Heme binding to NosP inhibits the activity of a c-di-GMP phosphodiesterase in Vibrio cholerae

Abstract

Heme, a complex of iron and protoporphyrin IX, plays an essential role in numerous biological processes including oxygen transport, oxygen storage, and electron transfer. The role of heme as a prosthetic group in bacterial hemoprotein gas sensors, which utilize heme as a cofactor for the binding of diatomic gas molecules, has been well studied. Less well known is the role of protein sensors of heme. In this report, we characterize the heme binding properties of a phosphodiesterase, CdpA, from Vibrio cholerae. We demonstrate that the N-terminal domain of CdpA is a NosP domain capable of heme binding, which consequently inhibits the c-di-GMP hydrolysis activity of the C-terminal phosphodiesterase domain. Further evidence for CdpA as a heme responsive sensor is supported by a relatively fast rate of heme dissociation. This study provides insight into an emerging class of heme-responsive sensor proteins.

Vibrio cholerae is a Gram-negative water dwelling bacterium that is the causative agent of the diarrheal disease cholera. V. cholerae has been well studied for its ability to adapt to changes during its transition from an aquatic environment to a human host.¹ Importantly, the pathogenicity of V. cholerae has been linked to the bacterium’s ability to switch between a motile and biofilm lifestyle during infection.² The production of virulence factors and intestinal colonization are coordinated with biofilm formation.

V. cholerae biofilm formation is largely regulated by the secondary messenger molecule cyclic diguanylate monophosphate (c-di-GMP).³ Several studies have implicated c-di-GMP in the regulation of bacterial signal transduction pathways including those responsible for motility, virulence and biofilm formation.⁴ High intracellular levels of c-di-GMP are consistent with a sessile or biofilm mode, while low levels of c-di-GMP are indictive of planktonic cells.⁴ The intracellular concentration of c-di-GMP is controlled by two enzymatic activities: diguanylate cyclase (DGC) activity is responsible for the synthesis of c-di-GMP through the cyclization of two GTP molecules, and is catalyzed by a GGDEF motif; the breakdown of c-di-GMP into pGpG or GMP is achieved by a phosphodiesterase (PDE), and is catalyzed by either an EAL or HD-GYP amino acid motif, respectively.

The importance of c-di-GMP metabolism in V. cholerae is supported by the quantity of c-di-GMP metabolizing enzymes that are encoded in its genome. V. cholerae encodes 22 EAL, 31 DGC, 9 HD-GYP, and 10 fused GGDEF-EAL domain proteins.⁵ One of the combined GGDEF-EAL domain proteins, Vc0130 (CdpA), has been previously implicated in biofilm formation and virulence in a mouse model of infection through the regulation of c-di-GMP.⁶ The C-terminal domain of CdpA is a functional PDE containing a catalytic ECL motif (Figure 1A). The DGC domain of CdpA is degenerate and contains a GVGEW motif. While the GVGEW motif does not possess DGC activity, the domain is necessary for the full activity of the PDE domain.⁶

Figure 1.

CdpA is a hemoprotein. (A) The domain architecture of CdpA consists of an N-terminal NosP domain (COG3287), a degenerate diguanylate cyclase domain containing the amino acid motif GVGEW, and a C-terminal phosphodiesterase containing the active residues ECL. (B) Purified hexaHis-tagged CdpA exhibits a faint yellow color indicating the presence of a small amount of bound heme. (C) UV/Vis spectra of CdpA as purified. A small Soret band around 410 nm is visible. (D) CdpA has the ligand binding properties of a hemoprotein, consistent with other NosP proteins. UV/visible spectra of ferric CdpA (red) with a λ_max at 410 nm, ferrous CdpA (purple) with a λ_max at 422 nm, NO-ligated CdpA (blue) with a λ_max at 397 nm, and CO-ligated CdpA (green) with a λ_max at 418 nm. (E) Spectral characterization of CdpA with varying heme (0 μM, 50 μM or 200 μM)

The N-terminal domain of CdpA (COG3287) is annotated as a F-box and intracellular signal transduction protein (FIST). FIST proteins have been characterized as a class of heme binding proteins that are members of the NosP family of proteins.⁷ NosP proteins, in several bacteria, have been demonstrated to be NO-sensitive heme proteins that mediate various signaling pathways involved in biofilm regulation. NosP domains are often found encoded in operons containing DGCs and PDEs, strongly indicating a role for NosP in regulating levels of c-di-GMP. NosP has been recently implicated in regulating c-di-GMP in both Shewanella oneidensis and Legionella pneumophila through multicomponent signaling pathways involving DGC, PDE, and DGC-PDE response regulators.^8,9 NosP has also been linked to the regulation of biofilm formation in Pseudomonas aeruginosa.⁷ In addition to CdpA, V. cholerae encodes a second NosP protein, VcNosP (Vc1444). VcNosP is a standalone NosP protein that has been found to be involved in the NO-dependent regulation of a histidine kinase, VpsS, which is involved in V. cholerae quorum sensing.¹⁰

The function of the N-terminal NosP domain of CdpA was not previous studied. Here, we demonstrate that the N-terminal domain of CdpA is a NosP domain, capable of binding heme. Further, heme binding alone is a strong regulator of the PDE activity of the C-terminal domain. This work provides further insight into the role of CdpA in V. cholerae, expands the putative roles of NosP in regulating bacterial biofilm formation, and provides an example for a newly emerging class of heme-responsive sensor proteins.¹¹

Results

CdpA is a heme binding protein

In order to characterize the heme binding properties of CdpA, CdpA was recombinantly expressed in E. coli and subsequently purified. Purified protein fractions exhibited a faint red color, indicating the presence of a small amount of bound heme (Figure 1B). Hemoproteins have characteristic absorption spectra due a π to π* transition of electrons in porphyrin ring orbitals. In particular, hemoprotein spectra feature strong absorbance around 400–430 nm called a Soret band. Shifts in the Soret maximum occur with changing ligation state. For this reason, UV/visible spectroscopy yields valuable information about the ligand binding properties of hemoproteins.

Purified CdpA exhibits a weak Soret peak around 410 nm, presumably due to a presence of a mixture of ferrous and ferric heme iron (Figure 1C). Exogenous heme was added, in excess, to the purified protein to facilitate heme binding. After removal of the excess unbound heme, heme bound CdpA samples were treated with the oxidizing agent, potassium ferricyanide, which yields the ferric heme complex with a Soret maximum of 409 nm (Figure 1D). Reduction of the ferric complex was achieved using sodium dithionite (Na₂S₂O₄) and resulted in a shift of the Soret maximum to 422 nm and the formation of sharp, split α/β bands. Carbon monoxide was added to the ferrous complex which resulted in the Fe(II)-CO complex and a Soret maximum of 418 nm. Addition of NO donor to the unligated ferrous complex, yielded the Fe(II)-NO complex, with a Soret band shift to 397 nm, and a broadening of the α/β bands.

A comparison of the spectral characteristics of the CdpA complexes to those of other known hemoproteins reveal similarities amongst NosP proteins across different bacterial species, as well as similarities to other well studied hemoproteins including sGC, CooA, and VcH-NOX (Table 1).

Table 1.

UV/visible Spectral Peak Positions for Known Hemoproteins

Heme complex	Soret maximum (nm)	β (nm)	α (nm)	Reference
Fe(III)
sGC	393	-	-	23
CooA	424	541	566	24
VcH-NOX	406	522	636	25
PaNosP	410	-	-	7
LpNosP	405	-	-	9
VcNosP	409	-	-	10
CdpA	410	-	-	This paper
Fe(II)
sGC	431	555		23
CooA	426	529	559	24
VcH-NOX	428	558		25
PaNosP	420	524	554	7
LpNosP	420	530	556	9
SoNosP	417	524	552	8
VcNosP	420	524	554	10
CdpA	424	526	557	This paper
Fe(II)-NO
sGC	398	537	572	23
VcH-NOX	398	538	572	25
PaNosP	396	534	574	7
LpNosP	396	-	-	9
SoNosP	397	-	-	8
VcNosP	396	-	-	10
CdpA	397	537	570	This paper
Fe(II)-CO
sGC	423	541	567	23
CooA	422	540	568	24
VcH-NOX	422	540	568	25
PaNosP	416	538	565	7
SoNosP	416	-	-	8
VcNosP	418	-	-	10
CdpA	418	540	570	This paper

CdpA binds heme weakly

To explore the possibility that CdpA uses heme as a signaling molecule, flexibly binding heme rather than using it as a tightly bound prosthetic group, we measured the rate at which ferric CdpA transfers its heme ligand to apomyoglobin. Apomyoglobin has a very high affinity for heme, with a K_d for heme binding of about 1 × 10¹⁴ M^-1.¹² We expect that a flexible heme binding protein will readily transfer heme to an acceptor protein such as apomyoglobin.

When heme-bound CdpA was mixed with apomyoglobin, a peak corresponding to the Soret band of myoglobin appeared at 408 nm (Figure 2A). The change in absorbance at 408 nm was plotted versus time to determine the apparent rate constant for NosP heme dissociation, assuming the rate constant of heme binding to apomyoglobin is faster than the rate constant of heme dissociation from NosP (Figure 2B). The data were fit to a double exponential equation in order to account for a second (fast) rate which can be attributed to the rate constant of residual unbound heme in the NosP solution at the start of the experiment that rapidly binds to apomyoglobin.

Figure 2.

Heme dissociation from ferric CdpA. (A) Absorbance difference spectra depicting the transfer of heme from ferric CdpA to apomyoglobin. A rise in λ_max at 408 is attributed to the formation of myoglobin. UV/Vis spectra were collected every 22 seconds. (B) Changes in absorbance at 408 nm were plotted versus time and fit to a double exponential equation. This figure is representative of the 4 biologically repeated experiments.

The rate constant for heme dissociation from CdpA was determined to be (1.8 ± 0.5) x 10⁻² min^-1. This rate constant is relatively fast, compared to those of the well-characterized obligated heme binding proteins sGC, which has a heme dissociation rate constant of (5.0 ± 1.0) x 10⁻⁴ min⁻¹ for the ferric protein, and myoglobin which has a heme dissociation rate constant of 8.4 × 10⁻⁷ s⁻¹ or 5.0 × 10⁻⁵ min^-1.^13,14 The rate for heme dissociation from CdpA is in line with other predicted heme sensors including the heme-regulated eukaryotic initiation factor 2α kinase (HRI), present in red blood cells, which has a k_off for heme dissociation of 1.5 × 10⁻³ s⁻¹ or 9.0 × 10⁻² min^-1.¹⁴

CdpA phosphodiesterase activity is inhibited by increasing heme

The C-terminal domain of CdpA contains the catalytic amino acid motif ECL, which is responsible for PDE activity. CdpA has previously been shown to have in vitro PDE activity; however, these activity assays were performed on whole cell lysate, rather than with purified protein.⁶ Because of this, it is not clear what role heme played in these assays.

Here, purified CdpA was incubated overnight with varying heme (0 μM, 50 μM or 200 μM), excess heme was removed by size exclusion, and then the ferrous form of the NosP domain in CdpA was generated by treatment with sodium dithionite. The ferrous form of CdpA was used for this experiment to replicate the predicted state of the protein complex in vivo. The spectral characterization of the ferrous CdpA complex revealed increasing Soret band intensity at 424 nm with increasing heme concentration, as is expected (Figure 1E). When CdpA was incubated without heme, c-di-GMP substrate was hydrolyzed into pGpG within 5 minutes, and complete turnover of c-di-GMP was observed within 60 minutes (Figure 3A). When CdpA was bound to heme, c-di-GMP turnover was substantially decreased, only ~25% of substrate was turned over within 60 minutes for protein incubated with 50 μM heme, and ~10% of substrate was converted within 60 minutes for protein incubated with 200 μM heme. (Figure 3B, C). Thus, apparently, the PDE activity of CdpA is dependent on heme loading.

Figure 3.

Heme binding inhibits the phosphodiesterase activity of CdpA. HPLC traces monitoring the turnover of c-di-GMP (16.4 min) to pGpG (13.9 min) by CdpA Fe(II) bound to (A) 0 μM, (B) 50 μM, or (C) 200 μM heme over the course of 60 minutes. This figure is representative of independent experiments performed in duplicate.

Discussion

Iron is required for bacterial growth and has emerged a key player in bacterial pathogenesis.¹⁵ During infection, bacteria rely on acquisition of iron from the human host. Since 95% of the total iron in the human body is present in heme, pathogens have evolved ways to acquire iron from heme.^16,17 In many pathogens, including V. cholerae, hemolysins are secreted to release hemoglobin from red blood cells. Heme uptake into the cell is achieved by surface receptors of the TonB-dependent outer transport family, which bind and transport heme into the periplasm.¹⁸ Receptors on the inner membrane, which transport heme into the cytoplasm, have also been characterized.¹⁹ Heme in the cytoplasm is either degraded to release iron, or is used as a prosthetic group for hemoprotein gas sensors.

While many heme-based gas sensors have been characterized, there is increasing interest in identifying sensors which respond directly to the binding of heme. In this study, we have identified heme binding as a regulator of CdpA activity; CdpA is an enzyme which has been previously found to be involved in V. cholerae biofilm and virulence.⁶ The N-terminal domain of CdpA is characterized as a NosP domain, and exhibits similar ligand binding characteristics as previously studied NosP proteins as well as other hemoproteins (Table 1).

The relatively fast rate constant for heme dissociation from CdpA, however, supports its role as a flexible heme binding protein. We report here that heme binding to the N-terminal NosP domain of CdpA inhibits the activity of the C-terminal phosphodiesterase domain. Considering that inhibition of phosphodiesterase activity is consistent with higher levels of c-di-GMP and increased biofilm, this evidence suggests that heme detection by V. cholerae could contribute to an increase in intracellular c-di-GMP and consequently increased biofilm. Importantly, these results for CdpA activity are consistent with what has been previously observed about the role of this protein in V. cholerae; inactivation of CdpA was shown to result in a four-fold increase in intracellular c-di-GMP, and a three-fold increase in biofilm, compared to the WT strain.⁶

Further, regulation of biofilm in response to heme detection may give V. cholerae an advantage during infection of a host. As iron is a scarce yet vital commodity, it would be advantageous for V. cholerae to disperse from biofilm under iron limiting conditions to scavenge for iron/heme elsewhere. It has been shown that virulence gene expression in V. cholerae is suppressed by c-di-GMP²⁰ and that cpdA is expressed during the late stages of V. cholerae infection.⁶ Therefore, heme regulation of CdpA phosphodiesterase activity may be a mechanism to control virulence; when heme is in abundance, inhibition of c-di-GMP breakdown by CdpA may serve to keep V. cholerae in biofilm until heme is scarce; biofilm dispersal and onset of virulence pathology would commence upon heme dissociation.

Interestingly, other Vibrios, including V. harveyi and V. parahaemolyticus, contain a protein homologous to CdpA with annotated NosP and GGDEF/PDE domains, suggesting that heme regulation of c-di-GMP might occur in other bacterial species. This observation requires further exploration.

Experimental procedures

CdpA was cloned into a pET20b vector appending a 6-His tag at the C terminus by the use of NdeI and XhoI restrictions enzymes. The resulting plasmid was transformed into BL21 (DE3) pLysS cells. Transformed cells were grown in TB media (1.2% Tryptone, 2.4% Yeast extract, 0.04% glycerol with 17 mM KH₂PO₄ and 72 mM K₂HPO₄, pH 7.5). Expression was induced at OD₅₉₅ of 0.8 with 25 μM IPTG and carried out overnight at 16 °C. For purification, cells were lysed by sonication in buffer containing 20 mM Tris-HCl, 250 mM KCl, and 10% glycerol and cleared lysate was loaded onto nickel-NTA beads (Qiagen). Two wash steps were performed: 100 ml of 10 mM imidazole and 50 ml of 20 mM imidazole. After washing, the proteins were eluted in buffer with 250 mM imidazole and desalted with a PD-10 column (GE) in buffer containing 50 mM Tris, 200 mM KCl, 10% glycerol, 1 mM dithiothreitol, pH 8.0.

UV/visible spectra were recorded on a Cary 100 spectrophotometer equipped with a constant temperature bath. CdpA complexes were prepared in an oxygen-free glove bag as described previously.²¹

For the heme dissociation assay, purified heme bound ferric CdpA was mixed in a 1:5 ratio with apomyoglobin in a cuvette, as described previously.²² UV/visible spectra were recorded every 22 seconds at 23.5 °C for at least 120 minutes. Changes in absorbance at 408 nm were plotted versus time as two parallel exponentials of the form f(x) = Ax(1 – e^–kx).

PDE assays were performed at 37 °C in an assay buffer containing 50 mM Tris-HCl, 5 mM MgCl₂, 20 μM c-di-GMP and pH 8.0. Reactions were initiated by the addition of 500 nM of purified protein. Reactions were quenched by addition of 10 mM CaCl₂ and subsequent heating to 95 °C for 5 min to precipitate protein. Precipitated protein was removed by centrifugation and the resulting supernatant was filtered through a 0.22 μm membrane and analyzed by HPLC on a reverse phase C18 column (Shimazu) with an ion pairing buffer system.