The Phenylacetic Acid Catabolic Pathway Regulates Antibiotic and Oxidative Stress Responses in Acinetobacter

ABSTRACT The opportunistic pathogen Acinetobacter baumannii is responsible for a wide range of infections that are becoming increasingly difficult to treat due to extremely high rates of multidrug resistance. Acinetobacter’s pathogenic potential is thought to rely on a “persist and resist” strategy that facilitates its remarkable ability to survive under a variety of harsh conditions. The paa operon is involved in the catabolism of phenylacetic acid (PAA), an intermediate in phenylalanine degradation, and is the most differentially regulated pathway under many environmental conditions. We found that, under subinhibitory concentrations of antibiotics, A. baumannii upregulates expression of the paa operon while simultaneously repressing chaperone-usher Csu pilus expression and biofilm formation. These phenotypes are reverted either by exogenous addition of PAA and its nonmetabolizable derivative 4-fluoro-PAA or by a mutation that blocks PAA degradation. Interference with PAA degradation increases susceptibility to antibiotics and hydrogen peroxide treatment. Transcriptomic and proteomic analyses identified a subset of genes and proteins whose expression is affected by addition of PAA or disruption of the paa pathway. Finally, we demonstrated that blocking PAA catabolism results in attenuated virulence in a murine catheter-associated urinary tract infection (CAUTI) model. We conclude that the paa operon is part of a regulatory network that responds to antibiotic and oxidative stress and is important for virulence. PAA has known regulatory functions in plants, and our experiments suggest that PAA is a cross-kingdom signaling molecule. Interference with this pathway may lead, in the future, to novel therapeutic strategies against A. baumannii infections.

phenylalanine catabolism, specifically the steps downstream of the intermediate PAA (7). Two enzymes involved in the conversion of phenylalanine to PAA are the monoamine oxidase encoded by mao and the phenylacetaldehyde dehydrogenase encoded by feaB (Fig. 1A) (29). Neither mao nor feaB was significantly upregulated during TMP/SMX treatment according to RNA sequencing (RNA-Seq), suggesting that this response is not simply an increase in phenylalanine metabolism (7). To confirm our previous transcriptomic analyses, we performed reverse transcription-quantitative PCR (RT-qPCR) and determined that paaA and paaB levels are increased approximately 7-fold at 2 h post-TMP/SMX treatment compared to Ab17978 grown in LB (Fig. 1B). In contrast, levels of expression of mao and feaB were not significantly altered under these conditions. The observed increase in paa operon expression was coupled with an ;15-fold decrease in expression levels of csuA/B, which encodes the CsuA/B major pilin subunit of Csu pili (Fig. 1B). This is consistent with our previous work demonstrating repression of Csu pilus expression under TMP/SMX treatment (7). These results prompted us to hypothesize that environmental signals, such as antibiotic treatment, cause an increase in the expression of the paa operon, impacting PAA levels in the cell and leading to downstream responses, such as the repression of Csu pili (Fig. 1C).
PAA levels regulate biofilm formation and Csu expression. We confirmed, by Western blotting using an antibody for the CsuA/B major pilin, that TMP/SMX treatment almost completely abolished Csu expression in Ab17978, consistent with previously published results ( Fig. 2A) (7). We found that the TMP/SMX-mediated effect on Csu expression can be overcome by the addition of exogenous PAA in a concentration-dependent manner, reaching expression levels comparable to that of the LB control at 0.5 mM PAA ( Fig. 2A). Similar results were obtained in Acinetobacter nosocomialis M2, a strain of a medically relevant species, showing that this phenotype is not specific to Ab17978 (Fig. 2B). Moreover, addition of increasing concentrations of fluoro-phenylacetic acid (4F-PAA), a 4-fluorinated-derivative of PAA (Fig. 2C), also restored CsuA/B  2D). Both PAA and 4F-PAA restore Csu expression to LB-like levels at 0.5 mM (see Fig. S1A in the supplemental material), and growth of Ab17978 is not inhibited under these conditions, suggesting that these molecules are not toxic at these concentrations (Fig. S1B). Ab17978 can grow in M9 minimal medium supplemented with PAA, but not 4F-PAA, as the only carbon source, indicating that 4F-PAA is not metabolized into TCA cycle intermediates (Fig. S1C). These results indicate that restoration of Csu levels by PAA was not due to PAA being used as an additional carbon source. Csu pili are one of the main determinants of biofilm formation in Ab17978 (8,9). We found that, in agreement with Csu Western blot data, exogenously added PAA or 4F-PAA induced biofilm formation in Ab17978, as measured by crystal violet binding (Fig. 2E). Biofilm formation in a DcsuD mutant was completely abolished in the presence and absence of exogenous PAA, confirming that increased biofilm formation in response to PAA is dependent on Csu expression (see Fig. S2 in the supplemental material). Together, these results implicate PAA in a putative signaling pathway induced by antibiotic stress and resulting in physiological changes (e.g., Csu expression). PAA induces significant proteomic changes. Our results show that Ab17978 modulates expression of Csu pili in a PAA-dependent fashion. To identify other proteins that are differentially regulated in response to PAA, we performed quantitative differential proteomics. Wild-type (WT) Ab17978 was grown in LB, LB1TMP/SMX, LB1PAA, or LB1TMP/ SMX1PAA for 4 h before whole cells were harvested and analyzed via tandem mass spectrometry (MS/MS). In agreement with our previous results, proteomic analysis showed that levels of expression of proteins involved in the biogenesis of Csu pili were significantly Stress Responses in Acinetobacter mBio downregulated in TMP/SMX-treated cells, while levels of expression of proteins involved in PAA degradation were significantly upregulated. Addition of PAA overcame the effect of TMP/SMX on Csu proteins (Fig. 3). Several other proteins followed a similar pattern of expression, where addition of PAA reverted or overcame the effects of TMP/SMX treatment (see Table S1 in the supplemental material). The most prominent of these are 4 proteins encoded together in an operon that are strongly induced by PAA (ACX60_RS12735 to ACX60_RS12750). These proteins have putative hydrolase and oxidoreductase activities; however, their functions remain uncharacterized. Additional proteins include those with putative efflux pump activities (ACX60_RS16500 and ACX60_RS13840), a putative GTPbinding protein (ACX60_RS13705), and a TetR family regulator (ACX60_RS11270). Although further work is necessary to characterize the roles of these proteins in the PAA-induced response of A. baumannii, our results suggest that Csu downregulation is one of several responses to folate stress that can be reversed by PAA.
Mutagenesis of paaB alters the expression of multiple genes in the presence of antibiotics. Our experiments show that exogenous addition of PAA to A. baumannii cultures increases Csu expression, reversing the repression of Csu caused by TMP/SMX. To enable our investigation into the role of the PAA pathway in vivo, we sought to manipulate PAA levels by generating a paaB mutant in Ab17978. paaB is part of the multicomponent monooxygenase complex that is involved in one of the first steps of PAA catabolism (13,21). Previous work in Burkholderia cenocepacia has shown that paaB is essential for PAA catabolism, and paaB mutants accumulate PAA (23). We confirmed that the Ab17978 DpaaB mutant grows similarly to the WT in LB (see Fig. S3A in the supplemental material) but is unable to grow on PAA as the sole carbon source, and in trans complementation of paaB (paaB1) restored growth on PAA (Fig. S3B). As predicted, the accumulation of PAA in the DpaaB strain mimicked the effect of exogenous addition of PAA, leading to increased biofilm formation (Fig. S3C). WT, DpaaB, and paaB1 strains grown in LB exhibited comparable levels of Csu, as determined by Western blotting (Fig. 4A). However, levels of Csu in the DpaaB strain were considerably higher than in the WT strain when cells were grown in TMP/SMX (Fig. 4A). The paaB1 complemented strain was able to partially repress Csu expression, as the normalized Csu intensity of this strain was not significantly different from that of the WT (Fig. 4B). We next investigated how overexpression of the paa operon affects Csu expression. Thus, we generated a DpaaX mutant in Ab17978. PaaX is known to be the main regulator of the paa operon in Escherichia coli, where it was characterized to be a repressor (30). Csu expression in the DpaaX mutant was approximately 50% of that in the WT strain, further supporting the hypothesis that reduced PAA levels are partially responsible for repressed Csu expression in Ab17978 ( Fig. 4C and D). Together, these data further support our hypothesis that PAA catabolism is important for the response of Ab17978 to folate stress.
Proteomic changes may not necessarily reflect changes in gene expression, so we performed transcriptomic experiments on whole cells of the Ab17978 WT and DpaaB mutant strains grown in LB or LB1TMP/SMX (see Table S2 in the supplemental material). We found that, in the absence of antibiotic stress, 19 genes were significantly upregulated in the DpaaB strain relative to the WT. In agreement with our proteomic data, these included the 13 genes present in the paa operon (Table S2). Induction of the paa operon is consistent with a metabolic response to the accumulation of PAA, which occurs in DpaaB strains (21,23). Additionally, adeIJK genes, which encode an RND family efflux pump, were also increased in the paaB mutant. The AdeIJK efflux pump of A. baumannii is thought to be a broad-spectrum pump with a preference for amphiphilic compounds (31). Previously, both csu and paa genes were shown to be downregulated in a DadeIJK mutant of strain AB5075 (32). This suggests a regulatory connection between these pathways that is conserved between A. baumannii strains. Future work is needed to understand the exact mechanisms behind the cross talk between the PAA pathway, Csu pili, and efflux pumps in Acinetobacter. Several genes related to CoA metabolism were also downregulated in the paaB mutant compared to WT (ACX60_RS09235, ACX60_RS09240, and ACX60_RS06820). This is consistent with Heat map of proteins detected by mass spectrometry to be differentially expressed in Ab17978 whole cells treated with PAA, TMP/SMX (T/S), or TMP/SMX1PAA compared to the LB control. To prepare protein samples, Ab17978 was grown for 4 h under each of the 4 conditions before being pelleted, lysed, and acetone precipitated. TMP and SMX were used at 3 mg/mL and 15 mg/mL, respectively, and PAA was used at 0.5 mM. Paa proteins are highlighted in red, and Csu is highlighted in green. Other proteins were labeled in gray if they were (i) divergently regulated between TMP/SMX and TMP/SMX1PAA conditions and (ii) had putative annotated functions. Each colored block represents the mean of 4 biological replicates. Student t tests were used to compare groups, with a cutoff P value of ,0.05. our proteomic data which shows these proteins downregulated in response to exogenous PAA addition ( Fig. 3; Table S1). Contrasting the changes seen in LB, the WT and DpaaB mutant displayed a broader repertoire of differentially expressed genes in the presence of folate stress. In agreement with our model, the csu operon was more highly expressed in the DpaaB strain. Remarkably, many other genes were differentially regulated in the presence of the antibiotics (see Table S3 in the supplemental material). Although the function of most of these genes needs to be experimentally determined, we propose that genes that are differentially expressed in the WT and paaB mutant constitute the "PAA-dependent regulon." PAA catabolism is important for growth in antibiotics. In the previous experiments, we employed TMP/SMX to induce antibiotic stress. TMP/SMX is a cocktail of drugs that synergistically inhibits synthesis of tetrahydrofolate (THF), which is essential for nucleotide synthesis, leading to folate stress (7,33). However, PAA could be part of a general response to antibiotic stress rather than being specific to folate stress. Supporting this concept, we found that Csu expression was repressed in Ab17978 grown in the presence of either kanamycin or gentamicin at a concentration that was half of the MIC. Exogenous addition of 0.5 mM PAA overruled Csu repression (Fig. 5A).

Stress Responses in Acinetobacter mBio
Similarly, the DpaaB mutant strain grown in sub-MICs of these aminoglycosides was unable to repress Csu expression (Fig. 5B). Importantly, sub-MICs of kanamycin or gentamicin caused a significant inhibition of growth in the paaB mutant compared to the WT strain, indicating that the paaB mutant strain cannot properly respond to aminoglycosideinduced stress ( Fig. 5C to E). To understand if this PAA-dependent response is limited to aminoglycosides and TMP/SMX, we performed Csu Western blots on WT and DpaaB cultures grown in sub-MICs of diverse antibiotics. We found that antibiotics targeting the bacterial cell wall (carbenicillin and imipenem) or outer membrane (colistin) did not alter Csu expression in either strain ( Fig. 5F) In contrast, antibiotics targeting DNA replication (ciprofloxacin) or protein synthesis (erythromycin and tetracycline) repressed Csu in the WT strain, but not the DpaaB mutant (Fig. 5F). Our results indicate that PAA can mediate the response of A. baumannii to stress induced by multiple classes of antibiotics, specifically those with cytoplasmic targets. It has also been shown that MumR, a transcriptional regulator involved in manganese uptake and hydrogen peroxide (H 2 O 2 ) tolerance, influences the expression of the paa operon, and an Ab17978 mutant in the paa operon displayed increased sensitivity to H 2 O 2 and oxidative stress (34). Together with our observations, this suggests that PAA may play a more general role in responding to different types of stressors. PAA degradation is important for stress tolerance and virulence of the uropathogenic strain UPAB1. Although Ab17978 is a well-studied model strain used to study Acinetobacter physiology, it lacks many of the virulence factors and antibiotic resistance cassettes that are present in recent clinical isolates of A. baumannii (35,36). To investigate antibiotic sensitivity in a modern strain, we generated a DpaaB mutant in Stress Responses in Acinetobacter mBio the recent clinical A. baumannii isolate UPAB1, a urinary isolate that is virulent in multiple mouse models (37). As seen in Ab17978, UPAB1 DpaaB does not exhibit any growth defects in LB compared to WT (see Fig. S4A in the supplemental material), but the DpaaB strain is not able to grow on M9 medium with PAA as the sole carbon source. Importantly, chromosomal insertion of paaB (paaB1) complemented this phenotype (Fig. S4B). Similar to Ab17978 DpaaB, UPAB1 DpaaB also formed higher levels of biofilm compared to the WT and paaB1 strains, demonstrating that PAA catabolism plays an important role in biofilm formation in multiple Acinetobacter strains (Fig. S4C). UPAB1 is highly resistant to many antibiotics, including kanamycin and gentamicin at high concentrations (MIC of .400 mg/mL). Therefore, we used two antibiotics to which UPAB1 is not resistant, the aminoglycoside apramycin and the glycopeptide zeocin. We found that the paaB mutant displayed no growth under both apramycin and zeocin treatments compared to the WT and paaB1 strains, while all three strains grew similarly in LB ( Fig. 6A to C). This prompted us to measure the antimicrobial susceptibility profiles of WT and paaB mutants in UPAB1 and Ab17978. Indeed, we found that the MIC of apramycin is 4 times lower in UPAB1 DpaaB than WT UPAB1, consistent with previous growth curves (Table 1). Furthermore, modest decreases in MIC were observed in paaB mutants for ciprofloxacin, erythromycin, and zeocin in both UPAB1 and Ab17978 (Table 1). This piece of data further supports our hypothesis that a PAA pathway-mediated antibiotic stress response is conserved among A. baumannii strains and that interfering with this response can decrease antibiotic resistance in a clinical isolate. As mentioned previously, an Ab17978 mutant in the paa operon displayed increased sensitivity to H 2 O 2 (38). In agreement, we found that survival of UPAB1 DpaaB was reduced by 2 logs compared to the WT strain after treatment with H 2 O 2 ( Fig. 6D and E). These results further support the notion that PAA is also involved in the response of A. baumannii to oxidative stress. Thus, we hypothesized that the inability of the paaB mutant to properly respond to this kind of stress may impact the virulence of UPAB1. While systemic models can serve as a useful benchmark for changes in virulence, they do not adequately represent the clinical manifestations of Acinetobacter disease. Because many A. baumannii Stress Responses in Acinetobacter mBio infections occur in immunocompromised hosts, who often have medical implants, we employed a previously established catheter-associated urinary tract infection (CAUTI) murine model of infection that utilizes UPAB1, a urinary isolate, as a model (37). Briefly, 6-to 8week-old female C57BL/6 mice were infected with 2 Â 10 8 CFU after a small silicone catheter was placed in each mouse urethra, and organs were plated for CFU 24 h postinfection. We found that the paaB mutant showed a significant reduction in recoverable CFU from the catheters and the bladders of infected mice (Fig. 7). The paaB1 complemented strain showed CFU burdens comparable to WT levels. This confirms that PAA degradation is a critical pathway for the virulence of a modern clinical isolate in a CAUTI model.

DISCUSSION
A. baumannii is a threatening nosocomial pathogen that is highly resistant to antibiotics and tolerant to a wide range of stressors. Although the repertoire of virulence factors of A. baumannii remains largely uncharacterized, we know that it does not have the typical weapons like T3SS or T4SS that allow bacteria to interfere with the host immune response. In fact, Acinetobacter's pathogenic potential is thought to rely on a "persist and resist" strategy that allows it to survive under a variety of harsh conditions (36). Therefore, investigating the mechanisms used by A. baumannii to adapt to stress will aid in our understanding of its pathogenesis and may lead to novel therapeutic discoveries. We and others have shown that the paa operon, which is responsible for PAA degradation, is one of the most differentially regulated pathways in response to numerous stress conditions and is strongly upregulated under antibiotic treatment (15,16,18,39,40). In this study, we show that the metabolite PAA interferes with the response of A. baumannii to various stressors. This activity is mimicked by its nonmeta- Stress Responses in Acinetobacter mBio bolizable derivative 4F-PAA, which indicates this role is independent of PAA catabolism. We determined that increased concentrations of PAA, either through exogenous treatment or through mutation of the PAA degradation pathway, induce expression of Csu pili, overcoming antibiotic-mediated repression of Csu. Our results also confirmed that PaaX, the repressor of the paa operon, is involved in the regulation of Csu. Furthermore, we established that paaB mutant strains, which are known to accumulate PAA (21,23), are sensitive to oxidative stress and exhibit reduced antibiotic tolerance and resistance. Finally, we demonstrated that a functional PAA catabolism pathway is critical for virulence of a clinical isolate in a murine CAUTI model. Taken together, this is the first evidence that PAA acts as a regulatory signal in multiple Acinetobacter strains. We propose a model in which environmental cues lead to an increase in paa operon expression, which controls cellular PAA levels and ultimately leads to changes in gene expression to control pilus expression, biofilm formation, and the response to antibiotics and oxidative stress Although folate stress and certain conditions increase paa expression (15)(16)(17)(18)39), reports have shown that paa expression is strongly repressed as a response to other environmental changes (14,38). Therefore, it is tempting to speculate that cells respond to both high and low concentrations of PAA by regulating the expression of different gene subsets (Fig. 8).
It is not yet understood how the PAA signaling cascade fits into the global regulatory networks in A. baumannii. However, there are several regulatory proteins that have been shown to act upstream of the paa operon. We confirm that PaaX is important for regulation of Csu, which is consistent with overexpression of the paa operon. It is possible that PaaX interacts with other regulatory components to dynamically regulate paa operon expression under a variety of environmental conditions. In fact, the paa operon is also strongly regulated by the GacSA TCS in A. baumannii. GacS is an important virulence-associated histidine kinase that has roles in regulating biofilm formation and motility, both in A. baumannii and in Pseudomonas species (22,41). Interestingly, a gacS mutant of A. baumannii showed up to 200-fold repression of paa genes compared to the WT, indicating that GacS is a major regulator of the paa operon and could be responsible for regulating paa genes in response to environmental changes (22). Additional regulators that act upstream of the paa operon include BlsA, a photoreceptor that negatively regulates expression of paa genes in response to blue light (14), and MumR, a manganese-responsive transcriptional regulator of the oxidative stress response in A. baumannii (34), further suggesting that PAA degradation plays a role in adaptation to a variety of environmental conditions.
We have shown that the paa operon has downstream effects on the expression of Stress Responses in Acinetobacter mBio several factors, including Csu pili. It is well established that Csu expression is also regulated by the BfmRS TCS (42), a TCS involved in antibiotic tolerance, membrane homeostasis, and desiccation resistance (43,44). While multiple factors likely affect Csu regulation, further work is necessary to determine if PAA interacts with components of the Bfm system and how PAA induces Csu expression. The involvement of multiple regulators in paa operon regulation supports our model that PAA is an important player in multiple signaling cascades. Previous studies in the Gram-negative bacteria P. aeruginosa and B. cenocepacia, have shown that PAA might interact with the quorum-sensing machinery of these strains (23,45). However, in B. cenocepacia, it was discovered that PAA is converted to PAA-CoA to interfere with quorum-sensing-mediated virulence of this strain (46). We cannot rule out that PAA-CoA, rather than PAA, is the signaling molecule that is responsible for the appreciable phenotypes in Acinetobacter. PAA could potentially be converted to PAA-CoA by the paaK-encoded PAA-CoA ligase before exerting its effects on the cell. However, for this to happen, paaK should be regulated independently from the rest of the paa operon to allow for dynamic control over PAA-CoA levels. Our transcriptomic and proteomic data indicate that this is not the case. Furthermore, PAA was shown to accumulate in a paaABCDE mutant (21). A metabolomic analysis would aid in the determination of the exact levels of the different metabolites in the PAA pathway under different environmental conditions.
Previous reports have linked PAA to virulence in the model strains Ab17978 and ATCC 19606 (21,22). It was shown that PAA produced by A. baumannii is sensed by zebrafish neutrophils, which are attracted to the site of infection, leading to greater bacterial killing and improved host survival (21). Here, we show that inactivation of PAA catabolism attenuates UPAB1 virulence in a murine CAUTI model. We propose that attenuation is due to the reduced capacity of the DpaaB strain to respond to the stresses the bacterium encounters during infection, which may include oxidative, osmotic, and other types of stress. More experiments are required to determine how host and bacterial factors contribute to the phenotype of paa mutants in vivo and how this plays a role in Acinetobacter pathogenesis.
We have observed for the first time that PAA is involved in a signaling pathway in A. baumannii. The biggest remaining question is how PAA mediates these physiological changes. PAA behaves as a signaling molecule in multiple organisms from plant to bacteria; however, little is understood about PAA-receptor interactions. In plants, PAA is a phytohormone that regulates many important processes involved in growth and development (25). In fact, PAA can interact with different auxin coreceptors from the transport inhibitor response1/auxin signaling F-box (TIR1/AFB) family, leading to changes in gene expression in plant cells (25). In bacteria, the only characterized PAAreceptor interaction is in the rhizobacterium P. putida, where PAA, mostly produced by plants, is sensed by the methyl-accepting chemotaxis receptor Aer2 (27). In this system, P. putida senses PAA gradients using Aer2 and swims toward PAA to use it as a carbon source. While this cascade is for metabolic purposes, it does suggest that bona fide PAA receptors may exist in other bacteria, such as A. baumannii, that sense and respond to endogenously produced PAA. Therefore, we hypothesize that PAA is an interspecies, cross-kingdom signaling molecule.
Understanding the molecular basis underlying the PAA-mediated response to stress could open new avenues for drug development and research. In fact, some PAA derivatives are already part of a group of FDA-approved nonsteroidal anti-inflammatory drugs, some of which have been used to treat bacterial infections in murine models (47,48). The therapeutic potential of PAA or PAA derivatives in interfering with the stress response of drug-resistant bacteria is extremely relevant to the growing field of antimicrobial resistance, warranting further research into their potential effects on Acinetobacter stress tolerance and pathogenesis.

MATERIALS AND METHODS
Bacterial strains and growth conditions. The bacterial strains used in this study are listed in Table S4 in the supplemental material. Unless otherwise noted, strains were grown in lysogeny broth Stress Responses in Acinetobacter mBio (LB) liquid medium at 37°C with shaking (200 rpm). When necessary, strains were grown in kanamycin (30 mg/mL), kanamycin (7.5 mg/mL), zeocin (50 mg/mL), gentamicin (15 mg/mL), trimethoprim (6 mg/ mL), and/or sulfamethoxazole (30 mg/mL). Overnight cultures used for phenotypic assays were always grown without antibiotics. Construction of paa mutants and complemented strains. The primers used in this study are listed in Table S5 in the supplemental material. The paaB mutant of UPAB1 was constructed as done previously (49). Briefly, an FLP recombination target (FRT) site-flanked zeocin resistance cassette was amplified from a variant of pKD4 with primers harboring 18 to 25 nucleotides of homology to the flanking regions of the target gene, paaB. Upstream and downstream flanking regions of paaB were also amplified, and the 3 fragments were assembled via overlap extension PCR to form linear DNA (50). For Ab17978, a kanamycin cassette was amplified with 150-bp oligonucleotide primers with flanking regions of paaB or paaX to form the linear PCR product. Linear PCR products were then electroporated into competent Ab17978 or UPAB1 carrying pAT04, which expresses the RecAB recombinase (49). Mutants were selected on zeocin (50 mg/mL) for UPAB1 or kanamycin (7.5 mg/mL) for Ab17978 and confirmed by PCR. To remove the antibiotic cassettes, mutants were transformed with the plasmid pAT03, which expresses the FLP recombinase. Clean mutants were confirmed by PCR and sequencing.
Complementation of paaB mutants was done using a mini-Tn7-based, four-parental conjugation technique for chromosomal insertion that was previously described (51). Briefly, paaABC and 300 bp upstream of paaA were amplified from the Ab17978 chromosome and inserted into pUC18-miniTn7-Gm (51) using Hi-Fi DNA assembly mix (New England Biolabs) to make pUCT18-miniTn7-Gm-paaABC. For UPAB1, paaABCDEFG and 300 bp upstream of paaA were amplified from the chromosome of UPAB1 and inserted into pUCT18-miniTn7-Zeo (52). This vector was then linearized and religated to include only paaAB to create pUCT18-miniTn7-Zeo-paaAB. To create paaB1 complemented strains, overnight cultures of either Ab17978 DpaaB or UPAB1 DpaaB recipients, HB101(pRK2013), EC100D(pTNS2) for UPAB1 or SM10(pTNS3) for Ab17978, and either TOP10(pUCT18-miniTn7-Zeo-paaAB) for UPAB1 or TOP10(pUCT18-miniTn7-Gm-paaABC), were normalized to an optical density at 600 nm (OD 600 ) of 2.0, and 100 mL of each culture was added to 600 mL LB. The bacterial mixtures were washed 2 times with 1 mL LB, resuspended in 25 mL LB, spotted onto a low-salt Luria agar plate, and incubated overnight at 37°C. For UPAB1, resuspended bacteria were plated on Luria agar with chloramphenicol (12 mg/mL) to select against E. coli strains and zeocin (50 mg/mL) to select for mini-Tn7 recipients. For Ab17978, Vogel-Bonner minimal medium with 0.2% glycerol, chloramphenicol (12 mg/mL), and gentamicin (100 mg/mL) was used for selection, similar to what was previously described (53). For UPAB1, the paaB1 strain was then transformed with pAT03 to remove the zeocin cassette and make a clean UPAB1 paaB1 strain. Complemented strains were verified by PCR and sequencing.
Csu Western blotting and quantification. Ab17978 overnight cultures were diluted in 10 mL fresh LB to an OD 600 of 0.05 in 50-mL conical tubes and grown at 37°C with shaking for 4 h until they reached an OD 600 of 1.0. For growth under TMP/SMX conditions, the concentrations of 3 mg/mL for TMP and 15 mg/mL of SMX were used for Ab17978, and TMP/SMX (1/4 mg/mL) was used for M2. For other antibiotics, bacteria were treated with antibiotic concentrations under which the final culture OD 600 values were between 65 and 85% of the untreated control. The cells were then pelleted by centrifugation and resuspended in Laemmli buffer to a final OD 600 of 20. Then, 5 mL of whole-cell samples was loaded onto 15% acrylamide SDS-PAGE gels for separation, transferred to a nitrocellulose membrane, and probed with polyclonal rabbit anti-CsuA/B (1:2,000) (42) and monoclonal mouse anti-RNA polymerase (anti-RNAP) (1:2,600) (BioLegend, San Diego, CA). Western blots were then probed with IRDye-conjugated anti-mouse and anti-rabbit secondary antibodies (both at 1:15,000) (LI-COR Biosciences, Lincoln, NE) and visualized with an Odyssey CLx imaging system (LI-COR Biosciences). Western blot signals were quantified using LI-COR Image Studio software, and the ratio of Csu to RNA polymerase intensity was calculated for at least 4 biological replicates.
Biofilm assays. Overnight cultures were normalized to an OD 600 of 0.05 in LB and inoculated into sterile, round-bottom polystyrene 96-well plates (Corning, Inc.) with a final volume of 200 mL per well. Plates were incubated for 24 h under static conditions at 37°C with humidity to prevent evaporation. After incubation, culture was removed and transferred to another 96-well plate, and OD 600 readings were taken using a BioTek microplate reader. Wells were washed 5 times with water and stained with 300 mL 0.1% crystal violet solution dissolved in water for 15 min. After staining, wells were washed 5 times with water and left to completely dry overnight. Crystal violet staining was dissolved in 200 mL 30% acetic acid, and absorbance was measured at 550 nm.
Reverse transcription-quantitative PCR. Overnight cultures of Ab17978 were inoculated into 50 mL LB or LB plus TMP and SMX (3 mg/mL and 15 mg/mL, respectively) to a final OD 600 of 0.05 and grown for 2 h at 37°C with shaking. Cells were quickly pelleted and treated with RNAprotect (Qiagen, Inc.), followed by total RNA isolation using the ZR Fungal/Bacterial RNA MiniPrep kit (Zymo Research Corp.). RNA samples were subjected to rigorous DNase treatment using a TURBO DNA-free kit to remove contaminating DNA. For reverse transcription (RT)-PCR, cDNA was prepared from 1 mg RNA using a high-capacity RNA-to-cDNA kit (Applied Biosystems), according to the manufacturer's protocol. The cDNA was diluted 1:10, and 1 mL was used as the template for quantitative PCR (qPCR) using PowerUp SYBR green master mix (Applied Biosystems) on a ViiA7 real-time PCR machine (Applied Biosystems), following the manufacturer's suggested protocol. The A. baumannii rpoB gene was used as a reference gene (41). All gene-specific primers used for qPCR were designed using IDT PrimerQuest and are listed in Table S5. Threshold cycle (C T ) values were normalized to rpoB, and fold changes and log 2 (fold changes) were calculated using the DDC T method.
Preparation of whole-cell lysates for comparative proteomics. Overnight cultures of WT Ab17978 were grown in LB and then subcultured to an OD 600 of 0.05 in 100 mL of one of the following 4 culture conditions: LB, LB1TMP and SMX (3 mg/mL and 15 mg/mL), LB1PAA (0.5 mM), or LB1PAA1TMP/SMX. Four individual 100-mL culture biological replicates were prepared for each condition. Cultures were grown for 4 h at 37°C with shaking, pelleted, washed 1Â with cold phosphate-buffered saline (PBS), and pelleted again. Pellets were flash frozen in liquid nitrogen and stored at 280°C until extraction. For protein extraction, pellets were thawed on ice and resuspended in 25 mL cold PBS plus Pierce protease inhibitor. Cells were lysed with two passages through a cell disruptor at 35,000 lb/in 2 (Constant Systems, Ltd., Kennesaw, GA). Unbroken cells were pelleted at 8,000 rpm for 10 min. Protein content of lysates was quantified using a DC protein assay kit (Bio-Rad), and 200 mg of protein was acetone precipitated twice. For acetone precipitation, 4 volumes of cold acetone was added to samples, and the samples were incubated overnight at 220°C and centrifuged to discard the supernatant. Protein pellets were resuspended in 200 mL of water, and acetone precipitation was repeated with 4 h of incubation at 220°C. Protein was pelleted at 16,000 Â g and dried at 70°C for 4 min before being stored at 220°C. No less than 100 mg was used for comparative proteomics.
Digestion of proteome samples. Precipitated whole-cell lysates were resuspended in 6 M urea, 2 M thiourea in 40 mM NH 4 HCO 3 and then reduced for 1 h with 20 mM dithiothreitol (DTT). Reduced samples were then alkylated with 50 mM chloroacetamide for 1 h in the dark. The alkylation reaction was then quenched by the addition of 50 mM DTT for 15 min, and samples were digested with Lys-C (1/200 [wt/wt]) for 3 h at room temperature. Samples were diluted with 100 mM NH 4 HCO 3 4-fold to reduce the urea/thiourea concentration below 2 M, then trypsin (1/50 wt/wt) was added, and the mixture was allowed to digest overnight at room temperature. Digested samples were acidified to a final concentration of 0.5% formic acid and desalted with homemade high-capacity StageTips composed of 1 mg Empore C 18 material (3M) and 5 mg of Oligo R3 reverse-phase resin (Thermo Fisher Scientific) as described previously (54,55). Columns were wet with buffer B (0.1% formic acid, 80% acetonitrile) and conditioned with buffer A* (0.1% trifluoroacetic acid [TFA], 2% acetonitrile) prior to use. Acidified samples were loaded onto conditioned columns and washed with 10 bed volumes of buffer A*, and bound peptides were eluted with buffer B before being dried then stored at 220°C.
LC-MS analysis of proteome samples. Dried proteome digests were resuspended in buffer A* and separated using a two-column chromatography setup composed of a PepMap100 C 18 20-mm by 75-mm trap and a PepMap C 18 500-mm by 75-mm analytical column (Thermo Fisher Scientific). Samples were concentrated onto the trap column at 5 mL/min for 5 min with buffer A (0.1% formic acid, 2% dimethyl sulfoxide [DMSO]) and then infused into an Orbitrap Fusion Lumos mass spectrometer (Thermo Fisher Scientific) at 300 nL/min via the analytical column using a Dionex Ultimate 3000 UPLC (Thermo Fisher Scientific). One hundred twenty-five-minute analytical runs were undertaken by altering the buffer composition from 2% buffer B (0.1% formic acid, 77.9% acetonitrile, 2% DMSO) to 22% B over 95 min, then from 22% B to 40% B over 10 min, and then from 40% B to 80% B over 5 min. The composition was held at 80% B for 5 min and then dropped to 2% B over 2 min before being held at 2% B for another 8 min. The Orbitrap Fusion Lumos mass spectrometer was operated in a data-dependent mode automatically switching between the acquisition of a single Orbitrap MS scan (300 to 1,600 m/z, a maximal injection time of 50 ms, an automated gain control [AGC] set to a maximum of 4 Â 10 5 ions, and a resolution of 120,000) and every Orbitrap tandem mass spectrometry (MS/MS) high-energy collisional dissociation (HCD) scan of precursors (normalized collision energy [NCE] of 35%, a maximal injection time of 100 ms, an AGC set to a maximum of 2 Â 10 5 ions, and a resolution of 15,000) every 3 s.
Proteomic analysis. Whole-proteome samples were processed using MaxQuant (v1.6.3.4) (56) and searched against the A. baumannii ATCC 17978 proteome (NCBI accession no. CP012004 [3,663 protein sequences]). Searches were undertaken using "trypsin" enzyme specificity with carbamidomethylation of cysteine as a fixed modification. Oxidation of methionine and acetylation of protein N termini were included as variable modifications, and a maximum of 2 missed cleavages were allowed. To enhance the identification of peptides between samples, the "match between runs" option was enabled with a precursor match window set to 2 min and an alignment window of 20 min with the label-free quantitation (LFQ) option enabled (57). The resulting outputs were processed within the Perseus (v1.6.0.7) analysis environment (58) to remove reverse matches and common protein contaminants prior to further analysis. For LFQ comparisons, biological replicates were grouped, and missing values were then imputed based on the observed total peptide intensities with a range of 0.3s and a downshift of 2.5s using Perseus. Student t tests were undertaken to compare the proteomes between groups, with the resulting data exported and visualized using ggplot2 (59) within R. The resulting MS data and search results have been deposited into the PRIDE ProteomeXchange Consortium repository (60,61).
Preparation of samples for RNA sequencing. WT Ab17978 and Ab17978 DpaaB were grown overnight in LB before being diluted into 10 mL of LB or LB1TMP/SMX (3 mg/mL and 15 mg/mL, respectively) and grown for 2 h at 37°C with shaking. Four individual overnight and 10-mL culture biological replicates were prepared for each condition. Cultures were normalized, and the equivalent of 1 mL of an OD 600 of 1.0 was pelleted quickly at 8,000 rpm and treated with RNAprotect for 5 min at room temperature. Pellets were flash frozen and stored at 280°C until extraction. For RNA extractions, pellets were thawed on ice, resuspended in 600 mL TRIzol (Invitrogen) with 4 mL of glycogen at 5 mg/mL, and lysed via bead beating. Samples were pelleted, and supernatants were treated with chloroform. RNA was extracted from the aqueous phase using the RNeasy minikit (Qiagen, Inc.), and RNA quality was checked by agarose gel electrophoresis and A 260 /A 280 measurements. RNA was stored at 280°C with SUPERase-IN RNase inhibitor (Life Technologies) until library preparation.
RNA sequencing and analysis. RNA sequencing (RNA-Seq) was performed as done previously (62). Briefly, 400 ng of total RNA from each sample was used for generation of cDNA libraries following the RNAtag-Seq protocol. PCR-amplified cDNA libraries were sequenced on an Illumina NextSeq500, generating a high sequencing depth of ;7.5 million reads per sample. Raw reads were demultiplexed by 59 and 39 indices, trimmed to 59 bp, and quality filtered (96% sequence quality . Q14). Filtered reads were mapped to the corresponding reference genome using bowtie2 with the "very-sensitive" option (-D 20 -R 3 -N 0 -L 20 -i S, 1, 0.50). Mapped reads were aggregated by feature Count, and differential expression was calculated with DESeq2 (63,64). In each pairwise differential expression comparison, significant differential expression was filtered based on two criteria: jlog 2 fold changej of .1 and adjusted P value (padj) of ,0.05. All differential expression comparisons were made between the WT and paaB mutant either in LB or under TMP/SMX treatment. The reproducibility of the transcriptomic data was confirmed by an overall high Spearman correlation across biological replicates (R . 0.95). Raw RNA-Seq data sets are available at the Sequence Read Archive (BioProject accession no. PRJNA726989). Differential expression data can be found in Tables S2 to S3 and at https://www.ncbi.nlm.nih.gov/ Traces/study/?acc=PRJNA726989&o=acc_s%3Aa.
Growth assays. Growth curves were performed in sterile, round-bottom, polystyrene, 96-well plates. Bacterial strains were grown overnight in LB, washed in PBS, and diluted in fresh medium (LB or M9 minimal medium supplemented with PAA derivatives) to an OD 600 of 0.01. Bacterial suspensions were then inoculated into 96-well plates at a 150-mL final volume before being grown at 37°C under shaking conditions. OD 600 values were measured at 30-min intervals for 16 h with a BioTek microplate reader. For antibiotic growth curves, the same protocol was used, and bacteria were diluted in LB with the appropriate antibiotic concentration. All experiments were performed on 3 independent days with at least 3 wells per strain per condition.
Drug susceptibility assays (MIC determination). Antimicrobial susceptibility was determined using the 2-fold broth dilution microtiter assay as previously described (32,65). Briefly, cells were inoculated at a density of 10 5 cells/mL in a 96-well microtiter plate containing 2-fold dilutions of the appropriate antibiotics. MICs were visually determined after plates were incubated at 37°C under shaking conditions for 16 h.
H 2 O 2 killing assays. H 2 O 2 killing assays were performed as done previously (38). Briefly, overnight cultures of the WT, DpaaB, and paaB1 UPAB1 strains were diluted 1:1,000 into 10 mL LB in 50-mL conical tubes and grown to mid-exponential phase (OD 600 of 0.4 to 0.6). Cultures were then treated with 0 mM or 175 mM H 2 O 2 for 30 min before being serially diluted and spot plated onto LB agar to enumerate CFU. Experiments were performed at least 3 independent times and were plated in duplicate.
Murine CAUTI model. CAUTI infections were performed as previously described for A. baumannii (37). Bacterial strains were prepared for inoculation after static growth twice at 37°C by centrifugation at 6,500 rpm for 5 min. Cells were washed twice in 1Â PBS and resuspended in PBS to the final inoculum. Briefly, 6-to 8-week-old female C57BL/6 mice (Charles River Laboratories) were anesthetized by inhalation of 4% isoflurane, and a 4-to 5-mm piece of silicone tubing (catheter) was placed in the bladder via transurethral insertion. Mice were infected immediately following implant placement with ;2 Â 10 8 CFU bacteria in 50 mL via transurethral inoculation. At 24 h postinfection (p.i.), mice were euthanized, and bladders and catheters were aseptically removed. The bacterial load present in each tissue was determined by homogenizing each organ in PBS and plating serial dilutions on LB agar supplemented with antibiotics when appropriate. All CAUTI studies were performed in accordance with the guidelines of the Committee for Animal Studies at Washington University School of Medicine, and we have complied with all relevant ethical regulations. Mice were housed with a cycle consisting of 12 h each of light and dark with access to standard food and water ad libitum.
Data availability. The mass spectrometry proteomics data have been deposited into the PRIDE ProteomeXchange Consortium repository and can be accessed with the identifier PXD025651. Raw RNA-Seq data sets are available at the Sequence Read Archive (BioProject accession no. PRJNA726989).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.  access to MS instrumentation. We thank the members of the Feldman lab for critical reading of the manuscript. This work was supported by a grant from the National Institute of Allergy and Infectious Diseases (grant R01AI125363 and R01AI144120).
We declare no conflict of interest.