The Response of Acinetobacter baumannii to Hydrogen Sulfide Reveals Two Independent Persulfide-Sensing Systems and a Connection to Biofilm Regulation

Although hydrogen sulfide (H2S) has long been known as a respiratory poison, recent reports in numerous bacterial pathogens reveal that H2S and more downstream oxidized forms of sulfur collectedly termed reactive sulfur species (RSS) function as antioxidants to combat host efforts to clear the infection. Here, we present a comprehensive analysis of the transcriptional and proteomic response of A. baumannii to exogenous sulfide as a model for how this important human pathogen manages sulfide/RSS homeostasis. We show that A. baumannii is unique in that it encodes two independent persulfide sensing and detoxification pathways that govern the speciation of bioactive sulfur in cells. The secondary persulfide sensor, BigR, impacts the expression of biofilm-associated genes; in addition, we identify two other transcriptional regulators known or projected to regulate biofilm formation, BfmR and Crp, as highly persulfidated in sulfide-exposed cells. These findings significantly strengthen the connection between sulfide homeostasis and biofilm formation in an important human pathogen.

an effort to identify potential participants in this regulatory response that models those processes that may be triggered by fluctuations in endogenous sulfide/RSS inside the cell. Transcriptomic and label-free proteomic analysis reveal a complex regulatory response to sulfide, involving a primary (FisR) and a secondary (BigR) persulfide-sensing transcriptional regulator, each of which controls the expression of sulfide oxidation and transport enzymes, including two candidate persulfide dioxygenases (PDO1 and PDO2). Metabolite profiling of LMW thiols, sulfide, and RSS in these cultures before and after the addition of Na 2 S reveals that although cysteine levels rise, A. baumannii is relatively resistant to perturbation of the cell-associated RSS and H 2 S. A. baumannii encodes 3MST, the deletion of which specifically lowers LMW persulfide levels. Sulfide increases the transcription of genes encoding an alternative cytochrome bd oxidase cydAB, as shown previously in E. coli (28) and M. tuberculosis (9), which is refractory to inhibition by H 2 S (10,28). In addition, the cellular abundance of key ROS defense enzymes, including catalase, superoxide dismutase, and the alkylhydroperoxidase AhpCF, also increases. Protein S-sulfuration (persulfidation) profiling in A. baumannii reveals that ϳ13% of the proteome is persulfidated and identifies potential regulatory targets of H 2 S/RSS stress, including two highly persulfidated transcriptional regulators, BfmR and Crp, known or projected to function in biofilm regulation.

RESULTS
LMW thiols and persulfides in wild-type and ⌬MST cells. In previous studies in S. aureus and Enterococcus faecalis, we employed an isotope dilution, electrophile trapping method to estimate the concentrations of the major cellular thiols, thiol persulfides, and inorganic sulfide in cell lysates from mid-exponential-phase cells (11,23,29). In order to understand the regulatory response of A. baumannii ATCC 17978 to external sulfide stress, we used this approach to quantify endogenous levels of H 2 S, major cellular thiols and thiol persulfides in mid-log-phase wild-type (WT) cells and compare these levels to those in a strain lacking the 3MST (A1S_3379). 3MST is thought to be an important source of endogenously synthesized H 2 S from studies of E. coli (5,6,8). Note also that A. baumannii harbors complete glutathione (GSH) and coenzyme A (CoASH) biosynthetic pathways from cysteine, with homocysteine (HCys) derived from L-cystathionine (Fig. 1A).
We found that GSH is the major cellular thiol, followed by cysteine and CoASH, which are comparable in concentration to one another, with the concentration of homocysteine being about 10-fold lower than those of cysteine and CoASH and about equal to that of H 2 S (Fig. 1B). The addition of Na 2 S to WT cells results in a significant increase in cellular cysteine possibly as a result of increased flux through cysteine synthase (CysK) (Fig. 1A), with a corresponding increase in cysteine persulfide as well (Fig. 1B). A. baumannii harbors thiol persulfides for each of its major cellular thiols, with glutathione persulfide (GSSH) predominating (Fig. 1B). Exogenous Na 2 S perturbs these levels primarily by increasing GSSH, cysteine persulfide, and the inorganic species thiosulfate (S 2 O 3 2-[TS]). Interestingly, the fraction of thiol persulfide is significantly lower for GSSH and cysteine persulfide, which represent ϳ0.1% of their respective total thiol pools compared to that of CoASH persulfide which is ϳ2% (Fig. 1B).
The impact of 3MST on RSS in cells was also investigated (see Fig. S1 in the supplemental material). Interestingly, deletion of 3MST (ΔMST) does not directly impact cellular H 2 S levels, but instead results in small decreases in GSH and HCys levels, with correspondingly larger (Ն4-fold) decreases in endogenous GSSH, cysteine, and homocysteine persulfide levels ( Fig. S1A and B). The addition of Na 2 S to ΔMST cells results in increases in the LMW thiol persulfide pools within 30 min, notably GSSH, which increases ϳ8-fold to a level comparable to that of the WT strain (Fig. S1C). These data taken collectively reveal that 3MST significantly impacts endogenous sulfur speciation in cells and that some external Na 2 S is assimilated as TS and LMW thiol persulfides.
The 54 -transcriptional activator FisR controls the primary transcriptomic response to exogenous sulfide in A. baumannii. To further investigate the impact of H 2 S on A. baumannii, we investigated the transcriptomic response of A. baumannii to the addition of the 0.2 mM Na 2 S. We initially compared the transcriptomic profiles of three strains: the WT strain, the ΔbigR strain, and a ΔbigR mTn7(bigR C34A/C100A ) complemented strain. BigR (biofilm growth-associated repressor; A1S_1539) is an arsenicrepressor (ArsR)-superfamily transcriptional regulator expected to sense RSS on the basis of sequence similarity to other dithiol (C34, C100)-containing persulfide sensors characterized previously in Gram-negative bacteria, Rhodobacter capsulatus and Xylella fastidiosa, denoted SqrR and BigR, respectively (24,25). These results are summarized in Fig. 1C to G and also Table S1A in the supplemental material, with key aspects of these findings confirmed by real-time quantitative reverse transcription PCR (qRT-PCR) (Fig. 2).
Disodium sulfide induces the expression of several operons. The most highly upregulated genes are A1S_1708-1709-1710-1711, which encode a persulfide dioxygenase (PDO1), a sulfide:quinone oxidoreductase (SQR), a putative TS/sulfite effluxer TauE, and a candidate homoserine dehydrogenase ( Fig. 1C and F and Fig. 2A; see also Table S1A in the supplemental material). The first three genes correspond to those found in the RSS-regulated operon, cst, in S. aureus controlled by the RSS-sensing repressor, CstR (22,30,31). Other genes that are induced by sulfide include (i) the  terminal ubiquinol cytochrome bd-type oxidase, cydAB (A1S_1433-1434), as shown previously in E. coli and M. tuberculosis (9,28), (ii) A1S_0169-0171, the middle gene encoding a candidate outer membrane localized copper transporter, OprC (32), and (iii) A1S_3085-3086, encoding a flavohemoglobin/nitric oxide dioxygenase (hmp) and a 4Fe-4S-containing nitric oxide-sensing transcriptional repressor (nsrR) (33), also upregulated in S. aureus treated with Na 2 S (29). The A1S_1708-1711 operon was of particular interest since the sulfide-dependent induction of these genes is not dependent on BigR (Fig. 2B) nor are they BigR-regulated genes, as determined by a transcriptome sequencing (RNA-seq) experiment that compares gene expression in the ΔbigR and WT strains ( Fig. 1D; Table S1A). These data suggest an alternative mode of sulfide-inducible regulation of A1S_1708-1711. Upstream of this operon and transcribed in the opposite direction is A1S_1707, which encodes a canonical 54 -transcriptional activator, harboring an N-terminal regulatory domain, a central AAAϩ ATPase domain, and a C-terminal DNA-binding domain (26,34); further, the promoter upstream of A1S_1708 is predicted to be a 54 -dependent promoter. We denote A1S_1707 as fisR (Fis family transcriptional regulator), named after the sulfide-inducible 54 -dependent transcriptional activator previously characterized in Cupriavidus spp. (26). To test the hypothesis that FisR is a transcriptional activator of the A1S_1708-1711 operon, we created a deletion strain of fisR and determined the ability of this strain to grow when challenged with exogenous sulfide ( Fig. 3; Fig. S2A to C), while measuring expression of the BigR-regulated genes (A1S_1876-1878) and the A1S_1708-1711 operon by qRT-PCR (Fig. 2C). The ΔfisR strain exhibits a pronounced growth lag relative to the ΔbigR and WT strains that becomes more severe with increasing exogenous sulfide ( Fig. 3; Fig. S2). The sulfide-dependent induction of A1S_1708-1711 is completely ablated in the ΔfisR strain ( Fig. 2C) but not in the complemented strain (Fig. 2D); further, the sulfide-induced growth phenotype of the ΔfisR strain is also complemented in the ΔfisR::pfisR strain to WT growth levels ( Fig. 3; Fig. S2A to C).
Genes with reduced expression in the WT strain in the presence of exogenous sulfide are, as expected, largely limited to those genes associated with the uptake of    (35). These findings suggest that cells repress the uptake of other forms of sulfur when inorganic sulfide becomes available, while upregulating the uptake of these other forms of sulfur to protect against the effect of oxidative damage. BigR controls the expression of a secondary H 2 S/RSS-sensing and detoxification system. The genomic BigR regulon was mapped by comparing transcriptomic profiles of the ΔbigR strain to that of a chromosomally complemented strain encoding a mutant allele of BigR that lacks the persulfide-sensing dithiol pair [C34, C100; ΔbigR mTn7(bigR C34A/C100A )] ( Fig. 1E; Table S1A) (24,36). These data suggest that in addition to yeeE/yedE/pdo2-encoding sulfide detoxification operon found previously in Serratia spp. (37), genes predicted to be associated with assembly of the type 1 chaperoneusher pilus involved in epithelial cell adhesion and biofilm formation (A1S_1508-1510) (38), an uncharacterized short-chain dehydrogenase (sdr; A1S_0087), and A1S_1548-1549 may also be direct targets of BigR regulation, a finding consistent with the presence of candidate BigR operators upstream of these genes. Inspection of the qRT-PCR data obtained for the ΔfisR strain reveals that endogenous expression of the primary BigR-regulated operon (A1S_1878-1876) is ϳ10-fold higher than the untreated WT strain (Fig. 2C). This suggests that under these growth conditions, there is significant endogenous H 2 S/RSS that FisR is sensing (Fig. 1B), and when FisR is absent, i.e., in the ΔfisR strain, this leads to endogenous induction of BigR-regulated genes to a level comparable to that of the ΔbigR strain (Fig. 2B).
As expected for a secondary or backup (per)sulfide sensing and detoxification system, the ΔbigR strain and complemented ΔbigR mTn7(bigR) strains grow indistinguishably from the WT strain in the absence or presence of exogenous sulfide (Fig. S2D); however, the double ΔfisR ΔbigR mutant grows better than the ΔfisR strain, while the complemented ΔfisR ΔbigR mTn7(bigR) strain grows even better than the uncomplemented strain ( Fig. S2A to C). These experiments suggest that unregulated expression of the BigR-regulated genes in a ΔbigR strain protects A. baumannii against sulfide toxicity when FisR is not present and that BigR itself may also function as a persulfide "sink" under these conditions. These findings are consistent with a proteomics analysis of ΔfisR lysates, which reveals detectable BigR-regulated PDO2, in striking contrast to the WT strain (see below). Thiol and persulfide profiling of mid-log-phase ΔbigR and ΔfisR strains reveals that loss of the H 2 S/RSS-sensing activator (ΔfisR) or loss of the RSS-sensing repressor (ΔbigR) both perturb organic LMW thiol persulfide abundance relative to the WT strain in untreated cells, with virtually no impact on total thiol levels ( Fig. S3A and B). In both cases, GSSH levels are most significantly decreased, as expected from the expression of PDO2 in each strain as observed transcriptomically (Fig. 2), which is anticipated to specifically oxidize GSSH to sulfite (39,40). The addition of 0.2 mM Na 2 S to these strains simply increases cellular persulfide levels, with the ΔfisR strain characterized by an organic persulfide content that is indistinguishable from those concentrations found endogenously in the WT (Fig. S3C). We conclude that although FisR and BigR-regulated gene products clearly impact cellular abundance of organic persulfides, even under sulfide-stressed conditions, none reach a level that exceeds ϳ2% of the total thiol pool, as found for CoASH persulfide.
Global proteomics profiling of wild-type A. baumannii before and after addition of exogenous sulfide. We next employed a standard label-free "bottom-up" proteomics analysis (41) to estimate the degree to which cellular abundance of specific proteins change under conditions of exogenous sulfide stress (Table S1B). Four biological replicates of soluble lysates (cytoplasm and periplasm) were obtained from mid-logphase untreated (WT) and sulfide-treated (WTϩNa 2 S) cells ( Fig. 4; Fig. S4A and B). In total, 1,006 proteins were detected at least twice in the four replicates, with 47 proteins observed only in the WTϩNa 2 S cells and 17 observed only in WT untreated cells Response of A. baumannii to Hydrogen Sulfide ® (Fig. 4A). Consistent with the transcriptomic induction of the FisR-regulated genes, we observe high cellular abundance of the PDO1 and SQR, each of which rises to the top 50% of proteins detected ( Fig. 4B; Fig. S4B) and which are not detected in untreated cells. On the other hand, the nucleoid associated, nonspecific DNA binding protein Fis (A1S_2186) is only observed in untreated cells (Fig. 4B). In addition to these, there are several proteins whose cellular abundance significantly changes upon treatment with exogenous sulfide as depicted in a standard volcano plot format ( Fig. 4C and D). Of interest are the increased abundance of major ROS detoxification enzymes, including a heme-cofactored catalase (KatG) and putative alkylhydroperoxidase subunits AhpF, AhpC1, and AhpC2, two universal stress proteins (Usps) (42), as well as an uncharacterized multidomain sulfurtransferase (A1S_2708). Interestingly, AhpC2 is among the most abundant nonribosomal proteins found in A. baumannii lysates among two AhpC and five AhpF proteins encoded by the A. baumannii genome (35). Furthermore, AhpC1 contains a single Cys active site, in contrast to AhpC2, which when sulfenylated in M. tuberculosis is known to react with H 2 S to form a persulfidated Cys; this persulfidated intermediate is highly active in persulfide transfer to a model RSH acceptor (18). We note that although some ROS enzymes have been shown to process (catalase) H 2 S or biosynthesize (superoxide dismutase) RSS, there is no evidence to date to suggest these reactions take place in vivo (43). In fact, we find the cellular responses to Na 2 S and H 2 O 2 are largely independent of one another (Fig. S5) (35). This suggests either that OxyR does not respond directly to sulfane sulfur in A. baumannii as previously observed in E. coli (44) or that the induction of the OxyR regulon under these conditions is weak compared to the robust H 2 O 2 response. As a result, only a few members of the regulon are induced at or above our detection limits.
In addition, the cellular abundance of several proteins involved in metal homeostasis are also changed ( Fig. 4C and D). We observed increased abundance of CusF and CopZ, which are the periplasmic and cytoplasmic copper chaperones, respectively, and NlpE, an outer membrane lipoprotein that senses outer membrane stress, and is also involved in the response to periplasmic Cu stress (45). OprC, which is projected to bring Cu into the periplasm, is also transcriptionally upregulated by sulfide stress (Fig. 1C). In addition, two ferritin-like proteins, a heme-cofactored bacterioferritin (BfrB, A1S_3175) and a bacterial ferritin (FtnA, A1S_0800), both of which mineralize Fe for storage, are increased in the proteome ( Fig. 4D and C) and are transcriptionally upregulated ϳ3-fold by exogenous sulfide (Table S1A). These Fe storage proteins show high sequence similarity to BfrB and FtnA (formerly BfrA) in P. aeruginosa (46). In Pseudomonas aeruginosa FtnA exhibits constitutive expression during the exponential phase and Fe-dependent expression during the stationary phase, whereas BfrB is increased under excess Fe regardless of the growth phase (47,48). Upregulation of FtnA and BfrB in A. baumannii under Na 2 S may be indicative of conditions of excess bioavailable Fe but further studies are required to substantiate this. Taken together, these data suggest that exogenous sulfide impacts transition metal homeostasis in cells.
We also carried out a proteomics analysis of the ΔfisR strain before and after addition of sulfide ( Fig. S4C and D, Fig. S6, and Table S1C) in an effort to understand the impact of the weakly constitutive expression of the BigR-regulated operon in this strain (Fig. 2C), which gives rise to lower endogenous levels of GSSH (Fig. S3B). These findings make the prediction that the BigR-regulated PDO2 would be active in cells and detectable in the proteome under both conditions, but only in the ΔfisR strain. This is exactly what is observed, with no significant change in cellular abundance following the addition of Na 2 S to the ΔfisR strain (Fig. S6C). Aside from this, many of the perturbations in the global proteome in the WT versus ΔfisR strains are common to both strains, as expected (Fig. S6). A few proteins of note whose change in abundance with Na 2 S is significantly more in ΔfisR than WT include a ferric siderophore receptor protein (BfrD1, A1S_0474), a glutathione-dependent disulfide bond oxidoreductase (A1S_1411), and a nitrite/sulfite reductase (Sir, A1S_2846) ( Fig. S6D and Table S1C).
Mapping of proteome persulfidation sites. We previously developed a global enrichment-reduction strategy to identify cysteines in the proteome that are subject to persulfidation in the absence or presence of the exogenous Na 2 S (11). Briefly, in this method, like others (49,50), proteins in a lysate that harbor either a thiol or persulfide are enriched by alkylation with a biotinylated iodoacetamide (bioIAM) and adsorbed to neutravidin beads. Persulfide-containing peptides are eluted by reduction with TCEP [Tris(2-carboxyethyl)phosphine hydrochloride], capped with IAM, and subjected to liquid chromatography-tandem mass spectrometry (LC-MS/MS). This peptide fraction is therefore enriched for peptides from proteins that are persulfidated in cells before or after addition of exogenous sulfide, reflected in the parameter R , which is a measure of the extent to which a particular Cys residue is more heavily persulfidated in cells 30 min after the addition of sulfide versus prior to addition of sulfide (Fig. 5A). A R value of 1.0 identifies peptides that are not persulfidated prior to the addition of Na 2 S (23 proteins), a R ϭ 0.5 indicates equal peptide counts prior to and after the addition of Na 2 S, and a R value approaching 0 indicates peptides that are only persulfidated in the absence of exogenous sulfide (8 proteins).
We identify 513 proteins or 13% of the proteome that harbor at least one persulfidated thiol, with the vast majority of these (95%) persulfidated in both unstressed and stressed cells (Fig. 5; Table S1D). This finding is as expected from previous persulfidation profiling studies of S. aureus (11) and E. faecalis (B. Walsh, unpublished results) and suggests that the proteome provides a readily tapped reservoir of bioactive sulfane sulfur (11). These proteins map to several cellular processes, including fatty acid and amino acid biosynthesis and pyruvate, pyrimidine, and sugar metabolism. In an effort Response of A. baumannii to Hydrogen Sulfide ® to identify a persulfidation motif, all modified peptides were analyzed by the pLogo motif analyzer program (Fig. S7A). Although this analysis did not reveal conserved residues on either side of the persulfidated cysteine, hydrophilic residues are overrepresented, while hydrophobic and additional cysteine residues are underrepresented. This result is consistent with other studies, for protein persulfidation or S-sulfenylation but perhaps not for S-nitrosylation or electrophile-sensing cysteine thiolates (51,52). When we examines only peptides with high R , we still did not observe a consensus persulfidation motif, but the pattern of hydrophilicity vanished (Fig. S7B). We next recast our R plot in a way that is normalized to a change in protein cellular abundance as a ratio (log 2 ) of ratios of the persulfidated peptide yield versus change in cellular abundance in the presence versus absence of exogenous sulfide ( Fig. 5C; see also Fig. S7C and Table S1D in the supplemental material). Although this approach does not measure the actual fraction of persulfidated cysteine versus thiol in any one peptide, the normalized fold change in abundance of a persulfidated peptide must be indicative of an increase in this fraction. We found that the vast majority of the proteins with R values greater than ( R Ն 0.76) or less than ( R Յ 0.40) one standard deviation from the mean R value (0.58) are found in the wings of the normalized abundance plot as well (Fig. S7C) and correspond to a normalized change in persulfidation versus thiol status of greater than or less than 4-fold. Prominent exceptions to this trend are the FisR-activated PDO1 and SQR, which are only detected in the sulfide-treated cells and are thus characterized by a large increase in cellular abundance, and correspond to the group of 23 proteins with a R of 1 (Fig. 5B). In fact, four of the five Cys in PDO1 were detected as persulfidated, while two Cys in SQR were modified, including the presumed active site thiol, C248, expected to carry a persulfide as a catalytic intermediate (31). Cys-containing peptides found to be significantly enriched in persulfidation status are of interest since they represent protein targets for which persulfidation may be regulatory (Fig. 5D). Included in this group of enzymes is MnmA, the tRNA 2-thiouridylase, which incorporates s 2 U substitution at U34 in tRNAs. The persulfidated Cys is the resolving Cys in a two-Cys mechanism that donates the sulfur atom for thio-insertion (53,54).
In addition, there are five transcriptional regulators found in this group of proteins, and these include the zinc uptake repressor, Zur (55); the A. baumannii acetate operon repressor, IclR2; a cell-abundant catabolite regulatory protein (Crp)-like protein (A1S_1182); and the biofilm response regulator, BfmR (Fig. 5B). BfmR is part of the BfmRS two-component system and master regulator of biofilm formation and desiccation tolerance that controls the expression of the Csu chaperone-usher assembly system which produces the cell surface pili required for adherence to abiotic surfaces (56). The persulfidated Cys in BfmR is found in the receiver domain and close to the site of phosphorylation (D58), which regulates DNA binding (57). The persulfidated Cys in Zur is conserved in E. coli Zur as C17 and is located in the ␣1 helix, close to the bound DNA within the N-terminal DNA binding domain (58). The impact of persulfidation on IclR2 and Crp function is also not known, nor has A. baumannii Crp been functionally characterized in any detail (59). However, A. baumannii Crp is quite closely related to global regulators Crp from E. coli (50% identical) and Vfr from P. aeruginosa (54% identical), with the conserved Cys opposite the ligand binding pocket, which for E. coli Crp is cyclic AMP (cAMP) (60). Glutathione is an allosteric activator of Vfr transcriptional upregulation of the type III secretion system in P. aeruginosa (61); the same is true of the global regulator PrfA in Listeria monocytogenes (62,63). The small molecule activator of A. baumannii Crp-like protein is unknown.

DISCUSSION
This study presents a comprehensive analysis of the adaptive response of the major human pathogen A. baumannii to exogenous hydrogen sulfide stress as a model for intracellular sulfide accumulation and thus identifies essential features of sulfide/RSS homeostasis. Although it is not known under what particular niche A. baumannii might encounter exogenous sulfide, emerging evidence suggests that endogenous sulfide/ RSS accumulation may protect a wide range of bacterial pathogens against clearance by the immune system, while reducing its susceptibility to antibiotics (6, 8-10, 64, 65); in addition, increased cellular H 2 S may characterize cells in biofilms at low O 2 availability (66). We show here that A. baumannii contains the LMW thiols GSH, cysteine, HCys, and CoASH and detectable levels of their corresponding thiol persulfides. The addition of exogenous sulfide results in assimilation of this inorganic sulfide as organic thiol persulfides namely, GSSH, cysteine persulfide, as well as the inorganic RSS, TS. While 3MST is reported as a source of intracellular H 2 S in E. coli (5,8), we found that the deletion of 3MST in A. baumannii results in significantly decreased thiol persulfides with little detectable impact on measured H 2 S concentrations. This suggests that any H 2 S production by 3MST is readily converted to thiol persulfides, perhaps by FisR-regulated SQR, which may function as cellular antioxidants to reduce immune system clearance during infection (6). In addition, intracellular production of H 2 S or RSS suggests a need for tight regulation of these species to avoid H 2 S toxicity while not overclearing RSS, which impacts virulence in a variety of infection models in S. aureus and in E. faecalis (6,11,23).
We find that A. baumannii encodes primary and secondary regulatory mechanisms responsible for sulfide detoxification and RSS clearance ( Fig. 1 and 6). The primary mechanism is transcriptionally regulated by FisR, a 54 -dependent transcriptional activator, and detoxifies sulfide via SQR, PDO1, and a putative thiosulfate/sulfite effluxer, TauE, with the role of the homoserine dehydrogenase not known. The secondary mechanism involves transcriptional regulation by the ArsR-family RSS-sensing repressor BigR (36), whose operon includes a second PDO2 and two putative trans-

FIG 6
Cartoon summary of the findings presented here. In response to increased cellular sulfide, a program of H 2 S oxidation ensues (left), where sulfide is oxidized to sulfite by the combined action of the two FisR-regulated gene products, SQR and PDO1, with the intermediacy of an LMW thiol persulfide RSSH (GSSH is shown); thiosulfate (SSO 3 2-) may be formed by the uncatalyzed reaction of sulfite with RSSH as shown (39). Sulfite and/or thiosulfate may be exported through the exporter TauE as indicated (78), while an unknown form of sulfur (S*) is transported by YeeE/YedE (37). Meanwhile, the ubiquinol oxidase CydAB is upregulated and is refractory to inhibition by sulfide (9,28); the activities of SQR and CydAB are predicted to be linked though the ubiquinone pool (Q) (14). Evidence of increased bioavailable Cu (red spheres) derives from increased Cu importer expression (OprC) (41), increased expression of major Cu chaperones, and NlpE, linked to Cu(II)-linked lipoprotein stress and/or persulfidation of periplasmic thiolates (DsbC) (45). Increased bioavailable Fe (green spheres) is suggested by increased cell abundance of bacterioferritin (BfrB) and bacterial ferritin (FtnA). We also observed increased cellular concentrations of ROS enzymes (pink symbols). Finally, we observed downregulation of at least three ABC transporters involved in uptake of sulfur-containing molecules (other S; locus tags indicated), all via unknown regulatory mechanisms. Both primary and secondary H 2 S/RSS sensing and detoxification systems regulated by FisR and BigR, respectively, are shown below the gray dashed line. FisR is transcriptionally activated (ϩϩϩ) as a result of increased cellular H 2 S/RSS, while BigR-mediated RSS chemistry (36) drives transcriptional derepression (ϩϩϩ) of BigR-regulated genes. Genes indicated by a question mark (?) are candidate direct BigR targets.
Response of A. baumannii to Hydrogen Sulfide ® membrane proteins proposed previously in Serratia spp. to function in the transport of sulfur-containing molecules, and associated with the biosynthesis of the antibiotic prodigiosin (37). This secondary system is only active upon deletion of the repressor BigR, or upon deletion of the primary activator FisR (Fig. 2), suggesting this regulon acts as a backup system but only when FisR is unable to activate its regulon. Considering that A. baumannii FisR does not conserve the regulatory domain cysteines previously shown to contribute to sulfide-dependent gene expression in Cupriavidus, it is unclear how FisR senses H 2 S/RSS in A. baumannii (26). It is possible that FisR and BigR recognize a different small molecule or bind the same inducer but with different sensitivities. BigR, however, clearly performs thiol oxidative chemistry with organic persulfides (36).
In addition to the sulfide detoxification mechanisms, we have analyzed both the transcriptional and proteomic response of A. baumannii to exogenous sulfide (Fig. 6). This includes transcriptional upregulation of the alternative cytochrome oxidase cydAB, which is refractory to inhibition by H 2 S (10, 28), coupled with downregulation of the ABC transporters predicted to function in the uptake of taurine and TS/sulfate. This allows the organism to maintain respiratory flux and redox balance in the presence of H 2 S while limiting the uptake of other sulfur sources to utilize the inorganic sulfide at hand (9). Our proteomic analysis reveals increased abundance of ROS detoxification enzymes that are largely independent from the response of A. baumannii to H 2 O 2 , indicating that these enzymes may have roles in sulfide homeostasis in addition to the ROS response. Exogenous sulfide also appears to induce a high-Fe and high-Cu response in A. baumannii that is incompletely understood (Fig. 6). In addition, we noted significant increases in cellular abundances of metabolic enzymes aconitase, perhaps facilitated by apparently increased bioavailable Fe and S, isocitrate lyase, succinate semialdehyde dehydrogenase, and malate synthase, all of which suggests enhanced metabolic flux through the glyoxylate shunt (GS; Fig. S8) (67,68). The GS was previously shown to be critical for the growth of M. tuberculosis within the infected host (69) and for maintenance of Fe homeostasis in P. aeruginosa (68). Indeed, persulfidation of the acetate operon repressor IclR2 may function as a regulatory mechanism that controls flux through the GS under these conditions (70); however, GS and its regulation has not been extensively studied in A. baumannii.
We found that ϳ13% of the proteome in A. baumannii harbors persulfidated cysteines and the extent of persulfidation changes significantly with exogenous sulfide with little or no change in protein abundance for a number of these targets (Fig. 5). This, together with our sequence motif analysis, suggests that many of these modified cysteines must be surface exposed and potentially function as a sink for cellular sulfane sulfur, a finding consistent with previous studies in S. aureus (11). However, the persulfidation status of five transcriptional regulators changes significantly (Ն10-fold) under conditions of exogenous sulfide, and these therefore represent excellent candidate regulatory targets of protein S-sulfuration. Of particular interest is the master regulator of biofilm formation, BfmR, and a Crp (cAMP regulatory protein)-like regulator, each of which possess persulfidated Cys that may well be regulatory, given their physical positions within the known or predicted structures of these regulators (57,71). In addition, the Crp activity in other organisms, e.g., L. monocytogenes, is directly regulated by glutathione binding (62,63). Like BfmR, the Crp-like regulator in A. baumannii may also play a role in biofilm formation, specifically in the formation of the pellicle, a specialized form of biofilm characterized in Acetobacter aceti (72). In A. baumannii the cAMP phosphodiesterase CpdA was found to be crucial for pellicle formation (59) and in P. aeruginosa the Crp regulator, Vfr, transcriptionally regulates cpdA (73). In addition, a cpdA mutant A. baumannii strain exhibits decreased expression of the csu system that is regulated by BfmR positing a potential connection between these two regulatory systems (59).
The Crp-like regulator and BfmR in A. baumannii remain largely uncharacterized as to their mechanisms of regulation (57). Our data suggest that elevated H 2 S/RSSdependent persulfidation may impact their transcriptional programs. Such a connec-tion of sulfide homeostasis to the regulation of biofilm dynamics may also extend beyond the csu system, since BigR, directly or indirectly, regulates the expression of genes predicted to be associated with assembly of the type 1 chaperone-usher pilus (Fig. 1D) involved in epithelial cell adhesion and biofilm formation in other strains (38). We note that elevated H 2 S is sometimes associated with bacterial cells under anoxic or low O 2 tension that persist deep in a biofilm, e.g., in cystic fibrosis sputum samples (66), which suggests that H 2 S/RSS-driven persulfidation of one or more master regulators may constitute a regulatory switch to dynamically control biofilm formation. This regulatory hypothesis is analogous to our previous findings in S. aureus where widespread persulfidation of global virulence regulators was shown to alter the composition of the secretome, thus impacting immune cell killing potential (11). Current studies are directed toward understanding how H 2 S/RSS biogenesis and the dual RSS sensing and clearance systems described here impact host virulence in mouse models of infection (6).

MATERIALS AND METHODS
Bacterial strains and growth. Strains were grown in Luria-Bertani (LB) medium or on LB plates including 1.5% (wt/vol) agar. Antibiotics were included at the following concentrations: carbenicillin, 50 mg/liter for E. coli and 75 mg/liter for A. baumannii; tetracycline, 10 mg/liter; kanamycin, 40 mg/liter; chloramphenicol, 15 mg/liter; and sulfamethoxazole, 100 mg/liter. Each overnight culture was initiated from a single colony, followed by incubation at 37°C with shaking for 8 to 16 h. Overnight cultures were diluted 1:50 in fresh LB medium, followed by incubation for 1 h before inoculation at 1:100. For growth curves, stressors were added after the 1:100 dilution at the indicated concentrations, and growth was monitored based on the optical density at 600 nm (OD 600 ) using a GENESYS 20 Visible Spectrophotometer (Thermo Scientific).
Construction of plasmids, allelic exchange mutants, and genetic constructs. All strains, plasmids, and oligonucleotides used in this study are listed in Table S2A, Table S2B, and Table S2C and D, respectively. DNA was amplified with 2ϫ Phusion master mix (Thermo Fisher). All restriction enzymes were from NEB. pET3a-BigR was generated by amplification of bigR gene from wild-type A. baumannii ATCC 17978 genomic DNA and cloned into BamHI and NdeI sites via Gibson assembly (36). This vector was then used to construct BigR double Cys mutant pET3a-BigR C34A/C100A using a standard mutagenesis protocol.
To construct ΔfisR::Tc, ΔbigR::Kn, and ΔmstA::Kn strains, 1,000-bp regions flanking the genes to be deleted were amplified from A. baumannii ATCC 17978 genomic DNA. For the ΔfisR::Tc strain, the excisable tetracycline resistance gene tetA was amplified from pLDP55; for the ΔbigR::Kn and ΔmstA::Kn strains, the excisable kanamycin resistance gene aph was amplified from pKD4 (74). The PCR products were assembled into pFLP2 digested with BamHI and KpnI using NEB HiFi to generate pLDP79, pLDP39, and pLDP81, respectively. After electroporating A. baumannii ATCC 17978 and selecting for kanamycin resistance (Kn r ) or tetracycline resistance (Tc r ), colonies were patched to screen for carbenicillin resistance (Carb r ) and sucrose sensitivity (Suc s ). Suc s and PCR-confirmed merodiploids were counterselected on LB agar plates containing 10% sucrose and screened for Kn r /Tc r and Carb s . fisR complementation vectors were constructed in pLDP29, a derivative of pWH1266 (75), digested with BamHI and KpnI. To construct the bigR complementation strains, the bigR promoter and gene were amplified from wild-type A. baumannii ATCC 17978, or the gene was amplified from pET3a-bigR C34A/C100A . DNA fragments were ligated into pKNOCK-mTn7-Amp digested with BamHI and KpnI by HiFi assembly. The mini-Tn7 was introduced into the A. baumannii ΔbigR::Kn strain as previously described (76), using carbenicillin as selection and chloramphenicol to counterselect against E. coli. All strains were confirmed by PCR and checked for maintenance of the large conjugative plasmid pAB3 by resistance to sulfamethoxazole.
Quantitative RT-PCR. Overnight cultures of A. baumannii were diluted 1:50 in LB medium and grown for 1 h at 37°C. Cultures were then diluted 1:100 into 10 ml of LB medium and grown to an OD 600 of 0.2, followed by the addition of Na 2 S to a final concentration of 0.2 mM, and then incubated for an additional 30 min. Samples were then centrifugation at 4°C, washed with cold phosphate-buffered saline (PBS), and cell pellets were stored at 80°C. Pellets were thawed on ice and resuspended in 1 ml of TRI Reagent (catalog no. TR-118; Molecular Research Center). Resuspended cells were placed in tubes containing 0.1-mm silica beads (Lysing matrix B tubes, catalog no. 6911-100; MP Biomedicals) and lysed in a bead beater (Bead Ruptor 24 Elite; Omni) at a rate of 6 m/s for 45 s twice, with a 5-min cooling on ice between runs. Then, 200 l of chloroform was added, followed by vigorous mixing and centrifugation for 15 min at 13,200 rpm. The top aqueous layer was removed to a new tube, and 70% ethanol was added at a 1:1 volume ratio. RNA purification was completed using the RNeasy minikit (catalog no. 74104; Qiagen) following DNase I treatment (catalog no. 79254; Qiagen). Next, 5 g of total RNA was subsequently digested with the DNA-free kit (catalog no. AM1906; Ambion) and diluted 5-fold. First-strand cDNA was synthesized using random hexamers (Quanta Biosciences) and a qScript Flex cDNA synthesis kit (catalog no. 95049-100; Quanta Biosciences). Reactions contained 10 l of 2ϫ Brilliant III Ultra-Fast SYBR green QPCR master mix (catalog no. 600882; Agilent), 2 l each of 2 M PCR primers (Table S2D), 0.3 l of 2 M ROX reference dye, and 6 l of diluted cDNA. Relative transcript amounts were measured using an MX3000P thermocycler (Stratagene) running SYBR green with a dissociation curve program and nor-Response of A. baumannii to Hydrogen Sulfide ® malized to the amount of 16S rRNA. The thermal profile consisted of 1 cycle at 95°C for 3 min, followed by 40 cycles at 95°C for 20 s to 59°C for 20 s. Subsequently, a dissociation curve starting at 55°C and going to 95°C in 0.5°C increments with a dwell time of 30 s was applied to assess the specificity of the reactions. Fold changes were calculated from threshold cycle (C T ) values by comparing the ratio of treated to nontreated wild-type cells after normalization to 16S rRNA. Two technical replicates of three biological replicates were measured for each condition, and the means Ϯ the standard deviations (SD) are reported. Transcript amounts were compared using a Student t test.
RNA sequencing. RNA samples were prepared in the same way as for the qRT-PCR with samples collected after 10 min after the addition of 0.2 mM Na 2 S at an OD 600 of 0.2. Samples were submitted to the Center of Genomics and Bioinformatics at Indiana University to conduct RNA-seq analyses. The RNA integrity number was determined with TapeStation before the rRNA was removed using a Ribominus transcriptome isolation kit (Invitrogen, catalog no. K1550), and a library was generated using a TruSeq stranded mRNA library prep kit (Illumina). The results of all RNA-seq experiments have been deposited in the GEO database under GenBank accession number GSE148264.
Quantitation of cellular thiols and persulfides. Cellular profiling of RSS was performed as previously described (29). Briefly, overnight cultures were inoculated 1:50 in LB medium for 1 h prior to inoculation at 1:100 into fresh LB medium, followed by growth at 37°C with shaking until the OD 600 reached ϳ0.2; Na 2 S was then added to the cell cultures at the indicated concentrations, and the cells were collected at the indicated times (time zero is before the addition of Na 2 S), washed with PBS, and stored at 80°C until processing. The cell pellets were thawed, resuspended in 100 l of labeling buffer (25 mM Tris [pH 8.0], 1 mM monobromobimane, and 0.1% Triton-X in 50% acetonitrile), and freezethawed six times using liquid N 2 and a 37°C water bath. Cellular debris was pelleted by centrifugation, and the supernatant was transferred to a new 1.5-ml tube containing 100 l of 15 mM methane sulfonic acid to terminate the labeling reaction. The samples were then filtered in a 0.22-m spin filter prior to ultraperformance liquid chromatography/mass spectrometry, as previously described. Analysis was performed using MassLynx software, and quantitation was done by an internal standard synthesized as previously described (29).
Proteomic analysis of wild-type and ⌬fisR A. baumannii strains. Cell cultures were prepared as described above for the RT-PCR experiments. Quadruplicate 10-ml cultures of wild-type and ΔfisR strains were grown aerobically at 37°C to an OD 600 of 0.2. Cultures were then collected by centrifugation for untreated cells or treated with 0.2 mM Na 2 S for 30 min, followed by centrifugation. All cells were washed with cold PBS and stored at -80°C. Pellets were thawed on ice and resuspended in 600 l of lysis buffer (25 mM HEPES [pH 7.4], 150 mM NaCl, 5 mM TCEP, EDTA-free protease inhibitor cocktail [1:500 dilution]). Resuspended cells were transferred to lysing matrix B tubes and lysed in a bead beater at a rate of 6 m/s for 45 s three times, with a 5-min cooling on ice in between runs. Samples were centrifuged at 13,200 rpm for 15 min at 4°C. The supernatant was transferred to a new 1.5-ml tube, and the protein concentration was quantified by a Bradford assay using a standard protocol. Portions (20 g) of protein were dried in SpeedVac concentrator and resuspended in 8 M urea in 100 mM ammonium bicarbonate. To these solutions, 5 mM TCEP and 50 mM iodoacetamide were added for 45 min at room temperature to reduce and alkylate the cysteine residues. Proteins were then precipitated by 20% (vol/vol) trichloroacetic acid and let sit at -20°C for 1 h. Samples were centrifuged at 15,000 rpm for 20 min at 4°C with the supernatant removed. The pellets were washed with cold acetone and centrifuged twice at 15,000 rpm for 20 min at 4°C. Samples were dried in a SpeedVac concentrator, resuspended in 100 l of 1 M urea in 100 mM ammonium bicarbonate, and digested overnight at 37°C with the addition of a 1:100 (wt/wt) ratio of trypsin. Peptides were desalted by using a C 18 Omix Tip (Agilent Technologies) according to a standard protocol and analyzed as described below.
Enrichment and identification of S-sulfurated proteins in A. baumannii. Wild-type A. baumannii was grown as described above for proteomic analysis. Pellets from 400-ml cultures were resuspended in 1 ml of lysis buffer (25 mM MES [morpholineethanesulfonic acid], 200 mM NaCl [pH 6.0], degassed), and five freeze-thaw cycles were performed with liquid nitrogen and a 37°C water bath. The resuspended pellet was transferred to lysing matrix B tubes and lysed three times in a bead beater at a rate of 6 m/s for 45 s, with a 5-min cooling on ice between runs. Samples were centrifuged at 13,000 rpm for 15 min at 4°C. The supernatant was transferred to a new tube and spun at 13,000 rpm for 15 min at 4°C one additional time with the supernatant saved as the cell lysate. The protein concentration of the cell lysate was measured using a standard Bradford assay, and 8 mg of protein was used for the following workup, as previously described (11).
Briefly, lysates were buffer exchanged to 8 M urea in 100 mM ammonium bicarbonate and incubated with 2 mg of biotinylated iodoacetamide (b-IAM; Thermo Scientific) in the dark for 1.5 h at 25°C. The sample was then buffer exchanged to 1 M urea in 100 mM ammonium bicarbonate, digested with 60 g of trypsin at 37°C overnight, desalted over a C 18 column (Sep Pak; Waters), dried, and suspended in 1 ml of 100 mM ammonium bicarbonate. Then, 1 ml of neutravidin polyacrylamide resin (Thermo Scientific) was equilibrated with 100 mM ammonium bicarbonate and incubated with digested sample at room temperature for 1 h. The resin was then washed with 75 ml of 2 M NaCl in 100 mM ammonium bicarbonate, followed by a 15 ml of 100 mM ammonium bicarbonate wash at 37°C. The S-sulfurated peptides bound to the resin were then selectively removed by reduction with 1 ml of 20 mM TCEP at 37°C for 1 h, and the supernatant was collected as elution 1. This step was repeated, and the supernatant was collected as elution 2. The two elution samples were pooled and alkylated with 100 mM iodoacetamide in dark for 45 min at room temperature. The sample was then desalted with an Omix C 18 tip (Agilent Technologies) and subjected to LC-MS/MS analysis as described below. Three biological replicates were used as treated (WTϩNa 2 S) and untreated (WT) samples. Sigma ratios ( R ) were calculated for each protein using the total area for the cysteine peptides. The fold changes of S-sulfurated peptides (WTϩNa 2 S/WT) were normalized to the protein abundance from data acquired in the proteomic analysis described above. The S-sulfurated peptides only detected in one condition (WT or WTϩNa 2 S) or proteins not identified in our no-enrichment proteomic analysis were assigned peak areas equal to the lowest detected peptide or protein to eliminate division by zero during normalization.

LC-MS/MS analysis of proteome and S-sulfurated proteins.
Peptides were injected into a nano-ACQUITY high-pressure liquid chromatography (HPLC) system coupled to an Orbitrap Fusion Lumos mass spectrometer (Thermo Scientific, Bremen, Germany). The peptides were separated using a nano-ACQUITY HPLC HHS T3 analytical column (75 m ϫ 150 mm, 1.8 m) using an acetonitrile-based gradient (solvent A, 0% acetonitrile and 0.1% formic acid; solvent B, 80% acetonitrile and 0.1% formic acid) at a flow rate of 400 nl/min. A 90-min gradient was applied as follows: 0 to 0.5 min, 0 to 14% B; 0.5 to 72 min, 14 to 50% B; 72 to 74 min, 50 to 100% B; 74 to 77 min, 100% B; 77 to 78 min, 100 to 0% B; 78 to 90 min, isocratic flow at 0% B. The electrospray ionization was carried out with a nanoESI source at a 260°C capillary temperature and 1.8-kV spray voltage. The mass spectrometer was operated in data-dependent acquisition mode with mass range 350 to 2,000 m/z. The precursor ions were selected for MS/MS analysis in Orbitrap with 3-s cycle time using higher-energy collisional dissociation (HCD) at a 35% collision energy. The intensity threshold was set at 5e3. The dynamic exclusion was set with a repeat count of 1 and an exclusion duration of 30 s. The resulting data were searched against an A. baumannii database (UniProt UP000094982, 3,780 entries) in Proteome Discoverer 2.1. Carbamidomethylation of cysteine residues was set as a fixed modification. Protein N-terminal acetylation, oxidation of methionine, protein N-terminal methionine loss, protein N-terminal methionine loss and acetylation, and pyroglutamine formation were set as variable modifications. A total of three variable modifications were allowed. A trypsin digestion specificity with two missed cleavages was allowed. The mass tolerances for precursor and fragment ions were set to 10 ppm and 0.6 Da, respectively.
Statistical rationale and bioinformatics analysis. Proteins detected in fewer than two replicates were excluded from our statistical analysis. The statistical analysis of these data was completed using an unpaired, two-tailed Student t test with Welch's correction. Functional information for the selected proteins were gathered from the National Center for Biotechnology Information (NCBI; https://www.ncbi .nlm.nih.gov/), BioCyc (https://biocyc.org/web-services.shtml), and Kyoto Encyclopedia of Genes and Genomes (KEGG; https://www.genome.jp/kegg/pathway.html) databases. Metabolism pathway information for the selected proteins were obtained from KEGG and pathway analysis of S-sulfurated proteins was performed using KEGG pathway mapper (https://www.genome.jp/kegg/mapper.html). Motif analysis of S-sulfuration sites was performed using pLogo (http://plogo.uconn.edu) (77). For each modified cysteine, a 21-amino-acid sequence containing the S-sulfurated cysteine, with 10 amino acids flanking sequences on either side of the Cys, was selected. A background data set was constructed similarly using all cysteines identified in the reference genome for A. baumannii ATCC 17978 (GenBank accession no. NZ_CP018664.1).
Data availability. All RNA-seq experiment results have been deposited in the GEO database under GenBank accession number GSE148264.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.

ACKNOWLEDGMENTS
Work presented in this article was supported by grants R35 GM118157 to D.P.G. and R01 AI101171 to E.P.S. and D.P.G. from the National Institutes of Health (NIH). B.J.C.