Article ID: CJ-15-0025
Soluble guanylyl cyclase (sGC) is the physiological receptor for nitric oxide (NO) and NO-releasing drugs, and is a key enzyme in several cardiovascular signaling pathways. Its activation induces the synthesis of the second messenger cGMP. cGMP regulates the activity of various downstream proteins, including cGMP-dependent protein kinase G, cGMP-dependent phosphodiesterases and cyclic nucleotide gated ion channels leading to vascular relaxation, inhibition of platelet aggregation, and modified neurotransmission. Diminished sGC function contributes to a number of disorders, including cardiovascular diseases. Knowledge of its regulation is a prerequisite for understanding the pathophysiology of deficient sGC signaling. In this review we consolidate the available information on sGC signaling, including the molecular biology and genetics of sGC transcription, translation and function, including the effect of rare variants, and present possible new targets for the development of personalized medicine in vascular diseases.
The major components of the nitric oxide (NO)/cyclic guanosine-3’,5’-monophosphate (cGMP) pathway were identified in the late 1980s.1,2 NO is biosynthesized endogenously through sequential oxidation of the amino acid L-arginine by an enzyme family called the NO synthases (NOS).3 Three isoforms of NOS with different tissue distributions are known: neuronal NOS (nNOS/NOS1), inducible NOS in macrophages (iNOS/NOS2) and endothelial NOS (eNOS/NOS3).4 In the cardiovascular system, eNOS is the main source of NO production.5,6 The formation of NO by eNOS is increased by various stimuli such as platelet-derived factors, shear stress, acetylcholine, and cytokines.7
NO mediates its functions through its primary receptor: soluble guanylyl cyclase (sGC). Together with the adenylate cyclases, sGC belongs to the class III purine nucleotidyl cyclase family.8 Binding of NO to the heme moiety of sGC induces the transition from basal to activated sGC. Activated sGC converts guanosine-5’-triphosphate (GTP) to cGMP and pyrophosphate (PPi).9 PPi is emerging as a major factor in preventing vascular calcification.10 cGMP acts as a ubiquitous second messenger in intracellular signaling cascades, which serves to regulate the activity of a number of downstream proteins, including cGMP-dependent protein kinase G (PKG), cGMP-dependent phosphodiesterases (PDE) and cyclic nucleotide gated ion channels. The signals propagated through cGMP are varied and include vascular smooth muscle cell (VSMC) relaxation,11 inhibited platelet aggregation12 and modified neurotransmission,13 for example.
sGC is composed of 2 subunits: α and β.14 In 1981 Gerzer et al showed that sGC contains a heme in form of ferroprotoporphyrin IX.15 However, it was unknown for a long time whether this prosthetic heme group is sandwiched between the α and β subunits or whether it exclusively binds to the β subunit. In 1997, Zhao and Marletta demonstrated the ferrous heme to be ligated to the N-terminal part of the β subunit at His105.16
In humans, 2 types of each subunit exist: α1 and α2 for the α subunit and β1 and β2 for the β subunit. These 4 proteins are encoded by 4 distinct genes: GUCY1A3 (α1), GUCY1A2 (α2); GUCY1B3 (β1), and GUCY1B2 (β2). The genes for human α1 and β1 have been mapped to chromosome 4,17 and those encoding α2 and β2 to chromosomes 1118 and 13,19 respectively (Table 1). Dimerization of the enzyme is a prerequisite for its catalytic activity.20 Both α subunits give rise to a functional enzyme when coexpressed with β1 (ie, both α1/β1 and α2/β1 heterodimers are activated by NO).21 Although heterodimers seem to be the preferred form in cells, the features that determine this preference, and any possible role for homodimers, are not fully defined. Koglin et al could show that β2 does not exhibit cyclase activity when expressed with either α1 or α2 but that β2 is active in the absence of an α subunit.22 This implies that the β2 protein can function as a homodimer ex vivo. However, the physiological role of the β2 subunit in cGMP signaling remains elusive. By contrast, Zabel et al were able to overexpress α1/α1 and β1/β1 homodimers in Sf9 cells, both being catalytically inactive.23
| Gene | Chromosomal location |
Transcript variant | Exons | Isoform | Protein size (aa) |
|---|---|---|---|---|---|
| GUCY1A3 | 4q31.1-q31.2 | GUCY1A3-Tr1 | 11 | α1-IsoA | 690 |
| GUCY1A3-Tr2 | 10 | ||||
| GUCY1A3-Tr3 | 10 | ||||
| GUCY1A3-Tr4 | 8 | ||||
| GUCY1A3-Tr5 | 10 | α1-IsoB | 455 | ||
| GUCY1A3-Tr7 | 9 | α1-IsoD | 624 | ||
| GUCY1A3-Tr8 | 10 | α1-IsoA | 690 | ||
| GUCY1A2 | 11q21-q22 | GUCY1A3-Tr1 | 9 | α2i | 763 |
| GUCY1A3-Tr2 | 8 | α2 | 732 | ||
| GUCY1B3 | 4q31.3-q33 | GUCY1B3-Tr1 | 15 | β1-Iso1 | 641 |
| GUCY1B3-Tr2 | 14 | β1-Iso2 | 619 | ||
| GUCY1B3-Tr3 | 15 | β1-Iso3 | 599 | ||
| GUCY1B3-Tr4 | 16 | β1-Iso4 | 594 | ||
| GUCY1B3-Tr5 | 14 | β1-Iso5 | 586 | ||
| GUCY1B3-Tr6 | 13 | β1-Iso6 | 551 | ||
| GUCY1B2 | 13q14.3 | – | 17 | – | 617 |
Diminished expression and function of sGC contributes to the pathogenesis of several cardiovascular disorders such as coronary artery disease (CAD), atherosclerosis and hypertension.24 Most recently, the chromosomal locus harboring GUCY1A3 (α1) and GUCY1B3 (β1) was shown to contain genetic variants with genome-wide significant association to CAD and hypertension.25–27 Thus, the GUCY1A3/GUCY1B3 locus adds to the growing list of those contributing to CAD and myocardial infarction (MI) risk.28,29
The expression of sGC subunits is modulated at different levels, including inhibition of transcription,30,31 destabilization of mRNA,32,33 and protein degradation.34 The role of alternative splicing in this process still needs to be uncovered. Precise understanding of sGC splicing regulation could serve as a target for new therapeutic interventions and help to personalize sGC-targeting therapies in the treatment of vascular disease.
As demonstrated by several studies, human α1, α2 and β1 exist as different isoforms because of alternatively spliced transcript variants35–37 (Table 1). Besides several predicted sequences, NCBI nucleotide database research reveals a total of 7 alternatively spliced human transcript variants for GUCY1A3, 2 for GUCY1A2 and 6 for GUCY1B3. In the case of GUCY1A2 (α2 and α2i) and GUCY1B3 (β1-Iso1 to Iso6), each transcript variant codes for 1 unique isoform. However, the 7 alternatively spliced variants of GUCY1A3 only code for 3 different isoforms: canonical full-length 690 aa α1-IsoA (GUCY1A3-Tr1 to Tr4 and GUCY1A3-Tr8), 455 aa α1-IsoB (N-term∆235aa; GUCY1A3-Tr5) and 624 aa α1-IsoD (C-term∆66aa; GUCY1A3-Tr7). GUCY1A3-Tr1 to Tr4 coding for the identical α1-IsoA differ from each other only in their 5’- and 3’-UTR sequences, which likely affects mRNA stability.
In mammals, a splice variant of α2 (α2i) generates a dominant negative variant when forming a dimer with β1 because α2i contains an in-frame insertion of 31 amino acids within the catalytic domain.37 Concerning β2, the NCBI nucleotide database just provides information on a single human β2 transcript (Table 1). Because at least 3 different β2 transcripts have been observed in humans38–40 and different β2 isoforms have been cloned from rats,41 there is evidence that β2 also exists as different isoforms in humans.
Recently, Martin et al36 conducted a study of the role of alternative splicing of GUCY1A3 and GUCY1B3 sGC in healthy and diseased human vascular tissue. As splicing regulation of sGC and its biological role in vascular tissue has not been previously examined in vivo, they were the first to show splicing diminishing sGC function. Using a semiquantitative reverse transcriptase PCR approach, they uncovered various GUCY1A3 and GUCY1B3 splice variants in human aorta. Quantifying the total levels of GUCY1A3 and GUCY1B3 sGC transcripts using quantitative PCR revealed a 3.2- and 2.3-fold increase in, respectively, GUCY1A3 and GUCY1B3 mRNA in aortas with aneurysms compared with healthy control aortas. Interestingly, the aortas with aneurysms demonstrated decreased sGC activity that correlated with increased expression of dysfunctional sGC splice variants, because the composition of the splice forms in the aortas with aneurysms differed from that in control aortas (Figure 1).
Expression of different transcripts coding for sGC α1 and β1 in aneurysm and control aortas. Splice variants visualized on agarose-gels after RT-PCR. (A) GUCY1A3-Tr1 and GUCY1A3-Tr7 were higher in diseased aortas, GUCY1A3-Tr7 almost exclusively; GUCY1A3-Tr5 was more abundant in controls. (B) All GUCY1B3 transcripts shown were predominantly expressed in aortas with aneurysm. RT-PCR, reverse transcriptase polymerase chain reaction; sGC, soluble guanylyl cyclase. (Reprinted with permisson from Martin E, et al.36 Alternative splicing impairs soluble guanylyl cyclase function in aortic aneurysm.)
GUCY1A3-Tr1 coding for α1-IsoA and GUCY1A3-Tr7 coding for α1-IsoD were higher in diseased aortas. α1-IsoD lacks 66 C-terminal amino acids and has impaired enzymatic activity. In contrast, the level GUCY1A3-Tr5 coding for α1-IsoB was lower in the aneurysm samples. α1-IsoB is 235 amino acids shorter at the N-terminus and oxidation resistant.42 The immunoprecipitation studies and activity evaluation presented by Martin et al clearly demonstrated that α1-IsoB forms a functional heterodimer with β1 subunit in aorta in vivo. Concerning the expression of β1 isoforms, they found GUCY1B3-Tr1, GUCY1B3-Tr4 and GUCY1B3-Tr6 being more highly expressed in the aneurysm group than in controls. β1-Iso1 contains a 22 amino acid insertion in the N-terminal regulatory heme-NO/oxygen domain (H-NOX) domain, so heme-NO binding might be negatively affected. β1-Iso6 with a large 68 amino acid deletion in the H-NOX domain would also have impaired heme function. The β1-Iso4 not only carries the same 68-residue deletion, but also a 43 amino acid insertion close to the catalytic domain.
Comparison of sGC activity measurement of aortic lysates from control samples that predominantly expressed either α1-IsoA or α1-IsoB proteins showed α1-IsoB compensating for low levels of canonical α1-IsoA. In short, NO-induced sGC activity was comparable, regardless of whether α1-IsoA or α1-IsoB was predominant. By contrast, the sGC heterodimer containing α1-IsoD exhibited a diminished activation. Martin et al evaluated the enzymatic properties of the recombinant α1-IsoD subunit in Cos7 cells. Despite a detectable basal level of cGMP-forming activity confirming the formation of a catalytically active heterodimer, stimulation with NO only increased the cGMP amount marginally in contrast to α1-IsoA/β1 sGC. The diminished function of α1-IsoD could arise from the close proximity of the 66 amino acid deletion to the catalytic domain.
The functionality of α1-IsoB was already shown in 2003 by Koglin and Behrends.43 They analyzed N-terminal deletion mutants of the human α1 subunit after co-expression with the human β1 subunit. They observed that deleting the first 258 amino acids of the α1 subunit exerted an effect on neither sensitivity to NO nor heme binding. Unfortunately this finding is inconsistent with Wagner et al reported in 2005.44 They tested the effect of different truncations at the N-terminus of α1 and β1 on dimerization and found the amino acids 61–462 in α1 to build up the shortest possible fragment exhibiting wild-type-like dimerization. In a follow-up study in 2011, Kraehling et al showed that that the translation initiation site in GUCY1A3-Tr5 at position 259 is dominant over ATG at position 236 in the human sequence.42 Therefore, translation of GUCY1A3-Tr5 should lead to a subunit with a N-term∆258aa and not N-term∆235aa deletion. Our working group also investigated the translation initiation at both ATG236 and ATG259 of α1-IsoB but in a different way. Kraehling et al generated their construct coding for α1-IsoB-ATG259 by exchange of a nucleotide in the triplet coding for ATG236.42 Our coding sequence for α1-IsoB-ATG259 directly started with the triplet coding for ATG259, lacking the 69 “non-coding“ nucleotides upstream. When transfecting HEK293E cells with both the sequences coding for α1-IsoB-ATG236 and α1-IsoB-ATG259 the corresponding proteins resulted in bands of the same size on western blotting (unpublished data). Therefore, the hypothesis of ATG259 being dominant over ATG236 was confirmed.
Each sGC subunit is a multi-domain protein comprising 4 functionally different parts: (1) H-NOX, (2) a Per/Arnt/Sim-like domain (PAS), (3) an α-helical region capable of forming coiled-coils involved in dimerization (CC), and (4) a C-terminal catalytic domain (CAT) where the GTP binding and conversion takes place45 (Figure 2). These 4 specified domains form 2 rigid units within the sGC: the smaller unit comprises the dimeric catalytic domain, and the larger one is built from the clustering of the PAS and H-NOX domains. The helical domains form a dimeric parallel coiled-coil that flexibly connects the 2 modules.46
sGC domain organization and X-ray crystallographic models. Each subunit contains 4 modular domains; α1 domains are shown in shades of gray, and β1 domains are shown in color. The H-NOX domain of the β1 subunit contains the heme cofactor, shown in red. H-NOX, heme-NO/oxygen domain; PAS, Per/Arnt/Sim-like domain; sGC, soluble guanylyl cyclase. (Reprinted with permission from Campbell MG, et al.46 Single-particle EM reveals the higher-order domain architecture of soluble guanylate cyclase.)
Crystal structures of the independent domains have already been reported.47–50 Recently, Campbell et al were the first to show the 3D structure of Rattus norvegicus sGC holoenzyme using negative-stain electron microscopy.46 Still, no high-resolution 3D structure of the complete human holoenzyme is available to date. Determining the structure of full-length sGC is a prerequisite to understanding its function and for the design and improvement of therapeutics for treatment of related diseases.
Analogous to the transmembrane guanylyl cyclases, where binding of ANP is transmitted across the transmembrane helices leading to an active conformation of the 2 intracellular domains,51 it was assumed that binding of NO to the N-terminal H-NOX domain is transmitted to the C-terminal CAT across the coiled-coil domain.52 This linear transmission model disagrees with the findings from Winger et al,53 who showed that the isolated H-NOX domain can directly interact with the isolated catalytic region of sGC. Consistent with those results, Haase et al demonstrated the N-termini of sGC being in close proximity to the C-termini using fluorescence resonance energy transfer (FRET).54 In a very recent study, Busker et al55 studied the conformational change of full-length sGC under NO-stimulated conditions. As sGC contains 5 tryptophane residues distributed evenly over all 4 functional domains, Busker et al used these as donors for FRET. The substrate analog 2’-Mant-3’-dGTP was used as acceptor, making it possible to identify movements of the functional domains relative to the substrate-binding catalytic region. Their FRET signals indicated Trp-22 and Trp-466 were in close proximity to the catalytic domain upon activation of NO, which means that activation of sGC by binding of NO to the β1 H-NOX domain is transmitted to the catalytic domain both through the α1 coiled-coil domain and by direct interdomain interaction between the H-NOX and catalytic domain forcing the catalytic domain into the NO-activated conformation. In addition, Campbell et al46 described a second direct allosteric control mechanism through interaction between H-NOX and the PAS domain, as previously observed by others.56,57 Campbell et al showed that these 2 domains form a tight cluster, sharing large surfaces of interactions, and allowing each H-NOX domain to interact with both the α1 and β1 PAS domains, allowing small-scale changes in the H-NOX domain to be quickly recognized by the adjacent PAS. Contrary to Busker et al’s observations, Campbell et al did not describe a dramatic conformational change between the NO-bound and unbound states, which led them to assume that ligand binding only induces small-scale intradomain conformational changes.
Budworth et al investigated the localization of the subunits in humans and found α1 and β1 to be expressed in most tissues. The α2 subunit is found in fewer tissues, but is highly expressed in the brain, lung, colon, heart, spleen, uterus, and placenta.58 Pharmacological and biochemical kinetic studies conducted by Russwurm et al demonstrated that the naturally occurring sGC isoforms, α1/β1 and α2/β1, exhibit similar sensitivities to NO in vitro.21 Further studies by Bellingham and Evans59 showed that the differential biological effects of the 2 forms are based on their localization. Although α1/β1 sGC is primarily localized in the cytosol, thus producing an unfocussed source of cGMP, α2/β1 has a tendency to localize at the membrane, providing a localized pool of cGMP at this site.59 Bellingham and Evans measured the functional properties of α2/β1 by utilizing the NO-dependent activation of the ion channel cystic fibrosis transmembrane conductance regulator (CFTR), which occurs by phosphorylation via the membrane-bound PKGII isoform. They found that cGMP generated by α2/β1 activates CFTR far more effectively than the cytoplasmically located α1/β1, despite near identical catalytic properties. This suggests α2/β1 to be of general importance in mediating the membrane effects of NO and a potentially important selective drug target. However, the 150-kDa α1/β1 heterodimer is regarded as the most physiologically relevant isoform and therefore the most extensively studied one. The functional importance of α1/β1 sGC was demonstrated by the significantly decreased relaxing effects of major vasodilators such as acetylcholine, NO, YC-1 and BAY 41-2272 in α1 sGC knockout mice.60
The formation of the NO-heme complex is responsible for an up to 250-fold increase in the GTP cyclase activity rate of the enzyme.61 NO binds to the heme of sGC forming an unstable 6-coordinate complex, which rapidly converts into a 5-coordinate complex because of disruption of the coordinating bond between His105 and the heme.62 Once NO dissociates from sGC, basal cGMP production is restored,45 which ensures sGC activity is quickly up- and downregulated. However, not only is sGC activity itself regulated but also the amount of cGMP is controlled by certain PDEs that break the phosphodiester bond within cGMP, hydrolyzing it to GMP.63 In total, 11 different types of isoenzymes, each with several isoforms, exist. Whereas some PDEs are said to be cGMP-selective, because of their 100-fold substrate preference for cGMP over adenosine-3’,5’-monophosphate (cAMP), others are specific for hydrolyzing cAMP, and some PDEs can hydrolyze both cAMP and cGMP.64 For cGMP cleavage, PDE5 is considered to be the most important player in humans, with regard to its catalytic affinity being in the physiologically range and PDE5 being expressed in most peripheral tissues (ie, VSMC, plateletes, heart65).
Growing evidence indicates that imbalance of intracellular cGMP levels from dysregulation of either sGC or PDE5 plays a role in the risk for CAD and MI.
Recently, various mutations in the coding sequence of GUCY1A3 were found by our group to be associated with CAD/MI.66 The starting point was a family with 32 members of whom 22 had CAD under the age of 60 years. Exome-sequencing in 3 distantly related family members with MI revealed a digenic mutation in GUCY1A3 and CCT7. CCT7 codes for CCTη, which acts as a chaperonin folding the α and β subunits.67 Both mutations were absent in over 3,000 controls and in over 3,000 unrelated MI cases. However, analysis of exome-sequencing data of 252 young MI cases uncovered another 5 rare missense variants associated with MI.66 Moreover, sequencing GUCY1A3 in 48 patients from another 22 additional extended MI families revealed p.Gly537Arg, a rare mutation cosegregating with the disease in the family. These overall 7 rare variants of GUCY1A3 are currently under investigation by our working group in consideration of levels of protein expression, dimerization and enzyme activity (Table 2). Preliminary data have already revealed reduced protein expression, as well as reduced sGC activity by measuring NO-induced cGMP in p.Leu163Phefs*24 and p.Gly537Arg mutants compared with the wild type (Figure 3).66
| Mutation | Where identified |
|---|---|
| p.Leu163Phefs*24 | MI family |
| p.Gly537Arg | MI family |
| p.Lys53Glu | 252 young MI cases |
| p.Thr64Ala | 252 young MI cases |
| p.Thr229Met | 252 young MI cases |
| p.Ser478Gly | 252 young MI cases |
| p.Val587Ile | 252 young MI cases |
MI, myocardial infarction.
Reduced expression and enzymatic activity of the identified sGC mutants in HEK293 cells. (A) GUCY1A3 mutants were expressed together with GUCY1B3 in HEK293E cells. α1-sGC protein was strongly reduced in mutants (numbers in bars are independent transfections performed in duplicates). (B) Accordingly, NO-induced concentration-dependent cGMP formation was significantly attenuated after transfection with mutant proteins. AU, arbitrary units; GSNO, S-nitrosoglutathione; sGC, soluble guanylyl cyclase; WT, wild type. (Reprinted with permisson from Erdmann J, et al.66 Dysfunctional nitric oxide signalling increases risk of myocardial infarction.)
In this review we highlighted the fundamentals of sGC from it numerous forms of appearance at the mRNA level through its diversity in expression to the functional aspects. In the vasculature, the most important source of NO is eNOS. NO again is the main stimulus for sGC to transform GTP into the ubiquitous second messenger cGMP, which regulates many downstream proteins finally influencing vessel tonus and platelet aggregation, for example. Decreased cGMP production plays a decisive role in the pathogenesis of several disorders, including cardiovascular diseases. In their recent study, Martin et al were the first to nail down the relationship between splice variants of GUCY1A3 and GUCY1B3 and a phenotype in human aortas. Moreover, a recent genetic study revealed that alternative splice variants might henceforth play an important role as targets in the treatment of cardiovascular diseases.
The study was supported by the German Federal Ministry of Education and Research (BMBF) in the context of the e:Med program (e:AtheroSysMed) and the FP7 European Union project CVgenes@target (261123). Further grants were received by the Fondation Leducq (CADgenomics: Understanding Coronary Artery Disease Genes, 12CVD02) and the Deutsche Forschungsgemeinschaft (DFG SFB 1123).
Conflict of Interest: The authors declare no conflicts of interest.
