Recent evidence for multifactorial biofilm regulation by heme sensor proteins NosP and H-NOX

Abstract

Heme is involved in signal transduction by either acting as a cofactor of heme-based gas/redox sensors or binding reversely to heme-responsive proteins. Bacteria respond to low concentrations of nitric oxide (NO) to modulate group behaviors such as biofilms through the well-characterized H-NOX family and the newly discovered heme sensor protein NosP. NosP shares functional similarities with H-NOX as a heme-based NO sensor; both regulate two-component systems and/or cyclic-di-GMP metabolizing enzymes, playing roles in processes such as quorum sensing and biofilm regulation. Interestingly, aside from its role in NO signaling, recent studies suggest that NosP may also sense labile heme. In this Highlight Review, we briefly summarize H-NOX-dependent NO signaling in bacteria, then focus on recent advances in NosP-mediated NO signaling and labile heme sensing.

Introduction

Heme (iron protoporphyrin IX complex) is well-known for its critical and diverse roles as cofactors of hemoproteins. These roles include oxygen transport (hemoglobin)¹, short-term oxygen storage (myoglobin)^{2, 3}, enzymatic redox catalyzation (heme peroxidase and cytochromes P450 superfamilies), and electron transfer (cytochrome c).^{4, 5} Heme is also involved in signal transduction through heme-based gas/redox sensing proteins and heme-responsive proteins, collectively called heme sensor proteins.⁶ Briefly, heme-based gas/redox sensors detect the ligation of diatomic gaseous molecules (including molecular oxygen, carbon monoxide, and NO) to heme iron or oxidation state changes of the heme iron (ferric and ferrous) as signals. Heme-responsive sensors reversely bind heme itself as a stimulus to trigger a downstream response. Heme sensor proteins can either be standalone proteins that interact with an effector protein, or heme-binding domains that are part of the same polypeptide chain as an effector domain. A widely accepted mechanism of signal transduction in heme sensor proteins is that signal detection at the heme binding site causes a conformational perturbation that is propagated to the corresponding effector domain, hence altering the activity of the effector and leading to a cellular response.⁶

Heme-Based NO Signaling

NO, a diatomic, free radical gas has been understood to play signaling roles, including regulation of vasodilation and neurotransmission, in mammalian cells for many years. NO is endogenously synthesized in eukaryotes from L-arginine by the enzyme NO synthase (NOS).⁷ NO regulates cellular processes through ligation to its heme sensor protein, soluble guanylyl cyclase (sGC). sGC contains an N-terminal heme nitric oxide/oxygen binding (H-NOX) domain which selectively binds NO at the heme center.⁸ NO ligation to the H-NOX domain leads to the activation of the C-terminal catalytic domain of sGC, which then promotes the synthesis of an important secondary messenger cyclic-GMP (cGMP) from GTP.⁷

Emerging studies have shown that NO also has physiological roles in bacteria. At low concentrations (< ~μM)⁹, NO acts as a signaling molecule to modulate bacterial physiological processes including biofilm formation and quorum sensing.^{10, 11} A well-studied bacterial NO sensor is H-NOX, a family of typically standalone hemoproteins homologous to the H-NOX domain of eukaryotic sGCs and encoded in the genomes of at least 200 unique species of bacteria.^{11, 12} More recently, a unique heme-containing nitric oxide sensing protein (NosP) was identified, first in the pathogenic bacterium Pseudomonas aeruginosa.¹³ NosP is almost exclusively encoded in bacterial genomes. NosP is able to sense sub-nanomolar NO and appears to share other functional similarities with H-NOX, but NosP lacks a canonical H-NOX domain, based on sequence alignment.¹² To avoid the cytotoxic consequences of high NO concentrations (> ~μM), bacteria have also evolved NO detection machinery to detoxify NO through mechanisms such as nitric oxide reductase and flavohemoglobins.¹⁴

The physiological source of NO to which bacteria respond has been a long-standing open question. Several possible sources have been described. Bacteria endogenously produce NO through denitrification, in which nitrate or nitrite serves as a terminal electron acceptor in respiration under hypoxic conditions.¹⁵ Bacteria that form pathogenic or symbiotic relationships with eukaryotes are exposed to exogenous NO produced by host NOS enzymes.^16–19Some bacteria encode NOS enzymes and thus may produce NO endogenously.²⁰ Finally, there are many possible environmental sources of NO, including acid- or photo-induced degradation of nitrite, for example.²¹

Overview of H-NOX-Dependent NO Sensing

H-NOX proteins are heme-based gas sensor proteins. Some H-NOX domains are capable of selectively binding NO in the presence of molecular oxygen (O₂) while others bind both NO and O₂.⁴ The structure of H-NOX domains are well characterized and consist of a helical N-terminal and C-terminal subdomains. The heme prosthetic group is deeply buried between the subdomains. The heme iron ion is coordinated axially to a conserved histidine residue.^{22, 23} Also highly conserved is a YxSxR motif that serves to stabilize heme binding by forming salt bridges with the propionate groups.^{24, 25} In terms of ligand specificity, the distal heme pocket of NO-selective H-NOX domains contain hydrophobic residues that cannot participate in hydrogen bonding, whereas the O₂-binding H-NOX domains contain residues capable of hydrogen bonding with bound gas molecules.^{8, 26, 27} Ligation of NO causes the proximal histidine ligand to lengthen, which in turn causes the protein to undergo a conformational change that propagates signal transduction.²⁸

Bacterial hnoX genes are frequently encoded next to genes for H-NOX associated cyclic di-GMP processing enzymes (HaCE) or two-component systems (TCSs), suggesting H-NOX functions either through directly regulating c-di-GMP concentrations or through regulating the phosphotransfer activity of co-cistronic TCSs.

C-di-GMP is a bacterial secondary messenger that influences various cellular functions, principally related to the switch between a motile, single-cell state and an adhesive, surface-attached, multicellular state called biofilm.^{29, 30} The estimated intracellular concentration of c-di-GMP is usually in the sub- to low- micromolar range. ^{29, 31, 32} Low concentrations of intracellular c-di-GMP are associated with increased motility, primarily realized through the expression of flagellar motor genes.^33–36 High concentrations of c-di-GMP promote the generation and secretion of exopolysaccharides, thus contributing to biofilm formation.^36–39 The concentration of c-di-GMP is regulated by diguanylate cyclase (DGC) and phosphodiesterase (PDE) enzymes. DGC domains synthesize c-di-GMP from two molecules of GTP while PDE domains hydrolyze c-di-GMP. DGCs have a GGDEF motif,⁴⁰ and PDEs either have an EAL or an HD-GYP motif.^{41, 42} Although both motifs degrade c-di-GMP to pGpG, HD-GYP motifs further hydrolyze pGpG to GMP.⁴²

Shewanella woodyi, Agrobacterium vitis, and Legionella pneumophila each have genes coding for an H-NOX protein that is associated with a bifunctional HaCE containing both a DGC and a PDE domain in the same polypeptide. In S. woodyi, purified SwHaCE has both activities in the absence of H-NOX. When Fe^II-unligated SwH-NOX directly interacts with SwHaCE, the DGC activity is enhanced while the PDE activity remains unaffected, relative to the activity in the absence of H-NOX, leading to a significant increase in c-di-GMP levels and ultimately the formation of biofilm. When Fe^II-NO SwH-NOX is present, the PDE activity is enhanced while the DGC activity drops back to its basal level, leading to an overall decrease in c-di-GMP levels which in turn, leads to biofilm dispersal (Figure 1A).⁴³

Figure 1.

Interactions between H-NOX and partner proteins in various organisms. H-NOX and its partner protein are frequently encoded in the same operon. H-NOX can influence the catalytic activities of HaCEs and thus the relative c-di-GMP levels. (A). In S. woodyi, purified SwHaCE exhibits both DGC and PDE activities. Fe^II-unligated SwH-NOX enhances DGC activity but does not affect PDE activity. When the Fe^II-NO SwH-NOX complex is formed, PDE activity is upregulated and DGC activity returns to basal levels. B) In A. vitis, purified AcHaCE has both DGC and PDE activities. Fe^II-unligated AvH-NOX slightly inhibits both DGC and PDE activities to a similar extent. When Fe^II-NO AvH-NOX is present, PDE activity is increased. C) In L. pneumophila, LpgHaCE as purified has a degenerate PDE domain, indicated by quotation marks around the motif, which is catalytically inactive. LpgHaCE has DGC activity, and Fe^II-unligated LpgH-NOX does not affect its baseline activity. Fe^II-NO LpgH-NOX, however, inhibits DGC activity. Key: Regular black arrows, baseline activity; bold black arrows, upregulation of activity; grey arrows, downregulation of activity. Abbreviations: HaCE, H-NOX associated c-di-GMP processing enzyme; GGDEF, diguanylate cyclase (DGC) domain motif; EAL/HD-GYP, phosphodiesterase (PDE) domain motifs.

In A. vitis, AvHaCE, as purified, also has both DGE and PDE activities. In the presence of Fe^II-unligated AvH-NOX, both the DGC and PDE activities are somewhat reduced compared to the activity when H-NOX is absent. In the presence of Fe^II-NO AvH-NOX, however, the PDE activity is enhanced, causing the complex to act mainly as a phosphodiesterase and leading to an overall decrease in c-di-GMP levels (unpublished data; Figure 1B).

In L. pneumophila, the DGC activity of LpgHaCE is dominant in vivo; in fact PDE activity is not detected in purified LpgHaCE, presumably due to a catalytically inactive PDE domain. The activity of DGC is not affected by Fe^II-unligated H-NOX, however, it is inhibited by Fe^II-NO H-NOX, hence lowering the intracellular levels of c-di-GMP in response to NO and leading to biofilm dispersal (Figure 1C).⁴⁴ Thus, the overall regulatory effect of NO on c-di-GMP concentrations and biofilm formation through H-NOX/HaCE signaling is similar in all three of these organisms.

H-NOX is also involved in signaling through TCSs. TCSs use a network of signal transducing and response regulating proteins that allow bacteria to sense and respond to environmental stimuli such as pH, nutrients, and host-derived molecules.⁴⁵ TCSs typically have a histidine kinase (HK) and a cognate response regulator (RR). HKs have two highly conserved domains, a dimerization/histidine phosphotransfer domain (DHp) and a catalytic/ATP-binding domain (CA), while RRs contain a conserved receiver domain and an output domain. Some HKs also have a C-terminal receiver domain, making them hybrid HKs.^{46, 47} When a signal is detected, the HK is activated and undergoes autophosphorylation on a histidine residue of the DHp domain. The phosphoryl group is then transferred to an aspartate residue on the receiver domain of the RR which outputs a response. In the case of a hybrid HK, the phosphoryl group is intramolecularly transferred from the DHp domain to the aspartate of the C-terminal receiver domain, which then is transferred to the RR.⁴⁸ Another protein called a histidine phosphotransfer (Hpt) protein can also be present to mediate the phosphotransfer between the hybrid HK and the RR.⁴⁷ H-NOX associated TCSs can also influence c-di-GMP metabolism, which is functionally consistent with H-NOX/HaCE systems.⁴

Pseudoalteromonas atlantica encodes H-NOX and H-NOX associated histidine kinase (HahK) genes in the same operon. The HahK has autophosphorylation activity and phosphate is transferred to the receiver domain of its cognate response regulator, an H-NOX-associated response regulator (HarR) with a degenerate HD-GYP motif found in an upstream operon. The HarR exhibits autophosphatase activity as a means of controlling signal duration, but the degenerate HD-GYP domain appears to be indirectly involved in c-di-GMP metabolism rather than direct hydrolysis of c-di-GMP. Fe^II-NO H-NOX inhibits the autophosphorylation activity of HahK, suggesting that H-NOX is capable of regulating c-di-GMP metabolism through this TCS (Figure 2A).⁴⁹

Figure 2.

Interactions between H-NOX and partner proteins in various organisms. H-NOX also participates TCS pathways. A) In P. atlantica, Fe^II-NO PaH-NOX inhibits the autophosphorylation activity of HahK, which in turn, inhibits the forward phosphoryl flux to HarR. B) In S. oneidensis, Fe^II-NO SoH-NOX inhibits the autophosphorylation of HnoK. This halts the phosphoryl flux to HnoB, HnoC and HnoC, effectively decreasing c-di-GMP hydrolysis. C) In the HqsK-involved quorum sensing pathway, Fe^II-unligated H-NOX interacts with HqsK, which functions primarily as a kinase, leading to autophosphorylation and subsequent phosphoryl flux to LuxU and LuxO. In the presence of Fe^II-NO H-NOX, the autokinase activity of HqsK is inhibited. Instead, HqsK switches to a phosphatase, effectively reversing the phosphoryl flow away from LuxU and LuxO. Key: The phosphatase activity of HqsK is labeled in blue. Abbreviations: EAL, phosphodiesterase domain motif; “HD-GYP”, degenerate phosphodiesterase domain motif; HTH, helix-turn-helix DNA-binding domain; TCS, two-component system.

Shewanella oneidensis also encodes an H-NOX gene in the same operon as a HahK called HnoK. Fe^II-unligated H-NOX has both been reported to have no effect on,⁵⁰ and to promote,⁵¹ the autophosphorylation activity of HnoK. When NO binds to H-NOX however, there is a conformational change that inhibits the kinase.⁵¹ HnoK has three cognate response regulators: HnoB (an EAL c-di-GMP phosphodiesterase), HnoC (a HtH-type transcription factor), and HnoD (a degenerate HD-GYP phosphodiesterase). This multi-component network is involved in modulating c-di-GMP levels.⁵⁰ Hydrolysis of c-di-GMP through HnoB is favored when HnoK is active and HnoB is phosphorylated. Fe^II-NO H-NOX leads to inhibited HnoK activity, thus dephosphorylated HnoB. Dephosphorylated HnoD also fine-tunes HnoB by inhibiting its PDE activity.^{4, 51} HnoC itself regulates the transcription of hnoX, hnoK, and hnoBCD, acting as a repressor when unphosphorylated and allowing transcription when phosphorylated (Figure 2B).⁵²

The H-NOX TCS in Vibrio cholerae is similar to that of S. oneidensis. Fe^II-NO H-NOX regulates the kinase activity of a H-NOX associated histidine kinase, that respectively, transfers phosphate to homologues of HnoB and HnoD (V. cholerae does not have a homologue of HnoC), suggesting NO/H-NOX regulation of c-di-GMP concentrations in V. cholerae (Figure 2B).⁵³

In Vibrio harveyi and Vibrio parahaemolyticus, H-NOX interacts with a hybrid H-NOX-associated quorum sensing kinase (HqsK) to participate in quorum sensing.¹² In V. harveyi, HqsK is active as a kinase in the presence of Fe^II-unligated H-NOX, leading to phosphotransfer from HqsK to the histidine phosphotransfer protein LuxU, and then to the response regulator LuxO, a transcription factor that regulates hundreds of genes related to group behaviors in V. harveyi, including bioluminescence genes, as a function of its phosphorylation state. Fe^II-NO H-NOX inhibits the autokinase, but not autophosphatase activity of HqsK, resulting in phosphoryl flux away from LuxU/LuxO.⁵⁴ V. parahaemolyticus is similar in its use of the H-NOX/HqsK pathway to regulate quorum sensing genes in response to NO.⁵⁵ Vibrio fischeri and Silicibacter sp. TrichCH4B also use NO/H-NOX in quorum sensing pathways that regulates symbiosis with E. scolopes⁵⁶ and T. erythraeum⁵⁷, respectively (Figure 2C).

NosP-Dependent NO Sensing

As briefly described above, H-NOX is a bacterial NO sensor. However, many bacteria exhibit a NO-responsive phenotype but do not encode an hnoX gene, suggesting alternate NO sensory machinery. We recently discovered a novel heme-binding NO sensor called NosP. The tertiary structure of NosP is unknown; on the primary structure level, it consists of annotated N-terminal and C-terminal FIST (F-box and intracellular signal transduction) domains. Resonance Raman data collected on purified NosPs sequences from several different bacterial genomes suggest the existence of a distorted heme cofactor⁵⁸, which is also observed in H-NOX and is considered to be a common character of heme-based NO sensors.²⁸ The resonance Raman data also support the presence of a histidine amino acid residue coordinating the heme iron. The formation of the Fe^II-NO compound yields a 5-coordinate complex, presumably resulting from the cleavage of this Fe – N_His bond, which represents a possible mechanism for downstream propagation of the NO signal.⁵⁸

All NosP sequences purified to date exhibit slow NO dissociation rate constants (~ 10⁻⁴ s⁻¹)^{13, 59, 60}, are capable of binding carbon monoxide (CO), but do not bind O₂.^{11, 61} Like H-NOX, NosP genes are frequently co-cistronic with genes coding for TCSs, DGCs, and PDEs, suggesting functional homology between NosP- and H-NOX-regulated signaling pathways.¹³

NosP was first identified in the opportunistic pathogen P. aeruginosa. P. aeruginosa is a bacterium that has a well-characterized response to NO, i.e., NO-triggered biofilm dispersal, but does not encode an hnoX gene. P. aeruginosa encodes NosP (Pa1975) in the same operon as a hybrid histidine kinase (Pa1976). As with H-NOX-associated kinases (HahK), NosP-associated histidine kinases are generally referred to as NahKs.¹³ Pa1976 (PaNahK) has been previously demonstrated to be one of four kinases able to phosphorylate histidine phosphotransfer protein B (HptB/PA3345), that ultimately contributes to the regulation of the switch between mobile and sessile growth in P. aeruginosa.⁶²

Genetic disruption of PaNahK results in loss of the NO-induced biofilm dispersal phenotype. From in vitro experiments, it has been shown that when Fe^II-NO PaNosP interacts with PaNahK, the autokinase activity is inhibited relative to the Fe^II-unligated state. This results in decreased phosphoryl flux from PaNahK to HptB upon NO ligation to NosP (Figure 3).¹³ Interestingly, this mechanism for NO/NosP regulation of biofilm dispersal through HptB is contradictory to what has been proposed with regard to HptB function. HptB phosphorylates a checkpoint response regulator HsbR, which is a dual-function phosphatase/kinase. Phosphorylated HsbR exhibits primarily phosphatase activity which facilitates motility and biofilm dispersal. Dephosphorylated HsbR acts primarily as a kinase, which promotes c-di-GMP synthesis, resulting in biofilm formation (Figure 3).^62–64 In the presence of NO, decreased phosphoryl flux into HptB should result in a less phosphorylated HsbR, thus leading to biofilm formation, which is not consistent with the NO-induced dispersal phenotype observed.¹³

Figure 3.

In P. aeruginosa, PaNosP is involved in a complicated MCS. Two hybrid HKs, PA1611 and SagS, along with PaNahK, transfer phosphoryl to HptB. HptB, in turn, transfers phosphoryl to HsbR, a checkpoint RR involved in motility and biofilm behavior, and to RetS, another hybrid HK. RetS is also involved in the GacS/GacA signaling pathways, acting as a negative regulator of GacS, decreasing phosphoryl flux to GacA, and in turn, affecting the transcription of small regulatory RNAs rsmY and rsmZ. Abbreviations: TCS, two-component system; MCS, multi-component system; Hpt, histidine phosphotransferase domain; HK, histidine kinase; RR, response regulator; HTH, helix-turn-helix DNA-binding domain; PP2C, protein phosphatase 2C; Ser/Thr, serine/tyrosine kinase.

This apparent contradiction suggests a more complicated role for PaNosP than the simple model first proposed.¹³ As mentioned above, instead of functioning as a simple TCS, the PaNosP/PaNahK complex is one member of a multi-component system (MCS) including several other hybrid sensor kinases (PA1611, SagS, and RetS) that all contribute to determining the phosphorylation state of HptB.⁶² Therefore, the phosphorylation state of HptB is not determined simply by NO action on the PaNosP/PaNahK branch. Further, this HptB MCS is very complicated; it has been shown that PA1611 inhibits RetS activity through kinase-kinase interaction^{65, 66} and that the HptB MCS crosstalks the with the core GacS/GacA network that controls the acute/chronic virulence switch in P. aeruginosa.⁶⁷ RetS, which, in addition to being in the HptB MCS, also belongs to the GacS/GacA network, where it serves as a negative regulator of GacS and decreases phosphoryl flux to GacA.⁶⁸ Phosphorylated GacA favors chronic virulence by promoting the transcription of small regulatory RNAs rsmY and rsmZ (Figure 3).⁶⁹ It is possible that there is additional crosstalk between PaNahK and GacS/GacA, analogously to that with Pa1611. These possibilities, as well as crosstalk between PaNahK and other unknown partners, could explain the apparent contradiction in initial PaNosP/PaNahK characterization. Indeed, unpublished data from our lab support the possibility that PaNosP/PaNahK plays an underappreciated role in regulating P. aeruginosa motility and virulence.

L. pneumophila encodes a NosP (Lpg0279) that is in the same operon as a TCS consisting of a NahK (Lpg0278) and a NosP associated response regulator, NarR (Lpg0277). As purified, LpgNahK has autophosphorylation activity and phosphotransfer activity towards LpgNarR, a response regulator with active DGC and PDE domains (DGC is dominate). Basal autophosphorylation and phosphotransfer activities are maintained in the presence of Fe^II-unligated LpgNosP. When NO binds to NosP, however, there is an increase in NahK autophosphorylation, leading to increased phosphoryl flux to NarR. This leads to an increase in PDE activity and a decrease in DGC activity, effectively decreasing the local concentration of c-di-GMP, leading to biofilm dispersal.⁶¹ This is consistent with results from a study performed with a NarR deletion strain (∆lpl0325) of L. pneumophila. Without the NO-responsive output from NarR, the mutant exhibited a hyperbiofilm phenotype (Figure 4A).⁷⁰

Figure 4.

NosP and TCS pathways in various organisms. NosP, like H-NOX, is encoded in genes that are co-cistronic with HK-encoding genes. A) In L. pneumophila, LpgNahK exhibits basal autokinase activity when bound to Fe^II-unligated LpgNosP. This leads to a basal LpgNarR activities - although both activities are detectable, DGC activity dominates. Fe^II-NO LpgNosP leads to increased LpgNahK autophosphorylation and increased phosphoryl flux to LpgNarR. This results in the upregulation of PDE activity and the downregulation of DGC activity, which together, leads to decreased local c-di-GMP levels. B) In V. cholerae, a quorum sensing pathway is regulated by VcNosP. In the presence of Fe^II-unligated VcNosP, VpsS displays kinase autophosphorylation activity, thus directing phosphoryl flux towards LuxU and LuxO. Phosphorylated LuxO leads to the expression of regulatory small RNAs (Qrrs 1-4) and (indirectly) the expression of vps genes, which eventually leads to virulence and biofilm formation, respectively. In the presence of Fe^II-NO VcNosP, the autokinase activity of VpsS is inhibited; instead, the phosphatase activity is promoted and causes a reversed phosphoryl flux from LuxO to LuxU. Dephosphorylated LuxO does not allow the RNAs and the vps genes to be expressed and thus, curtails virulence and biofilm formation. Key: Bold arrows means upregulated activity while grey arrows indicate downregulated activity. The phosphatase activity of VpsS is labeled in blue. Abbreviations: TCS, two-component system; NahK, NosP-associated histidine kinase; Hpt, histidine phosphotransferase domain; NarR, NosP-associated response regulator; GGDEF, diguanylate cyclase domain motif; EAL, phosphodiesterase domain motif; HTH, helix-turn-helix DNA-binding domain.

V. cholerae encodes two NosP domains. One of these domains is a standalone protein (VcNosP/Vc1444) encoded in the same operon as a quorum sensing hybrid kinase called VpsS (Vc1445).⁵⁹ VpsS participates in quorum sensing together with three membrane-bound hybrid kinase receptors, which all direct the flow of phosphoryl to LuxU at low cell density. LuxU transfers phosphoryl to LuxO, a transcription factor that regulates the expression of regulatory small RNAs (Qrrs 1-4). This leads to an inhibition of HapR expression which results in derepression of AphA and vps genes, ultimately increasing virulence and biofilm formation.^{71, 72} At high cell density, phosphate flux is reversed and the receptor kinases exhibit phosphatase activity. In the presence of NO, NosP inhibits VpsS autophosphorylation and decreases phosphoryl flux to LuxU and LuxO. Here, NO acts similarly to an autoinducer, mimicking a high cell density state to elicit biofilm dispersal (Figure 4B).⁵⁹ The second NosP is encoded as a domain within a c-di-GMP phosphodiesterase enzyme. While this NosP is capable of binding heme and exhibits similar spectral features to the rest of the NosP family, it shows a distinct function as a heme-responsive sensor, which will be discussed in the next section.

S. oneidensis encodes NosP (SoNosP/So_2542) and a co-cistronic NosP-associated histidine kinase (SoNahK/So_2543). Interestingly, SoNahK phosphorylates same response regulators HnoB, HnoC and HnoD (with a kinetic preference for HnoC) as the SoH-NOX-associated histidine kinase (SoHnoK) discussed above (Figure 2B).⁶⁰ This suggests that SoH-NOX/SoHnoK and SoNosP/SoNahK are integrated into an NO-responsive MCS regulating intracellular c-di-GMP concentration.^{53, 60} As described above, studies looking at the H-NOX system in isolation^{50, 53} conclude that NO up-regulates biofilm formation by NO-bound SoH-NOX mediated inhibition of SoHnoK activity, resulting in decreased phosphoryl flux to HnoB and HnoD, and hence diminished PDE activity. However, a revised model incorporating SoNosP/SoNahK is now available. Recent data from in vivo and in vitro studies indicate that SoNosP/SoNahK functions as a master regulator of the SoH-NOX/SoHnoK system. In the absence of NO, SoNosP strictly inhibits the autokinase activity of SoHnoK, which leads to a dephosphorylated, and thus inactivated HnoB, resulting in accumulation of c-di-GMP due to diminished PDE activity (Figure 5A). In the presence of NO, Fe^II-NO SoNosP relieves this inhibitory effect on SoHnoK. Therefore, although NO binding to SoH-NOX inhibits SoHnoK activity compared to its activity in the absence of SoH-NOX, the overall effect of NO is to increase HnoB PDE activity through increased phosphoryl flux from SoHnoK (Figure 5B). This results in an NO-mediated increase in c-di-GMP hydrolysis and biofilm dispersal.⁶⁰

Figure 5.

The NO-responsive MCS present in S. oneidensis. Both SoNosP and SoH-NOX participate in this system, with SoNosP functioning as a master regulator (indicated by bold arrows). Both SoHnoK and SoNahK phosphorylate HnoD, HnoB and HnoC, with SoNahK having a kinetic preference for HnoC. A) In the absence of NO, Fe^II-unligated SoNosP strictly inhibits the autokinase activity of SoHnoK, and hence decreases phosphoryl flux into HnoB and HnoD, which results in diminished PDE activity and the accumulation of c-di-GMP. B) In the presence of NO, Fe^II-NO SoNosP leads to an overall increased phosphoryl flux into HnoD and HnoB and ultimately causes c-di-GMP hydrolysis and biofilm dispersal. Abbreviations: MCS, multi-component system; EAL/HD-GYP, phosphodiesterase domain motif; HTH, helix-turn-helix DNA-binding domain.

Heme-Responsive Signaling

Iron is a vital requirement that bacteria must obtain from their environment. Free and soluble iron is highly restricted in the hosts by high-affinity iron-sequestering proteins, such as transferrin, lactoferrin and hemoglobin. Therefore, bacteria secrete high-affinity iron chelator siderophores to compete for iron resources.^73–75 Iron starvation also triggers the expression of heme/hemoglobin utilization systems, in which pathogenic bacteria produce hemolysin to rupture red blood cells and capture released heme/hemoglobin through secreted hemophores or membrane-bound heme receptors.^76–79 The transcription of iron and heme utilization genes is controlled by ferric uptake regulator (Fur), which serves as a global negative regulator that utilizes ferrous iron as a corepressor.⁸⁰

While heme is a valuable nutrient, its homeostasis must be carefully monitored and maintained, as heme at > 1 μM is cytotoxic.⁶ Though the mechanisms of heme internalization have been well studied⁸¹, the fate of heme upon entering into the cytoplasm is still not well understood. It is known that in the cytoplasm, heme is either degraded by heme oxygenase to release Fe²⁺ or incorporated into hemoproteins as a cofactor. However, emerging evidence has shown that heme itself can also act as a signaling entity to be sensed by certain proteins.

For example, heme-regulated initiation factor 2α kinase (HRI) coordinates heme availability with globin biosynthesis to prevent globin accumulation in eukaryotes. HRI is involved in phosphotransfer to the α-subunit of eukaryotic initiation factor eIF2. Phosphorylated eIF2 inhibits the translation of globin. When heme is deficient, HRI is an active kinase that readily phosphorylates eIF2 to turn off globin translation. Under the condition of heme abundance, HRI is inactive, leading to globin synthesis.⁸² NO also modulates the activation of HRI by binding to HRI’s N-terminal heme binding domain to promote both the autokinase and eIF2α kinase activities of the heme-repressed HRI, suggesting that NO may serve as a potent activator of HRI as heme deficiency.⁸³

Heme sensor systems have been discovered in bacteria to regulate heme homeostasis through TCSs. For example, in the human pathogen Staphylococcus aureus, HssS is a membrane-anchored sensor kinase that recognizes extracellular heme. HssS becomes activated by exogenous heme, then subsequently phosphorylates the transcriptional factor HssR to promote the transcription of hrtAB, which results in production of heme efflux pumps. HssS/HssR helps sense extracellular heme stress and pump out surplus heme when pathogens are causing infections within heme-rich blood and tissues. An increased Staphylococcal hepatic virulence in mice caused by ΔhrtA and ΔhssR strains has been observed, presumably due to the accumulation of cytoplasmic heme that triggers cellular stress and eventually causes overexpression of virulence factors. Similar heme-regulated systems are also identified in other Gram-positive pathogens, suggesting a conserved regulatory role for HssS/HssR in heme homeostasis.⁸⁴

In another human pathogen Corynebacterium diphtheriae, heme has been implicated in regulating both heme utilization and detoxification through ChrS/ChrR TCS. ChrS is a membrane-bound kinase that binds heme directly and its autokinase activity is stimulated by heme.⁸⁵ Activated ChrS then relays the phosphoryl group to activate the transcriptional factor ChrA, which subsequently upregulates both heme oxygenase gene hmuO and heme efflux pump hrtAB, respectively.^{86, 87}

In non-pathogenic Gram-positive Lactococcus lactis, heme has been shown to directly bind a cytoplasmic transcriptional factor HrtR to derepress the transcription of hrtRAB operon. Since non-pathogenic gram-positive commensals are less likely to be in contact with heme-rich blood, HrtR is more likely to sense intracellular rather than extracellular heme stress.⁸⁸

In Gram-negative P. aeruginosa, heme has been characterized to participate in cell-surface signaling (CSS) through heme acquisition systems Has and Hxu. The presence of heme in extracellular media is sensed by outer membrane receptors HasR and HxuA to initiate the proteolysis of intramembrane anti-σ factors HasS and HxuR, subsequently leading to free σ^ECFs HasI and HxuI to promote the transcription of hasS hasR-hasAp and hxuA. This positive feedback allows the bacterium to produce more outer membrane receptors to better sense and utilize extracellular heme.⁸⁹ Heme has also been shown to overpower Fur repression in the condition of iron sufficiency to promotes the transcription of the Pseudomonas heme utilization (Phu) genes, though the mechanism of this regulation remains unclear.⁹⁰

Interestingly, recent studies suggest NosP may also function as a heme sensor in Gram-negative bacteria. Judging from the fact that NosP is cytoplasmic and is frequently found in the same operon as TCSs and c-di-GMP metabolizing enzymes, it is possible that intracellular heme availability indirectly controls c-di-GMP metabolism through NosP-mediated pathway, thus leading to bacterial lifestyle changes such as biofilm and virulence in response to iron/heme. This may be analogous to how S. aureus preferentially utilizes heme iron at the initiation stage of infection and switches to other iron-sequestering proteins during the late stage.⁹¹

NosP-Mediated Heme Sensing

As mentioned above in the NosP-dependent NO sensing section, V. cholerae encodes for a second NosP domain that is fused with a PDE domain in a single polypeptide chain (Figure 5). The domain arrangement of Vc0130 (CdpA) is an N-terminal NosP domain followed by both DGC and PDE domains. CdpA has only PDE activity; the DGC domain is degenerate. The DGC domain is required to achieve the optimal PDE activity, however. Inactivation of CdpA, which correlates with a higher level of intracellular c-di-GMP, results in a 3-fold increase in biofilm.⁹²

As purified, full-length CdpA has a very faint yellow color and a weak Soret absorbance band, indicating mostly apo CdpA with trace heme-bound holo-protein. Apo CdpA can be readily reconstituted with heme. Heme-bound CdpA has ligand binding (NO, CO) properties and spectral features that are consistent with other characterized members of the NosP family. However, CdpA binds heme rather weakly in comparison to obligate hemoproteins such as H-NOX and myoglobin. Further characterization revealed that the C-terminal PDE function of CdpA is inhibited by heme in a dose-dependent manner, indicating that CdpA may sense labile heme as a signaling molecule to regulate the hydrolysis of c-di-GMP and biofilm dispersal (Figure 6A).⁹³ This is consistent with the literature suggesting that c-di-GMP represses virulence factor expression and that cdpA is expressed during the late stages of infection in V. cholerae.⁹² Together, these data suggest heme sufficiency prevents c-di-GMP hydrolysis through CdpA, so that V. cholerae remains in biofilm. Upon heme deficiency, which is a sign of iron starvation, heme would dissociate from CdpA, triggering c-di-GMP hydrolysis and biofilm dispersal, releasing V. cholerae to colonize elsewhere, presumably under more iron/heme rich conditions.⁹³

Figure 6.

NosP might also be a heme sensor involved in heme-responsive pathways. (A) V. cholerae encodes CdpA, a protein with a NosP domain fused to a degenerate DGC domain and a catalytically active PDE domain. When heme is absent, the PDE domain hydrolyzes c-di-GMP to pGpG. The PDE activity is completely inhibited upon heme binding to CdpA. (B) B. thailandensis NosP also appears to be heme-regulated. Apo-BtNosP does not affect BtNahK autophosphorylation, but upon heme binding and formation of holo-BtNosP, the autophosphorylation activity is inhibited. Upon NO ligation, the inhibition of BtNahK is alleviated. Neither of these NosPs bind heme as tightly as obligate heme-binding proteins. Abbreviations: DGC, diguanylate cyclase; PDE, phosphodiesterase; “GGDEF”, degenerat diguanylate cyclase domain motif; “HD-GYP”, degenerate phosphodiesterase domain motif; EAL, phosphodiesterase domain motif; TCS, two-component system.

Purified NosP from Burkholderia thailandensis also suggests a heme-responsive function for NosP. Like CdpA, BtNosP has a faster heme dissociation rate constant than would be expected for an obligate heme-binding protein. In B. thailandensis, nosP is in the same operon as a TCS composed of a NosP-associated histidine kinase (BtNahK) and a degenerate HD-GYP domain-containing response regulator. Preliminary results from our laboratory indicate the most significant difference in BtNahK activity is between samples where apo- and holo-BtNosP are incubated with BtNahK. NO exerts regulatory effects on BtNahK activity as well. The inhibitory effect caused by heme binding is alleviated upon NO ligation to holo-Fe^II-unligated BtNosP (Figure 6B). Further studies are needed to determine if BtNosP is primarily a sensor for heme, NO, or both (unpublished data).

Conclusions

In this review, the most recent discoveries relating to NosP have been described. Like the better characterized bacterial sensor H-NOX, NosP modulates biofilm formation and quorum sensing through TCSs and c-di-GMP metabolism in response to NO. A distinct function of NosP is its potential to act as a heme sensor in addition to, or instead of, NO, although this potential function for NosP requires further study. Also, it appears that NosP is frequently engaged in MCSs, suggesting that NosP is involved in the integration of- and response to- many different environmental stimuli. NosP characterization is in its infancy; many aspects of its function are unknown, but the data thus far suggest it is important for the regulation of biofilm in response to environmental cues that are crucial for the virulence in many bacteria.