Inactivation of soluble guanylyl cyclase in living cells proceeds without loss of haem and involves heterodimer dissociation as a common step

Abstract

Background and Purpose:

Nitric oxide (NO) activates soluble guanylyl cyclase (sGC) for cGMP production, but in disease, sGC becomes insensitive towards NO activation. What changes occur to sGC during its inactivation in cells is not clear.

Experimental Approach:

We utilized HEK293 cells expressing sGC proteins to study the changes that occur regarding its haem content, heterodimer status and sGCβ protein partners when the cells were given the oxidant ODQ or the NO donor NOC12 to inactivate sGC. Haem content of sGCβ was monitored in live cells through use of a fluorescent-labelled sGCβ construct, whereas sGC heterodimer status and protein interactions were studied by Western blot analysis. Experiments with purified proteins were also performed.

Key Results:

ODQ- or NOC12-driven inactivation of sGC in HEK293 cells was associated with haem oxidation (by ODQ), S-nitrosation of the sGCβ subunit (by NOC12), sGC heterodimer breakup and association of the freed sGCβ subunit with cell chaperone Hsp90. These changes occurred without detectable loss of haem from the sGCβ reporter construct. Treating a purified ferrous haem-containing sGCβ with ODQ or NOC12 caused it to bind with Hsp90 without showing any haem loss.

Conclusion and Implications:

Oxidative (ODQ) or nitrosative (NOC12) inactivation of cell sGC involves sGC heterodimer dissociation and rearrangement of the sGCβ protein partners without any haem loss from sGCβ. Clarifying what changes do and do not occur to sGC during its inactivation in cells may direct strategies to preserve or recover NO-dependent cGMP signalling in health and disease.

LINKED ARTICLES:

This article is part of a themed issue on cGMP Signalling in Cell Growth and Survival. To view the other articles in this section visit http://onlinelibrary.wiley.com/doi/10.1111/bph.v179.11/issuetoc

1 |. INTRODUCTION

Soluble guanylyl cyclase (sGC, EC 4.6.1.2) is a heterodimer composed of an α and β subunit (Montfort et al., 2017). Nitric oxide (NO) activates cGMP production by sGC through binding to the ferrous haem located in its β subunit. However, when cells and tissues undergo oxidative or nitrosative stress, their sGC becomes insensitive towards NO activation and unable to participate in NO-driven cGMP signalling (Horst & Marletta, 2018; Shah et al., 2018). This occurs in a variety of diseases including pulmonary arterial hypertension (Stasch & Evgenov, 2013), heart failure (Breitenstein et al., 2017), asthma (Ghosh et al., 2016), chronic obstructive pulmonary disease (Glynos et al., 2013) and cancer (Mujoo et al., 2010). Preventing or reversing NO insensitivity in sGC is a current biomedical and pharmacologic goal (Sandner et al., 2019).

Our understanding of sGC inactivation is incomplete and could benefit from further research. We know that NO insensitivity develops in tissues when reactive oxygen species (ROS) are generated due to immune activation, or in cells when they are immune-stimulated or have low MW oxidants like 1H-[1,2,4]oxadiazolo[4,3-a] quinoxalin-1-one (ODQ) added (Bellamy & Garthwaite, 2002; Garthwaite et al., 1995; Horst & Marletta, 2018; Nakane et al., 2002; Shah et al., 2018; Zhao et al., 2000). Mechanisms invoked for ROS-induced sGC inactivation include the oxidation of its ferrous haem to the ferric form, which no longer responds to NO, and loss of the haem from sGC (Fritz et al., 2011; Pan et al., 2016). However, it is notable that sGC haem oxidation and haem loss have never been directly demonstrated in cells or tissues. Instead, these changes are inferred based on the ability of sGC to be activated by low MW agonists like 4-[[(4-carboxybutyl)[2-[2-[[4-(2-phenylethyl)phenyl]methoxy]phenyl] ethyl]amino]methyl]benzoic acid hydrochloride (BAY 58-2667, BAY 58) or by 2-[[6-amino-3,5-dicyano-4-[4-(cyclopropylmethoxy)phenyl]-2-pyridinyl]thio]-acetamide (BAY 60-6583, BAY 60), which are known to displace the oxidized haem from purified sGC and bind within its haem pocket (Ghosh et al., 2014; Kollau et al., 2018; Martin et al., 2010; Mendes-Silverio et al., 2012; Stasch et al., 2006). sGC also becomes NO insensitive when cells or tissues undergo excessive NO exposure (Beuve, 2017; Ghosh et al., 2016; Sayed et al., 2007). In this case, cysteine groups in sGC are known to become S-nitrosated (SNO), and some have been implicated in cause and effect (Beuve, 2017; Beuve et al., 2016; Sayed et al., 2007). Importantly, NO-based sGC inactivation is also associated with a significant change in the protein associations of the sGCβ subunit, namely, the breakup of the sGC heterodimer and binding of the freed sGCβ subunit with cell chaperone heat shock protein 90 (Hsp90) (Ghosh et al., 2016; Sayed et al., 2007). However, it is currently unknown if this change is a common feature of sGC inactivation.

Based on the above, we addressed the following questions concerning sGC inactivation by oxidant ODQ or by NO in living cells: does either process involve haem loss from sGC? Is sGC heterodimer breakup a common mechanism of inactivation? What changes during sGC inactivation drive the sGCβ subunit to associate with Hsp90? Are agonists like BAY 58 neutral indicators of haem-free sGC in cells, or do they also promote its formation? Our study utilized HEK293 cells expressing either wild-type sGCβ or the reporter construct tetra-cysteine sGCβ (TC-sGCβ) whose fluorescence emission after labelling with the biarsenical FlAsH reagent (4’,5’-bis(1,3,2-dithioarsolan-2-yl) fluorescein) can indicate gains or losses in its haem content in real time in living cells (Griffin et al., 1998). The TC-sGCβ construct is capable of forming a fully functional sGC heterodimer when it is expressed along with sGCα in cells (Dai et al., 2020; Hoffmann et al., 2011). We also performed experiments using purified proteins to assess how ODQ or NO treatments affect haem oxidation or loss from sGCβ and its interaction with Hsp90. Our findings clarify what changes do or do not take place in sGCβ and in an sGC heterodimer, as a consequence of its oxidant- or NO-driven inactivation in living cells.

2 |. METHODS

2.1 |. General methods and materials

The TC-FlAsH in-cell tetra cysteine tag detection kit was obtained from Invitrogen (Carlsbad, CA, USA). All other reagents and materials were obtained from sources reported elsewhere. 3-(4-Amino-5-cyclopropylpyrimidin-2-yl)-1-(2-fluorobenzyl)-1H-pyrazolo[3,4-b]pyridine (BAY 41-2272, BAY 41) was a generous gift from Dr. Peter Sandner (Bayer AG, Leverkusen, Germany). sGC activator BAY 58-2667 (BAY 58) was purchased from Sigma-Aldrich (St. Louis, MO, USA). Other materials were obtained as previously reported (Dai et al., 2019, 2020).

2.2 |. Molecular biology

pCMV5 mammalian expression plasmids (RRID:Addgene_15002) containing rat sGCα1(1–690), sGCβ1(1–619), sGCβ1(1–619) with a tetra-cysteine motif (CCPGCC) at residue 239–244 have been reported previously (Dai et al., 2019). pET20b bacterial expression plasmids (RRID:Addgene_50723) containing C-terminus His6 tagged rat sGCβ1 (1–385), sGCβ1(1–385) with a tetra-cysteine motif (CCPGCC) at residue 239–244 (TC-sGCβ1(1–385)) have been reported previously (Dai et al., 2020). pet15MHL bacterial expression plasmid containing full-length human Hsp90β has been reported previously (Sarkar et al., 2015).

2.3 |. Protein purification

Haem-free forms of His6-tagged rat sGCβ1(1–385) and TC-sGCβ1(1–385) were expressed and purified using methods reported previously with modifications (Dai et al., 2019, 2020). Briefly, BL21(DE3) cells transformed with sGC expression plasmids were first grown in 37°C in Terrific Broth until the OD reached 1.5 and then underwent cold shock for 15 min before adding 0.5 mM IPTG. Culture was continued at room temperature for 24 h, after which the cells were pelleted by centrifugation and resuspended in 20 ml of Buffer A (40 mM HEPES, 150 mM NaCl, 10% glycerol, pH 7.6) supplemented with protease inhibitors, benzonase and lysozyme. Resuspended cells were lysed by sonication and clarified by centrifugation. The apo-sGC proteins were purified by gravity Ni-NTA chromatography on a 25-ml column pre-equilibrated with Buffer A. The column was washed with three column volumes of Buffer A containing 40 mM imidazole, and the sGC protein was eluted with Buffer A containing 160 mM imidazole. Eluted apo-sGC proteins were pooled, concentrated and dialysed into Buffer A.

Holo-sGC proteins (ferrous sGC and ferric sGC) were obtained by reconstituting the apo-sGC proteins with haem as reported elsewhere with modifications (Martin et al., 2003). Briefly, ferrous sGC was made by mixing apo-sGC with 1.1 mole ratio of haem in Buffer A then passing the mixture through a PD-10 desalting column (Sigma-Aldrich, St. Louis, MO, USA) to remove any unbound haem. Ferric sGC was made by mixing apo-sGC with 1.1 mole ratio of haem and ODQ in Buffer A then passing the mixture through a PD-10 desalting column.

His6-tagged human Hsp90 was expressed and purified using methods reported previously (Sarkar et al., 2015), using gravity Ni-NTA chromatography on a 25-ml column pre-equilibrated with phosphate buffered saline (PBS). The column was washed with three column volumes of phosphate-buffered saline (PBS) containing 80 mM imidazole, and the sGC protein was eluted with Buffer A containing 200 mM imidazole.

2.4 |. Cell culture and transient transfection

HEK293 cells (CCLV Cat# CCLV-RIE 0197, RRID:CVCL_0045) were cultured in six-well plates or fluorescent 96-well plates as described previously (Dai et al., 2020). sGC plasmids were transfected into HEK293 cells using lipofectamine 2000 (Thermo Fisher Scientific, Waltham, MA, USA) as reported previously (Dai et al., 2020). Briefly, 1 μg or 100 ng of DNA were mixed with 5 or 0.5 μl lipofectamine 2000 in 50 or 5 μl Opti-MEM medium then added into each well in the 6- or 96-well plate, respectively. In some cases, the haem biosynthesis inhibitor succinyl acetone was added 72 h prior to transfection at 400 μM to enable HEK293 cell accumulation of apo-sGCβ species (Dai et al., 2020). Cells were harvested for supernatant production with methods described previously (Ghosh et al., 2014). Briefly, cells were washed with PBS and then resuspended in Buffer A with 1 mM phenylmethylsulfonyl fluoride (PMSF) on ice. Two freeze-thaw cycles were applied to the resuspended cells, followed by clarification by centrifugation. Protein concentration in cell supernatant was quantified using a ND-2000-US-CAN Spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA).

2.5 |. Gel analysis, Western blotting and immunoprecipitation

To detect and compare cell protein expression levels, samples of HEK293 cell supernatants (30 μg protein) were run 7.5% SDS-PAGE, electroblotted onto PVDF membranes and then probed with the respective sGCβ, sGCα, Hsp90 or β-actin antibodies using standard procedures as described previously (Dai et al., 2019). Primary antibodies and their dilutions used for Western blotting in this study were polyclonal sGCβ1 antibody, 1:1,000 dilution (Abcam Cat# ab84955, RRID:AB_1925048, 1:1,000); polyclonal sGCα antibody, 1:1,000 dilution (Cayman Chemical Cat# 160895, RRID:AB_10079373); polyclonal human Hsp90 antibody, 1:500 dilution (Cell Signaling Technology Cat# 4874, RRID:AB_2121214); mouse monoclonal β-actin antibody, 1:1000 dilution (Sigma-Aldrich Cat# A5441, RRID: AB_476744). Goat anti-mouse IgG (H + L)-HRP conjugate (Bio-Rad Cat# 170-6516, RRID:AB_11125547) was used as secondary antibody for Western blots with 1:1000 dilution. The Western Lightning Plus-ECL kit from PerkinElmer Life Sciences (Waltham, MA, USA) was used to visualize the protein bands. Western blotting or immunohistochemistry has been conducted the experimental detail provided conforms with BJP Guidelines (Alexander et al., 2018).

Immunoprecipitations were carried out using methods described previously with modifications (Dai et al., 2019). Briefly, 500 μg of total cell supernatants was pre-cleared with 20 μl of protein G Sepharose beads (GE, Chicago, IL, USA) at 4°C for 1 h to remove substances non-specifically binding to protein G Sepharose. The remaining supernatants were then incubated with 3 μg of anti-sGCβ1 antibody (Cayman Chemical Cat# 160897, RRID:AB_10080042) at 4°C overnight. Protein G Sepharose beads (20 μl) were then added to the solutions and incubated for 1 h at 4°C. Afterward, the Protein G beads were pelleted by centrifugation at 448 x g, washed three times with 40 mM HEPES, 150 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40, pH 7.6 and then boiled with 100 μl SDS-PAGE loading buffer to elute the bound proteins. sGCβ, sGCα and Hsp90 contained in the elutions were then probed by SDS-PAGE and Western blotting as described above.

2.6 |. Detection of sGC activity

cGMP production by sGC in the HEK293 cell supernatants was measured by a previously described method (Dai et al., 2019). Briefly, GTP (0.25 mM), MgCl₂ (0.5 mM), phosphodiesterase inhibitor 3-isobutyl-1-methylxanthine (0.5 mM), sGC simulator BAY 41 (10 μM) or sGC activator BAY 58 (10 μM) were mixed with cell supernatant samples and incubated for 30 min at 37°C. The cGMP produced was quantified using a cGMP ELISA kit (Cell Signaling Technology, Danvers, MA). Absorbance values at 450 nm were read on a Flexstation 3 96-well plate reader (Molecular Devices, San Jose, CA, USA). cGMP production values were normalized according to the total protein amounts in the cell supernatants.

2.7 |. Detection of Cys S-nitrosation in sGC with the biotin switch assay

The biotin switch assay was performed on cell supernatants as described previously with modifications (Jaffrey & Snyder, 2001). Briefly, HEK293 cell supernatants were first incubated with 10 mM methylmethane thiosulfonate (MMTS) and 2% SDS in 250 mM HEPES pH 7.7 buffer containing 1 mM EDTA and 0.1 mM neocuproine at 50°C for 20 min. Reaction mixtures were then incubated with 10 volumes of −20°C acetone for 20 min and centrifuged at 22,042 x g for 10 min at 4°C. The pellet was resuspended in Buffer A and incubated with 10 mM EZ-Link™ HPDP-Biotin (Thermo Fisher Scientific, Waltham, MA, USA) and 1 mM ascorbate for 1 h at room temperature to label protein SNO modifications with biotin. Afterwards, 3 μg of anti-sGCβ1 antibody (Cayman Chemical Cat# 160897, RRID: AB_10080042) was incubated with the reaction mixture for 1 h at 4°C, followed by another 1 h incubation with 20 μl Protein G Sepharose beads at 4°C. Protein G beads were pelleted by centrifugation at 448 x g, washed three times with 40 mM HEPES, 150 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40, pH 7.6 and then boiled with 100 μl SDS-PAGE loading buffer. The samples were then subject to SDS-PAGE and Western blot analysis using HRP-conjugated streptavidin (Thermo Fisher Scientific, Waltham, MA, USA) to detect the biotinylated sGCβ1.

2.8 |. In-cell and in vitro FlAsH labelling of TC-sGCβ1(1–619) and TC-sGCβ1(1–385)

TC-sGCβ(1–619) expressed in HEK293 cells was labelled with FlAsH using the method previously described with modifications (Dai et al., 2019). Briefly, HEK293 cells expressing TC-sGCβ(1–619) grown in fluorescence 96-well plates were washed three times with phenol red-free DMEM containing 1 g/L glucose. Cells were then incubated with a mixture of FlAsH and 1,2-ethanedithiol made in Opti-MEM (final concentrations 0.5 and 12.5 μM, respectively) for 30 min at 37°C. Afterwards, cells were washed three times with 250 μM 1,2-ethanedithiol in phenol red-free DMEM with 10% FBS and then suspended in phenol red-free DMEM with 10% FBS and 15 mM HEPES prior to measurements.

Purified TC-sGCβ(1–385) was labelled with FlAsH by a previous reported method with modifications (Dai et al., 2020). Briefly, TC-sGCβ(1–385) was diluted to 100 μM and then incubated with FlAsH and 1,2-ethanedithiol (final concentrations 120 and 500 μM) in Buffer A for 1 h at room temperature in the dark and then applied to PD10 desalting column in Buffer A.

2.9 |. Monitoring the haem content of TC-sGCβ in cells

Haem content of the FlAsH-TC-sGCβ(1–619) expressed in HEK293 cells was monitored using method reported previously with modifications (Dai et al., 2020). Briefly, after labelling the TC-sGCβ in HEK293 cells with FlAsH, 10 μM ODQ or 30 μM N-ethyl-2-(1-ethyl-2-hydroxy-2-nitrosohydrazino)ethanamine (NOC12) was added into medium. The fluorescence was then monitored in a Flexstation 3 96-well plate reader over time at 37°C using excitation at 508 nm and emission at 528 nm. Traces were fitted using OriginLab 8 (Origin Lab Corp., Northampton, MA, USA).

2.10 |. UV–VIS and fluorescence emission measurements

UV–Vis spectra of purified sGCβ(1–385) proteins were obtained using a Shimadzu UV-2401 PC UV–Vis spectrophotometer (Kyoto, Kyoto, Japan). Proteins were diluted to 1 μM in Buffer A. Hitachi model F-2500 Spectrofluorometer (Chiyoda City, Tokyo, Japan) was used to obtain fluorescence emission spectra (480–600 nm) of 10 μM TC-FlAsH-sGCβ proteins in Buffer A with excitation at 450 nm.

2.11 |. Fluorescence polarization measurements

FITC labelling of purified sGCβ1(1–385) proteins and fluorescence polarization measurements were performed at room temperature using methods described previously (Sarkar et al., 2015). Briefly, the sGCβ1(1–385) proteins were first labelled with FITC in 0.1 M sodium bicarbonate, pH 9.3 by incubation for 1 h at room temperature in the dark. The reaction mixture was passed through a PD-10 column to remove excess FITC and to exchange the protein sample into 40 mM HEPES, 150 mM NaCl, pH 7.6 buffer. In some cases, the reaction mixture was incubated with 10 μM ODQ or 30 μM NOC12 before passing through the PD10 column. FITC-labelled proteins were diluted in fluorescence 96-well plates to a final concentration of 0.5 μM and mixed with different amounts of purified Hsp90. After 30 min incubation, the fluorescence polarization was measured in a Flexstation3 96-well plate reader instrument using excitation at 495 nm and emission at 520 nm. Plots were fitted using OriginLab 8.

2.12 |. Data and statistical analysis

In this study, kinetic trace recordings were presented as mean ± standard deviation. GraphPad Prism 5 (RRID:SCR_002798) was used to evaluate the difference between groups with one-way ANOVA, and P < 0.05 was considered to be statistically significant in all analysis. For quantification of SNO modifications on sGC, the intensity of the SNO-sGCβ band at 3 h in each repetition was considered to be 100% value and was then used to score the band intensities at other time-points. Statistical analysis was undertaken only for studies where each group size was at least n = 5. The declared group size is the number of independent values, and that statistical analysis was carried out using these independent values. To analyse S-nitrosation in sGC, the band intensity at 3 h was set as 100% to normalize the result. Outliers were included in data analysis and presentation. This manuscript complies with the BJP’s recommendations and requirements on experimental design and analysis (Curtis et al., 2018).

2.13 |. Nomenclature of targets and ligands

Key protein targets and ligands in this article are hyperlinked to corresponding entries in the IUPHAR/BPS Guide to PHARMACOLOGY (http://www.guidetopharmacology.org) and are permanently archived in the Concise Guide to PHARMACOLOGY 2019/20 (Alexander, Christopoulos et al., 2019; Alexander, Fabbro et al., 2019; Alexander, Kelly et al., 2019).

3 |. RESULTS

3.1 |. A model system to study sGC inactivation in cells

To investigate sGC inactivation, we adopted methods we used to study sGC haem insertion and heterodimer formation in cells (Dai et al., 2019; Ghosh et al., 2016). These include monitoring cGMP production in response to BAY 41 and BAY 58 to judge the relative levels of sGC present as the ferrous versus the ferric or haem-free forms, using a FlAsH-labelled TC-sGCβ reporter protein to monitor its relative haem level in live cells or in vitro, and performing antibody pulldown and Western analyses on cell supernatants to judge the association of sGCβ with its protein partners sGCα or Hsp90.

We performed experiments, as shown in Figure 1a, to establish an sGC inactivation protocol. HEK293 cells were co-transfected to express sGCα and sGCβ, and after 24 h, we added either vehicle, 10 μM ODQ, or 30 μM NOC12 (NO donor half-life of ~6 h) to the cells. Cells were harvested at various time points, and we measured the cell supernatant cGMP production in response to BAY 41, which only activates the NO-responsive form of sGC (ferrous sGC heterodimer), or in response to BAY 58, which only activates NO-unresponsive forms of sGC (ferric sGC and apo-sGC; Ghofrani et al., 2013; Krieg et al., 2009; Sandner et al., 2019). Figure 1c shows that for the transfected cells grown in media alone their supernatant cGMP production was activated by both BAY 41 and BAY 58 in a ratio that remained constant over time. This confirmed that the cells express a mixture of sGC in its ferrous heterodimer form and in a haem-free (apo) form under typical culture conditions (Ghosh et al., 2014). ODQ treatment of the cells eliminated the BAY 41 activation response of their sGC almost completely within the first hour and caused a reciprocal gain in the BAY 58 activation response (Figure 1d), as reported previously (Bellamy & Garthwaite, 2002; Kollau et al., 2018; Schrammel et al., 1996; Stasch et al., 2006; Zhao et al., 2000), consistent with ODQ oxidizing the ferrous sGC to the ferric form in the cells, which is known to allow BAY 58 activation. The NOC12 treatment initially caused an increase in BAY41 response and a reciprocal decrease in BAY 58 response, as previously reported (Ghosh et al., 2016), and then ultimately caused a change in the sGC activation response towards BAY 41 and BAY 58 that matched what we observed with ODQ and took 2 h to reach near equilibrium (Figure 1e). The NOC12 treatment also caused a time-dependent buildup of SNO modifications in sGCβ (Figure 1b), as observed in previous studies (Ghosh et al., 2016; Sayed et al., 2007). The amount of SNO-sGCβ increased in concert with the loss of the BAY 41 activation response, consistent with SNO modifications in sGC being temporally linked to its development of NO insensitivity (Ghosh et al., 2016). Together, these results defined a framework to study the ODQ- and NOC12-based inactivation of sGC in live HEK293 cells.

FIGURE 1.

Effects of ODQ and NOC12 on the activity of sGC transiently expressed in HEK293 cells. (a) Experimental scheme. HEK293 cells co-transfected with sGCα and sGCβ were treated with NO donor NOC12 or ODQ and then were subjected to the following tests. (b) Representative Western blot showing SNO-sGCβ build-up in the cells during NOC12 treatment. (c-e) cGMP production in response to BAY 41 or BAY 58 by HEK293 supernatants made from cells given vehicle, ODQ or NOC12. Panel e also shows build-up of SNO-sGCβ at different time points after adding NOC12. The cGMP production activity of the un-transfected cell supernatants towards BAY 41 and BAY 58 were 0.17 ± 0.06 and 0.45 ± 0.11 pmol mg⁻¹, respectively. Data are plotted as scatter plots, with solid symbols representing the mean ± S.D; n = 5

3.2 |. Relationship between sGC inactivation and haem loss

It is often thought that haem loss accompanies sGC inactivation, but whether this occurs to sGC in cells is unknown. To investigate, we expressed a full length TC-sGCβ protein in HEK293 cells and labelled it with FlAsH so that its haem content could be monitored by following the FlAsH fluorescence signal of the live cells in a plate reader (haem loss from TC-sGCβ increases the FlAsH fluorescence, whereas haem acquisition decreases the fluorescence; Dai et al., 2020). In cells given media alone, the FlAsH-labelled TC-sGCβ fluorescence signal remained steady over time, indicating that the sGCβ maintained a constant level of haem content (Figure 2a). We added 0.5% (v/v) Tween 20 to the cells as a positive control for haem loss, because it is known to disrupt cells and to extract the haem from sGC (Kollau et al., 2018; Martin et al., 2003). Tween 20 addition led to a time-dependent increase in fluorescence (Figure 2b), suggesting it caused haem loss from the TC-sGCβ. In comparison, an identical Tween 20 treatment of HEK293 cells made haem deficient by pre-culture with succinyl acetone and that were expressing the FlAsH-labelled apo-TC-sGCβ protein (Dai et al., 2020) caused only a weak fluorescence increase (Figure 2c), and Tween 20 treatment of a bacterially expressed and purified FlAsH-labelled shorter apo-TC-sGCβ(1–385) protein caused no fluorescence increase (Figure S1). This confirmed that the fluorescence gain that is shown in panel b was due to Tween 20 causing haem loss from the TC-sGCβ expressed in the cells. The cell expression levels of TC-sGCβ protein were found to be similar in the different experiments (Figure 2i). Together, these data show that the haem content of TC-sGCβ remained stable over the experimental time period under normal culture conditions and confirm that we can detect haem loss from the TC-sGCβ protein that is expressed in live cells.

FIGURE 2.

Effect of ODQ and NOC12 on the haem content of TC-sGCβ expressed in live cells. The haem content of FlAsH-TC-sGCβ1 expressed in HEK293 cells was monitored by fluorescence emission in real time under the various experimental conditions. In some cases, the cells had been made haem-deficient before use by culture with succinyl acetone (SA). (a) Cells grown in normal media given vehicle at time = 0 as a negative control. (b) Cells grown in normal media given Tween 20 at time = 0 as a positive control. (c) Haem-deficient cells given vehicle or Tween 20 at time = 0 as controls. (d) Cells grown in normal media given ODQ. (e) Cells grown in normal media given NOC12. (f) Haem-deficient cells given haem and then given vehicle or Tween 20 at the indicated times as controls. (g) Haem-deficient cells given haem and ODQ at the indicated times. (h) Haem-deficient cells given haem and NOC12 at the indicated times. (i) Representative Western blots showing the relative expression levels of TC-sGCβ and β-actin proteins in the normal (−SA) or haem-deficient (+SA) cells under the different experimental conditions. Data shown are means ± S.D.; n = 5

When the cells expressing the FlAsH-labelled TC-sGCβ protein were incubated with either 10 μM ODQ (Figure 2d) or 30 μM NOC12 (Figure 2e), we observed no subsequent fluorescence increases over time. This indicated that there was negligible haem loss from TC-sGCβ. Considering that sGC activity becomes inhibited after adding ODQ or NOC12 for these same time periods (see Figure 1), our results imply that oxidation of the TC-sGCβ ferrous haem to ferric did not cause it to lose measurable haem in the live cells. Similarly, they imply that the NOC12 treatment did not cause haem loss from TC-sGCβ.

To investigate further, we performed experiments in which 5 μM haem was added to the cells in order to deliberately increase the haem content of the FlAsH-labelled TC-sGCβ protein, as we have done previously when studying sGC haem incorporation (Dai et al., 2020; Ghosh et al., 2014). Figure 2f shows that a fluorescence decrease occurred during the first hour after adding haem, confirming that the cells took up the haem and inserted it into their apo-TC-sGCβ protein subpopulation (Dai et al., 2020). We then removed the haem-containing medium and replaced it either with fresh medium alone or with medium containing Tween 20 (0.5%, v/v), 10 μM ODQ or 30 μM NOC12 and then continued to follow the cell fluorescence signals to detect any loss of haem from the haem-replete TC-sGCβ. The Tween 20 addition caused the expected increase in fluorescence emission due to its causing haem loss from the TC-sGCβ (Figure 2f). Conversely, there was no increase in fluorescence emission after adding either ODQ or NOC12 to the cells (Figure 2g,h). Western analysis showed that the cells expressed similar levels of TC-sGCβ protein under the different culture conditions (Figure 2i). Thus, the ODQ or NOC12 treatments caused no detectable haem loss from TC-sGCβ even after it had been made fully haem-replete in the live cells.

As a further check, we performed similar experiments using cells that were transfected to express both TC-sGCβ and sGCβ so that we could determine the effects of the ODQ and NOC12 treatments on haem retention by TC-sGCβ when it was part of an sGC heterodimer, rather than alone. Under this circumstance, the FlAsH-labelled TC-sGCβ retained a constant level of haem in cells cultured with medium alone (Figure 3a). Adding ODQ or NOC12 to the cells did not cause detectable haem loss from the TC-sGCβ when it was in heterodimer form (Figure 3b) despite both treatments causing an inactivation of sGC heterodimer cGMP production in response to BAY 41 and causing a reciprocal increase in its cGMP production in response to BAY 58 (Figure 3c). Western analysis showed that the cells expressed similar levels of TC-sGCβ and sGCα subunits under the different culture conditions (Figure 3d). Together, our findings indicate that TC-sGCβ retained its haem, following treatment with ODQ or NOC12, regardless of whether the TC-sGCβ was present on its own or was present as part of a functional sGC heterodimer in the cells.

FIGURE 3.

Effect of ODQ and NOC12 on the haem content of TC-sGCβ in live cells co-transfected with sGCα. The haem content of FlAsH-TC-sGCβ co-expressed with sGCα in HEK293 cells was monitored by fluorescence emission in real time under the various experimental conditions. (a) Cells grown in normal media given vehicle at time = 0 as a negative control. (b) Cells grown in normal media given ODQ or NOC12. (c) cGMP production in response to BAY 41 or BAY 58 by HEK293 supernatants made from cells given vehicle, ODQ, or NOC12 at time 0 and 3 h. (d) Representative Western blots showing the relative expression levels of TC-sGCβ, sGCα and β-actin proteins in cells under the different experimental conditions. Traces and activity measures represent mean ± S.D.; n = 5. *P ≤ 0.05, significantly different as indicated

3.3 |. Effects of BAY 58 on the sGC haem content in vitro and in living cells

Although sGC activators like BAY 58 are known to displace haem from purified sGCβ proteins or from the purified sGC heterodimer in vitro (Fritz et al., 2011; Martin et al., 2010; Pan et al., 2016; Zhao et al., 2000), whether they can change the haem content of sGCβ in living cells is unclear. To address this, we first studied how BAY 58 could change the haem content of a bacterially expressed and purified ferrous haem-containing FlAsH-labelled TC-sGCβ (1–385) protein in the absence or presence of Tween 20, ODQ or NOC12. The UV–visible spectra of the purified TC-sGCβ(1–385) protein in its ferrous haem-containing form with or without a FlAsH label are shown in Figure 4a. Adding BAY 58 alone to the ferrous haem-containing FlAsH-labelled TC-sGCβ(1–385) protein did not cause haem loss after a 1 h period, as indicated by the fluorescence emission spectral traces reported in Figure 4b. This is consistent with previous studies that showed BAY 58 causes negligible haem loss from a purified ferrous haem-containing sGCβ (Fritz et al., 2011; Martin et al., 2010; Zhao et al., 2000). However, when ODQ or NOC12 was added to the ferrous haem-containing FlAsH-labelled TC-sGCβ(1–385) protein along with BAY 58, we did observe haem loss from the protein, which based on the fluorescence increases approached a level that was more than half of the haem loss that occurred when the protein was incubated with Tween 20 (Figure 4b). Incubating the ferrous haem-containing FlAsH-labelled TC-sGCβ(1–385) protein with either NOC12 or ODQ alone caused no significant haem loss (Figure 6c; see below). Together, our findings show that ODQ-driven haem oxidation in the purified FlAsH-labelled TC-sGCβ(1–385) protein weakened its haem binding and made it susceptible to displacement by BAY 58, identical to that reported when purified ferric haem-containing sGCβ was treated with BAY 58 (Fritz et al., 2011; Zhao et al., 2000). Our findings also reveal that NOC12 treatment weakened the haem binding affinity of the purified FlAsH-labelled TC-sGCβ(1–385) protein such that it then allowed BAY 58 to displace the haem, which had not been previously reported.

FIGURE 4.

Capacity of BAY 58 to displace haem from TC-sGCβ in vitro. (a) UV–visible spectra of 1 μM TC-sGCβ(1–385) and FlAsH-TC-sGCβ(1–385) after reconstitution with haem. (b) Fluorescence emission traces were collected after incubating 1 μM of a purified ferrous haem-containing FlAsH-labelled TC-sGCβ(1–385) protein with either nothing, 0.5% Tween 20 (v/v), 10 μM ODQ, 30 μM NOC12, 10 μM BAY 58 or their indicated mixtures (2 h for NOC12 + BAY 58 group, 1 h for all other groups). Data representative of n = 5 trials

FIGURE 6.

ODQ and NOC12 cause sGC heterodimer breakup and an association of the sGCβ subunit with Hsp90 in cells. (a) Representative Western blots of sGCβ immunoprecipitations of supernatants from HEK293 cells co-transfected with sGCα and sGCβ. Cell supernatants were immunoprecipitated with anti-sGCβ antibody then blotted for bound sGCα and Hsp90 showing the relative levels of sGCα or Hsp90 found associated with sGCβ. (b) quantitative analysis of the band intensities in a. (c) Representative Western blots of sGCα reverse immunoprecipitations of supernatants from HEK293 cells co-transfected with sGCα and sGCβ. Cell supernatant were immunoprecipitated with anti-sGCα antibody then blotted for bound sGCβ. (d) Quantitative analysis of the band intensities in c. Data are presented as scatter plots with the solid symbols indicating the mean value ± S.D.; n = 5. *P ≤ 0.05, significantly different as indicated

We next examined how BAY 58 would affect the haem status of the FlAsH-labelled TC-sGCβ expressed in live HEK293 cells that had or had not undergone an inactivating regimen (pre-culture with 10 μM ODQ for 1 h or with 30 μM NOC12 for 2 h). The fluorescence emission trace in Figure 5a shows that BAY 58 addition caused no haem displacement from the TC-sGCβ over a 5 h monitoring period in cells cultured in medium alone. However, in cells that had undergone an inactivation treatment by ODQ or by NOC12, we observed a slow and gradual fluorescence emission increase upon BAY 58 addition (Figure 5b,c), indicating that haem was being displaced from the TC-sGCβ. This haem displacement began immediately in the ODQ-treated cells and began after a brief delay in the NOC12-treated cells. Western analysis showed that the cells expressed similar levels of TC-sGCβ under the different culture conditions (Figure 5d). Fitting the traces of fluorescence gain in Figure 5b,c to a single exponential function gave estimated rates of BAY 58-driven TC-sGCβ haem loss of 8.0 ± 0.2 × 10⁻³ min⁻¹ and 9.0 ± 0.3 × 10⁻³ min⁻¹ in the ODQ- and NOC12-treated cells, respectively. These rates are within the range of rates by which BAY 58 was reported to displace ferric haem from purified sGCβ proteins (Fritz et al., 2011; Pan et al., 2016; Zhao et al., 2000). Thus, we found that BAY 58 could cause haem displacement from TC-sGCβ but only in cells that had been subject to the ODQ or NOC12 treatments. Together, our results provide the first demonstration of haem displacement from sGCβ in living cells and define the conditions that are required for it to occur.

FIGURE 5.

Capacity of BAY 58 to displace haem from TC-sGCβ expressed in living cells. (a–c) Fluorescence emission traces monitoring the haem content in FlAsH-labelled TC-sGCβ expressed in HEK293 cells that had been pre-incubated with either vehicle (a), 1 h ODQ (b) or 2 h NOC12 (c) prior to adding BAY 58 at indicated time = 0. A single exponential fitting of the fluorescence traces are shown as red lines in b and c. The fluorescence time traces in a–c represent the mean ± S.D.; n = 5. (d) Representative Western blots showing the relative expression levels of TC-sGCβ and β-actin proteins in the cells under the different experimental conditions

3.4 |. Relationships between sGC inactivation, heterodimer breakup and sGCβ association with Hsp90

We next investigated how the ODQ and NOC12 inactivation treatments would affect the sGC heterodimer level in cells by determining the sGCβ subunit associations with sGCα compared with its association with Hsp90. Binding of sGCβ to Hsp90 or sGCα are mutually exclusive (Dai et al., 2019; Ghosh et al., 2014; Stuehr et al., 2021), and in healthy cells the level of sGCβ that is bound to either sGCα or Hsp90 reflects the levels of mature sGC heterodimer or immature sGCβ respectively (Dai et al., 2019; Ghosh et al., 2014). We co-transfected HEK293 cells with sGCα and sGCβ and after 24 h treated the cells with vehicle or with 10 μM ODQ for 1 h or 30 μM NOC12 for 2 h to inactivate sGC. We then performed sGCβ antibody pulldowns with the cell supernatants to compare the levels of sGCα and Hsp90 that were bound to sGCβ before and after the inactivation regimens. Data in Figure 6a,b show that in cells receiving vehicle, the sGCβ was found to be associated with both sGCα and Hsp90 as expected, indicating they contained a mixture of mature and immature sGC. Treating the cells with either ODQ or NOC12 diminished the level of sGCβ associated with sGCα and increased its association with Hsp90. We also performed sGCα pulldowns on replica cell supernatants that confirmed sGCα association with sGCβ was diminished following the ODQ or NOC12 treatments (Figure 6c,d). Together, this shows that sGC inactivation by either ODQ or NOC12 caused the sGC heterodimer to dissociate and caused the freed sGCβ subunit to bind with Hsp90.

3.5 |. ODQ and NOC12 act directly on ferrous sGCβ to promote its Hsp90 binding

To better understand how the ODQ and NOC12 treatments cause sGCβ to change its protein partners in the cells, we studied how treating purified ferrous haem-containing sGCβ(1–385) or FlAsH-labelled TC-sGCβ(1–385) proteins with ODQ (10 μM, 1 h) or NOC12 (30 μM, 2 h) would affect their Hsp90 binding affinities. Treating the sGCβ(1–385) protein with ODQ or NOC12 caused UV–visible spectral changes that indicated perturbations of its bound haem (Figure 7a). ODQ shifted the sGCβ haem Soret peak from 431 to 407 nm, indicating the ferrous haem became oxidized to ferric (Fritz et al., 2011; Zhao et al., 2000; Zhong et al., 2010) without causing loss of haem, because free haem has maximum Soret absorption at 385 (Karnaukhova et al., 2014). Likewise, NOC12 shifted the sGC haem Soret peak from 431 to 418 nm, possibly indicating it led to formation of a six-coordinate NO complex with the bound haem in sGCβ (Tsai et al., 2011). Replica experiments using the FlAsH-labelled ferrous haem-containing TC-sGCβ(1–385) protein and a FlAsH-labelled apo-TC-sGCβ(1–385) protein control showed that the individual ODQ and NOC12 treatments had little ability to increase the fluorescence emission intensity (Figure 7b), indicating that neither treatment caused haem loss from the purified protein.

FIGURE 7.

Relationship between the ODQ or NOC12 impacts on the sGCβ haem and on their abilities to influence sGCβ binding to Hsp90. Experiments were performed using purified sGCβ(1–385) or FlAsH-labelled TC-sGCβ(1–385) proteins and Hsp90. (a) UV–visible spectra of ferrous haem-containing sGCβ1(1–385) after it was subject to the indicated treatments. (b) Fluorescence emission spectra of apo- FlAsH-labelled TC-sGCβ(1–385) or ferrous haem-containing FlAsH-labelled TC-sGCβ(1–385) after it was incubated under the different conditions. (c) UV–visible spectrum of apo-sGCβ(1–385) and FITC-labelled apo-sGCβ1(1–385) after being reconsisituted with haem (holo-sGC). Each spectrum shown is the average of 5 runs. (d–f) Change in fluorescence polarization values recorded during Hsp90 titrations of 0.5 μM of a FITC-labelled apo-sGCβ (1–385), or a FITC-labelled holo-sGCβ given either nothing (d), ODQ treatment (e) or NOC12 treatment (f) respectively. Data shown are scatter plots with the solid symbol being the mean value ± S.D.; n = 5. Red lines indicate the computer-fitted binding curves as explained in Section 2

We then studied Hsp90 binding to the purified haem-containing sGCp(1–385) protein using an established fluorescence polarization method (Sarkar et al., 2015). The apo- and ferrous haem-containing sGCβ(1–385) proteins were labelled with FITC (Figure 7c) and then incubated with buffer alone or with ODQ (10 μM, 1 h) or NOC12 (30 μM, 2 h) prior to being titrated with increasing amounts of purified Hsp90. The traces in Figure 7d indicate that Hsp90 bound well to the FITC-labelled apo-sGCμ(1–385) control protein, as indicated by the - concentration-dependent increase in residual polarized fluorescence, whereas Hsp90 showed negligible binding with the FITC-labelled ferrous haem-containing sGCβ(1–385) protein, as we have reported before (Sarkar et al., 2015). Notably, treatment with ODQ or NOC12 of the FITC-labelled ferrous haem-containing sGCβ(1–385) allowed it to bind with Hsp90 (Figure 7e,f). Together, our results indicate that the ODQ and NOC12 treatments increased the affinity of purified sGCβ(1–385) for Hsp90, although these treatment did not displace its bound haem.

4 |. DISCUSSION

When sGC becomes insensitive towards NO it cannot generate cGMP in NO-driven signalling cascades and is considered inactive. NO-insensitivity can develop through oxidative or nitrosative pathways and is thought to contribute to several diseases (Horst & Marletta, 2018; Mayer et al., 2009; Shah et al., 2018). We studied sGC inactivation driven by oxidant ODQ or by the NO donor NOC12, in cells in order to clarify missing or misunderstood aspects of the inactivation process, focusing on possible changes in sGCβ haem content, sGC heterodimer stability and in the protein interactions of the sGCβ subunit. Our fluorescent TC-sGCβ reporter protein allowed us to track its haem content when it was present alone or in heterodimer form in live cells before, during and after the sGC inactivation treatments, which had not been previously accomplished. Overall, our findings reveal that although ODQ and NOC12 inactivate sGC through different means, they both promote a similar trajectory of events in cells that culminate in sGC heterodimer dissociation and sGCβ subunit binding with Hsp90, all occurring without any significant haem loss from sGCβ. Thus, our findings clarify what happens to sGC when it becomes inactivated by either oxidative or nitrosative means in biological settings.

It was unexpected to find that no haem was lost from TC-sGCβ expressed in cells during or following either inactivation regimen. This was the case regardless of whether the TC-sGCβ was expressed alone in cells or was expressed along with sGCα and shown to form a functional heterodimer. Our result contradicts current dogma that posits haem loss is a common feature of sGC inactivation in tissues and cells (Sandner et al., 2019; Thoonen et al., 2015). Indeed, the only circumstance where we observed TC-sGCβ to lose haem in live cells was when the cells were incubated with BAY 58 while undergoing an ODQ or NOC12 inactivating regimen. But this represents an artificial condition, because to date, no natural, cell-derived, low MW compounds or proteins have been found to mimic BAY 58 in being able to displace haem from inactive sGC in a similar manner. Knowing that haem loss is not normally an outcome of sGC inactivation in cells shifts the focus onto the other changes that do occur to sGC during its inactivation. Moreover, having haem remain bound in the inactive sGC is advantageous from a biomedical standpoint, because it makes it somewhat easier for the cells to reverse the sGC inactivation either on their own (Brunner et al., 1996) or through pharmacological means.

Another finding that was unexpected was that ODQ treatment caused the sGC heterodimer to dissociate in cells. ODQ is known to directly oxidize the sGCβ ferrous haem to the ferric state in the context of an sGC heterodimer and in isolated sGCβ proteins as well (Schrammel et al., 1996; Sobey & Faraci, 1997; Stasch et al., 2006). Thus, our results imply that simple oxidation of the sGCβ ferrous haem to ferric, on its own, may be enough to prompt the sGC heterodimer to dissociate in cells. A connection between the sGC haem redox state and its heterodimer status has not been considered before in the literature. Importantly, an ODQ-driven sGC heterodimer dissociation eliminates any chance of activating sGC to produce cGMP, in response to NO.

We also saw that the ODQ treatment caused the sGCβ subunit to associate with Hsp90 in cells. In our experiments with the bacterially expressed and purified ferrous haem-containing sGCβ(1–385) protein, we found that oxidizing its ferrous haem to ferric with ODQ enhanced its binding affinity towards Hsp90, despite causing no loss of haem. This change in affinity towards Hsp90 may help to explain why we observed the sGCβ subunit to associate with Hsp90 in cells after the ODQ treatment. Together, our findings imply a model where sGCβ haem iron oxidation to the ferric state by ODQ causes structural changes in the sGCβ subunit that on one hand destabilize its interaction with the sGCα subunit and lead to heterodimer dissociation and, on the other hand, enhance its affinity towards Hsp90, resulting in formation of a sGCβ-Hsp90 complex. Having a haem protein’s interactions be influenced by changes in the haem iron redox state is not unprecedented. For example, a haem redox change has been reported to influence the subunit interactions within bacterial di-haem cytochrome c peroxidases (Bewley et al., 2013) and to regulate the binding affinity of cytochrome b5 towards its cytochrome P450 partner (Noble et al., 2007). Moreover, ODQ-driven ferrous haem oxidation in sGC increased its interaction with its reductase protein in cells (Rahaman et al., 2017), much like we see here for it increasing the sGCβ-Hsp90 interaction. It will be interesting to study what structural changes in sGCβ drive such consequential changes to its protein partner associations.

The NOC12 inactivation treatment also caused no loss of bound haem from TC-sGCβ when it was expressed in cells alone or in the context of a sGC heterodimer but still caused sGC heterodimer dissociation and formation of the Hsp90–sGCβ complex. However, NOC12 is likely to inactivate sGC by a different mechanism from that used by ODQ. During prolonged NOC12 exposure of the purified ferrous haem-containing sGCβ(1–385) protein, a Soret peak at 417 nm was observed, which is distinct from ferric sGCβ and close to the peak position assigned to the six-coordinate ferrous sGC-NO complex (Tsai et al., 2011). Indeed, NOC12-driven sGC inactivation is unlikely to proceed through simple haem oxidation, as it does for ODQ, and may instead involve additional protein changes such as SNO modifications in sGC, which built up in the cells during the NOC12 treatment, as observed previously (Ghosh et al., 2016; Sobey & Faraci, 1997) and as observed in the present study for sGCβ. It will be useful to explore if specific SNO modifications destabilize the sGC heterodimer or promote complex formation between the haem-containing sGCβ subunit and Hsp90. By resolving the molecular changes to sGCβ and the sGC heterodimer that take place during NOC12 inactivation in cells, our work helps to explain how NO can engage in non-canonical feedback to self-limit NO-driven cGMP signalling in health and disease (Dao et al., 2020).

Figure 8 presents a scheme based on results to date that indicates what forms of sGC build up in cells before and after sGC inactivation treatments. In healthy cells, two forms of sGC exist that are in equilibrium: an immature form in which the apo-sGCβ subunit is complexed with Hsp90, and the mature sGC heterodimer whose sGCβ subunit contains ferrous haem and is complexed with sGCα (Ghosh et al., 2014). Notably, other possible sGC forms such as a haem-free sGC heterodimer (Thoonen et al., 2015) or an Hsp90-bound sGC heterodimer (Papapetropoulos et al., 2005) have not been directly shown to build up in cells, although such forms have been proposed. When the sGC heterodimer is inactivated by haem oxidation (i.e., by ODQ or possibly by cell-derived ROS), a distinct sGC species ultimately builds up in cells following heterodimer breakup, namely, the ferric haem-containing sGCβ subunit in complex with Hsp90. When NO or related reactive nitrogen species (RNS) inactivate sGC heterodimer, a somewhat different species ultimately builds up following heterodimer breakup, namely, the SNO-modified, haem-containing sGCβ in complex with Hsp90. Curiously, the changes in sGCβ protein associations due to the inactivation regimens are the reverse of what occurs during maturation of the sGC heterodimer in cells (Dai et al., 2019; Ghosh et al., 2014). Among the four forms of sGC that build up in cells, three are haem-containing, but only one of these three is functional for NO-driven cGMP production (the ferrous haem-containing heterodimer), whereas all others are NO-unresponsive and therefore non-functional, unless pharmacological sGC agonists like BAY 58 are provided. Together, these four forms of sGC represent the minimal number of known species found in cells. Their relative levels should vary depending on the health of the cell, its levels of internal and external oxidative or nitrosative stress, and its capacity to reverse any haem redox, covalent and protein structural changes that accompany sGC inactivation. These issues can now be further addressed.

FIGURE 8.

sGC species present in cells and the changes caused by sGC inactivation. In healthy cells, sGC exists as a mix of an immature and mature form in which the sGCβ subunit differs in its protein partners, haem content and capacity to function in cGMP production. Oxidants like ODQ or reactive oxygen or nitrogen species (ROS or RNS) inactivate the sGC heterodimer in conjunction with their causing oxidation of its ferrous haem to ferric and/or causing Cys SNO protein modifications. This in turn causes a dissociation of the sGC heterodimer and the binding of the haem-containing sGCβ subunit with Hsp90. These inactivation processes occur without any significant haem loss from the sGCβ. A subpopulation of immature apo-sGCβ remains present in both healthy and damaged cells (green dashed box). It can bind BAY 58, and this causes it to form an active sGC heterodimer with sGCα

Previously, the only form of sGC that had been directly shown to bind Hsp90 was the immature apo-sGCβ (Dai et al., 2019; Ghosh & Stuehr, 2012). Thus, it was assumed that the increase in Hsp90 association seen after sGC inactivation would involve a haem-free form of sGCβ. However, our results imply that it is haem-containing forms of sGCβ that bind to Hsp90 as a consequence of sGC inactivation and heterodimer breakup. This finding makes it even less likely that a haem-free sGC heterodimer can ever build up in cells, due to the demonstrated lack of haem loss from TC-sGCβ and the tendency of the inactivated sGC heterodimer to dissociate.

Finally, our study helps to clarify how pharmacological sGC activators may function in living systems. To activate sGC, compounds like BAY 58 must bind within its haem pocket (Martin et al., 2010), and when BAY 58 is given to cells or animals, it can stimulate cGMP production and downstream responses (Ghosh et al., 2014; Krieg et al., 2009; Sandner et al., 2019). In healthy cells, we saw that BAY 58 was unable to displace haem from TC-sGCβ. This finding agrees with previous in vitro findings using purified forms of sGCβ (Fritz et al., 2011; Zhao et al., 2000) and is likely to exclude haem displacement as a possible mechanism of BAY 58 activation, under these conditions. This leaves BAY 58 to act by binding in the empty haem pocket of the immature apo-sGCβ subpopulation that is complexed with Hsp90 and is present in both healthy and stressed cells (Figure 8). Indeed, BAY 58 is known to drive the apo-sGC-Hsp90 species to form an active heterodimer in cells (Ghosh et al., 2014). In comparison, the NOC12 or ODQ treated cells additionally contain damaged forms of sGC, and we found that BAY 58 can displace haem from the TC-sGCβ under this circumstance. Such haem displacement from the damaged sGCβ forms may provide an additional way for BAY 58 to generate active sGC in the stressed cells. These concepts can now be investigated further and considered when designing or interpreting translational studies using sGC activators.

What is already known

• Cell oxidants or NO render soluble guanylyl cyclase (sGC) unable to participate in NO-sGC-cGMP signalling.

What does this study add

• sGC inactivation by oxidant ODQ or by NO involves sGC heterodimer dissociation without haem loss.

• ODQ and NO both drive free sGCβ subunits to bind with heat shock protein 90.

What is the clinical significance

• Understanding how sGC changes during inactivation can tailor strategies to preserve or rescue NO-sGC-cGMP signalling.

Abbreviations:

sGC

soluble guanylyl cyclase

RNS

reactive nitrogen species

ODQ

1H-[1,2,4]oxadiazolo[4,3-a]quinoxalin-1-one

NOC12

N-ethyl-2-(1-ethyl-2-hydroxy-2-nitrosohydrazino) ethanamine

BAY41

3-(4-amino-5-cyclopropylpyrimidin-2-yl)-1-(2-fluorobenzyl)-1H-pyrazolo[3,4-b]pyridine

BAY 58

4-[[(4-carboxybutyl)[2-[2-[[4-(2-phenylethyl)phenyl]methoxy]phenyl]ethyl] amino]methyl]benzoic acid hydrochloride

BAY 60

2-[[6-amino-3,5-dicyano-4-[4-(cyclopropylmethoxy)phenyl]-2-pyridinyl]thio]-acetamide

Hsp90

heat shock protein 90

SNO

S-nitrosated

FlAsH

4′,5′-bis(1,3,2-dithioarsolan-2-yl)fluorescein

TC

tetra-cysteine motif CCPGCC

MMTS

methylmethane thiosulfonate

DATA AVAILABILITY STATEMENT

The authors confirm that the data supporting the findings of this study are available within the article and its supporting information.