Localisation And Characteristics of Natriuretic Peptide Receptors in the Gills of the Atlantic Hagfish Myxine Glutinosa (Agnatha)

Natriuretic peptides (NPs: ANP, BNP and CNP) are a family of peptide hormones found principally in the heart and central nervous system. They reduce blood pressure and hypervolaemia in mammals by decreasing cardiac output, reducing peripheral vascular resistance (partly by relaxation of vascular smooth muscle) and decreasing intravascular volume. In addition, blood volume is reduced directly by the potent diuretic and natriuretic effects of NPs in the kidney and secondarily by the inhibition of the release of aldosterone from the adrenal glands and renin from the juxtaglomerular cells (Brenner et al. 1990; Ruskoaho, 1992). The functions of NPs in the central nervous system appear to be the control of water intake and salt preference and the inhibition of vasopressin secretion (Samson, 1990).

The effects of NPs are mediated through two membrane-bound receptor types (NPRs); guanylate-cyclase-coupled receptors (approximate molecular mass 130 kDa) that activate the guanosine 3’,5’-cyclic monophosphate (cyclic GMP) intracellular second-messenger system (Martin et al. 1989; Koller and Goeddel, 1992), and the ‘clearance’ receptor (NPR-C, a homodimer of a 65 kDa protein) that is not coupled to guanylate cyclase activity. NPR-C was originally named because it was believed to modulate circulating concentrations of NPs by their removal from the blood (Chinkers and Garbers, 1991; Maack, 1992). Some studies suggest that NPR-C is linked with second-messenger systems other than cyclic-GMP-mediated mechanisms (Levin, 1993). At least two guanylate-cyclase-linked receptors have been identified to date: NPR-A and NPR-B. NPR-A appears preferentially to bind ANP, but will also bind BNP and to a lesser extent to CNP; whereas NPR-B binds CNP preferentially (Koller and Goeddel, 1992). These receptors require an intact cysteine ring for NPs to bind. However, NPR-C can bind a diversity of NPs, including truncated and ring-deleted analogues (Brenner et al. 1990).

Since the role of ANP in mammalian salt and water balance was discovered, interest has developed in the possible action of NPs in fish osmoregulation (for a review, see Evans, 1990, 1993, 1995; Evans and Takei, 1992). Immunohistochemical studies and radioimmunoassays indicate that all classes of fishes have NP systems (Reinecke et al. 1985, 1987; Evans et al. 1989; Vallarino et al. 1990; Donald and Evans, 1992; Donald et al. 1992), and various NPs have been isolated and sequenced from both teleosts and elasmobranchs (Takei et al. 1989, 1990, 1991; Price et al. 1990; Schofield et al. 1991; Suzuki et al. 1991, 1992). There is some evidence to suggest that NPs may function as a saltwater hormone in fishes, since plasma NP immunoreactivity is greater in some fishes adapted to higher salinities (Westenfelder et al. 1988; Evans et al. 1989; Freeman and Bernard, 1990; Smith et al. 1991); the eel (Anguilla japónica), however, appears to be an exception since ANP plasma concentrations decline in higher salinities (Takei and Balment, 1993).

There have been a few studies on NP binding sites in teleosts and elasmobranchs. ¹²⁵I-ANP binding sites have been found in the heart of Conger conger (Cerra et al. 1992), in the kidney, gills and heart of two species of Antarctic fish (Uva et al. 1993) and in chondrocytes of gill cartilage and parenchymal cells of the secondary lamellae of the Japanese eel Anguilla japonica (Sakaguchi et al. 1993). ¹²⁵I-CNP binding has been observed in the rectal gland of the dogfish shark Squalus acanthias, and CNP stimulates cyclic GMP production in this tissue (Gunning et al. 1993). ANP has been found to stimulate cyclic GMP production in isolated rainbow trout nephrons (Perrott et al. 1993). In a series of studies, Duff and Olson (1992) and Olson and Duff (1993) have identified NPR-C-like receptors in the gills of the rainbow trout. These binding sites are capable of removing 60% of ¹²⁵I-ANP from the circulation in a single pass through the gills.

Unlike all other fishes that are obligatory osmoregulators, the hagfish, whose blood is virtually iso-osmotic with sea water, does not have an osmotic problem and behaves physiologically like an osmoconforming marine invertebrate (Hardisty, 1979). However, hagfish do have an NP system: the heart, brain and plasma of Myxine glutinosa have NP-like immunoreactivity (Reinecke et al. 1987; Evans et al. 1989; Donald et al. 1992), and NP binding sites have been located in the archinephric ducts and glomeruli of the kidney and in the ventral aorta (Kloas et al. 1988). In addition, rat ANP and CNPs, which dilate the ventral aortic vascular smooth muscle of both the toadfish Opsanus beta and the dogfish shark Squalus acanthias, also dilate that of the hagfish Myxine glutinosa (Evans et al. 1989, 1993; Price et al. 1990; Evans, 1991). Dilation of the ventral aorta could alter the dynamics of gill blood perfusion. Because hagfishes diverged from other vertebrate groups at least 500 million years ago (Forey and Janvier, 1993), they are an interesting group in which to study the evolution of NPs and their receptors. Similarities between the NP system of hagfish and higher vertebrates are presumably indicative of the ancestral vertebrate condition. However, the converse is not necessarily true: hagfish characteristics that are not shared with higher vertebrates may not be primitive, but may be derived characteristics resulting from the separate NP evolution since the separation of myxinoids from the main vertebrate line. The present study examines the presence and character of NP binding sites in the gill of the Atlantic hagfish Myxine glutinosa using autoradiography, radioligand binding assays and guanylate cyclase assays.

Hagfish were collected from the Bay of Fundy, supplied by Huntsman Marine Laboratory, St Andrews, NB, Canada, and maintained in running sea water (SW, 12–14 °C) at the Mount Desert Island Marine Laboratory, Maine, or in 10 °C tanks aerated through charcoal/fibre filters at the University of Florida, Gainesville. Animals were allowed to equilibrate for 3 weeks before experimentation. Hagfish were anaesthetised in MS 222 (1:1000, Sigma, St Louis, MO) before dissection.

After dissection, tissues were freeze-mounted in Tissue Tek (Miles Inc. Elkhart, Indiana) and sectioned in a microtome cryostat (Minotome, IEC, Massachusetts). 18 μm sections were mounted on gelatin-chromium-aluminium-coated slides and dried overnight under vacuum at 4 °C. The sections were stored in sealed boxes at –20 °C until used.

Sections were preincubated for 15 min at room temperature (22-24 °C) in 50mmoll⁻¹ Tris-HCl buffer (pH 7.4), 50 mmol l⁻¹ NaCl, 5 mmol l⁻¹ MgCh, 0.1% bovine serum albumin (BSA) and 0.05 % bacitracin. The sections were then incubated for 90 min in the same buffer supplemented with 4 μ gml ¹ leupeptin, 2 μgml⁻¹ chymostatin, 2 μgml⁻¹ pepstatin, 1 μmoll^-1 PMFS (phenylmethylsulphonylfluoride) and rat (3-[¹²⁵I]iodotyrosol²⁸) atrial natriuretic peptide (74 TBq mmol ^-1; Amersham, Illinois) or human, porcine, rat (¹²⁵I-[Tyr⁰]) C-type natriuretic peptide-22 (55.5TBqmmol ^-1; Peninsula Laboratories, California). Nonspecific binding was determined in adjacent sections in the presence of 1 μmol l⁻¹ unlabelled rat 3-28 ANP (rANP, Bachem, California) for ¹²⁵I-ANP-incubated sections and 1 μmol ⁻¹ porcine CNP (pCNP; Bachem, California) for ¹²⁵I-CNP-incubated sections. The ability of 1 μmoll^{− 1} pCNP (¹²⁵I-ANP-labelled sections), 1 μmoll⁻¹ rANP (¹²⁵I-CNP-labelled sections) and 1 μmoll ⁻¹ rat des[Gln¹⁸, Ser¹⁹, Gly²⁰, Leu²¹, Gly²²]ANP-(4-23)-NH2 (C-ANF; Bachem, California) to displace specific radiolabelled NP binding was also determined in adjacent sections. C-ANF is a truncated ANP that binds only to NPR-C in mammals (Maack et al. 1987).

Following incubation, the slides were washed (2×10 min at 4 °C) in 50 mmol l⁻¹ Tris-HCl buffer, fixed for 20 min in 4% formaldehyde in 0.1 moll⁻¹ phosphate buffer (pH 7.4, 4 °C), washed in 0.1 mol l⁻¹ phosphate buffer (pH 7.4, 4 °C) and then in distilled water (1 min), dehydrated through alcohols and dried overnight at 60 °C. Sections were apposed to Hyperfilm-βmax (Amersham, Illinois) for 5 days at room temperature. The film was processed using Kodak GBX developer (4 min), rinsed in water (2 min) and fixed with Kodak GBX fixer (5 min).

For examination of binding sites with light microscopy, some sections were dipped in nuclear track emulsion (Kodak NTB.2) at 43 °C. After drying, the sections were stored for 10 days at 4 °C and then developed in Kodak D 19 (3 min), washed in water and fixed in Kodak Rapid Fixer diluted 1:1 (7 min). Subsequently, they were stained in 1% Toluidine Blue, examined with an Olympus BH-2 microscope and photomicrographs made with a Wild Leitz MPS 46 Photoautomat camera on Kodak T-max 100 black and white film.

Gill membranes were prepared from individual hagfish for saturation and competition binding studies, affinity cross-linking followed by SDS-polyacrylamide electrophoresis and for guanylate cyclase assays. The gill pouches were removed from anaesthetised hagfish, placed in a 50 ml centrifuge tube in 5 ml of ice-cold 50 mmol l⁻¹ Tris-HCl and 1mmoll⁻¹ NaHCO3 (pH 7.4) and quickly homogenised with a Tissue-Tearor (Biospec, Bartlesville, Oklahoma). The homogenate was diluted with 5 ml of 50 mmol l⁻¹ Tris-HCl, 1mmoll⁻¹ EDTA and 1 mmol l⁻¹ MgCb (pH 7.4) and centrifuged at 800 g for 10 min at 4 °C. The supernatant was collected and centrifuged at 30 000 g for 20 min. The pellet was washed with 50 mmol l⁻¹ Tris-HCl (pH 7.4) and 250 mmol l⁻¹ sucrose and resuspended in 400 μl of the same sucrose buffer. Protein concentration was determined with a BCA protein assay kit (Pierce), calibrated against BSA standards. Membranes were stored at –70 °C until use.

For the ANP saturation curves, 75 μg of gill membrane protein was incubated in 250 μl of incubation buffer (the same as was used for sections, see above) in the presence of increasing concentrations (5-300 pmoll⁻¹) of ¹²⁵I-rANP. Binding reactions were performed for 90 min at 23 °C. The reaction was terminated by the addition of 2 ml of ice-cold 50 mmoll⁻¹ Tris-HCl (pH7.4) and the preparation was filtered through 1 % polyethylenimine-treated Whatman GF/C filters. Filters were washed with 5 ml of 50 mmol l⁻¹ Tris-HCl (pH 7.4), and the radioactivity was measured in a Beckman gamma counter with 78 % efficiency. For competition binding studies, 75 μg of gill membrane protein was incubated in 250 μl of the incubation buffer and 25 pmoll⁻¹ of ¹²⁵I-rANP or ¹²⁵I-pCNP with the unlabelled peptides rANP, C-ANF and pCNP present in increasing concentrations (10⁻¹² to 10⁻⁶moll⁻¹). Binding reactions were performed as above.

Hagfish gill membranes were isolated as described above and 100–125 μg of protein was placed in incubation buffer with 0.25 nmol l⁻¹ iodinated peptide in the presence or absence of excess unlabelled rANP, pCNP or C-ANF. The final incubation volume was 250 μl. Affinity cross-linking was performed according to Martin et al. (1989). Following incubation, the covalent cross-linking agent disuccimidyl suberate (DSS; Pierce) in dimethylsulphoxide was added to a final concentration of 1 mmol l^-1 and the mixture was shaken gently for 20 min at 23 °C. The cross-linking reaction was stopped by the addition of an equal volume of quench buffer (400mmoll^-1 EDTA and 1moll^-1 Tris-HCl, pH6.8). Membranes were centrifuged in an Eppendorf centrifuge at 13 000 g for 20 min to separate unbound hormone, and the pellet was resuspended in 30 μl of sample buffer, for SDS-polyacrylamide gel electrophoresis (SDS-PAGE), containing 62.5 mmol l⁻¹ Tris base, 2% SDS, 5% glycerol, 0.01% Bromophenol Blue, 2% β-mercaptoethanol, pH 6.8. The suspended material was then boiled for 4 min. Samples, including one of molecular mass markers (30–200 kDa), were loaded onto a 7.5 % unidimensional polyacrylamide slab gel and electrophoresed at 200 V. Gels were stained with Coomassie Brilliant Blue (Bio-Rad), dried and apposed to Hyperfilm MP (Amersham) with intensifying screens for 1–2 weeks at –70 °C. Films were developed as for section autoradiography. Molecular masses of subsequent bands were determined as predicted values from the regression equations of the negative logarithm of relative mobility versus molecular mass standards for each gel.

Gill membranes were isolated as described above and used immediately. For determination of guanylate cyclase activity, 50 μg of gill protein was added to 50 mmol l⁻¹ Tris-HCl, 2 mmol l⁻¹ isobutylmethylxanthine (IBMX), 10 mmol l⁻¹ creatine phosphate, 1000 i.u. ml⁻¹ creatine phosphokinase, 4 mmol l⁻¹ MnCl2, 1 mmol l⁻¹ GTP and increasing concentrations of rANP, pCNP and C-ANF in a final volume of 100 μl. The basal rate of cyclic GMP generation was determined in tubes without ligand. The incubations were performed for 15 min at 23 °C and were terminated by the addition of 4 mmol l⁻¹ EDTA. The tubes were boiled for 3 min and centrifuged at 2300g for 15 min. The supernatant was collected and frozen and the cyclic GMP content was determined by radioimmunoassay (cyclic GMP RIA kit, Amersham, Arlington Heights, Illinois).

The values of K_d and B_max were determined from the saturation binding data using EBDA and LIGAND computer programs (McPherson, 1985). Additional statistics were computed using the Statview SE program (Abacus Concepts, 1988). Data are presented as means ± S.E.M.

The gross morphology of hagfish gills is different from that of other fishes. The gill lamellae are contained in ovoid muscular pouches through which water flows in countercurrent to the blood; M. glutinosa has six gill pouches on either side of the oesophagus. Fig. 1 shows longitudinal sections through a whole gill pouch; there are three main regions, afferent, lamellar and efferent (Fig. 1A). The afferent section (relative to blood flow) includes a multilayered epithelium containing ionocytes (cells structurally similar to the ion-transporting chloride cells of higher fish) separated by connective tissue from the arterio-arterial vasculature. The vasculature consists of a network of radial vessels and afferent cavernous tissue surrounded by smooth muscle (Fig. 1A: a). Arranged between the afferent and efferent sections of the gill are the respiratory lamellae (Fig. 1A: indicated by arrows). These are characterised by a bilayered epithelium in which ionocytes are present. There is no smooth muscle in this section of the gill, although there are pillar cells. The lamellar portion of the gill is drained by efferent lamellar arterioles and the efferent cavernous tissue. The efferent portion of the gill is also characterised by a multilayer epithelium; however, ionocytes are absent here (Fig. 1A: e) (Elger, 1987).

Both ¹²⁵I-ANP- and ¹²⁵I-CNP-specific binding were observed on the respiratory lamellae of the hagfish gill (Fig. 1A; Fig. 1 shows autoradiographs of gill sections incubated with ¹²⁵I-ANP, ¹²⁵I-CNP-specific binding is not shown, but was similar); specific binding of radioligands was displaced by both 1 mmol l⁻¹ rANP and 1 μmol l⁻¹ pCNP (Fig. 1B,C, ¹²⁵I-CNP not shown). 1 μmoll⁻¹ C-ANF did not completely displace ¹²⁵I-ANP binding, but displaced virtually all ¹²⁵I-CNP binding (Fig. 1D, ¹²⁵I-CNP not shown). Examination of radiotrack emulsion-dipped sections indicated that specific binding sites for both radioligands were scattered generally in the lamellar region over the thin bilayered epithelium (Fig. 2A,B, ¹²⁵I-CNP binding only shown, ¹²⁵I-ANP binding was similar). The exact cell types to which the radioligands were binding could not be determined. Specific binding sites were not observed in either the afferent or efferent gill regions.

Both ¹²⁵I-ANP and ¹²⁵I-CNP bound specifically and saturably to hagfish gill membranes. Maximum binding for both radioligands was reached by 200pmoll⁻¹ (Figs 3 and 4). EBDA and LIGAND analysis indicated that ¹²⁵I-ANP binding fitted a single site model with an apparent K_d of 15.4±1.6pmoll⁻¹ and a Bmax of 45.9±3.0fmolmg⁻¹ protein or, alternatively, multiple sites with equal affinities (Fig. 3). Analysis of the ¹²⁵I-CNP binding site indicated a two-site model with a high- and a low-affinity site (Fig. 4). The high- affinity site was not statistically different from the ¹²⁵I-ANP site, with an apparent Kd of 12.9±4.7pmoll⁻¹ and a B_max of 23.4±6.5fmolmg⁻¹ protein. The low-affinity site had an apparent K_d and B_max of 380±80pmoll⁻¹ and 120±21fmol mg⁻¹ protein, respectively.

1nmoll⁻¹ unlabelled rANP, and 20 and 30 nmol l⁻¹ unlabelled pCNP and C-ANF, respectively, competed for 50 % of ¹²⁵I-ANP-specific sites. 100 nmoll⁻¹ rANP bound virtually all ANP sites. 1 μmoll⁻¹ pCNP and 1 μmoll⁻¹ C-ANF competed for all but 5 % and 10 % of binding sites, respectively (Fig. 5A). 0.1 nmoll⁻¹ rANP and pCNP and 8nmoll⁻¹M C-ANF competitively inhibited 50% of ¹²⁵I-CNP-specific binding. 1 nmoll⁻¹ rANP, 10 nmoll⁻¹ pCNP and 1 μmoll⁻¹ C-ANF bound 100% of ¹²⁵I-CNP-specific sites (Fig. 5B). Rat ANP and pCNP competed equally for ¹²⁵I-CNP binding sites above 50 % binding. However, below 50 % binding, rANP was a more effective competitive inhibitor than pCNP, suggesting that these sites are low-affinity CNP binding sites (in accordance with the saturation analysis above) that bind ANP with greater affinity.

Affinity cross-linking followed by SDS-PAGE under reducing conditions of ¹²⁵I-ANP and ¹²⁵I-CNP binding to gill membranes indicated an apparent single binding site with an approximate molecular mass of 150 kDa (Fig. 6, lane 1; only ¹²⁵I-ANP binding shown, results of ¹²⁵I-CNP binding were identical). Binding was prevented by the addition of 0.1 μmoll⁻¹ rANP and pCNP to the incubation reaction (Fig. 6, lanes 2 and 3). The addition of 0.1 μmol l⁻¹ C-ANF did not completely inhibit radioligand binding (Fig. 6, lane 4) although, in other experiments (not shown here), visible binding was prevented by 1 μmol l⁻¹ C-ANF. The mean apparent molecular masses from three experiments for each ligand were 150±2.5kDa for ¹²⁵I-ANP and 153±16kDa for ¹²⁵I-CNP. In neither case did a second lower band appear, contrasting with mammals and teleosts, both of which show a lower band indicative of the NPR-C homodimer breaking into the monomeric species under reducing conditions (Martin et al. 1989; Donald et al. 1994).

The basal cyclic GMP accumulation rate was 2.9±0.4 pmoll⁻¹ cyclic GMP mg⁻¹ protein min⁻¹. Both rANP and pCNP stimulated cyclic GMP production in a dose-dependent manner; 0.1 nmol l⁻¹ rANP and above, and 10 mmol l⁻¹ pCNP and above, significantly stimulated cyclic GMP production above the basal rate (Fig. 7). A maximum rANP-stimulated rate of 6.2±1.6pmoll⁻¹cyclicGMPmg⁻¹protein min⁻¹ was reached in the presence of 0.1 μmol l⁻¹ rANP. The maximum pCNP-stimulated rate, 4.2±0.6 pmol l⁻¹ cyclic GMP mg⁻¹proteinmin⁻¹, was approached at 1 μmoll⁻¹ pCNP. C-ANF did not stimulate cyclic GMP production at any concentration.

The present study shows that there are both ANP and CNP receptors on the lamellar epithelium of the gills of M. glutinosa (Figs 1 and 2). Because the epithelium in the lamellar region does not contain smooth muscle, receptors may be involved in epithelial cell function (perhaps ionocyte function) or in NP clearance from the circulation, rather than in the regulation of blood flow. However, blood flow in the lamellar regions could be influenced by possible pillar cell contraction: teleost pillar cells contain myosin, but whether they have contractile properties has yet to be demonstrated (Farrell, 1993). Although, at this point, there is no evidence that NPs are directly regulating local lamellar blood flow, the presence of NPRs in the ventral aorta (Kloas et al. 1988) and the ability of NPs to dilate this tissue (Evans, 1991; Evans et al. 1993) suggest that Nps can regulate blood flow to the branchial vasculature as a whole.

The presence of NP receptors in the gill agrees with teleost studies; binding is predominantly located on the chloride cells of two species of Antarctic fishes, Chionodraco hamatus and Pagothenia bernacchii (Uva et al. 1993), and Broadhead et al. (1992) suggest that binding occurs on chloride cells of the eel Anguilla anguilla. In the Japanese eel Anguilla japonica, NP binding is localised mainly on chondrocytes of gill cartilage and on parenchymal cells, which include chloride cells, of the secondary lamellae (Sakaguchi et al. 1993). Other studies have shown binding in the arterio-arterial gill vasculature in trout, where receptors appear to perform a clearance function (olson and Duff, 1993) and on the afferent and efferent branchial arteries and arterioles of the gulf toadfish Opsanus beta (Donald et al. 1994). In the latter study, affinity cross-linking experiments showed high (140 kDa) and low (75 kDa) molecular mass binding sites, suggesting a population of NPR-C and guanylate cyclase receptors similar to those found in mammals.

Saturation analysis indicates that there are saturable ANP and CNP binding sites in the gill (Figs 3 and 4). Interpretation of the Scatchard plots suggests that ANP binds to a minimum of one high-affinity site, whereas CNP binds to two sites of differing affinity. The competition data demonstrate that ANP and CNP are capable of competing for the binding sites of both radioligands, but with different efficiencies. CNP is less able to compete with ANP for ¹²⁵I-ANP sites (Fig. 5A), but ANP and CNP compete equally well for ¹²⁵I-CNP sites (Fig. 5B), suggesting that there are two ANP sites, one that binds ANP in preference to CNP (the low-affinity CNP site) and one that binds ANP and CNP with equal affinity (the high-affinity CNP site). In addition, the similarity in the upper 50% of the competition curves of rANP and pCNP for ¹²⁵I-CNP binding sites (Fig. 5B) is consistent with the presence of a high-affinity ANP/CNP site, whereas the dissimilarity between the competing ligands in the lower 50 % of the curves suggests the presence of a high-affinity ANP/low-affinity CNP site. A similar situation was found in shark rectal gland competition studies, where CNP, rather than ANP, displaced specific binding more readily in the lower 50 % of the curve (Gunning et al. 1993). The stimulation of cyclic GMP production by ANP (Fig. 7), and to a lesser extent by CNP, indicates that it is probably the high-affinity ANP/low-affinity CNP receptor that is coupled to guanylate cyclase activity. The high-affinity ANP/low-affinity CNP receptor appears to resemble the NPR-A of mammals. Mammalian guanylate-cyclase-linked NPR-A receptors bind ANP with much greater affinity than they bind CNP (Koller and Goeddel, 1992). There is no evidence in the hagfish gill of a guanylate-cyclase-linked NPR-B-like receptor that preferentially binds CNP (Koller and Goeddel, 1992).

C-ANF competitively inhibits the majority of both ¹²⁵I-ANP and ¹²⁵I-CNP binding (Fig. 5), but it is unclear to which site this NP analogue preferentially binds. It appears only partially to displace ¹²⁵I-ANP, but to displace all of ¹²⁵I-CNP from tissue sections. Therefore, it is tempting to suggest that C-ANF binds mainly to the high-affinity ANP/CNP receptor (Fig. 1D). C-ANF has been constructed to bind specifically to the clearance (NPR-C) receptor in mammals (Maack et al. 1987); however, whether this specificity for NPR-C is maintained in other vertebrate classes is unknown.

Both receptor types in the hagfish gill have an apparent molecular mass of approximately 150 kDa (Fig. 6). This molecular mass is slightly heavier than that of the mammalian NPR-A, and the homodimer form of the NPR-C, which both appear at approximately 130 kDa. In the presence of reducing conditions, mammalian NPR-C, and toadfish NPR-C-like receptors, separate into the monomeric species and are visible as lower molecular mass bands at 65 kDa (mammals, Brenner et al. 1990) and 75 kDa (toadfish, Donald et al. 1994). Sakaguchi et al. (1993) discovered a 68 kDa band under reducing conditions in Japanese eel gill membranes; biochemical characterisation demonstrated that this receptor was of the NPR-C type. The absence of a lower molecular mass band in hagfish gills strongly suggests that, unlike the situation in mammals and teleosts, the high-affinity ANP/CNP receptor is not a homodimeric NPR-C type. Consequently, it appears unlikely that this ANP/CNP receptor is homologous in structure to the NPR-C of mammals. It is also unlikely that this receptor is linked to guanylate cyclase activity, because CNP did not stimulate cyclic GMP production as effectively as did ANP (Fig. 7). Whether the high-affinity ANP/CNP hagfish receptor functions as a clearance receptor has yet to be determined; however, it is possible that the hagfish gill contains receptors to modulate circulating NP concentrations, since teleost gills perform a clearance function for NPs and other humoral factors (Olson and Duff, 1993). It is also unknown whether the ANP/CNP receptor is linked to other second-messenger pathways, as has recently been demonstrated for the mammalian NPR-C (Levin, 1993).

It is now evident that hagfish have a well-developed NP system. Previous research has shown that not only do the heart, brain and plasma of Myxine glutinosa have NP-like immunoreactivity (Reinecke et al. 1987; Evans et al. 1989; Donald et al. 1992), but also that NPs are vasoactive, dilating the vascular smooth muscle of the ventral aorta (Evans, 1991; Evans et al. 1993), where binding sites have also been located (Kloas et al. 1988). The discovery of NP immunoreactivity and the localisation of NP binding sites in the hagfish brain indicate that NPs are neuropeptides, functional in the central nervous system (Donald et al. 1992; J. A. Donald and T. Toop, unpublished observations). The presence of binding sites in the glomeruli and archinephric ducts shows that the hagfish kidney is also a target organ (Kloas et al. 1988; T. Toop, unpublished observations).

The present study extends NP function to the gills of hagfish. Although the present study utilizes heterologous mammalian peptides and not the native hagfish NPs, this research clearly indicates that guanylate-cyclase-linked activity in NPRs is a phylogenetically ancient vertebrate characteristic. In addition, the presence of ANP-specific receptors, rather than CNP-specific ones, suggests that CNP is not the primitive circulating vertebrate NP, in spite of its importance as a heart hormone in elasmobranchs (Schofield et al. 1991; Suzuki et al. 1991; Solomon et al. 1992; Karnaky et al. 1992). The ancestral presence of the NPR-C, or clearance receptor, is less clear. Further investigations are needed before the relationship between the hagfish ANP/CNP receptor and the NPR-C can adequately be determined.

This research was supported by National Science Foundation Grant DCB 8916413 to D.H.E. and by NIH EHS-P30-ESO3828 to the Center for Membrane Toxicity Studies at MDIBL. We thank Daryl Harrison for assistance with the illustrations.