Guanylyl cyclases (GCs) are responsible for the production of the secondary messenger cyclic guanosine monophosphate (cGMP). cGMP plays important roles in a variety of physiological responses such as vision, olfaction, muscle contraction, homeostatic regulation, cardiovascular and nervous function 1. GCs are multi-domain proteins, which occur in two forms: receptor guanylyl cyclases (rGCs) and soluble guanylyl cyclases (sGCs). The phylogenetic relationships among GC isoforms of vertebrates and invertebrates have not been thoroughly investigated. We have used whole genome sequence data to investigate the evolution of GCs across a diverse range of animal phyla. Our analyses reveal that the GC family has undergone several lineage specific gene expansions, most notably in nematodes, echinoderms and vertebrates.

rGCs were first isolated in the echinoderms, where they are involved in chemotaxis between egg and sperm cells 2. Seven different classes of rGC genes have been found in mammals, each represented in humans by a single gene. Two of these genes (GC-D and GC-G), are considered to be pseudogenes in humans 3, while the remaining five genes encode functional rGCs. Two of these, GC-A and GC-B, are targets for the natriuretic peptides – a family of polypeptide hormones that act to reduce blood volume by stimulating natriuresis and diuresis in the kidney 4. The GC-C receptor was first described as the target for heat-stable enterotoxin secreted by pathogenic strains of Escherichia coli 5. GC-C is expressed in the intestine where it is involved in the regulation of fluid and electrolyte balance and its endogenous ligands have been identified as uroguanylin and guanylin 6. Retinal rGCs (GC-E and GC-F) play a critical role in vision, as they enhance the synthesis of cGMP in a negative Ca²⁺ modulated feedback loop. These retinal rGCs are regulated by small Ca²⁺-binding proteins that detect changes in cytoplasmic Ca²⁺ concentration and act through the cytoplasmic domain of the protein 7. Homologues of all the mammalian rGCs have also been detected in teleost fish 8.

The genome of Drosophila melanogaster contains six predicted rGC genes and the Anopheles genome is predicted to have an orthologue of each of these six genes 9. One rGC gene has been cloned from the tobacco hornworm Manduca sexta 10 and one from the silkmoth Bombyx mori 11. The physiological roles of rGCs in insects are not well characterised, but it has been established that the B. mori rGC, BMGC-1 is regulated in the flight muscles in a circadian fashion 12; that the M. sexta neural-specific rGC, MSGC-II is most similar to the vertebrate retinal guanylyl cyclases and is inhibited by Ca²⁺ 13 and that the rGC protein, Gcy76C, is required for axonal repulsion in D. melanogaster 14. Thus it seems likely that in insects, as in vertebrates, rGCs play important roles in a variety of physiological responses. The Caenorhabditis elegans genome contains at least 25 predicted rGC genes 1516. The expression patterns often of these genes have been investigated, and all ten are expressed in subsets of C. elegans sensory neurons. Mutant phenotypes have been described for two of these rGC genes: odr-1 16 and daf-11 17 and in both cases chemosensory signalling is affected.

rGCs contain an extracellular binding domain, a single membrane-spanning domain, a protein kinase homology domain (KHD) and intracellular coiled-coil dimerization and catalytic domains (Figure 1). The KHD has significant similarity with known protein kinases but contains no kinase activity; it functions as a negative regulatory element whose deletion by mutagenesis gives rise to a constitutively active GC receptor 18. The region between the cyclase domain and the KHD forms a coiled-coil domain, enabling the formation of dimeric proteins which are considered to be the minimal catalytic unit for human rGC enzymes 19.

sGCs are ubiquitously expressed in mammalian cells where they affect a variety of important physiological functions including smooth muscle relaxation, vasodilation, neuronal signal transduction, blood platelet reactivity and phototransduction 20. They are activated by nanomolar concentrations of nitric oxide (NO), a freely diffusible membrane permeant gas. In mammals sGC typically forms a heterodimer composed of an α- and a β-subunit, each of which contains a regulatory domain, a coiled-coil domain and a cyclase domain (Figure 1). The human genome encodes two sGC α-subunit and two sGC β-subunit genes. The N-terminal portion of the β-subunit constitutes the heme-binding domain that confers NO sensitivity to the enzyme. Upon activation of the sGC by NO the GC activity is accelerated by 100–300 fold 21. At the C-terminus of each subunit is a well-conserved catalytic domain. Contained between the heme-binding and catalytic regions is a dimerization domain responsible for heterodimer formation. A prokaryote heme binding protein family with significant sequence identity to the heme binding domain of eukaryotic sGCs has been identified 22. This heme-binding family was found in various bacterial lineages, but among the eukaryotes was detectable only in the animal lineage. Among the residues conserved in the prokaryote sequences are the histidine residue which covalently binds the heme prosthetic group and a YxS|TxR motif which has also been implicated in heme binding 23. sGC heme-like domains from the obligate anaerobe Thermoanaerobacter tengcongenesis and the facultative anaerobe Vibrio cholerae have been cloned 24. That study found that V. cholerae protein bound NO, whereas the T. tengcongenesis protein is capable of forming a stable O₂ complex and has NO binding characteristics similar to myoglobin and other O₂ sensors This heme-binding family has therefore been named H-NOX (Heme-Nitric oxide/Oxygen binding) 24 and is made up of two domains, the H-NOB (Heme NO Binding) and H-NOBA (Heme NO Binding Associated) 22. The H-NOBA domain occurs between the HNOB and the cyclase domains in animal sGCs 22.

The aim of the work reported here was to utilise whole genome data from vertebrate and invertebrate animals to describe a molecular phylogeny for both the soluble and receptor GCs. The most notable features of the resulting phylogenies are the number of lineage specific rGC and sGC expansions that have occurred during metazoan evolution. These lineage specific expansions have resulted in great diversity within the signal transduction and cellular communication pathways which are regulated by this functionally diverse multi-domain GC protein family.

Guanylyl- and adenylyl cyclases are multi-domain proteins that display a wide range of structural forms. Prokaryotes possess six classes (I-VI) of adenylyl cyclases (ACs), five of which exhibit a variety of structurally unique catalytic domains that are absent from eukaryotes 34. The class III purine nucleotide cyclase domain is present in all eukaryotes, however the diversity observed in eukaryotes for this class is only a small subset of the class III multi-domain cyclase proteins found in prokaryotes 35. In prokaryotes GCs are less abundant than ACs. To date only a single candidate prokaryote GC gene has been found (Cya2) in the cyanobacterium Synechocystis 36. Some authors postulate that the relative stability of cGMP, in comparison to cAMP, may have contributed to the absence of cGMP in bacteria, where a rapid turn-over of signalling molecules is required 35. There is no evidence for GC genes in any of the complete yeast genomes, nor have they been reported so far in other fungi 34. Similarly, while class III cyclases appear to be abundant in the chlorophyte algal antecedents of land plants, they are absent in flowering plants 34. Five class III purine nucleotide cyclase genes have been detected in the Dictyostelium discoideum 37. Class III cyclases have also been detected in several other phylogenetically divergent protistan phyla, however the diversity of structural forms displayed by these cyclases suggests that they are of paraphyletic origin 3438. The catalytic domains of class III ACs and GCs from metazoans appear to be monophyletic 3839. Functional similarities between the catalytic domains of metazoan ACs and GCs have also been demonstrated. For example, targeted replacement of two amino acids in the guanine-binding pocket of the rat retinal GC-1 receptor changed its specificity from GTP to ATP, while retaining its capacity to be activated by Ca²⁺-binding proteins 40.

Based on whole genome data and individual studies it is apparent that rGCs are a highly successful and diverse protein family which are utilised by vertebrate and invertebrate animals for a variety of roles in signal transduction and organismal homeostasis. The accretion of additional domains to the class III cyclase domain (especially the extracellular receptor, transmembrane and regulatory domains, Figure 1) has resulted in a very successful multi-domain rGC protein. This modular arrangement has facilitated the evolution of novel physiological rGC signalling pathways in different animal lineages through modifications to the receptor and regulatory domains. Subsequent gene duplications have further expanded and refined these pathways leading to several lineage specific gene expansions of rGC genes. Among these expansions is a large nematode specific rGC clade comprising 21 genes in C. elegans alone; a vertebrate specific expansion in the natriuretic receptors GC-A and GC-B; a distinct vertebrate specific expansion in the guanylyl GC-C receptor and an echinoderm specific expansion in the sperm rGC genes. Similar domain accretions and lineage expansions have also occurred in the sGCs, but the intracellular localisation of sGCs restricts the diversity of extracellular ligands which can activate them to membrane permeant molecules such as NO and O₂. Despite this, NO is an important signalling molecule involved in the regulation of diverse physiological mechanisms in the vertebrate cardiovascular, nervous and immune systems. The synthesis of NO by nitric oxide synthases (NOS) is controlled by complex intrinsic and extrinsic factors such as post translational modification, co-factor and substrate compartmentalization, phosphorylation and specific interactions with other proteins such as calmodulins 4142.

The D. melanogaster genome contains five genes encoding sGC subunits. Two of these genes, Gycα-99B and Gycβ-100B, encode α- and β-subunits which form a conventional heterodimeric NO-sensitive sGC, whereas the remaining three genes encode subunits with reduced sensitivity to NO 43. An sGC subunit from Manduca sexta was the first example of a sGC which exhibited enzyme activity without the need for co-expression of additional subunits. This M. sexta sGC is insensitive to NO 26 and forms active homodimers 44. The genome of C. elegans contains seven predicted sGC genes, but unlike D. melanogaster, a nitric oxide synthase gene has not been detected in the C. elegans genome 45. The sequences of all seven C. elegans sGCs are more similar to the β-subunits of mammalian guanylyl cyclases than to the α-subunits. However, while the C. elegans sequences conserve the heme-binding histidine residue, they lack two critical cysteine residues required for NO activation in mammalian sGC β-subunits 28. GCY-35 is required by C. elegans for avoiding hyperoxic conditions and unlike canonical NO-sensitive sGCs, it can bind oxygen 29. Similarly the D. melanogaster atypical sGCs have also been shown to function as molecular oxygen sensors 27.

In invertebrates with external fertilization, chemotaxis is a key event in guiding sperm to conspecific eggs. The jelly coat of echinoderm eggs releases species specific sperm-activating peptides 46 and echinoderm sperm have specific rGCs for these chemotactic molecules. Activation of sperm rGCs leads to a rapid, large but transient rise in cGMP and mediates ion fluxes across the sperm membrane. This in turn affects flagellar motion and the direction of movement 4748. According to our phylogenetic hypothesis (Figure 2) echinoderm sperm rGCs form a distinct lineage-specific clade within a larger clade of invertebrate rGCs. Whether other invertebrates with external fertilization use rGC signalling has not been established. In mammals with internal fertilization sperm chemotaxis is also a critical component of the fertilisation process, but here chemotactic responses depend on G protein coupled chemoreceptors. Interestingly, evidence is accumulating that sperm maturation and the acrosome reaction are induced in mammalian sperm by stimulation of an NO-sensitive sGC 49.

The natriuretic receptors GC-A and GC-B appear to be a vertebrate specific novelty – each represented by a single gene in mammals. Their ligands, the natriuretic peptides (NP), also comprise a small protein family. Available data indicate that there is a single NP and a single NP receptor in the jawless Agnathan fish 5051. Both the ligand and the receptor family differentiated during fish evolution in response to selection pressure to achieve body fluid homeostasis in osmotically variable aquatic environments. The medaka fish and puffer fish genomes contain six NP genes 51 together with two GCA and one GCB receptor genes 52. This level of complexity in natriuretic peptide signalling has not been retained in mammals, as mammalian genomes possess three NP genes and two rGC natriuretic receptor genes. It has been proposed that a reduction in the number of natriuretic ligands and their receptors in amniotes was associated with the transition from an osmotically variable aquatic environment to dry land, where the most important aspect of body fluid regulation became the retention of water 51. These vertebrate natriuretic rGCs are found within a larger clade, which also contains insect rGC sequences (Figure 2). Based on the available data it is impossible to determine if these insect receptors are involved in insect natriuresis. However, the Bombyx mori rGC found in this clade is expressed in the antennal lobe 11, therefore its function is more likely to be involved in chemoreception. The GC-C receptor, although also involved in salt regulation in the intestine, seem to have an independent phylogenetic origin from the natriuretic GC receptors.

In mammals the functional sGC unit is an α/β heterodimer which binds one heme group per dimer. Selective binding of NO at the heme iron activates the enzyme to convert GTP to the second messenger cGMP. The selectivity displayed by NO sensitive sGC is remarkable, considering that the heme in sGC is identical to that in the O₂ storage and transport proteins and that the concentration of O₂ is higher by 3 orders of magnitude than NO in eukaryotic cells 30. Our phylogenetic reconstruction shows the existence of a basal group of NO insensitive insect and nematode sGCs which are activated by O₂, implying that the primordial eukaryotes probably utilized sGC as an O₂ sensor. The M. sexta MsGC-3 subunit from the basal clade forms active sGC homodimers 2644, as also does its D. melanogaster homologue Gyc-88E 4353, while the C. elegans genome lacks sGC α-subunit genes. However, there is genetic evidence that the β-subunit like sGC proteins GCY-35 and GCY-36 of C. elegans can function as α/β-like heterodimers 54. Thus the initial form of the eukaryote O₂ sensitive sGC may have been a homodimer, with subsequent evolution and diversification of the α and β lineages leading to the formation of heterodimers, a shift to NO sensitivity and diversification of function.

Phylogenetic analyses suggest that that the sGC H-NOB domain was acquired by horizontal transfer from a cyanobacterial source 22. The H-NOB domain of facultative aerobic bacteria and the cyanobacteria is predicted to bind NO 30 and it is clear from Figure 7 that both Anabaena and Nostoc lack the Tyr-140 residue which is essential for the stabilisation of O₂ binding in the T. tengcongensis H-NOB. Thus the primordial eukaryotes probably obtained a cyanobacterial H-NOB domain which was sensitive to NO; substitution of one of the hydrophobic residues in the distal heme-binding pocket by a Tyr residue in the primordial animal sGC H-NOB domain would have changed the ligand specificity to O₂. Subsequent replacement of the Tyr residue in the distal heme pocket by an I1e residue in the β-1 and β-2 sGC lineages would have regenerated an NO sensitive sGC. Once NO sensitive sGCs evolved, they acquired additional physiological functions, including regulation of vascular and non vascular smooth muscle relaxation and vascular homeostasis 55, antimicrobial and anti tumor activity 56 as well as roles in neuronal survival and synaptic maintenance 57.

Lineage specific expansion of both the rGC and sGC gene families has occurred in nematodes, the largest of these is a rGC expansion comprising 21 genes (Figure 2). All seven C. elegans sGC are expressed in sensory neurons 1529 but in addition gcy-35 has a wider distribution, being also expressed in pharyngeal and body wall muscles and the excretory cell 29. The available expression patters for the rGC genes also implicate them in nervous system function. Previously it has been demonstrated that five rGCs are specifically expressed in sensory neurons or interneurons in C. elegans 15; expression of the rGCs gcy-5, gcy-6 and gcy-7 was observed in the ASE neurons which detect water soluble cues. The rGC gene, odr-1, is expressed in a subset of chemosensory neurons and is essential for responses to all volatile odorants sensed by the AWC neurons 16. Similarly the rGC gene daf-11 is expressed in a number of sensory neurons and daf-11 mutants have defects in dauer pheromone response and in their ability to detect certain odors. The sGC, GCY-35, has been shown to mediate oxygen sensation in C. elegans 29. Thus both sGCs and rGCs have been shown to have central roles in chemosensation in C. elegans. The chemosensory system of nematodes displays many differences from the olfactory systems of vertebrates and insects 58, largely resulting from the relatively small number of olfactory neurons in nematodes. C. elegans has only twelve pairs of sensory neurons within each of its two olfactory amphid organs. However the relative lack of anatomical complexity in the nematode sensory nervous system appears to have been compensated during nematode evolution by an increased functional complexity and multitasking capacity of individual sensory neurons 58. For example, individual olfactory neurons express multiple odor receptors, multiple heterotrimeric G protein α subunits, multiple GCs and they display a wide range of other, often-novel, mechanisms for signal integration within individual neurons. In consequence, lineage specific gene expansions are particularly noticeable in nematodes for neuronal gene families. For example the largest and most diverse nicotinic acetylcholine receptor gene family is that of C. elegans 59; novel families of potassium channels have been identified in C. elegans 60; a nematode specific expansion in the heterotrimeric G protein α-subunit gene family has been documented 61 and G protein coupled chemoreceptor genes comprise the largest gene family in C. elegans 62. Additionally, asymmetric expression patterns of neuronal genes increases the discriminatory power and olfactory potential of C. elegans 63. One such example is the asymmetric expression of rGC genes in the bilaterally symmetrical ASE taste receptor neurons 64. In adult worms the rGC genes gcy-6 and gcy-7 are only expressed in left sided ASE neurons, whereas gcy-5 is expressed only in right hand sided ASE neurons. This asymmetry of rGC expression correlates with a functional asymmetry of the left and right ASE neurons and thereby increases the odor discrimination capacity of the nematodes 65.

The chromosomal position of all "paranome genes" in C. elegans has recently been reported 66. The "paranome" is defined as the set of all duplicate genes in a genome 66. These authors found that duplications within the C. elegans genome are generally intrachromosomal while in S. cerevisiae they are usually interchromosomal. Chromosome V appears to have undergone a high degree of self-duplication in C. elegans, as 48.9% of its 4,792 genes are paranome members. C. elegans chromosomes II and IV also have a high percentage of paranome genes: 31.6% and 33% respectively 66. Three of the seven sGCs and seven of the 25 receptor GCs reside on chromosome V, a finding that supports these previous observations 66. The chromosomal location of all GCs in C. elegans is closely correlated with phylogenetic position on our reconstructed trees (this is especially true for rGCs).

Analysis of eukaryotic proteomes has shown that all but a small proportion of the eukaryotic protein repertoire is formed from protein domains which have been extant since the origin of eukaryotes 67. This trend is also apparent for the sGC and rGC proteins where the prokaryote progenitors of the Class III purine nucleotide cyclase, the H-NOX family of domains and the kinase homology domains have been identified. The conservation of the linear order of the individual domains of the rGC and sGC proteins, respectively, together with the extent of sequence identity across the entire lengths of these proteins from both vertebrate and invertebrate animals strongly suggests that each protein family is of monophyletic origin. The rGC and sGC proteins detected in the protistan systems investigated to date are more closely related in terms of sequence identity and domain topology to adenylyl cyclases (ACs) 386869. Metazoan membrane bound ACs are composed of two membrane domains, each consisting of six transmembrane helices. Each membrane domain is followed by a catalytic domain and the two catyalytic domains function as a functional heterodimer with a single catalytic pocket 70. In Dictyostelium discoideum the GC gene, DdGCA, encodes a protein with 12 transmembrane helices and two cyclase domains, 67, a configuration also found in the GCs of malaira parasite Plasmodium falciparum and the ciliates Paramecium and Tetrahymena 38. By contrast, membrane bound GCs of metazoans have only one single helix transmembrane domain and one cyclase domain. Thus the animal rGC and sGC families appear to have evolved after the divergence of the animal and protistan lineages. No rGC and sGC sequence information is currently available for the parazoa, so it is not known if the animal GCs evolved before the divergence of the parazoa and eumetazoa. The evolution of novel genes by the amalgamation of individual functional domains has been a frequent route for the emergence of new signal transduction and cell communication mechanisms in metazoans 7172737475. The modular arrangement of the sGC and rGC proteins has facilitated the evolution of novel signalling pathways in animal lineages through modifications to the receptor domains and by combining the cGMP product of GC activation with distinct downstream effectors such as cGMP dependent protein kinases, cGMP-gated ion channels and phosphodiesterases. The sGC, while sensitive only to membrane permeant NO or O₂, has coevolved with a very sensitive and complex NO production system which provides a very effective local cell-to-cell signalling system. Our phylogenetic analysis reveals that once the rGC and sGC multidomain proteins had evolved in the animal lineage subsequent gene duplications, tissue specific expression patterns and lineage specific expansions resulted in the evolution of new networks of interaction and new biological functions associated with the maintenance of organismal complexity and homeostasis.

GCs are responsible for the production of the secondary messenger cGMP, which plays important roles in a variety of physiological responses such as vision, olfaction, muscle contraction, homeostatic regulation, cardiovascular and nervous function. There are two types of GCs in animals, soluble sGCs which are found ubiquitously in cell cytoplasm, and receptor GC forms which span cell membranes. We have reconstructed molecular phylogenies for both sGC and rGC proteins. The most notable features of the resulting phylogenies are the number of lineage specific rGC and sGC expansions that have occurred during metazoan evolution. Among these expansions is a large nematode specific rGC clade; a vertebrate specific expansion in the natriuretic receptors GC-A and GC-B; a vertebrate specific expansion in the guanylyl GC-C receptor, an echinoderm specific expansion in the sperm rGC genes and a nematode specific sGC clade. The nematode specific GC genes identified within this study have expression and localisation patterns specific to sensory neurons. This expansion of the molecular diversity in individual neurons may compensate for the relative lack of anatomical complexity in the nematode sensory nervous system.

Our phylogenetic reconstruction also shows the existence of a basal group of nitric oxide (NO) insensitive insect and nematode sGCs which are activated by O₂. This suggests that the primordial eukaryotes probably utilized sGC as an O₂ sensor, and that the ligand specificity of sGC later switched to NO which provides a very effective local cell-to-cell signalling system.

Our phylogenetic analysis of animal and bacterial H-NOB domain sequences supports the hypothesis 22 that this domain originated from a cyanobacterial source. It has been shown that the introduction of a polar Tyr residue into the non polar distal pocket of the H-NOB domain is sufficient to change its binding specificity from NO to O₂. Our alignment of H-NOB domain sequences shows that non-polar residues only line the predicted distal pocket region of the β-1 and β-2 sGC sequences, which bind NO. However, all sequences from the basal sGC clade and the nematode specific clade which bind O₂ have a Tyr residue in the predicted distal pocket region (with the exception of GCY-37 which has a conservative Phe substitution). These observations support the hypothesis that the presence of a Tyr residue in the distal pocket of the H-NOB domain is necessary for O₂ binding, and is used to kinetically distinguish between NO and O₂ 3176.

All authors were involved in the design phase. DMO sourced all known nematode guanylyl cyclase proteins and DF located subsequent proteins from GenBank, completed genomes and EST databases. DF and DMO performed all EST database searches. DF and DMO performed the phylogenetic analysis and chromosomal locations of nematode genes. AMB directed the project and provided advice on all analyses. All authors were involved in the drafting of the manuscript and approved the final manuscript.