was obtained. However, Southern blot analysis suggested that complementation did not result in integration into the chromosome as expected, as evident by the presence of both the 3.4-kb fragment observed in the ⌬trxA deletion mutant and an additional band approximately 5 kb in size following HindIII digest (Fig. 6B). Similarly, XbaI digest produced both the 5.1-kb fragment associated with the ⌬trxA mutant and a fragment approximately 10 kb in size. Whole-genome sequencing failed to pinpoint an integration site, suggesting that the ⌬trxA c strain may maintain trxA ectopically on a plasmid. Despite this, the presence of the Ci77 trxA gene in the ⌬trxA c strain was confirmed through SalI digest of PCR-amplified trxA (Fig. 6C). PCR targeting trxA in the ⌬trxA mutant resulted in a 971-bp product corresponding to the erythromycin resistance cassette (Fig. 6C). SalI digestion of the trxA PCR product from WT Ci79 produced fragments of 178 bp and 199 bp in length (seen in Fig. 6C as a single band of approximately 200 bp). However, cleavage was not observed with the ⌬trxA c strain, resulting in a 370-bp band. The 971-bp PCR product was observed in the comple- mented strain as well. Most importantly, the lack of TrxA protein expression in the ⌬trxA mutant and subsequent restoration in the ⌬trxA c strain were confirmed by Western blot analysis (Fig. 6D). Deletion of trxA had unexpected effects on A. baumannii growth in LB medium. We observed an increase in the lag phase of growth associated with ⌬trxA compared to WT and ⌬trxA c isolates (Fig. 7A). However, log-phase growth appeared largely unaffected. Additionally, ⌬trxA colonies were smaller than WT (Fig. 7B). Similar observations have been reported in yeast following disruption of thioredoxin genes and were attributed to an inability to reduce glutathiolated proteins during transition from lag-to log-phase growth (39). While the presence of secreted reductase enzymes has been reported in the extracellular proteome (40), the inability to reduce membrane-impermeant DTNB indicated that secreted redox activity was greatly reduced following deletion of trxA (Fig. 7C). These data suggest not only that A. baumannii secretes TrxA but that this protein is the primary thiol-reducing protein responsible for reduction of disulfide bonds in the extracellular environment. Although lacking a typical amino-terminal signal sequence, thioredoxins have been shown to be secreted by Helicobacter pylori (31) as well as various normal and tumor mammalian cells via a nonclassical pathway (41,42).
Effect of thioredoxin A on GI tract colonization. An ex vivo intestinal bacterial attachment assay and in vivo oral bacterial challenge experiments were used to elucidate the role of TrxA in A. baumannii colonization in the GI tract. Two thioredoxin inhibitors, DTNB (a competitive inhibitor of disulfide reductase enzymes [43]) and 1-methylpropyl-2-mercaptoimidazolyl disulfide (PX-12; an irreversible thioredoxin inhibitor [44]), were shown to block the WT Ci79 mediated-release of SC from SIgA A. baumannii GI Colonization Mediated by SIgA and TrxA ® (Fig. 8A). Additionally, Ci79 attachment to excised intestinal sections was significantly reduced in the presence of these two TrxA inhibitors (Fig. 8B). For the in vivo study, mice were challenged orally with PSVue-794-labeled bacteria to monitor bacterial clearance from the GI tract via in vivo live imaging. A significant decrease in bacterial colonization of the GI tract was evidenced by both diminished fluorescent intensity and reduced fecal shedding of the ⌬trxA strain by 24 h relative to mice challenged with either the WT or ⌬trxA c strain (Fig. 9). These results demonstrated the important role of TrxA in A. baumannii GI colonization. Loss of TrxA also resulted in decreased mortality following systemic A. baumannii infection. As shown in Fig. S3, while all mice succumbed to infection following a 2-LD 50 (50% lethal dose; 1 ϫ 10 6 CFU) intraperitoneal challenge with both WT and ⌬trxA c strains, 100% of mice challenged with equivalent  doses of the ⌬trxA strain survived. In fact, mortality was observed only after administration of Ͼ5 ϫ 10 7 CFU/mouse, a nearly 100-fold increase in the associated LD 50 .

DISCUSSION
Data presented here suggest that SIgA contributes to A. baumannii colonization of the GI tract. We observed a significant reduction in bacterial colonization ( Fig. 1 and 3) and a nearly 80% decrease in bacterial attachment in the absence of SIgA (Fig. 2). While we observed breakdown of SIgA by A. baumannii (Fig. 4A), initially suggesting possible proteolysis similar to that described with Clostridium spp. (27), the absence of secreted protease activity (Fig. 4B) in combination with observed secretion of thiol-reducing activity (Fig. 4C) suggested that breakdown was mediated instead by reduction of disulfide bonds within SIgA by a thiol-reducing enzyme. Indeed, prior studies have demonstrated the importance of disulfide bonds to SIgA structural stability (45). Supporting this hypothesis, Helicobacter pylori has been reported to secrete TrxA in response to host immunoglobulin for protection from the host immune system (31). Similarly, our RNA-seq analysis specifically demonstrated upregulation of trxA gene expression in A. baumannii following exposure to SIgA (Fig. 5). While thiol reductase activity was detected in the absence of SIgA, upregulation of trxA gene expression suggests that A. baumannii mounts a similar response to H. pylori. Additionally, inhibition of TrxA resulted in ablation of SIgA reduction and a significant reduction in A. baumannii attachment to the intestinal mucosal surface (Fig. 8).
Thioredoxin has been implied to have many cellular functions beyond protection from oxidative stressors, including cell cycle regulation and protein folding/transportation (46). Utilizing a tandem affinity purification (TAP)-tagged Escherichia coli TrxA, Kumar et al. (46) identified 80 TrxA-associated proteins involved in distinct cellular processes that include transcription regulation, cell division, energy transduction, and A. baumannii GI Colonization Mediated by SIgA and TrxA several biosynthetic pathways. Similarly, in our study, the loss of TrxA expression resulted in a multitude of effects on the bacterium, including decreased colony size and an extended lag phase (Fig. 7B). Additionally, lack of TrxA protein expression ablated secreted disulfide reductase activity (Fig. 7C). Mice challenged with A. baumannii Ci79 ⌬trxA also survived challenge doses nearly 100-fold higher than the LD 50 associated with the WT Ci79 strain (see Fig. S3 in the supplemental material) and exhibited significantly increased mutant clearance and decreased fecal shedding (Fig. 9). As complementation restored all WT phenotypes, our data indicate two potential functions of TrxA secreted by A. baumannii with respect to colonization of mucosal surfaces. As shown in our working model of A. baumannii GI tract colonization (Fig. S1), TrxA secreted by A. baumannii reduces disulfide bonds within the immediate microenvironment, including those within SIgA, neutralizing protective barrier functions associated with the immunoglobulin. Second, reduced SIgA provides A. baumannii a means of colonizing mucous layers though increasing concentrations of available free SC within the mucous lining. However, the mechanism(s) of SC-mediated A. baumannii GI colonization remains elusive. While reduction of SIgA by TrxA clearly contributes to this process, immunoglobulin may not be the only target of the protein as thioredoxin also has been implicated in bacterial mucosal colonization by other mechanisms. For example, Helicobacter pylori uses secreted thioredoxin to reduce mucin molecules to their monomeric form, decreasing mucus viscosity and allowing the organism to colonize as well as facilitate migration to the epithelial surface (31). Deletion of either the TrxA or TrxC gene in H. pylori impairs the organism's ability to colonize the stomach following oral bacterial inoculation (47).
While TrxA has now been identified as a potential mechanism by which A. baumannii colonizes the GI tract, the viability of this target for chemotherapy may be questioned due to apparent widespread distribution of TrxA gene homology among commensal bacterial species. However, the use of A. baumannii-specific TrxA as a subunit vaccine may provide at least partial protection while preventing speciesnonspecific inhibition. Additionally, attenuation resulting from loss of TrxA expression by A. baumannii may allow the ⌬trxA mutant to serve as a useful live-attenuated vaccine strain. Current studies by our laboratory are focused on evaluating this possibility, as immunization with the ⌬trxA mutant provided marked protection against systemic infection with the WT strain (48). Bacteria were grown at 37°C to desired concentrations based on optical density of broth cultures measured at 600 nm (OD 600 ).

MATERIALS AND METHODS
Mice. Eight-to 10-week-old pathogen-free C57BL/6 mice were purchased from Charles River Laboratories (Frederick, MD). Homozygous pIgR knockout mice (pIgR Ϫ/Ϫ ; B6.129P2-Pigr tm1Fejo/Mmmh ) were obtained from the Mutant Mouse Resource and Research Center (University of Missouri, Columbia, MO). Male pIgR Ϫ/Ϫ mice were bred with female WT C57BL/6 mice, producing litters consisting of 100% heterozygous offspring (pIgR ϩ/Ϫ ), ensuring that resulting offspring received SIgA in breast milk (see Fig.  S4 in the supplemental material). Female pIgR ϩ/Ϫ (SIgA-producing) offspring were backcrossed with homozygous pIgR Ϫ/Ϫ mice, producing litters of approximately 50% pIgR ϩ/Ϫ and 50% pIgR Ϫ/Ϫ mice. Genotyping was performed by PCR utilizing the following primer combinations: pIgRKO_FW/pIgRKO_RV ( In vivo live imaging. Stocks of A. baumannii clinical isolate Ci79 (200-l aliquots) were prepared from log-phase growth and frozen at Ϫ80°C in LB broth supplemented with 10% glycerol. Titers of frozen stocks were determined on three separate days and averaged. Prior to challenge, bacteria were thawed and pelleted at 3,000 ϫ g. Bacterial pellets were suspended in PSVue-794 (Molecular Targeting Tech Ex vivo intestinal attachment assay. Bacterial attachment to intestinal mucosa was performed as described by Guentzel and Berry (33), with some modifications. Briefly, bacteria were washed once in sterile phosphate-buffered saline (PBS) and diluted to a concentration of 1 ϫ 10 8 CFU/ml. The small intestine was dissected from humanely euthanized 6-to 10-day-old infant mice, unless otherwise stated, and cut into sections measuring approximately 1 cm in length. Using fine scissors, the intestinal lumen was exposed and sections were placed in the bacterial suspension for 30 min with regular agitation. In some experiments, the bacterial suspensions were prepared with either 1 mM DTNB (Fisher Scientific, Pittsburgh, PA) or 15 g/ml PX-12 (Tocris, Minneapolis, MN). Sections were placed in 500 volumes of sterile PBS and inverted 7 times. This was repeated once before sections were placed in 200 volumes of sterile PBS and allowed to soak for 5 min. Each section was transferred to 10 ml sterile PBS and homogenized. Dilution plating was performed on LB agar containing 50 g/ml chloramphenicol and 10 g/ml cycloheximide to inhibit growth of resident microbiota.
Analysis of secreted protease activity. Bacterial suspensions of either A. baumannii or P. aeruginosa were prepared from log-phase growth (OD 600 of 0.7), and 10 l was spotted onto LB agar supplemented with 2% skim milk.
Disulfide reductase assay. Indicated bacteria were grown at 37°C to an OD 600 of 0.7. Each strain was pelleted at 3,000 ϫ g and washed three times in M9MM (48 mM Na 2 HPO 4 , 167 mM KH 2 PO 4 , 8.5 mM NaCl, 19 mM NH 4 Cl, 2 mM MgSO 4 , 100 M CaCl 2 , 0.4% glucose). Bacterial pellets were suspended in either sterile M9MM alone or M9MM supplemented with 1 mM DTNB and grown for 24 h at 37°C, at which point supernatants were collected. Reductase activity was observed through yellow color development detected at 450 nm.
SIgA reduction assay. Bacterial broth cultures were pelleted at 3,000 ϫ g for 5 min, washed three times in equal volumes of sterile DMEM (Life Technologies, Inc., Grand Island, NY) to remove trace amounts of reducing substances, and suspended in an equal volume of sterile DMEM. In parallel, lyophilized SIgA from human colostrum or serum IgA from human plasma (Athens Research, Athens, GA) was diluted to a concentration of 100 g/ml in sterile DMEM. Once prepared, 250 l of the bacterial suspension was combined with 250 l SIgA or serum IgA and incubated for 24 h. For experiments examining inhibition of SIgA reduction, individual preparations of SIgA were prepared containing either 2 mM DTNB or 30 g/ml PX-12, resulting in final concentrations of 1 mM and 15 g/ml, respectively, and incubation was reduced from 24 hours to 2 hours. Supernatants were collected and mixed 1:4 in native sample buffer (40% glycerol, 250 mM Tris-HCl, 0.015% bromphenol blue) before being subjected to nonreducing SDS-PAGE on a 12% polyacrylamide gel. Following electrophoresis, proteins were transferred to a polyvinylidene difluoride (PVDF) membrane (Bio-Rad, Hercules, CA) for Western blot analysis using goat anti-human SC primary antibody (1:1,000 dilution; Sigma-Aldrich, St. Louis, MO) followed by rabbit anti-goat horseradish peroxidase (HRP)-conjugated secondary antibody (1:3,000; KPL, Gaithersburg, MD).
RNA-seq assay. A. baumannii clinical isolate Ci79 was grown to log phase, and two 1-ml aliquots were obtained. Aliquots were washed three times in sterile DMEM, and the resulting pellets were suspended in either 1 ml sterile DMEM alone or 1 ml sterile DMEM with 100 g/ml SIgA before incubation at 37°C. Bacterial cells were pelleted and snap-frozen at Ϫ80°C prior to mRNA extraction at indicated intervals using the Ambion PureLink RNA minikit (Life Technologies, Waltham, MA). rRNA depletion was performed utilizing the Ambion MICROBExpress bacterial mRNA enrichment kit (Life  Table 2) and subsequently cloned into pMAL-C2X, resulting in incorporation of a 3= maltose binding protein (MBP) tag. Rosetta-gami E. coli cells were transformed with the pMAL-trxA construct for generation of an MBP-TrxA fusion protein (rTrxA). rTrxA expression in the transformed E. coli was induced in the presence of glucose (final concentration, 2 g/liter) and IPTG (isopropyl-␤-D-thiogalactopyranoside; 1 mM) overnight in a 16°C shaking incubator. Isolation of rTrxA by amylose affinity chromatography was conducted according to the manufacturer's recommendations (New England BioLabs, Ipswich, MA).
To generate anti-TrxA antibody, rTrxA protein was diluted to a concentration of 0.2 mg/ml and mixed 1:1 with TiterMax Gold adjuvant (Sigma-Aldrich, St. Louis, MO). BALB/c mice were vaccinated with two 50-l subcutaneous (s.c.) injections at the base of the tail, resulting in a total dose of 10 g rTrxA per mouse. A booster was administered 14 days after initial vaccination consisting of a single 50-l s.c. injection of recombinant protein (5 g). At 28 days after initial vaccination, mice were anesthetized and sera were collected.
Generation of a targeted thioredoxin deletion mutant. A. baumannii Ci79 DNA was extracted from a 3-ml volume of bacterial suspension using the Wizard Genomic DNA purification kit (Promega, Madison, WI). Using primer combinations Up_Fw/Up_Rv and Dn_Fw/Dn_Rv (Table 2) in combination with TaKaRa Ex Taq polymerase (Clontech, Mountain View, CA), a 973-bp region located directly upstream of trxA and a 988-bp region located directly downstream of trxA were amplified by PCR (Fig.  S5, step 1). PCR products were processed using the Wizard SV gel cleanup kit (Promega, Madison, WI), and short overlap extension PCR (SOE-PCR) was performed under conditions described by Liu et al. (50), using primers Up_Fw and Dn_Rv and TaKaRa Ex Taq polymerase. The resulting PCR product (Fig. S5, step 2), containing a BamHI restriction site in place of the trxA gene, was subsequently ligated into pGEM-T Easy (Promega, Madison, WI) using T4 DNA ligase (Promega, Madison, WI) to make plasmid pBPA001. Transformants were identified through antibiotic selection on LB agar supplemented with 100 g/ml ampicillin (Fig. S5, step 3). The erythromycin resistance cassette and promoter (Fterm r ) were extracted from pKEK887 by double digest with BglII and BamHI (50). In parallel, pBAP001 was subject to single restriction digest by BamHI. A 3:1 mix of Fterm r to pBPA001 was ligated to form pBPA002 and transformed into JM109 E. coli cells. Transformants were selected by antibiotic selection on LB agar supplemented with 100 g/ml ampicillin and 100 g/ml erythromycin (Fig. S5, step 4). A. baumannii Ci79 was grown at 37°C to an OD 600 of 0.7 from a 1:100 dilution of an overnight broth culture. Once obtained, 5 ml was pelleted and washed in ice-cold 200 mM RbCl 3 times on ice for 5 min. Bacteria were suspended in 50 l 200 mM ice-cold RbCl, and 20 l (1 g/l) pBPA002 was added with 50 l cryotransformation buffer (10 mM HEPES, 0.1 M CaCl 2 , 10 mM RbCl, 10% glycerol at pH 6.5) (50). The mixture was incubated on ice for 30 min and snap-frozen for 5 min in a dry-ice-ethanol bath. Bacteria were thawed at room temperature, suspended in 900 l sterile LB, and incubated at 37°C for 90 min. Bacteria were pelleted and suspended in 100 l sterile LB before plating on LB agar containing erythromycin. Isolated colonies were subcultured twice on selective agar and then three times on nonselective agar to ensure stable integration.
Complementation of the deletion mutant. The trxA gene was amplified from A. baumannii Ci77 by PCR utilizing the primer combination Up_Fw/Dn_Rv and TaKaRa Ex Taq polymerase to produce a 2,326-bp product from extracted genomic DNA (Fig. S5, step 5). The PCR product was cloned into pGEM-T to generate pBPA003 (Fig. S5, step 6). An overnight culture of A. baumannii Ci79 ⌬trxA was diluted 1:20 in 250 ml fresh LB and grown at 37°C to an OD 600 of 0.8. Bacteria were placed on ice for 10 min and washed three times in sterile ice-cold 10% (vol/vol) glycerol. The pellet was suspended in 500 l 10% glycerol, and 100-l aliquots were frozen at Ϫ80°C. Then, 20 l pBPA003 (1 g/ml) was added to frozen aliquots thawed on ice. Thawed bacteria and plasmid were transferred to prechilled electroporation cuvettes (1-mm gap; Bio-Rad, Hercules, CA) and electroporated using a Bio-Rad Gene Pulser (18 kV, 200 ⍀, 25 F) (Bio-Rad, Hercules, CA). Transformed bacteria were suspended in sterile LB broth, incubated at 37°C for 1 h, and plated as before. Analysis of disulfide reductase activity revealed enhanced growth of the parental WT A. baumannii Ci79 strain in the presence of reduced DTNB, suggesting that it may utilize TNB as a carbon source (Fig. 7D). In contrast, growth of the ⌬trxA mutant remained static due to its inability to reduce DTNB. As such, complemented mutants were screened on M9MM agar with 1 mM DTNB. Isolated colonies were subcultured twice on M9MM plus DTNB and then three times on nonselective LB agar to ensure stable expression.
Confirmation of genetic manipulation. Growth was assessed through measurement of the optical density at 600 nm for A. baumannii Ci79, ΔtrxA, and ΔtrxA c strains grown in LB broth lacking antibiotics over 24 h. Differences in colony size for each strain were subsequently assessed following subculture on LB agar lacking antibiotics. Genetic manipulation was confirmed by Southern blot analysis. Genomic DNAs from Ci79, ΔtrxA, and ΔtrxA c strains were isolated using the GeneJET genomic DNA purification kit (Thermo Scientific, Rockford, IL), digested with XbaI or HindIII (New England BioLabs), and run on an 0.8% agarose gel at 16 V until completion. DNA was transferred to a nylon Biodyne B membrane (Thermo Scientific, Rockford, IL) and UV-cross-linked for 5 min. A DNA probe amplified from WT A. baumannii Ci79 genomic DNA using the DN_Fw/DN_Rv primer set was biotin labeled using the North2South biotin random prime labeling kit (Thermo Scientific, Rockport, IL). Hybridization and detection were performed using the North2South chemiluminescent hybridization and detection kit (Thermo Scientific, Rockport, IL) per manufacturer's instructions. Additionally, the trxA gene was amplified from genomic extracts obtained from A. baumannii Ci79, ⌬trxA, and ⌬trxA c strains with the TrxA-Fw/TrxA-Rv primer set (Table  2) and digested with SalI (Fig. 6D) to distinguish trxA gene origin (i.e., Ci77 or Ci79). Loss and subsequent restoration of TrxA protein expression were assessed by Western blot analysis using the mouse anti-rTrxA antibody (1:1,000) described above.
Fecal shedding assay. Fecal pellets were collected at 24-h intervals from WT C57BL/6 mice challenged by oral gavage with the A. baumannii Ci79, ⌬trxA, or ⌬trxA c strain at a concentration of 1 ϫ 10 8 CFU per mouse. Pellets were homogenized in sterile PBS to a concentration of 0.1 mg/ml, and 100 l was spread onto CHROMagar Acinetobacter medium (DRG International, Inc., Springfield, NJ). Fecal pellets were collected prior to challenge to ensure that the mice were not precolonized with A. baumannii.
Statistics. For most experiments, statistical differences were assessed by one-way analysis of variance (ANOVA) with Holm-Sidak correction or the Welch t test. Differences in gene expression observed by RNA-seq were assessed by Student's t test. For fecal shedding, statistical differences were assessed by both Kruskal-Wallis test with Dunn's correction and Fisher's exact test. All statistics and graphs were compiled using GraphPad Prism statistical software.

ACKNOWLEDGMENTS
This work received research support and funding provided by the Army Research Office of the Department of Defense (W911NF-11-1-0136), the Alvarez Graduate Research Education Excellence Fund and by an appointment to the Postgraduate Research Participation Program at the U.S. Army Institute of Surgical Research administered by the Oak Ridge Institute for Science and Education through an interagency agreement between U.S. Department of Energy and USAMRMC. Computational support from the UTSA Computational System Biology Core was funded by the National Institute on Minority Health and Health Disparities (G12MD007591) from the National Institutes of Health. This study also was supported with the Jane and Roland Blumberg Professorship in Biology for B. P. Arulanandam.
The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the views of the Department of the Army or the Department of Defense.