Probing Domain Interactions in Soluble Guanylate Cyclase

Abstract

Eukaryotic nitric oxide (NO) signaling involves modulation in cyclic GMP (cGMP) levels through activation of the soluble isoform of guanylate cyclase (sGC). sGC is a heterodimeric hemoprotein that contains a Heme-Nitric oxide and OXygen binding (H-NOX) domain, a Per/ARNT/Sim (PAS) domain, a coiled-coil (CC) domain, and a catalytic domain. To evaluate the role of these domains in regulating the ligand binding properties of the heme cofactor of NO-sensitive sGC, chimeras were constructed by swapping the rat β1 H-NOX domain with the homologous region of H-NOX domain-containing proteins from Thermoanaerobacter tengcongensis, Vibrio cholerae, and Caenorhabditis elegans (TtTar4H, VCA0720, and Gcy-33, respectively). Characterization of ligand binding by electronic absorption and resonance Raman spectroscopy indicates that the other rat sGC domains influence the bacterial and worm H-NOX domains. Analysis of cGMP production in these proteins reveals that the chimeras containing bacterial H-NOXs exhibit guanylate cyclase activity, but this activity is not influenced by gaseous ligand binding to the heme cofactor. The rat-worm chimera containing the atypical sGC Gcy-33 H-NOX domain was weakly activated by NO, CO and O₂, suggesting that atypical guanylate cyclases and NO-sensitive guanylate cyclases have a common molecular mechanism for enzyme activation. To probe the influence of the other sGC domains on the mammalian sGC heme environment, heme pocket mutants (Pro118Ala and Ile145Tyr) were generated in the β1 H-NOX construct (residues 1–194), the β1 H-NOX-PAS-CC construct (residues 1–385), and the full-length α1β1 sGC heterodimer (β1 residues 1–619). Spectroscopic characterization of these proteins shows that inter-domain communication modulates the coordination state of the heme-NO complex and the heme oxidation rate. Taken together, these findings have important implications for the allosteric mechanism of regulation within H-NOX containing proteins.

Soluble guanylate cyclases (sGCs) are heterodimeric hemoproteins that respond to gaseous signaling molecules. The best-characterized eukaryotic sGCs contain a heme cofactor that rapidly binds nitric oxide (NO), but does not bind oxygen (O₂). Additionally, the heme iron is stable in the ferrous oxidation state. This selective binding of NO enables the protein to function as an essential NO sensor in mammals. Worms and flies are known to contain atypical sGCs that are distinct from mammalian sGC as they bind and respond to O₂ in addition to NO and carbon monoxide (CO). Despite these differences in ligand selectivity, both non-O₂ binding and O₂ binding sGCs perform critical physiological roles by synthesizing cyclic GMP (cGMP) (1–4).

The best-characterized non-O₂ binding sGC heterodimer is the rat α1β1 protein. The α1 and β1 subunits are highly homologous, and both proteins consist of four distinct domains (reviewed in (5)). The β1 subunit, which is the heme-binding subunit, contains an N-terminal Heme-Nitric oxide and OXygen (H-NOX) binding domain, a Per/ARNT/Sim (PAS) domain, a coiled-coil (CC) domain and a C-terminal catalytic domain (Figure 1A). Predicted O₂-binding sGCs from Drosophila and C. elegans also contain H-NOX, PAS, CC and catalytic domains based on primary sequence analysis.

Figure 1.

Domain architecture of sGC α1β1, β1(1–385) and β1(1–194). H-NOX, PAS, coiled- coil (CC) and catalytic (C) domains are shown (A). Chimeras contain the T. tengcongensis, V. cholerae or C. elegans H-NOX domain fused to the rat β1 PAS, CC and C domains (B). Homology model of the rat β1 H-NOX domain (1U55.pdb) (C). Residues in the proximal pocket (His105 and Pro118) and in the distal pocket (Ile145) are shown.

Previously characterized heme binding truncations of the β1 subunit include the H-NOX construct β1(1–194) (6) and the H-NOX-PAS-CC construct β1(1–385) (7). Several bacterial H-NOX proteins with homology to β1(1–194) have been characterized. These homologs are not fused to guanylate cyclase domains, but are thought to be involved in gaseous signaling. Some H-NOX homologs are fused to predicted methyl-accepting chemotaxis proteins, like the O₂-binding H-NOX domain from Thermoanaerobacter tengcongensis, while others are associated with histidine kinases, like the H-NOX protein from Vibrio Cholerae, or diguanylate cyclases. To date, there is no crystal structure of an H- NOX domain from a guanylate cyclase; however, the structures of three bacterial H-NOX proteins with high homology to β1(1–194) have been determined (8–11). These structures have guided many proposals on sGC activation and regulation, yet the precise function of each domain in the mammalian enzyme remains to be determined. Additionally, the functional and structural homology between the bacterial H-NOX proteins, the eukaryotic O₂-binding H-NOX proteins, and the eukaryotic non-O₂ binding H-NOX proteins is unknown. To address these questions, chimeric proteins were constructed by replacing the rat β1 H-NOX domain with bacterial or eukaryotic H-NOX domains. We found that enzyme sensitivity to gaseous ligand binding (NO, O₂ and CO) is observed when the β1 H-NOX is replaced with an atypical guanylate cyclase H-NOX domain from Caenorhabditis elegans (Gcy-33), but this sensitivity is abolished by replacement of the sGC heme domain with two of its bacterial homologs (TtTar4H H-NOX and VCA0720 H-NOX). This suggests that NO-sensitive and atypical guanylate cyclases have a common mechanism of communication between the H-NOX domain and catalytic domains and/or contain domains that exhibit a relatively high degree of structural homology. To further probe the role of the PAS, CC and catalytic domains in modulating the ligand binding properties of sGC, two residues (Tyr145 and Pro118, Figure 1) known to be important for O₂ binding and/or heme conformation in Tt H-NOX (12, 13) were mutated in the full-length α1β1 heterodimer, β1(1–385), and β1(1–194). Our results with these mutants highlight the allosteric influence that the α1 subunit and the β1 PAS, CC and catalytic domains have over the heme environment.

MATERIALS AND METHODS

Reagents

Primers were obtained from Elim Biopharmaceuticals. Sf9 cells were obtained from the Department of Molecular and Cell Biology Tissue Culture Facility, University of California, Berkeley. Diethylammonium (Z)-1-(N,N-diethylamino)diazen-1-ium-1,2-diolate (DEA/NO) was from Cayman Chemical Co.

Generation of sGC chimeras, protein expression, and purification

Alignments of various H-NOX domains were done with MegAlign (LaserGene, DNAStar Inc.) to guide chimera construction. The N- terminal residues of TtTar4H (1–180), VCA0720 (1–178), and Gcy-33 (1–184) were fused to rat β1 residues 187–619 for Tt_β1 and Gcy33_β1 or residues 181–619 for Vc_β1, and then cloned into pFastBac (Invitrogen). All constructs were verified by sequencing (University of California, Berkeley DNA Sequencing Facility). For the chimera expression the Bac-to-Bac baculovirus expression system (Invitrogen) was used to generate recombinant baculovirus according to the manufacturer’s instructions. Sf9 cells were coinfected with wild-type α1 and chimeric β1 constructs. α1/Tt_β1, α1/Vc_β1 and α1/Gcy33_β1 were purified using a previously reported procedure (14) with a few modifications. After elution from the Ni affinity resin, α1/Tt_β1 was exchanged into a 0 mM NaCl buffer and then applied to an anion-exchange column. After washing, the column was developed with a gradient of 0–150 mM NaCl in elution buffer. α1/Vc_β1 purified with a substoichiometric amount of heme and was reconstituted. To reconstitute the protein, 1.5–2 equivalents of hemin were added to the protein and the sample was left on ice for 12 hrs to allow the reaction to equilibrate. Excess hemin was then removed by applying the protein to a PD-10 column equilibrated with 50 mM Hepes, pH 7.4, 50 mM NaCl, 5 mM dithiothreitol (DTT). The protein was characterized both as isolated and after heme reconstitution to ensure the procedure did not affect heme ligand binding.

Generation of mutants, protein expression, and purification

Mutants of rat β1 (P118A and I145Y) were generated using the QuikChange XL site-directed mutagenesis kit (Stratagene) according to the manufacturer’s instructions. The accuracy of each substitution was verified by sequencing (University of California, Berkeley DNA Sequencing Facility).

Rat β1(1–194) and β1(1–385) were expressed and purified by a method that was slightly modified from an existing protocol (6). Specifically, constructs were transformed into E. coli Tuner(DE3) competent cells. Expression cultures for β1(1–194) and β1(1–385) were grown at 37 °C to an Abs₆₀₀ of 0.6 - 0.7 and were then cooled to 20 °C. Protein expression was induced by the addition of IPTG to a final concentration of 0.5 mM and cultures were supplemented with aminolevulinic acid to a final concentration of 0.1 mM for all constructs except β1(1–385) P118A which was induced with 1 mM IPTG in the absence of aminolevulinic acid. Cells were harvested by centrifugation 15–18 h post- induction, and cell pellets were stored at −80 °C. Frozen cell pellets from 2 or 3 L of culture were thawed and resuspended in 50 mL of buffer A [50 mM DEA, pH 8.5, 20 mM NaCl, 5 mM DTT, 1 mM Pefabloc (Pentapharm), and 5% glycerol]. Resuspended cells were lysed with an Emulsiflex-C5 high-pressure homogenizer and centrifuged for 90 min at 100000 g. The supernatant was applied to a Toyopearl SuperQ 650 M (Tosohaas) anion-exchange column and a gradient was developed from 20 to 500 mM NaCl. Fractions containing the protein of interest were concentrated and then applied to a prepacked Superdex S75 HiLoad 16/60 gel-filtration column (Pharmacia) for β1(1–194) or a prepacked Superdex S200 HiLoad 16/60 gel-filtration column (Pharmacia) for β1(1–385). The gel filtration buffer was 50 mM TEA, pH 7.5, 150 mM NaCl, 5 mM DTT, and 5% glycerol. Protein was then pooled, concentrated, and stored at −80 °C. Protein concentrations were determined using the Bradford Microassay (Bio-Rad Laboratories). β1(1–385) P118A, β1(1–385) P118A/I145Y and β1(1–194) P118A were isolated with a substoichiometric amount of heme and were reconstituted using the method described above for the α1/Vc_β1 heme reconstitution. These proteins were characterized both as isolated and after heme reconstitution to ensure the procedure did not affect heme ligand binding.

The Bac-to-Bac baculovirus expression system (Invitrogen) was used to generate recombinant baculovirus according to the manufacturer’s protocol for the expression of β1 P118A. Sf9 cells were cultured and recombinant α1β1 P118A was purified according to a previously published protocol (14) with the following modification. sGC collected from the Ni affinity resin was treated with 1.5 equivalents of hemin in 25 mM TEA, pH 7.5, 50 mM NaCl, 5 mM DTT, and placed on ice for 12 hrs. After this procedure the protein was applied to an anion-exchange column as reported for the purification of wild-type α1β1 (15). Protein purity was assessed by SDS-PAGE (purity > 95%) and concentrations were determined using the Bradford Microassay (Bio-Rad Laboratories). The protein was characterized both as isolated and after heme reconstitution to ensure the procedure did not affect enzyme activity or ligand binding.

Enzyme assays

Duplicate end-point assays were performed for sGC chimeras at 25 °C and for α1β1 P118A at 37 °C as previously described (16). Briefly, sGC complexes were formed with DEA/NO (100 μM) or CO gas (Praxair, Inc.) and confirmed with electronic absorption spectroscopy. Assay mixtures for α1β1 P118A contained 0.2 μg of protein, 50 mM Hepes, pH 7.4, 1 mM DTT, 3 mM MgCl₂, 1.5 mM GTP and 150 μM YC-1 (in DMSO) where indicated. Assay mixtures for chimeras contained 0.2 - 1 μg of protein, 50 mM Hepes, pH 7.4, 50 mM NaCl, 1 mM DTT, 4 mM MgCl₂, 2 mM GTP and 150 μM YC-1 (in DMSO) where indicated. Chimera assays of Fe^II or Fe^II-CO complexes were done in the presence of 100 μM sodium dithionite and limited O₂. All assays were in a final volume of 100 μL and contained 2% DMSO, which was shown not to affect enzyme activity. Reactions were quenched after 2 minutes for α1β1 P118A or 5 minutes for sGC chimeras by the addition of 400 μL of 125 mM Zn(CH₃CO₂)₂ and 500 μL of 125 mM Na₂CO₃. cGMP quantification was carried out using a cGMP enzyme immunoassay kit, Format B (Biomol), per the manufacturer’s instructions. Each experiment was repeated 3 times to ensure reproducibility.

Kinetics

The oxidation rates of β1(1–385) P118A and β1(1–194) P118A were determined using a stopped-flow spectrophotometer (TgK Scientific) at 10 °C and 37 °C, respectively. Samples of anaerobic Fe^II protein were combined with an equal volume of air-saturated (21% O₂) buffer (50 mM Hepes, pH 7.4, 50 mM NaCl). For β1(1–385) P118A, a small amount of sodium dithionite (2 equivalents) was present during mixing which was found to be necessary to keep the protein fully reduced during subsequent manipulations. The observed O₂-binding rate and oxidation rate of β1(1- 385) P118A/I145Y was measured using electronic absorption spectroscopy at 37 °C. Spectra were collected after addition of 150 μL of O₂ saturated buffer to the Fe^II protein in 150 μL of anaerobic buffer. The change in the absorbance maximum versus time was plotted and the data were fit to a single exponential equation. Dissociation of NO from the heme of sGC was measured at 25 °C using the CO/dithionite trapping method described previously (17). The change in the absorbance maximum versus time was plotted and the data were fit to a double exponential equation. The time courses shown are representative results of experiments repeated 2–6 times for each construct.

Resonance Raman spectroscopy

RR spectra were collected for samples in spinning NMR tubes via backscattering geometry at room temperature. The protein (~10 μM heme) was in 50 mM Hepes, pH 7.4, 50 mM NaCl, 1 mM DTT and 150 μM YC-1, where indicated. The excitation wavelengths at 430 nm (for ligand-free samples) and 413 nm (for CO-bound samples) were obtained by frequency doubling, using a nonlinear lithium triborate crystal, of a Ti:sapphire laser (Photonics International TU-UV), which was pumped by the second harmonic of a Q-switched Nd:YLF laser (Photonics Industries International, GM-30-527). For CO-bound samples, laser power at the sample was kept to a minimum (less than 1 mW) by using a cylindrical lens to avoid the photolysis of bound CO. For ligand-free samples, laser power at the sample was 2–3 mW. Scattered light was collected and focused onto a single spectrograph (SPEX 1269) equipped with a CCD detector (Roper Scientific) operating at −110 °C. Spectra were calibrated with dimethyl formamide and DMSO-d₆. A Grams A/I software (Thermo- Galactic) was used to analyze the spectra.

RESULTS & DISCUSSION

sGC is a heterodimeric hemoprotein where each protein subunit consists of four distinct domains (Figure 1A). Proteins with homology to each of the four domains have been elucidated by X-ray crystallography, but there is no structure of the full-length, multi-domain protein. In the work reported here, sGC chimeras were designed and characterized to evaluate the structural and functional similarity between sGC and prokaryotic and eukaryotic homologs of sGC and to illuminate aspects of inter-domain communication in lieu of a full-length sGC structure.

Characterization of sGC chimeras

Previous studies have shown that the β1 PAS, CC and catalytic domains are essential to protein dimerization, while the N-terminal H-NOX domain is essential to gaseous ligand sensing (reviewed in (18)). When NO binds to the rat sGC H-NOX domain, cGMP synthesis increases several hundred-fold. To determine if guanylate cyclase activity can be regulated by β1 H-NOX homologs, the β1 H-NOX domain in the rat sGC was replaced with bacterial and worm H-NOX domains. Specifically, three different H-NOX proteins were fused to the β1 PAS-CC-C domain (Figure 1B); TtTar4H H-NOX, V. cholerae VCA0720 H-NOX, and C. elegans Gcy-33 H-NOX (Tt_β1, Vc_β1, and Gcy33_β1). The β1 sequence was replaced with Tt H-NOX, Vc H-NOX and Gcy-33 H- NOX sequences with 17, 25 and 27% identity, respectively (determined with Lasergene, DNAStar, Inc.). Tt H-NOX has the advantage of being an O₂-binding H-NOX whose structure has been solved, while V. cholerae is a non-O₂ binding prokaryotic H-NOX with greater sequence homology to the β1 H- NOX domain, although no structure has yet been determined. The heme domain from Gcy-33 has not been biochemically characterized, but it is known to be an atypical guanylate cyclase (reviewed in (19)) that mediates an O₂-dependent behavior in worms (20). This observation, in addition to the presence of a tyrosine at position 145 (rat β1 numbering), strongly suggests the protein binds O₂ (21).

The fusion proteins were cloned and then co-expressed with the rat α1 subunit in a Sf9/baculovirus system. The resulting chimeric heterodimer was purified using Ni-NTA metal affinity chromatography, where the tag is on the α1 protein, followed by anion exchange chromatography. All constructs purified as heterodimers and were greater than 95% pure based on SDS-PAGE. Both prokaryotic chimeras were expressed and purified with yields that were 2–3-fold greater than rat α1β1, suggesting that domain swapping might be a useful method to increase protein yields. The α1/Tt_β1 heterodimer as isolated had an electronic absorption maximum between 418–420 nm (Table 1). This species is likely the Fe^II- O₂ complex based on its similarity to other O₂-bound heme proteins (22, 23). After reduction with sodium dithionite, the ferrous-unligated protein was found to form 6-coordinate complexes with O₂, CO and NO (Figure 2, Table 1). The ability to bind O₂ and the formation of a 6-coordinate Fe^II-NO complex demonstrates that the α1/Tt_β1 chimera has properties more like the bacterial Tt H-NOX protein (23) than mammalian β1 H-NOX (6). The α1/Vc_β1 heterodimer was isolated as a mixture of apo and heme-bound protein. After reconstitution with heme, the protein formed a 6-coordinate complex with CO and a 5-coordinate complex with NO, similar to the wild-type Vc and β1 H-NOXs (6, 23). α1/Gcy33_β1 was isolated heme-bound and required reduction with sodium dithionite to form a reduced Fe^II–unligated species. After reduction, the Soret maximum shifted in the presence of NO, CO and O₂ to 422, 421, and 418 nm, respectively (Figure 2). Biochemical characterization of full-length Gcy-33 has not been reported so these results cannot be compared to the wild-type protein; however, the deviations of these Soret values from wild-type α1β1 suggests that the C. elegans atypical cyclase forms 6-coordinate complexes with NO, CO and O₂. This indicates that Gcy-33 has biochemical properties similar to the related Drosophila atypical guanylate cyclase, Gyc-88E (24) and supports in vivo data that suggests the protein functions as an O₂ sensor in worms (20).

Table 1.

Electronic absorption peak positions for sGC chimeras at 25 °C

Protein	Ligand	Coor	Soret (nm)
α1/β1	As isolated	5	431
	FeII-unligated	5	431
	FeII-NO	5	399
	FeII-CO	6	423
α1/Tt_ β1	As isolated	6	418–420
	FeII-unligated	5	432
	FeII-NO	6	420
	FeII-CO	6	423
	FeII-O2	6	420
α1/Vc_ β1a	As isolated	NDb	402/420
	FeII-unligated	5	426
	FeII-NO	5	401
	FeII-CO	6	420
α1/Gcy33_ β1	As isolated	NDb	424
	FeII-unligated	5	433
	FeII-NO	6	422
	FeII-CO	6	421
	FeII-O2	6	418

^aa1Vc_ β1 reconstituted with heme.

^bND; not determined.

Figure 2.

Spectroscopic characterization of sGC chimeras. Electronic absorption spectra of α1/Tt_β1 (A) and α1/Gcy33_β1 (B) at 20 °C. Fe^II-unligated (black solid line), Fe^II-CO (black dashed line), Fe^II-NO (gray solid line) and Fe^II-O₂ (black dotted line) complexes are shown.

To determine if the mammalian H-NOX domain homologs can regulate guanylate cyclase activity, cGMP synthesis was measured in the presence of various heme ligands. The α1 and β1 catalytic domains formed a functional dimer since each chimera exhibited a basal activity (Table 2). Interestingly, α1/Gcy33_β1 exhibited a basal activity that was significantly greater than the α1β1 basal activity (326 nmol min⁻¹ mg⁻¹ versus 50–100 nmol min⁻¹ mg⁻¹ for α1β1). The atypical cyclase Gyc-88E homodimer also exhibited a relatively high basal activity when compared to α1β1 (24). While all of the chimeras exhibited a basal activity, NO did not activate α1/Tt_β1 and α1/Vc_β1. Additionally α1/Tt_β1 was not activated by CO or O₂ and YC-1 did not activate the protein in the unligated, Fe^II-CO or Fe^II-NO bound states (Table 2). YC-1 is a small molecule that synergistically activates sGC in the presence of CO and NO (25). Thus, replacement of the β1 H-NOX domain with either the T. tengcongensis or V. cholerae bacterial H-NOXs abolished the sensitivity of these chimeric sGCs to NO. However, cGMP synthesis of α1/Gcy33_β1 increased 1.5-2-fold in the presence of NO, CO and O₂ (Figure 3). Therefore, this atypical guanylate cyclase H-NOX domain maintains some β1-like function when fused to the β1 PAS, CC and catalytic domain.

Table 2.

Activity of sGC chimeras in various ligation states at 25 °C

Protein	FeII complex	Specific Activity (nmol cGMP min−1 mg−1)	Fold Change (FeII-unligated/FeII-X)	Fold Change (+YC-1/-YC-1)
α1/ β1	FeII-unligatedb	58 ± 21	1	5
	FeII-NO	6054 ± 2303	77	19
	FeII-CO	124 ± 20	2	31
α1/Tt_ β1	FeII-unligated	38 ± 2	1	1
	FeII-NO	36 ± 3	1	1
	FeII-CO	37 ± 0.3	1	1
	FeII + O2	37 ± 2	1	1
α1/Vc_ β1a	FeII-unligated	44 ± 1	1	1
	FeII-NO	55 ± 14	1	1
α1/Gcy33_ β1	FeII-unligated	326 ± 25	1	NDc
	FeII-NO	560 ± 28	1.7	ND
	FeII-CO	495 ± 24	1.5	ND
	FeII + O2	650 ± 76	2	ND

^aa1/Vc_ β1 reconstituted with heme.

^bFe^II-unligated data in the presence of 100 μM sodium dithionite.

^cND; not determined.

Figure 3.

Activity of sGC chimeras. Activity of α1/Tt_β1 (white bars) and α1/Gcy33_β1 (black bars) in the presence and absence of NO, CO, and O₂ at 37 °C. Samples of Fe^II-unligated protein contained 100 μM sodium dithionite.

The absence of enzyme sensitivity to NO, CO and O₂ in the rat-bacterial chimeras suggests that ligand binding to these proteins does not induce the same conformational change as is observed with the α1β1 heterodimer. Several factors could contribute to this loss of sensitivity; the communication between the heme binding pocket and the catalytic domains may be disrupted, the prokaryotic H-NOXs could utilize a different allosteric mechanism of domain regulation, or the prokaryotic H-NOX domains could undergo a conformational change upon NO binding that is distinct from that of the β1 H-NOX domain. To further address these possibilities, resonance Raman (RR) spectroscopy was used to evaluate how the mammalian sGC PAS, CC, and catalytic domains influence the H-NOX heme environment in the context of the chimeric proteins.

RR spectroscopy has been previously used to examine the heme environment of Tt H-NOX, Vc H-NOX and rat α1β1 (23, 26). The RR spectra of α1/Tt_β1 and α1/Vc_β1 were collected for this work and compared to the previously published reports. RR spectra of the Gcy 33 construct could not be obtained due to high background fluorescence of the sample. The Fe^II-CO spectrum of α1/Tt_β1 shows that the FeC and CO frequencies are observed as single bands at 489 cm⁻¹ and 1987 cm⁻¹, respectively (Figure 4, Table 3). These values are similar to those reported for the Tt H-NOX Fe^II-CO complex (490 and 1989 cm⁻¹, respectively) (23). Additionally, the heme skeletal markers in the high-frequency region and the porphyrin conformation-sensitive bands in the low-frequency region are similar to those observed with wild-type Tt H-NOX. The RR spectrum did not change upon addition of YC-1, indicating that the small molecule does not bind to this protein or YC-1 is no longer able to induce a conformational change within the heme pocket when it binds. This result is in agreement with the activity studies (vide infra), which showed that YC-1 did not activate the protein.

Figure 4.

Resonance Raman spectra of the α1/Tt_β1 and α1/Vc_β1 chimeric heterodimers in the Fe^II-CO state. Low- (left panel) and high- (right panel) frequency regions are shown for constructs in the absence and presence of YC-1 as indicated. The asterisk denotes bands from DMSO. The ν_Fe-CO and ν_CO stretching modes are indicated in the low- and high-frequency regions, respectively.

Table 3.

Resonance Raman frequencies and mode assignments for various heme proteins in the Fe^II-CO ligation state^a

Protein	ν2	ν3	ν4	ν(Fe-CO)	ν(C-O)	Ref.
α1β1	1582	1496	1371	473/493	1968/1988	(26)
Tt H-NOX	1580	1494	1369	490	1989	(23)
α1/Tt_β1	1581	1495	1370	489	1987	*
Vc H-NOX	1578	1492	1367	491	1985	(23)
α1/Vc_β1	1582	1503	1370	489	N.D.b	*

^aVibratins in cm⁻¹;

^bN.D., not determined;

^*This work.

The Fe^II-CO RR spectrum of α1/Vc_β1 shows that the νFeC frequency is observed at 489 cm⁻¹ (Figure 4), which is similar to the value reported for wild-type Vc H-NOX (491 cm⁻¹) (23). The weak νCO mode was not observed due to the high fluorescence background of the sample; however, based on the ν FeC frequencies of both α1/Tt_β1 and α1/Vc_β1 it seems that the anomalously low backbonding in α1β1 (26) cannot be induced in the prokaryotic H-NOX domains with fusion to the β1 subunit. Interestingly, some heme skeletal marker bands in the high-frequency region shifted from the Vc H- NOX values such that they are closer to the values for α1β1 (Table 3). Specifically, ν₂, ν₃ and ν₄ upshift by 4, 11 and 3 cm⁻¹, respectively, and this shows that the Vc H-NOX heme environment is indeed influenced by the presence of the PAS, CC and catalytic domains. YC-1 addition produces no spectral change, except for a broadening of the ν_FeC band, possibly signaling a minor species with an altered FeCO conformation.

YC-1 is thought to bind to the N-terminus of the α1 subunit (27–29) and this binding event is known to affect the sGC heme conformation (30, 31). Some communication between this binding site and the heme pocket also occurs in the α1/Vc_β1 chimera, but this communication is not sufficient to induce activity changes as evidenced by the lack of YC-1-induced activation (Table 2). Together these RR results suggest that the non O₂-binding prokaryotic H-NOX shares a more homologous structure to the rat β1 H-NOX domain than the O₂-binding prokaryotic H-NOX. This would enable the Vc H-NOX domain to respond to the β1 PAS, CC and catalytic domains and undergo a modest conformational change upon YC-1 binding.

The observation that the Tt H-NOX domain was apparently unaffected by fusion to the β1 protein, shows that some H-NOX proteins do not have the ability to retain even partial function with domain swapping. Perhaps the mechanism of regulation between O₂-binding and non O₂-binding prokaryotic H-NOXs is too divergent to allow for the retention of function. Conversely, when fused to the β1 protein the predicted O₂-binding eukaryotic H-NOX domain from a guanylate cyclase was responsive to the presence of NO, O₂ and CO. This argues for some commonality in the mechanism of activation in O₂-binding and non O₂-binding guanylate cyclases, but the varying degree of ligand induced activation in α1β1 and α1/Gcy33_ β1 highlights the mechanistic and/or structural differences between these two classes of sGCs.

Critical heme pocket residues

In order to further probe the influence of the sGC PAS, CC, and catalytic domains on H-NOX ligand binding properties and heme environment, site-directed mutagenesis experiments were carried out on mammalian sGC domain truncations – β1(1–194) and β1(1–385) – and full-length protein. The β1(1–194) construct contains the H-NOX domain alone, and the β1(1–385) construct contains the H-NOX, PAS, and CC domains (Figure 1B). Two residues in the sGC heme pocket were chosen for site-directed mutagenesis experiments aimed at probing the involvement of the PAS, CC and catalytic domains in heme ligand binding. Investigations with the O₂- binding H-NOX domain from T. tengcongensis have revealed a proximal pocket proline (Pro118 in the rat β1 numbering system) that is important for maintaining the heme conformation (12, 13) (Figure 1C). In addition to affecting the protein heme conformation, mutation of proline 118 to alanine in Tt H-NOX increases O₂ affinity (12) and increases the proximal Fe-His bond strength (13). Residue 145 (in the rat β1 numbering system) is another critical residue in H-NOX proteins (Figure 1C). Generally, in O₂-binding H-NOX proteins, position 145 is a tyrosine, while an isoleucine or leucine is most commonly present in non O₂-binding H-NOX proteins like β1. A kinetic study showed that Tyr145 is critical for O₂ binding in Tt H-NOX (32). To probe the function of this distal pocket residue in β1, Ile145 was replaced with a tyrosine in β1(1–385) (I145Y). This mutation produced a protein that was capable of binding O₂ (32); however, further studies showed that the same mutation in the full-length α1β1 heterodimer did not produce an O₂-binding protein (33, 34). In this work, a comparative study with different sGC β1 chain lengths (Figure 1A) was carried out with modification of the conserved heme pocket residues 118 and 145.

Ligand binding to β1 P118A and β1 I145Y mutants in full-length α1β1 and H-NOX domain constructs

Ligand binding and O₂ reactivity in Pro118 and Ile145 mutants of α1β1, β1(1–385) and β1(1–194) were examined. The electronic absorption maxima for the Fe^II, Fe^II-NO and Fe^II-CO complexes in α1β1 and H-NOX domain mutants are reported in Table 4. Unlike the Tt H-NOX P118A mutant, all β1 P118A mutants were isolated with a substoichiometric amount of heme, determined by the ratio of the Soret maximum (428 nm) to the total protein absorbance (280 nm) compared to that of the wild-type protein. Pro118 would appear to be important for heme affinity in sGC α1β1, but not as critical for that in Tt H- NOX. Mutants isolated with a substoichiometric amount of heme were successfully reconstituted. All mutants were then reduced with sodium dithionite and these reduced proteins exhibited absorbance maxima between 424–431 nm (Figure S1). After exposure to CO, the absorbance maxima shifted to 420–423 nm, indicative that all formed a 6-coordinate Fe^II-CO complex. Both α1β1 P118A and β1(1- 194) P118A form a stable 5-coordinate complex with NO after displacement of the proximal histidine residue, while β1(1–385) P118A oxidizes rapidly in the presence of either NO or O₂. Characterization of β1 P118A mutants before and after reconstitution confirms that this procedure did not affect the ligand binding properties of the protein.

Table 4.

Electronic absorption peak positions for various sGC β1 mutants in α1β1 and H-NOX domains at 25 °C^a

Protein	Mutation	FeII	FeII-NO	FeII-CO	FeII-O2	Ref.
α1β1	wt	431	399	423	-	(37)
	β1 P118A	428	399	420	-	*
	β1 I145Y	429	416/399	423	-	(33)
β1(1–385)	wt	431	399	423	-	(7)
	β1 P118A	424	398	420	-	*
	β1 I145Y	428	400/417	422	417	(32), this work
	β1 P118A/β1 I145Y	425	399	420	417	*
β1(1–194)	wt	431	398	423	-	(6)
	β1 P118A	428	398	421	-	*
	β1 I145Y	428	400	420	-	*

^aPeak positions in nm;

^*This work. In bold is the most abundant species of a mixture.

Enzyme assays were performed to test if alteration of the conserved Pro118 residue affects sGC activity. The mutant had a basal activity that was similar to that of the wild-type protein (Figure S2). This activity increased 40–50-fold in the presence of NO and 1–2-fold upon CO binding. A similar fold-increase was observed with the protein that was not reconstituted with heme, suggesting that the procedure did not affect catalysis. Thus, sGC α1β1 P118A exhibits a reduction in maximal NO, CO and YC-1-stimulated activity when compared to wild-type sGC. A reduction in maximal enzyme activation was also observed when conserved distal pocket residues were mutated in sGC α1β1 (17), so, not surprisingly, both distal heme pocket and the proximal heme pocket (Pro118) residues are important for enzyme activation.

In agreement with a previous report, a Fe^II-O₂ complex was observed in β1(1–385) after the introduction of a tyrosine at position 145 (32); however, the same mutation in either the full-length α1β1 heterodimer (17, 33) or the β1(1–194) construct did not produce an O₂-binding protein. Additionally, the I145Y mutation in the three sGC constructs led to different Fe^II-NO coordination states. Full-length α1β1 I145Y is mostly 6-coordinate (>85%) (17, 33) and β1(1–385) I145Y is mostly 5-coordinate (>90%) at 25 °C, but both are mixtures of 5- and 6-coordinate complexes. However, β1(1–194) I145Y exclusively formed a 5-coordinate complex (Figure S1), suggesting that the tyrosine is positioned differently in this protein compared to the full-length protein. Perhaps the distance from the distal pocket tyrosine to the bound ligand varies such that it is unable to form a hydrogen bond in the H-NOX protein. Alternatively, the PAS and CC domains may be involved in preventing the breaking of the Fe-His bond in the full-length protein.

Kinetic characterization of β1 P118A and β1 I145Y mutants

In addition to influencing ligand coordination states, distal and proximal heme pocket modifications are known to influence ligand- binding kinetics. Heme oxidation in the presence of O₂ and NO dissociation from the heme were examined in the β1 mutants. It was previously determined that the β1 C-terminal truncation β1(1–194) produces a protein that is more susceptible to oxidation than the full-length protein (6). In this report the β1(1–385) oxidation rate was 4-fold slower than β1(1–194) and the full-length protein has no observable oxidation rate under the experimental conditions (Table 5). This demonstrates that the other domains on the β1 subunit and possibly the α1 subunit contribute to the remarkable stability of the Fe^II heme state of the α1β1 heterodimer.

Table 5.

Comparison of kinetic parameters for various sGC β1 mutants in α1β1 and H-NOX domains at 37 °C

Protein	Mutation	obs Kon O2 (μM−1 s−1)	Kox (s−1)
α1β1	wt	N.O.a	N.O.
	β1 P118A	N.O.	N.O.
	β1 I145Y	N.O.	N.O.
β1(1–385)	wt	N.O.	0.00029 ± 0.00002
	β1 P118A	N.O.	17.6 ± 0.3b
	β1 I145Y	~0.00004c	0.00020c
	β1 P118AI145Y	0.00012 ± 0.00003	0.00062 ± 0.00025
β1(1–194)	wt	N.O.	0.00121d
	β1 P118A	N.O.	0.1950 ± 0.006
	β1 I145Y	N.O.	0.0022 ± 0.0005

^aN.O., not observed;

^bRate measured and reported at 10 °C;

^cRates from (32);

^dRates from (6).

The effect of mutation within the proximal and distal heme pocket on O₂ binding and oxidation was further examined. Whereas the oxidation of β1(1–194) and β1(1–385) significantly increased (≥100-fold) in the P118A constructs, the α1β1 P118A heme did not oxidize even after 4 hrs at 37 °C in the presence of O₂ (Table 5). Additionally, no transient Fe^II-O₂ complex was observed during the oxidation of either β1(1–194) P118A or β1(1–385) P118A. The effect of this mutation on the β1 H-NOX truncations interaction with O₂ is more consistent with Tt H-NOX than α1β1. The dissociation of O₂ from Tt H-NOX decreased ~6-fold upon mutation of the corresponding proline to alanine (12). Since the O₂ association rate was unaffected by this mutation the net effect produced a protein with an increased affinity for O₂. While an observed O₂ off-rate can not be measured in the β1 constructs, it is clear that unlike in full-length sGC, mutation of Pro118 in β1(1–194) and β1(1–385) appreciably altered the affinity of the protein for O₂ and/or lowered the heme reduction potential.

In contrast to the dramatic effect on the H-NOX oxidation rate in the Pro118 mutant, the β1 I145Y mutation only slightly increased (≤2-fold) the oxidation rate of β1(1–194) and did not affect the oxidation rate of β1(1–385) (Table 5). In agreement with previous reports, no heme oxidation was observed in the full-length α1β1 I145Y protein (17, 33). Thus, the Fe^II heme center is significantly influenced by the conformation of the full-length heterodimer. Perhaps the α1 subunit prevents O₂ from reaching the heme cofactor or influences the iron redox potential.

Since β1(1–385) I145Y is known to bind O₂ and mutation of proline 118 was shown to alter O₂ reactivity, the β1(1–385) P118A/I145Y double mutant was made to test if the mutations could synergistically alter the affinity for O₂. β1(1–385) P118A/I145Y purified with a substoichiometric amount of heme and was reconstituted as described. The reconstituted protein formed a 5-coordinate complex with NO and 6-coordinate complexes with both CO and O₂ (Table 4). Interestingly, the oxidation rate and observed O₂ association rate increased (3- and 5-fold, respectively) in the double mutant when compared to the I145Y mutant (Table 5). The presence of a tyrosine in the heme distal pocket significantly reduced the oxidation rate of β1(1–385) P118A (Figure 5), perhaps by stabilizing O₂ binding at the heme.

Figure 5.

Effect of heme pocket mutation on β1(1–385) oxidation rates. The change in absorbance of the Fe^III Soret maximum minus the absorbance of the Fe^II Soret maximum was plotted versus time. Data were fit with a single exponential. Time courses for wt (black solid line), P118A (black dashed line) and P118A/I145Y (gray solid line) are shown for β1(1–385). Inset shows the data for β1(1–385) P118A collected from 0 to 0.2 seconds.

To further examine these sGC heme pocket mutants, the observed NO dissociation rate was measured using the CO/dithionite trapping method (14). Table S1 shows that despite significant changes in the susceptibility of the protein to O₂-induced oxidation, the NO dissociation rate in β1(1–194) P118A was not very different from the wild-type rate. NO dissociation from the β1 I145Y mutants was also examined (Figure S3). Previous reports have shown that the dissociation rate significantly increased in α1β1 I145Y (33); however, there were only slight changes in both the β1(1–194) I145Y and the β1(1–385) I145Y NO dissociation rates (< 2.5-fold). Clearly the most significant effect of the I145Y mutation was in the NO coordination state and NO dissociation rate of the full-length protein, and this work shows that the heme domain truncations β1(1–194) and β1(1–385) do not mimic these effects. The varying properties of these mutants in the H-NOX constructs when compared to the α1β1 heterodimer suggest that the full-length protein has additional means to regulate ligand binding and heme reactivity. This regulation may involve a forced structural change in the heme pocket, which could alter potential hydrogen bonding contacts in the mutants.

Probing the β1 heme environment with resonance Raman spectroscopy

It is known that mutation of the proximal pocket proline influences the heme conformation and Fe-His bond strength of Tt H-NOX (13). Specifically, several out-of-plane low frequency modes known to be sensitive to ruffling and saddling deformations exhibited a reduction in intensity and the Fe-His stretching frequency upshifted by 6 cm⁻¹ (13). The resonance Raman spectra of wild-type, P118A, and I145Y mutants of β1(1–194) in the Fe^II-unligated state were collected (Figure S4). Unfortunately this data could not be compared to the resonance Raman spectra of mutants in full-length sGC due to a high fluorescence background in these samples. In β1(1–194) P118A a slight upshift (2 cm⁻¹) in the ν_Fe-His band was observed (35) whereas, here, a slight downshift (3 cm⁻¹) is seen in β1(1–194) I145Y when compared to the wild-type protein. Therefore, the Fe-His bond strength decreases after mutation of Ile145 and increases after mutation of Pro118 (analogous to Tt H-NOX). In agreement with a previous report, there is no evidence for changes in heme conformation in these mutants based on the relative signal intensities of the low-frequency bands (35).

The RR Fe^II-CO spectra of the β1(1–385) and β1(1–194) I145Y mutants were also collected (Figure 6). In β1(1–194) I145Y there is a shift in the ν_CO band from 1969 to 1949 cm⁻¹ and in β1(1–385) I145Y there is a shift in the wild-type 492 cm⁻¹ ν_FeC band to 498 cm⁻¹. These shifts are likely due to a positive polar interaction between the introduced tyrosine and the bound CO (36). This interaction may affect the polarity around the ligand, weaken the heme propionate contacts with β1, and/or change the Fe-C-O angle (35).

Figure 6.

Resonance Raman spectra of wt and I145Y β1(1–385) and β1(1–194) in the Fe^II-CO state. Low- (left panel) and high- (right panel) frequency regions are shown. The ν_Fe-CO and ν_CO stretching modes are indicated in the low- and high-frequency regions, respectively.

The binding of NO and O₂ is significantly influenced by the PAS, CC and catalytic domains based on electronic absorption spectroscopy and kinetic analysis of site-directed mutants. These domains decrease the heme oxidation rate, perhaps by a mechanism that modulates O₂ accessibility and/or the iron redox potential. Allosteric interactions from these domains induce structural changes in the heme-binding pocket as evidenced by variations in the heme coordination state upon NO binding. Mutational analysis of conserved heme pocket residues also highlights the importance of studying the full-length protein to confirm biochemical predictions based on sGC truncations as some residues, like P118, significantly influence the ligand binding properties of isolated H-NOX proteins but not full-length sGC.

In summary, the β1 PAS, CC and catalytic domains can influence the heme environment of Vc H-NOX highlighting the structural similarity between sGC and non O₂-binding bacterial sGC homologs. Additionally, it was determined that Gcy-33 binds oxygen, in addition to NO and CO, and that the α1/Gcy33_β1 chimera is responsive to varying heme ligation states (Figure 7A). This suggests that atypical guanylate cyclases and NO sensitive guanylate cyclases have a common mechanism of domain regulation. While the precise molecular mechanism of this regulation remains to be determined, mutational analysis indicates that the allosteric interaction of the β1 PAS, CC and catalytic domains on the α1 and β1 subunits affects O₂ reactivity and NO dissociation (Figure 7B) - properties which are known to be essential to the physiological function of sGC in mammalian cells.

Figure 7.

Model of inter-domain communication. Fusion of Vc or Tt H-NOX domains to sGC produces a protein that does not respond to ligand binding at the heme, but fusion of the C. elegans Gcy-33 H-NOX domain influences the H-NOX heme environment and enzyme sensitivity to gaseous ligands (A). Mutation within the heme-binding pocket is influenced by the β1 PAS, CC and C domains on the α1 and β1 subunits (B). These observations indicate that allosteric interactions regulate sGC heme binding properties and enzyme activation.

Abbreviations

sGC

soluble guanylate cyclase

NO

nitric oxide

H-NOX

Heme-Nitric oxide and OXygen binding domain

PAS

Per/ARNT/Sim

RR

resonance Raman

YC-1

3-(5′-hydroxymethyl-3′-furyl)-1-benzylindazole

DEA/NO

diethylammonium (Z)-1-(N,N-diethylamino)diazen-1-ium-1,2- diolate

DTT

dithiothreitol

Sf9

Spodoptera frugiperda

Hepes

4-(2-hydroxyethyl)-1-piperazineethane sulfonic acid

DMSO

dimethyl sulfoxide

EIA

enzyme immunoassay