H 2 S and reactive sulfur signaling at the host-bacterial pathogen interface

Bacterial pathogens that cause invasive disease in the verte-brate host must adapt to host efforts to cripple their viability. Major host insults are reactive oxygen and reactive nitrogen species as well as cellular stress induced by antibiotics. Hydrogen sulfide (H 2 S) is emerging as an important player in cytoprotection against these stressors, which may well be attributed to downstream more oxidized sulfur species termed reactive sulfur species (RSS). In this review, we summarize recent work that suggests that H 2 S/RSS impacts bacterial survival in infected cells and animals. We discuss the mechanisms of biogenesis and clearance of RSS in the context of a bacterial H 2 S/RSS homeostasis model and the bacterial transcriptional regulatory proteins that act as “ sensors ” of cellular RSS that maintain H 2 S/RSS homeostasis. In addition, we cover fluorescence imaging – and MS – based approaches used to detect and quantify RSS in bacterial cells. Last, we discuss proteome persulfidation ( S -sulfura-tion) as a potential mediator of H 2 S/RSS signaling in bacteria in the context of the writer-reader-eraser paradigm, and progress toward ascribing regulatory significance to this widespread post-translational modification.

Bacterial pathogens that cause invasive disease in the vertebrate host must adapt to host efforts to cripple their viability. Major host insults are reactive oxygen and reactive nitrogen species as well as cellular stress induced by antibiotics. Hydrogen sulfide (H 2 S) is emerging as an important player in cytoprotection against these stressors, which may well be attributed to downstream more oxidized sulfur species termed reactive sulfur species (RSS). In this review, we summarize recent work that suggests that H 2 S/RSS impacts bacterial survival in infected cells and animals. We discuss the mechanisms of biogenesis and clearance of RSS in the context of a bacterial H 2 S/RSS homeostasis model and the bacterial transcriptional regulatory proteins that act as "sensors" of cellular RSS that maintain H 2 S/RSS homeostasis. In addition, we cover fluorescence imaging-and MS-based approaches used to detect and quantify RSS in bacterial cells. Last, we discuss proteome persulfidation (S-sulfuration) as a potential mediator of H 2 S/RSS signaling in bacteria in the context of the writer-reader-eraser paradigm, and progress toward ascribing regulatory significance to this widespread post-translational modification.
Infectious disease is a global and significant threat to human health. There is an increasingly urgent need to develop new antimicrobial strategies to combat these increasingly drug-resistant and life-threatening pathogens (1,2). One important approach to do this is to understand bacterial adaptation to the myriad of host immune responses that have evolved to clear bacterial infections. For example, transition metal homeostasis (metallostasis) (3) effectively controls the metalation status of the proteome (4,5). Upon infection, the host actively disrupts metallostasis by restricting access (6)(7)(8) or intoxicating cells with metals (9, 10) to limit bacterial growth. Pathogens, in turn, adapt by employing specialized transcriptional regulators, metallosensors, that sense metals and regulate the expression of genes encoding proteins that collectively maintain bioavailable metal in a range compatible with physiological needs (Fig. 1A, top) (11)(12)(13)(14). In an analogous fashion, bacteria encode specialized transcriptional regulators that sense oxidized or "reactive" sulfur species (RSS), derived from hydrogen sulfide (H 2 S) (15)(16)(17)(18)(19)(20). As cellular concentrations of RSS rise, RSS sensors turn on the expression of genes that encode enzymes that reduce cellular loads of H 2 S/RSS to avoid H 2 S toxicity and overpersulfidation of the metabolome and proteome (Fig. 1A, bottom). These RSS sensors, like metallosensors, control H 2 S/RSS homeostasis, allowing bacterial cells access to these molecules at low concentrations to meet physiological needs. In the infected host, H 2 S and RSS are derived from host cell metabolism, from commensal bacteria in polymicrobial communities, or from the pathogen itself. Recent studies that build on prior work (21) suggest that bacterial H 2 S biogenesis may well be a clinically important adaptive response during infections (22)(23)(24)(25)(26).
A second feature that is common to metallostasis and H 2 S/ RSS homeostasis, beyond the sensors themselves, is the concept of speciation (Fig. 1B). In metallostasis, speciation defines the metallome, or all coordination complexes, both small molecule and protein, and oxidation states of all transition metals in the cell (Fig. 1B, top). Metallosensors surveil the cytoplasm for some specific feature of the metallome (e.g. zinc in exchange-labile complexes) and alter gene expression upon metal binding. In H 2 S/RSS homeostasis, speciation is defined by the components of the RSS pool, which encompasses organic and inorganic molecules containing sulfur in oxidation states higher than H 2 S, many of which contain sulfur-bonded or "sulfane" sulfur ( Fig. 1B, bottom) (27). Analogous to a metallosensor, known RSS sensors specifically surveil the cytoplasm for a particular feature of the RSS pool, in this case sulfane sulfur (15)(16)(17)(18)(19)(20).
In this review, we summarize the biogenesis and clearance of H 2 S/RSS and the potential role these molecules play in bacterial infections. In addition, we discuss the molecular mechanisms of RSS sensors that maintain H 2 S/RSS homeostasis in bacteria. Elucidation of how H 2 S/RSS are leveraged in bacteria at the host-pathogen interface relies on the development of molecular tools to identify, detect, and quantify H 2 S and RSS as well as small-molecule probes to generate these species in vitro or in vivo. Last, we discuss recent efforts to detect and understand the regulatory significance of protein persulfidation (S-sulfuration) in bacteria.
Hydrogen sulfide and reactive sulfur species in bacteria H 2 S is an electron-rich molecule historically well-known to drive photosynthesis (28) and energy metabolism in sulfideoxidizing and sulfate-or sulfite-reducing microorganisms (29,30). In 2011, Nudler and co-workers (21) reported that endogenously synthesized H 2 S or application of exogenous sulfide salts protected multiple bacterial pathogens against a broad array of mechanistically distinct antibiotics when grown in culture. This initial report, despite few insights into a possible mechanism, suggested that H 2 S might have beneficial properties in human disease-causing microorganisms and has thus inspired Figure 1. Set-point homeostasis models and speciation in bacteria. A, transition metal homeostasis (top) is orchestrated by a panel of metal-specific sensors that prevent metal starvation or toxicity by regulating the expression of proteins involved in the uptake, efflux, storage, or allocation of metals in cells (3,4,(11)(12)(13)(14). Transcriptional response curves are shown for a pair of sensors that detect a specific metal (e.g. Zn II ). These dual sensors collaboratively control metal bioavailability in a concentration range that is compatible with cellular physiology (gray box). H 2 S/RSS homeostasis (bottom) is achieved by a single RSS sensor that transcriptionally regulates the expression of enzymes involved in the biogenesis, clearance, transport, and assimilation of H 2 S/RSS (16)(17)(18)(19)(20). The transcriptional response of an RSS sensor detects a concentration range (gray box) that prevents cellular toxicity, while maintaining access to H 2 S/RSS that is physiologically beneficial at lower concentrations. B, metal speciation (top) of first row, late d-block transition metals is defined by the metallome, a descriptor of all oxidation states and coordination complexes in the cell, ranging from exchange-labile small-molecule metal complexes to protein cofactors (shown in cartoon form). Reactive sulfur speciation (bottom) is defined by all inorganic and organic small molecules that harbor sulfur atoms in oxidation states more positive than -2 (see key) (27) and are collectively termed reactive sulfur species (RSS).
Physiological conditions for the production, regulation, and signaling of H 2 S/RSS in bacteria H 2 S and the gut microbiome The gut microbiome is a complex, nutrient-rich environment and host to well over 100 bacterial species (65,66). The microorganisms that inhabit this niche are a significant endogenous source of sulfur-containing compounds and H 2 S, the Endogenous production of H 2 S and cross-talk between H 2 S/RSS and ROS/RNS in bacteria. A, the reverse transsulfuration pathway synthesizes L-cysteine from L-homocysteine via the intermediacy of L-cystathionine, which is catabolized by CAT, whose product is utilized by 3MST to generate pyruvate and H 2 S (31)(32)(33)(34)(35)(36)(37)(38). L-Homocysteine is also the immediate precursor to L-methionine. Both CBS and CSE catalyze a number of additional reactions with alternative substrates (red arrows) that generate endogenous H 2 S (43), illustrated in the dashed box. CBS, cystathionine-b-synthase, CSE, cystathionine-g-lyase; MAT, methionine adenosyltransferase; MS, methionine synthase; SAHH, SAH hydrolase. B, small-molecule cross-talk between H 2 S/RSS and RNS and ROS. C, direct reaction of HSwith protein nitrosothiols or disulfides and sulfenic acids induced by RNS and ROS, respectively, all result in the formation of the thiol persulfide with release of HNO, protein thiol (RSH), and water, respectively. Reaction of HSwith sulfenic acids is believed to protect these protein thiols from irreversible overoxidation, indicated by the block arrow (right).
latter estimated to range from ;0.2 to 2.4 mM (Fig. 3A) (67). Methionine catabolism to cysteine via the reverse transsulfuration pathway ( Fig. 2A) and the catabolism of organic sulfonates, notably taurine, are known to be catalyzed by gut microbiota (41,42,(68)(69)(70)(71)(72). This niche is also home to sulfate-reducing bacteria that are responsible for significant production of H 2 S (73,74), in addition to the reduction of tetrathionate and thiosulfate to H 2 S that also occurs in the gut (75). H 2 S produced by the gut microbiota is oxidized by gastrointestinal epithelial cells to thiosulfate and tetrathionate (76,77); these molecules in turn are utilized by the microbiota as electron acceptors, resulting in a symbiotic relationship derived from interconversion of sulfur species (Fig. 3A) (72). Perturbation of this symbiotic relationship results in the accumulation of H 2 S now linked to several gut-derived diseases (78)(79)(80). Interestingly, gut inflammation caused by the pathogen Salmonella enterica serovar Typhimurium results in increased production of tetrathionate from the oxidation of thiosulfate by gut inflammation-derived ROS, which this bacterium uses as an alternate electron acceptor, thus providing a growth advantage in this niche (81).

H 2 S at the host-pathogen interface
There are two major physiological conditions within the infected host where H 2 S and downstream RSS may enhance bacterial survival in what is otherwise a hostile microenvironment. These are (i) resistance against myriad oxidative stressors and antibiotic challenge and (ii) H 2 S/RSS-dependent regulation of biofilm dynamics. The host immune system produces diverse ROS, including O 2 (82)(83)(84), hydrogen peroxide (H 2 O 2 ) (85), and hydroxy radical (OH · ) (86) to combat bacterial infections (87). Antibiotics are also thought to induce generalized oxidative stress (88)(89)(90)(91), although this has been widely debated in recent years (92,93). The first documentation that H 2 S impacted antibiotic resistance in E. coli was published 47 years ago (94). Renewed interest came in 2011 when it was demonstrated that bacterially derived H 2 S conferred resistance to a broad range of antibiotics in several bacterial pathogens (21). Bacterially derived NO · has also been found to provide protection against antibiotics by the same group (95) perhaps because of H 2 S/NO · cross-talk, which has only been investigated in recent years (Fig. 2B) (53)(54)(55)(56). It was not until 2014 that RSS were shown to function as antioxidants in mammalian cells (47), which may partly explain H 2 S-enhanced bacterial resistance to antibiotics (21). RSS have been detected in a number of bacterial pathogens (17,20,53,96) and are the subject of ongoing work to better understand the role of RSS in the bacterial response to the host immune system.
Recent work has investigated the mechanism by which H 2 S/ RSS might confer antibiotic resistance and protection against ROS and sulfide toxicity (Fig. 3B). In E. coli, increased H 2 S results in a respiratory flux switch from that of the primary cytochrome bo oxidase to the alternate cytochrome bd oxidase, a copper-free enzyme that is far less susceptible to inhibition by H 2 S (26,97). Although cytochrome bd oxidase does not pump protons, it still enables aerobic metabolism and robust growth. This respiratory switch in response to H 2 S may well occur in several other bacterial pathogens, including Acinetobacter baumannii (20) and Mycobacterium tuberculosis; in the latter case, low levels of H 2 S enhance the respiration, energy production, and survival of M. tuberculosis in infected mice (25). Further studies are required to establish the generality of this adaptive response across a wider range of organisms. Others have posited that H 2 S-mediated cytoprotection occurs via sequestration of the prooxidant free Fe(II) (23,98); however, this remains incompletely understood.
A number of recent reports have described a potential role of H 2 S at the host-pathogen interface beyond protection against antibiotic and oxidative stress (22,24,25,99). In infected macrophages and in mice, Helicobacter pylori was found to induce the expression of the host transsulfuration pathway enzyme CSE ( Fig. 2A), resulting in increased cystathionine production that enhances H. pylori survival in these models (99). Any connection of cystathionine to host or bacterially derived H 2 S was not elucidated in this work. S. aureus and E. coli strains lacking enzymes involved in H 2 S biogenesis are more readily cleared in infected macrophages and are less resistant to leukocyte-mediated killing in a burn-infection model (22). In addition, E. coli strains lacking the H 2 S-generating enzyme 3MST, when challenged with antibiotics, give rise to a suppressor mutation that recovers H 2 S biogenesis via up-regulation of the single-domain sulfurtransferase PspE (24). Together, these studies suggest that H 2 S biogenesis reduces the efficacy of antibiotics and that up-regulation of H 2 S may be a clinically important adaptive response during infections. These studies support the proposal that H 2 S functions as an infection-relevant antioxidant or proantioxidant, in the latter case, as a precursor to oxidized RSS (47).
Whereas H 2 S/RSS-dependent regulation of biofilm dynamics remains largely unknown, recent studies suggest a potential connection. Biofilms are often polymicrobial communities that assemble on both abiotic (e.g. catheters and implants) and biotic (e.g. cells and cell debris) surfaces while conferring increased resistance to antibiotics (100)(101)(102). Cells near the base of biofilm structures are often nutrient-poor, and some reside at an oxic/anoxic boundary. Low-O 2 (hypoxic) conditions can also result from increased O 2 consumption by host immune cells to produce superoxide anion (O 2 ) (103, 104). In these low-O 2 regions, bacteria respire via reduction of nitrate (NO À 3 ), producing NO · on pathway to nitrous oxide (N 2 O) and dinitrogen (N 2 ) (105). These nitrogen-containing species have been reported to lead to biofilm dissemination of S. aureus (106) and P. aeruginosa (107,108), consistent with an impact on biofilm dynamics.
Redox homeostasis is also implicated in proper biofilm formation in P. aeruginosa (109), whereas cysteine and GSH-deficient uropathogenic E. coli exhibit dysregulated biofilm formation that is restored upon the addition of exogenous thiols (110). Although the connection between biofilm regulation and H 2 S/RSS homeostasis is largely speculation at this point, H 2 S has been detected in cystic fibrosis sputum, a complex biofilm (111), and H 2 S has been found to promote formation of biofilms by intestinal microbiota while reducing the proliferation of planktonic bacterial cells (112). We recently characterized the biofilm growth-associated repressor, BigR, in A. baumannii as an RSS sensor (20,113), as previously characterized in plant pathogens (114)(115)(116). That work also identified two transcriptional regulators in A. baumannii known or projected to be involved in biofilm regulation that were characterized by significantly increased protein persulfidation mediated by exogenous sulfide (20). Whereas these studies suggest that H 2 S/ RSS homeostasis impacts biofilm dynamics, more studies are needed to better understand this connection at a mechanistic level.

Biogenesis and clearance of organic RSS in bacteria
The endogenous production of H 2 S in bacteria suggests that more oxidized forms of sulfur may be present in cells and formed via enzymatic and possibly nonenzymatic mechanisms (Fig. 4, A-C). The extent to which these pathways, particularly nonenzymatic routes, contribute to RSS pools in bacteria is not known and may well differ among organisms. Emerging evidence in mammalian systems demonstrates the role of ferric (Fe III )-heme in the oxidation of H 2 S, which reduces the Fe III to Fe II and forms the one-electron oxidized radical HS · upon dissociation ( Fig. 4A) (117)(118)(119)(120)(121). Recent work reveals that this mechanism is used to reactivate enzymes requiring a catalytically active ferrous heme from the inactive ferric state, formed during turnover (121). Additionally, HS · can be formed by the reaction of H 2 S with superoxide radical anion O 2 or with cysteine-coordinated Zn(II) sites in proteins (122). Fe III -heme has also been shown to result in formation of thiosulfate and hydropolysulfide species in mammalian systems ( Fig. 4A) (117,118). Formation of organic thiyl radical, RS · , and related organic polysulfide species may occur via similar chemistry, supported by recent work using an LMW thiol for the reactivation of a catalytically active ferrous heme ( Fig. 4B) (121). The extent to which heme-based biogenesis of RSS occurs in bacteria is not yet known.
An important enzymatic route to the biogenesis of RSS in bacteria is the sulfide:quinone oxidoreductase (SQR) (76,(123)(124)(125). SQR catalyzes the two-electron oxidation of H 2 S to sulfane sulfur fixed as organic and inorganic per-and polysulfides ( Fig. 4B), concomitant with reduction of the quinone pool (126). This enzyme may well provide a source of electrons for the alternative cytochrome bd oxidase in organisms that encode this alternate oxidase, analogous to that observed for SQR with complex III/IV when the concentration of H 2 S is low ( Fig. 3B) (76). In organisms (e.g. Enteroccocus faecalis) that do not appear to encode an SQR but where RSS have been detected and quantified (17), the mechanism of RSS biogenesis is not known, and may well suggest a role for nonenzymatic or as yet uncharacterized enzymatic processes in these organisms. In addition to SQR, recent work in A. baumannii reveals that 3MST may also contribute to pools of LMW persulfides, although there are clearly other contributors (20).
"Orphan" STRs, which we define as not yet connected to any biosynthetic pathway, particularly those regulated by RSS sensors in bacteria, may well play roles in sulfide detoxification or assimilation (17,135), but their biological functions remain enigmatic. This remains a significant challenge in the field. RSS sensor-regulated STRs are often kinetically characterized in vitro as sulfurtransferases from a thiosulfate donor to a CN 2 acceptor; however, their physiological donors and acceptors, whether they be small molecules or proteins, have generally not been identified, and any role in targeted transsulfuration has not been established (Fig. 4C). Recently, a single cysteine peroxiredoxin (a major H 2 O 2 -detoxifying enzyme) characterized by a long-lived sulfenylated intermediate was shown to rapidly react with H 2 S to form a protein persulfide, which participated in persulfide transfer to a thiol acceptor (136). This suggests that peroxiredoxins may function in transsulfuration, but this requires further investigation (Fig. 4C). Similarly, the extent to which small-molecule RSS species, particularly those containing sulfane sulfur, participate in transsulfuration reactions with each other, LMW thiols, or even protein thiols is largely unknown (Fig. 4C).

RSS clearance enzymes
In addition to the biogenesis of RSS, a number of bacterial enzymes have been characterized that function in their clearance (Fig. 4D). A well-known player in the clearance of organic persulfides is persulfide dioxygenase (PDO), which harbors a mononuclear, nonheme Fe II center (76,(137)(138)(139)(140). In bacteria, PDOs have been characterized as single or multidomain enzymes, and the presence of additional domains appears to impact the distribution of products. All PDOs, regardless of their domain organization, use molecular oxygen to oxidize the terminal sulfur of an RSSH substrate to sulfite, which, for a single-domain PDO, is the final product. Some PDOs have an appended STR domain, and these have been designated PDOrhodanese fusion proteins (PRFs) (Fig. 4D). In the case of the PRF characterized in Burkholderia phytofirmans, the C-terminal rhodanese domain generates the GSH persulfide substrate that the PDO domain then oxidizes to sulfite (138). In contrast, the multidomain PRF CstB from S. aureus oxidizes two equivalents of persulfide substrate to thiosulfate as the final product; the C-terminal rhodanese domain also possesses transsulfuration and thiosulfate transferase activity (139). In contrast to the oxidative chemistry of PDOs, E. faecalis encodes a CoA disulfide reductase-rhodanese homology domain fusion protein  (43,51). In A, the biogenesis of thiosulfate shown is a schematic rendering only. C, transsulfuration reactions that may impact sulfur speciation and assimilation catalyzed by sulfurtransferases (STR) (126), peroxiredoxins (Prx) (136), or other thiol-containing enzymes or via nonenzymatic interconversion among organic LMW thiol/persulfides or inorganic RSS. D, enzymatic clearance of RSS known to occur in bacteria (17,76,138,139). Red sulfurs indicate inorganic sulfur additions. Characterized enzymes are represented by blue ovals, whereas incompletely understood enzymatic reactions are shown as gray ovals and dashed arrows. CoAPR, CoA persulfide reductase; CstB, S. aureus cst operon-encoded persulfide dioxygenase-rhodanese fusion; 3MST, 3-mercaptopyruvate sulfurtransferase; PDO, persulfide dioxygenase; PRF, persulfide dioxygenase-rhodanese fusion; Q/QH 2 , quinone; SQR, sulfide:quinone oxidoreductase.

Regulatory sensing of RSS in bacteria
The discovery of endogenous H 2 S production and pathways for the biogenesis and clearance of RSS in bacteria requires a mechanism to establish cellular H 2 S/RSS homeostasis. This is mediated by RSS sensors (Fig. 1A, bottom). We and others have discovered and characterized structurally diverse transcriptional regulators that react with RSS to drive transcriptional derepression or activation of genes encoding common sulfide detoxification or oxidation enzymes described above. These RSS sensors are widespread and have been identified in both Gram-positive and Gram-negative organisms. They include CstR from S. aureus (15,16) and E. faecalis (17), SqrR from Rhodobacter capsulatus (18,113), the SqrR homolog BigR from Xylella fastidiosa (114)(115)(116) and A. baumannii (20), and FisR from Cupriavidus pinatubonensis (19) and A. baumannii (20).

CstR
The CsoR-like sulfurtransferase repressor, CstR, is a member of the CsoR (copper-sensitive operon repressor) family of transcriptional repressors (144) and was first discovered in S. aureus (15,16). S. aureus CstR regulates the cst operon encoding a multidomain STR (CstA), a PRF (CstB), and a type II SQR, rather analogous to the well-studied mitochondrial sulfide detoxification system (76,123,135,139). The enzymes encoded by the S. aureus cst operon oxidize sulfide to thiosulfate via persulfide intermediates and have been extensively reviewed elsewhere (126). Recently, we characterized CstR from a second human pathogen E. faecalis that regulates a cst-like operon encoding two orphan STRs and a CoAPR (Fig. 5A, left) (17). CstR is homotetrameric in solution and is anticipated to harbor four peripherally arranged dithiol "sensing" sites found between protomers in a dimer-of-dimers D 2 -symmetric architecture, like CsoR (Fig. 5A, middle) (145). CstR represses transcription in the reduced state, whereas reaction with sulfane sulfur-containing RSS, but not H 2 S itself, results in a mixture of di-, tri-, and tetrasulfide interprotomer cross-links that negatively regulates DNA operator-promoter binding, allowing for transcription initiation (Fig. 5A, right) (16). In both S. aureus and E. faecalis, the operons are also inducible by the nitroxyl (HNO) donor Angeli's salt, but not by an NO · donor (17,53). In S. aureus this may be the result of increased cellular per-and polysulfides after the addition of nitroxyl to cells, supporting the notion of H 2 S/NO · cross-talk in this organism (Fig. 2B). Although CstR appears selective for H 2 S and RSS in cell-based transcription reactions (16), the structure/reactivity factors that enforce this apparent specificity are not yet known.

SqrR and homologs
The sulfide:quinone reductase repressor, SqrR, was originally characterized in the photosynthetic bacterium R. capsulatus and is responsible for the regulation of 45% of all sulfide-responsive genes in this organism, including an SQR (18). SqrR is a member of the arsenic repressor (ArsR) superfamily (146) in striking structural contrast to CstR. SqrR adopts the ArsR family a1-a2-a3-a4-b1-b2-a5 "winged-helical" dimeric fold, where one Cys in the a2 helix and one Cys in the a5 helix from the same subunit create a pair of dithiol RSS-sensing sites on the dimer (Fig. 5B, middle) (113). Other RSS-responsive ArsR family repressors include the biofilm growth-associated repressor BigR, characterized in X. fastidiosa (114)(115)(116) and A. baumannii (20), and PigS characterized in Serratia spp. (147). A. baumannii BigR regulates a secondary RSS detoxification system that includes a second PDO and two transmembrane proteins proposed to be involved in the transport of sulfur-containing molecules (20,147). Although PigS has not been functionally characterized as an RSS sensor, it regulates several enzymes known to function in H 2 S/RSS clearance, including a single-domain PDO and a CoAPR encoded by coaP (Fig. 5B, left) (147). It is interesting to note that PigS and its regulon are part of the larger PigP regulon involved in the biosynthesis of the antibiotic prodigiosin (Pig), thus implying a regulatory connection between antibiotic biosynthesis and H 2 S/RSS homeostasis. The ArsR-family RSS sensors that have been functionally characterized behave analogously to CstR, functioning as repressors in the reduced state and dissociating from the DNA upon reaction with sulfane sulfur-containing RSS, to readily form nearly exclusively tetrasulfide (SqrR) and pentasulfide (BigR) bridges, respectively (Fig. 5B, right) (113). In contrast to CstR, these (poly)sulfur bridges are formed within a subunit, and although other linkages are made, they are far less abundant compared with the mixture of products found in CstR (16).
Recent work from our laboratory utilized SqrR as a model dithiol transcriptional regulator to investigate the structural and reactivity features that govern its oxidant selectivity and specificity (113). Indeed, SqrR is specific for sulfane sulfur and only forms a disulfide when treated with potent diazirene electrophile TMAD (diamide), but not with more common cellular oxidants including GSH disulfide or H 2 O 2 . Whereas this low reactivity toward cellular oxidants can be partially explained by the relatively high apparent pK a of the dithiol pair, the high selectivity toward RSS is enforced by structural features of SqrR in various oxidation states. These structures reveal a high energetic barrier to form the disulfide because of large rearrangements that must occur in order to form the disulfide; in addition, the disulfide is not on pathway to form the major tetrasulfide product. In contrast, formation of the tetrasulfide does not require large structural rearrangements; on the contrary, this linkage results in the collapse of the dithiol pocket that completely shields the tetrasulfide linkage from solvent. This study demonstrates that SqrR-like dithiol-based repressors achieve high RSS specificity from the conformational landscape of the protein ensemble, which favors installation of a PTM that minimizes local structural frustration (113). It will be interesting to determine whether these principles apply to other structural classes of RSS sensors or if there are additional determinants that dictate their specificity.

FisR
A third structural class of RSS-sensing transcriptional regulators first characterized in C. pinatubonensis (19), and more recently in A. baumannii (20), is FisR (Fis family transcriptional regulator). In both bacteria, FisR transcriptionally activates the expression of a PDO and SQR and a putative sulfite/sulfonate effluxer, TauE (148), in only A. baumannii (Fig. 5C, left). In contrast to CstR and SqrR-like RSS sensors, FisR is a canonical s 54 -dependent transcriptional activator that harbors an N-terminal regulatory domain, a central AAA1 ATPase domain, and a C-terminal DNA-binding domain (Fig. 5C, middle) (149). In C. pinatubonensis FisR, reaction with inorganic RSS appears to result in the formation of di-and tetrasulfide cross-links between two cysteine residues in the regulatory domain, which in turn stimulates the ATPase activity of the central AAA1 domain, which likely activates hexameric assembly and promoter melting by s 54 -RNA polymerase holoenzyme (Fig. 5C, right) (19). In A. baumannii FisR, these cysteines are not present, and as a result, H 2 S/RSS is likely sensed using an alternate mechanism (20), which includes heme-based (117-121) and mononuclear, nonheme Fe-based RSS-sensing regulatory models (150)(151)(152).

Chemical tools for generation, detection, and quantification of RSS
To understand the role of H 2 S and RSS in signaling at the host-pathogen interface, tools must be available that allow for the generation, detection, and quantification of these species in vivo. The type and number of molecular probes used for the generation of H 2 S and RSS have substantially increased over the past several years, and they are now being used in bacteria to provide critical insights into H 2 S signaling in these organisms (22,25,26). Fluorescence-based probes provide rapid and sensitive detection of H 2 S or sulfane sulfur with several options now commercially available. In addition, recent efforts to quantify H 2 S and RSS in complex cellular mixtures have provided  (16). A structural model of CstR using Protein Data Bank entry 5FMN as a template reveals an all a-helical protein with four peripheral dithiol-sensing sites in the tetrameric structure (middle). The reaction products of CstR with RSS reveal a mixture of di-, tri-, and tetrasulfide interprotomer linkages (right). B, SqrR (operon not shown), like its homologs BigR and PigS, function as transcriptional repressors in their reduced forms and regulate the expression of typical sulfide oxidation and detoxification enzymes (left) (18,20,116,147). The homodimeric structure of SqrR reveals a pair of dithiol-sensing sites (middle, Protein Data Bank entry 6O8L), which readily form intraprotomer sulfur bridges of four (SqrR) or five (BigR) sulfur atoms upon reaction with RSS (right) (113). C, FisR is a s 54 -dependent transcriptional activator and activates the expression of a sulfide detoxification operon that is similar to that encoded by the cst operon (left). FisR is organized into three domains (regulatory, ATPase, and DNA-binding domain), but to date, there are no structures of a functionally characterized RSS-sensing FisR (middle). C. pinatubonensis FisR reacts with RSS in vitro to form a mixture of di-and tetrasulfide linkages that weakly stimulate the ATPase activity, resulting in transcriptional activation from s 54 -RNA polymerase (RNAP)-transcribed promoters (right) (19). The regulatory mechanism operative in A. baumannii FisR has not yet been determined (20). new insights into this process. As many of these molecular tools have been extensively reviewed elsewhere (153)(154)(155)(156)(157)(158)(159), here we provide only a summary of the available approaches, while pointing out specific challenges to their use.

H 2 S and RSS donors
H 2 S donors largely fall into three main classes. These are hydrolysis-based, thiol-dependent, and carbonyl sulfide (COS)based donors (Fig. 6A). Hydrolysis-based donors function over a wide range of pH, with GYY4137 as a widely used and commercially available donor employed by several groups to study the mechanism of H 2 S cytoprotection in bacteria (22,25,160).
Concerns over the relatively slow release rates have led to second-generation GYY4137 derivatives, including JK-2, that more efficiently release H 2 S at physiological pH (161). The use of hydrolysis-based donors requires careful consideration of the pH dependence and kinetics of H 2 S release, and many donors are not commercially available. Thiol-dependent H 2 S donors are attractive tools because of their use of typically cellabundant cellular reducing thiols, including GSH and cysteine (Fig. 6A). However, some of these probes are quite slow, require high concentrations of thiol, or are activated by a specific thiol (154). As the type and concentrations of thiols in bacteria have only recently been investigated and only in a small sampling of bacteria (17,20,53,96), this potentially limits the broad applicability of such probes in bacteria. In addition, the use of cellular thiols to activate compounds may disrupt the cellular thiol/disulfide redox balance, leading to physiological impacts not specifically due to H 2 S release. Furthermore, some thiolactivated donors proceed through a highly reactive persulfide intermediate, making it difficult to attribute a physiological impact to the persulfide intermediate or to H 2 S itself. Last, COS-based donors have been designed with various release mechanisms and rely on an endogenous carbonic anhydrase to produce H 2 S in high catalytic yields from probe-dependent release of COS (Fig. 6A) (162). The use of these compounds, like other enzyme-activated H 2 S donors (26), obviously relies on broad cell permeability and constitutive expression of a carbonic anhydrase (163), neither of which has been systematically investigated in bacteria.
Whereas H 2 S donors have been extensively developed over the last 10 years, RSS donors typically found in the form of persulfide donors are comparatively less so (153,158). For in vitro chemical reactivity studies, we and others have relied on in situ thiol persulfide generation from the reaction of excess H 2 S with a disulfide, resulting in a mixture containing the thiol persulfide that is not easily separated from the remaining reactants of H 2 S, disulfide, and thiol (45). Although in situ thiol persulfide generation may also result in formation of the more stable trisulfide species (164), we observed no reaction of SqrR (Fig. 5B) with cysteine trisulfide (113). Furthermore, whereas this work also revealed somewhat faster kinetics of SqrR with the in situ generated persulfide versus a persulfide donor analog of Snitrosoacetyl-penicillamine (SNAP), the SNAP analog persulfide donor has the added benefit that it is soluble in aqueous solution and spontaneously generates the persulfide species after S to N carbonyl transfer (Fig. 6B) (165). In addition, Wang and co-workers (166,167) recently developed a series of esterasesensitive persulfide donors, including GSH persulfide, whose physiological effects have thus far only been studied in mammalian cells. These cell-permeable persulfide donors generate persulfide species under physiological conditions with minimal disruption to cellular redox status (i.e. they are not thiol-activated, an important feature).
ROS-activated H 2 S and persulfide donors have also garnered recent attention (Fig. 6B) (154,158) because these types of probes are particularly useful for studying ROS/H 2 S cross-talk, which is likely relevant at sites of infection. These donors are specific for H 2 O 2 activation and react readily to generate H 2 S or a persulfide species. Recently, two groups have developed H 2 O 2 -specific persulfide-generating probes that exhibit greater cytoprotective effects in cells challenged with oxidative stress than H 2 S only-generating probes, such as GYY4137 (168,169). This finding is consistent with the idea that RSS rather than H 2 S per se are important effector molecules of H 2 S signaling and cytoprotection and merits further study in bacterial cells.

Detection of H 2 S/RSS
Although historically many groups have utilized methylene blue (170) or lead acetate (21) paper strips to measure H 2 S in growing bacterial cells, the latter is not quantitative, and both methods are unable to detect RSS. More recently, several groups have developed sensitive fluorescent probes for detection of H 2 S and RSS, and this has been comprehensively reviewed elsewhere (155,157,159). Of these tools, the WSP and SSP series of fluorescent molecules that detect H 2 S and sulfane sulfur, respectively, are most commonly used in both mammalian and bacterial cells (Fig. 6C) (24,77). When used by themselves, these probes cannot be used to perform absolute quantitation of H 2 S/RSS; however, one group recently coupled SSP4 cell labeling with MS to quantitate sulfane sulfur, thus establishing the possibility of making these optical methods quantitative (171). This approach, however, would only allow for quantitation of total sulfane sulfur without molecule-specific identification. Regardless of their inability to provide absolute quantitation, these probes provide a rapid and simple readout for production of H 2 S/RSS appropriate for comparing bacterial strains and growth conditions. In addition, the SSP series provide a rapid means for quantifying sulfane sulfur from enzymatic activity in vitro, a strategy recently used to measure RSS biogenesis and clearance activities of catalase and superoxide dismutase, respectively (172,173).

Quantification of H 2 S/RSS in cell lysates
Isotope-dilution LC-MS-based methods to detect and quantify H 2 S and RSS have considerable advantages over optical imaging modalities, including the ability to perform molecular identification and absolute quantitation; however, these methods are not easily applied to living bacterial cells and therefore rely on post-cell growth sample work-up and chemical derivatization of RSS. We and others have traditionally employed electrophilic trapping by monobromobimane (MBB) with quantitation originally via fluorescence detection and more recently via the addition of isotopically labeled internal standards necessary for LC-MS quantitation (Fig. 6D) (17,20,47,53,96). It was recently shown that quantitation of H 2 S by MBB is sensitive to small variations in pH, alkylation time, and temperature, thus emphasizing the importance of controlling reaction conditions when comparing samples (174). Bogdandi and co-workers (175) also showed that the electrophilicity of the alkylating agent also impacts quantitation of RSS to varying degrees. N-Ethylmaleimide, the most electrophilic agent tested, resulted in cleavage of per-and polysulfide chains, whereas b-(4-hydroxyphenyl)ethyl iodoacetamide (HPE-IAM) (Fig. 6D) showed little to no cleavage under the same experimental conditions. MBB has an intermediate impact resulting in slightly lower yields of per-and polysulfide species relative to HPE-IAM but exhibited significantly better quantitation than N-ethylmaleimide. As a result, many groups now use HPE-IAM for quantification of RSS (175)(176)(177).
A more general consideration of electrophile-based trapping approaches for RSS quantitation is that these methods provide only a snapshot of cellular RSS speciation, particularly in light of our incomplete understanding of sulfane sulfur "scrambling" that will conspire against any method of quantitative analysis. Better understanding of this will provide complementary information to quantitative techniques and valuable insights on the lifetime and speciation of RSS in cells.

H 2 S signaling via protein S-sulfuration
PTMs of cysteine thiols are known to impact signaling and play regulatory roles in proteins (43,(178)(179)(180). Persulfidation of cysteine residues can occur via reduction of nitrosated or sulfenylated protein thiols, while also protecting these latter thiols from overoxidation (59,60). Persulfidation may also function in H 2 S signaling as a regulatory modification by introducing altered chemistry or "blocking" of active-site thiols, thiols in a regulatory or allosteric domain, or those found in transcriptional regulators, as shown for the virulence regulator MgrA in S. aureus (96). However, the relevance and regulatory nature of this PTM remain largely unexplored in bacteria, leaving large gaps in our understanding of protein persulfidation as a regulatory modification in H 2 S/RSS signaling. For example, persulfidation of the active-site thiol in the cell-abundant glycolytic enzyme glyceraldehyde-3-phosphate dehydrogenase (GAPDH) has been reported by several groups; however, there are conflicting reports to whether this modification activates (58) or inhibits (96,181) the activity of this enzyme. Furthermore, the same active site thiol in GAPDH is the target of many other thiol modifications, including S-nitrosation (182), S-sulfenylation, S-sulfinylation (183), and S-thiolation (184). To better understand the role of protein persulfidation in H 2 S/RSS signaling, it becomes important to understand how this modification is installed and removed from proteins and to identify regulatory targets of this PTM.
It is instructive to discuss thiol-based PTMs, including persulfidation, S-sulfenylation, or S-nitrosation, in the "writer, reader, eraser" paradigm, which represents the formation, signal transduction, and removal of PTMs (Fig. 7A). The writer refers to the process(es) that install the PTM to the protein target, such as S-nitrosoglutathione for protein S-nitrosation (185). The reader includes PTM-dependent interacting partners, whether it be a protein or small molecule, thus transducing the signal in response to the PTM (e.g. S-nitrosation of procaspase-3 promotes its interaction with acid sphingomyelinase and prevents apoptosis) (186). Last, the eraser refers to the proteins or small molecules responsible for removing the PTM after signal transduction has occurred. In S-nitrosation, several denitrosylases have been characterized, including the thioredoxin system (187); other enzymes reduce S-nitrosoglutathione or S-nitroso-CoA directly to sulfenamides, thus lowering the steady-state pools of these small-molecule NO donors (188)(189)(190). For bacterial protein persulfidation, the writers and erasers are perhaps slightly better understood, whereas we have little to no knowledge of the readers. We note also that writers and erasers are typically restricted to enzymes (e.g. kinase-phosphatase and/or acetylase-deacetylase pairs), but here we explicitly consider nonenzymatic pathways as well.
The writers of protein persulfidation may include nonenzyme-catalyzed transsulfuration with RSS or direct reaction of H 2 S with more oxidized cysteines or, enzymatically, via transsulfuration by STRs or other enzymes (Fig. 7A, left). Although  question mark). B, overview of methodologies developed to detect global proteome persulfidation via alkylation/reduction or "tag-switch" approaches (43,198). Protein persulfides are indicated with a red S. Sulfur adducts generated in the first step are shown in blue, whereas green spheres indicate functional groups not shown here (e.g. a fluorescent moiety). C, schematic of the low-pH QTRP method that allows for the direct detection of protein persulfides (red S, blue sphere) (206). See section on H 2 S signaling via protein S-sulfiration for additional details.
the reactivity of STRs strongly suggests a role in protein transsulfuration, little is known regarding the rates of these reactions, specificity, and physiological impact of many STRs. These small-molecule and protein writers can also potentially function in reverse, as the erasers, by removing the sulfane sulfur in a process that is not fully understood. In S. aureus, the cst operon-encoded STR CstA was shown to facilely remove the persulfide moiety on the cysteine desulfurase SufS in support of the idea that STRs can also catalyze the removal of protein persulfides (135). In addition, the thioredoxin (Trx)/thioredoxin reductase cascade may be largely responsible for the removal of protein persulfides for which there is now evidence (Fig. 7A, right) (191,192). The characterization of two thioredoxins in S. aureus that have significant activities on protein persulfides relative to disulfides versus the canonical TrxA certainly suggests a role for Trx in this process, as well as the possibility that bacteria encode specific thioredoxin-like proteins for this purpose (96). Furthermore, a thioredoxin-based proteomic profiling strategy was used to identify potential cellular targets for these persulfide-reducing thioredoxins in S. aureus (193). Analogous strategies might be applied to STRs in an effort to identify protein targets that could function as donors or acceptors in transsulfuration reactions.

Detection of global proteome persulfidation
To evaluate proteome persulfidation, a number of methods have been developed over the past several years to identify candidate targets of this PTM. The first method proposed for detection of protein persulfides relies on S-methylmethane-thiosulfonate to selectively methylate protein thiols over persulfides, followed by persulfide capture by biotin-HPDP (N-[6-(biotinamido)-hexyl]-39-(2-pyridyldithio)propionamide) and enrichment by streptavidin beads (Fig. 7B) (58). However, the selectivity of S-methylmethanthiosulfonate toward thiols over persulfides has been challenged and shown not to be sufficient for general application (194). Several groups then developed approaches that rely on derivatizing both protein thiols and persulfides with a biotinylated alkylating agent (96,192,195,196). After enrichment with streptavidin, the mixed disulfide that characterizes only persulfide-containing peptides or proteins is selectively reduced, alkylated, and identified by LC-MS/MS. Enrichment of whole proteins versus peptides is far more prone to artifacts (192,197) and should be avoided. Enriching peptides rather than proteins will also tend to minimize the presence of false positives that derive from endogenous disulfide bonds in the lysate, although this may be expected to be low for cytoplasmic lysates, given the high reducing capacity of the cytoplasm (96,195,196).
Several "tag-switch" approaches have also been developed for identifying protein persulfides (191,(197)(198)(199)(200). These methods do not rely on the reduction of mixed disulfides from alkylated persulfides but instead exploit the unique reactivity of the mixed disulfide toward a specific nucleophile. The first tagswitch method reported utilized an electrophilic blocking reagent, methylsulfonylbenzo-thiazole, to react with both protein thiols and persulfides (199). These residues when blocked have very different reactivities toward nucleophiles (i.e. thiol adducts are unreactive, whereas persulfide adducts as mixed disulfides retain significant reactivity). The persulfide adduct is then reacted with a cyanoacetate-based reagent (e.g. CN-biotin) to "switch" the tag on these residues (Fig. 7B) (47,198,199). Several cyanoacetate-based reagents have been developed, including fluorescent adducts, and each relies on its high specificity toward the mixed disulfide of protein persulfides over protein disulfides (191). Most recently, Zivanovic et al. (200) reported a new tag-switch workflow where 4-chloro-7-nitrobenzofuran (201,202) is used to label thiols, persulfides, sulfenic acids, and free amines, followed by selective switching of the persulfidederived mixed disulfide by dimedone, a reagent commonly used to detect sulfenylated cysteines (Fig. 7B) (203,204). The advantage of this workflow is that the availability of many derivatized dimedones makes it possible to modify the workflow to fit the application, be it fluorescence imaging, streptavidin enrichment, or click chemistry-based enrichment approaches and downstream MS (205).
Although these workflows are powerful and are expected to be broadly applied, none achieve direct detection of protein persulfides. A recent study by Fu et al. (206) approached this challenge by leveraging the inherent pK a difference between persulfides and thiols in their development of low-pH quantitative thiol reactivity profiling (QTRP) (Fig. 7C). QTRP incorporates an initial alkylation step in low-pH conditions, where most thiols are protonated and unreactive, thus allowing selective alkylation of persulfides (206). Use of isotopically labeled reagents permits estimation of the fraction persulfide versus thiol for all Cys-containing peptides. An important shortcoming of this method is that pK a estimations for thiols and persulfides (44) are not generally known on a proteome scale and may well vary widely. Thus far, QTRP has only been applied in mammalian cells; it would be of interest to determine how many proteins can be detected with high fractional persulfidation levels in less complex bacterial proteomes, which may aid in the identification of regulatory persulfides, characterized by high relative quantitation of persulfide versus thiol.
Finally, regardless of the workflow used to detect bacterial proteome persulfidation, these studies seem to suggest that the proteome may function as a "sink" for sulfane sulfur, given that upward of ;15% of the proteins in a bacterial proteome are persulfidated (at some fraction) in a way that is not greatly impacted by the addition of exogenous sulfide to growing cells (20,96). Taken at face value, this suggests that most of these proteome persulfidation events are unlikely to be regulatory, consistent with a lack of strongly compelling persulfidation consensus sequence motif, implying that these cysteines may be persulfidated at random (20,206,207). The challenge we now face is not in the detection of sites of persulfidation in a proteome but to distinguish those regulatory sites from those that are persulfidated collaterally on solvent-accessible, nonconserved, and highly reactive cysteines. This is a key aspect of the "writer, reader, eraser" paradigm of H 2 S/RSS signaling that is thus far largely unexplored. We would expect regulatory persulfidation events to be found on conserved cysteines in enzymes that catalyze committed steps or define metabolic hubs or branch points in a metabolic pathway(s) (195), in transcriptional regulators (111), as allosteric modulators, or in enzyme active sites (e.g. GAPDH) (58,96,181). Indeed, a subset of these regulatory cysteines may be subject to other PTMs, which would increase the biological complexity of the regulatory response.

Future outlook
Whereas studies of the chemical biology of H 2 S and RSS have greatly expanded in recent years, our molecular understanding of the underlying mechanisms that drive regulatory signaling and cytoprotection of bacterial cells against antibiotic and host assault remains poorly defined. The high reactivity and potentially facile interconversion of RSS represents a particularly challenging analytical problem. Recent advances in methodologies used to quantify RSS will allow us to further elucidate their roles as potent antioxidants (47), modulators of antibiotic efficacy (21), and the "currency" of H 2 S signaling, the latter analogous to LMW organic nitrosothiols in S-nitrosation-based signaling pathways (185,(188)(189)(190).
Recent studies have begun to establish the physiological importance of H 2 S/RSS biogenesis in bacteria at the host-pathogen interface and the central role played by distinct classes of transcriptional regulators that act as molecular sensors that survey cellular H 2 S/RSS required, in some cases, to retain full pathogen virulence (17,96). We do not yet fully understand the mechanisms that make these sensors specific for RSS relative to other oxidants (16), although these data are beginning to appear (113). Finally, although global S-sulfuration profiling has successfully mapped hundreds of sites of protein persulfidation in many different contexts and systems, the next major challenge is to identify and characterize regulatory sites of protein persulfidation in bacteria, like that recently found in human aquaporin-8, a gated H 2 O 2 transmembrane channel (208). These thiols represent physiologically relevant targets of reversible persulfidation that impact enzyme activity, a metabolic pathway(s) or gene expression program (195), beyond simple detoxification and homeostasis of H 2 S/RSS. We suspect that as yet uncharacterized "orphan" STRs that are regulated by RSS sensors (17,135) and perhaps cell-abundant peroxiredoxins (136) and thioredoxins may be excellent candidates as writers and/or erasers in H 2 S/RSS signaling (Fig. 7A), where specific protein-protein interactions may well "direct traffic" in the dynamic flow of regulatory sulfur in cells. Addressing these challenges will better strengthen the foundation of H 2 S/RSS signaling as a bona fide signaling process in bacterial cells in the infected host, thus representing new physiology prime for the development of novel antimicrobial interventions.
Funding and additional information-This work was supported by National Institutes of Health Grant R35 GM118157 (to D. P. G.). B. J. C. W. was supported by a fellowship from the Graduate Training Program Quantitative and Chemical Biology (QCB) at Indiana University (Grants T32 GM131994 and GM109825) and the Kratz Fellowship awarded by the Department of Chemistry, Indiana University. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Conflict of interest-The authors declare that they have no conflicts of interest with the contents of this article.
Note added in proof-In a recent paper, Ng et al. (2020) Front. Microbiol. 11, 1875 showed that two multidrug-resistant strains, including one clinical isolate, of Acinetobacter baumannii, do not biosynthesize appreciable hydrogen sulfide and that exogenous NaHS appears to potentiate the killing efficacy of a number of diverse antibiotics, in contrast to nearly all other findings discussed here. Interestingly, neither A. baumannii strain used by Ng et al. encodes 3-mercaptopyruvate sulfurtransferase (3MST) or RSS sensors FisR or BigR (see Fig. 5) or FisR-and BigR-regulated genes as described in A. baumannii ATCC 17978 (20).