Conversion of big endothelin-1 by membrane-bound metalloendopeptidase in cultured bovine endothelial cells

We propose a candidate for the "putative" endothelin (ET) converting enzyme in the cultured endothelial cells (ECs) of bovine carotid artery. The enzyme is membrane-bound, soluble in 0.5% Triton X-100, and capable of converting human big ET-1 to ET-1 by a specific cleavage between Trp21 and Val22. The conversion reached 90% after a 5-hr incubation in the presence of DFP, PCMS and pepstatin A, but it was inhibited by EDTA, omicron-phenanthroline or phosphoramidon. The enzyme is very sensitive to pH, and active only between pH 6.6 and pH 7.6. Conversion of big ET-3 by this enzyme was only 1/9 that of big ET-1. From these results, ET-1 converting enzyme in the bovine EC is most likely to be a membrane-bound, neutral metalloendopeptidase, which is much less susceptible to big ET-3.     

Identification of endothelin converting enzyme in bovine lung membranes using a new fluorogenic substrate.

An enzyme partially purified from bovine lung membranes appears to be endothelin converting enzyme (ECE). This enzyme specifically cleaves big endothelin-1 (big ET-1) at the proper site, between Trp21 and Val22, with maximum activity at pH 7.5 and with a Km of roughly 3 microM, to produce endothelin-1 (ET-1) and C-terminal peptide (CTP). This same enzyme hydrolyzes the fluorogenic substrate succinyl-Ile-Ile-Trp-methylcoumarinamide to release the highly fluorescent 7-amino-4-methylcoumarin. The peptide derivative has the same amino acid sequence as big ET-1 and is a good substrate with a Km of about 27 microM. This enzyme is a metalloproteinase. It is not inhibited by five common proteinase inhibitors (pepstatin A, PMSF, NEM, E-64 and thiorphan) but it is inhibited by phosphoramidon and chelating compounds. The apoenzyme is restored to nearly full activity by a zinc-EDTA buffer with pZn = 13.     

Biochemical properties of endothelin converting enzyme in renal epithelial cell lines.

This is the first report clearly demonstrating the presence of endothelin (ET) converting enzyme (ECE) in non-vascular cells (renal epithelial cell lines, MDCK and LLC-PK1). ECEs derived from these epithelial cells were very similar to the endothelial ECE in the following biochemical properties: 1) The optimum pH was 7.0; 2) the Km value for big ET-1 was approximately 30 microM; 3) the enzyme was potently inhibited by EDTA, o-phenanthroline and phosphoramidon; and 4) the enzyme did not convert big ET-2 or big ET-3. These data suggest that phosphoramidon-sensitive ECE is involved in the processing of big ET-1 to ET-1 in the renal tubule.     

Endothelin converting enzyme of bovine carotid artery smooth muscles

This is the first report clearly demonstrating the presence of endothelin (ET) converting enzyme in vascular smooth muscle. Like cultured endothelial cells, noncultured vascular smooth muscle homogenate of bovine carotid arteries, converts human big ET- 1 to ET-1 at pH 3.0, pH 5.0 and pH 7.0, and the apparent ratio of these three activities is about 6:5:1, respectively. Peptides generated during incubation of the homogenate and big ET- 1 at the three pHs were identified as ET- 1 by radioimmunoassay and high performance liquid chromatography. The two acid enzymes are in the cytosol (103,000xg sup) and are inhibited by pepstatin A, while the neutral enzyme is sensitive to EDTA or phosphoramidon; 73% of the neutral enzyme activity was membrane-bound and the remainder (27%) cytosolic.     

Local formation and degradation of endothelin-1 in guinea pig airway tissues

Endothelin(ET)-1 and big ET-1 caused potent and sustained constriction of isolated guinea pig bronchus. The response to ET-1 was enhanced by phosphoramidon in a simple dose-related manner (0.01-1000 microM), while the response to big ET-1 was enhanced at lower doses (0.01-0.1 microM) but was suppressed at higher doses (100-1000 microM) of phosphoramidon. Big ET-1, when given intravenously (i.v.) to anesthetized guinea pigs, increased both bronchopulmonary inflation pressure and mean arterial blood pressure (2.5, 5, 10 nmol/kg i.v.). The pressor response to big ET-1 was attenuated by phosphoramidon dose-relatedly, while the pulmonary response was modified in a complex fashion composed of delayed onset and prolonged duration of action. These results suggest that ET converting as well as degrading enzymes coexist in the airway tissue and both enzymes are sensitive to phosphoramidon, so that phosphoramidon acts bifunctionally to reduce and stimulate the airway responses to big ET-1.     

Conversion of big endothelin-1 to endothelin-1 by two types of metalloproteinases derived from porcine aortic endothelial cells

Incubation of big endothelin-1 (big ET-1(1-39] with either the cytosolic or membrane fraction obtained from cultured endothelial cells, resulted in an increase in immunoreactive-endothelin (IR-ET), which was markedly inhibited by metal chelators. Phosphoramidon, a metalloproteinase inhibitor, specifically suppressed the membrane fraction-induced increase in IR-ET, whereas the increase in IR-ET observed with the cytosolic fraction was not influenced by phosphoramidon. Reverse-phase (RP)-HPLC of the incubation mixture of big ET-1 with the cytosolic or membrane fraction revealed one major IR-ET component corresponding to the elution position of synthetic ET-1(1-21). Simultaneously, immunoreactivities like the C-terminal fragment (CTF22-39) of big ET-1 were present, as deduced from the RP-HPLC coupled with the radioimmunoassay for CTF. Our results indicate the presence of two types of metalloproteinases, which convert big ET-1 to ET-1 via a single cleavage between Trp21 and Val22, in vascular endothelial cells.     

Effect of phosphoramidon on big endothelin-2 conversion into endothelin-2 in human renal adenocarcinoma (ACHN) cells. Analysis of endothelin-2 biosynthetic pathway.

The biosynthetic pathway of endothelin (ET)-2 was analyzed in cultured ACHN cells. In the supernatant, we detected three ET-2-related peptides, ET-2, big ET-2(1-38) and big ET-2(22-38). Phosphoramidon decreased the amount of ET-2 and increased that of big ET-2(1-38) dose-dependently. The amount of big ET-2(1-37) did not significantly change. These results suggest that big ET-2 is composed of 38 and not 37 amino acid residues, and that a putative ET-2-converting enzyme (ECE-2) should be classified as a phosphoramidon-sensitive neutral metalloprotease, bearing a resemblance to the putative ET-1-converting enzyme (ECE-1) in endothelial cells.     

Functional evidence for the presence of a phosphoramidon-sensitive enzyme in rat brain that converts big endothelin-1 to endothelin-1

Endothelin-1 (ET-1) is produced from its precursor, big endothelin-1 (BigET-1), by a putative endothelin-converting enzyme (ECE), but it is not known whether the enzyme is present in the brain. This study was conducted to examine the central hemodynamic effects of BigET-1 and to indirectly determine the presence of an ECE in rat brain. Cardiovascular effects of centrally administered BigET-1 and ET-1 were examined in anesthetized, ventilated rats. BigET-1 (100 pmol) or ET-1 (10 pmol) applied to the IV ventricle produced similar prolonged decreases in mean arterial pressure (MAP) and renal blood flow (RBF). Thus, peak decreases with BigET-1 were (mean +/- S.E.): MAP = -35 +/- 4%; RBF = -27 +/- 5%, while those with ET-1 were: MAP = -36 +/- 5%; RBF = -29 +/- 9%. Pretreatment with phosphoramidon, a metalloprotease inhibitor (90 nmol), abolished the hemodynamic responses elicited by BigET-1 (MAP = -9 +/- 2%; RBF = -3 +/- 2%) but not those produced by ET-1. These data indicate that; i) conversion of BigET-1 to ET-1 in the brain is essential for the expression of hemodynamic actions and ii) a metalloprotease capable of converting BigET-1 to ET-1 is present in rat brain.     

Do the structures of big ET-1 and big ET-3 adopt a similar overall fold? Consequences for endothelin converting enzyme specificity

Big ET-1 and big ET-3 are precursor peptides which render endothelin-1 (ET-1) and endothelin-3 (ET-3) relatively unreactive and resistant to proteolytic cleavage. Big ET-1 is cleaved in vivo by ECE-1 (endothelin-converting enzyme), and big ET-3 is also cleaved but apparently to a significantly lesser extent by this enzyme. To shed light on the relation between structure and function, circular dichroism (CD) spectroscopy and homology modeling were used to determine whether big ET-1 and big ET-3 adopt similar secondary and tertiary structures. Analyses of the CD spectra and thermal denaturation indicate they have similar secondary structures and thermal stabilities. Superposition of the modeled coordinates of both peptides indicates that they can adopt the same overall fold except in the C-terminal residues, 34-38 in big ET-1 and 34-41 in big ET-3. This region corresponds to an area of complete sequence heterogeneity between the two peptides. A model has been developed which has a loop for residues 27-30 (HVVP in big ET-1), which have previously been demonstrated to be essential for eliciting efficient hydrolysis of the W21-V22 bond in big ET-1 and which have the sequence QTVP in big ET-3. Differences in affinity between big ET-1 and big ET-3 for ECE-1 thus appear to be due solely to sequence variations in the local region of the cleavage site.     

The role of big endothelin-1 in colorectal cancer

INTRODUCTION: Changes in liver blood flow caused by an unknown splanchnic vasoconstrictor have been noted in colorectal cancer patients with liver metastases. This prospective study was performed to assess whether plasma levels of big endothelin-1 (big ET-1) were raised in patients with colorectal cancer. METHODS: Plasma samples from peripheral vein of patients who underwent surgery for primary colorectal cancer (n=60) and those with known colorectal liver metastases (n=45) for a period of 15 months were taken prior to treatment and compared to age- and sex-matched controls (n=20). Plasma samples were analysed by using a single-step sandwich enzyme immunoassay. Immunohistochemistry and in situ hybridisation were also performed on tumour sections to investigate the expression of ET-1 by cancer cells. RESULTS: The median (range) plasma concentration of big ET-1 in controls was 2.1 pg/mL (1.2-13.4 pg/mL). The median (range) plasma concentration of big ET-1 in colorectal cancer patients with no overt hepatic metastases was 3.8 pg/mL (1.2-15.8 pg/mL), p=0.002, and the median (range) plasma concentration of big ET-1 in colorectal cancer patients with hepatic metastases was 5.2 pg/mL (1.7-30 pg/mL), p=0.0001; both were significantly elevated compared to the control group. A significant difference in immunostaining for big ET-1 was noted between paired normal colonic mucosa (median score-1) and tumour sections (median score-3), p=0.01. CONCLUSION: This study has demonstrated elevated concentrations of big ET-1 in colorectal cancer patients, especially in those with hepatic metastases. Upregulation of ET activity in colorectal cancer could be inferred by the increased immunostaining of big ET-1 in cancer cells. Therefore, plasma big ET-1 levels should be evaluated as a potential tumour marker for the identification of hepatic metastases at an earlier stage.     

Importance of the C-terminal region of big endothelin-1 for specific conversion by phosphoramidon-sensitive endothelin converting enzyme

Based on our previous findings that phosphoramidon-sensitive endothelin (ET) converting enzyme (ECE) converts human big ET-1 but does not big ET-3, we investigated structural requirement for substrate peptide. We prepared shorter peptides of big ET-1 and measured hydrolysis of the Trp-Val bond of these peptides. Relative hydrolysis ratios of big ET-1(1-38), (1-37), (16-37), (1-31) and (17-26) were 1, 1.15, 3.71, 0.01 and 0, respectively. In addition, big ET-2 and big ET-3 were not significantly converted by ECE. These results suggest that the carboxyl-terminal sequence at residues 32-37 of big ET-1 is important for conversion, whereas the amino-terminal disulfide loop structure appears to interfere with access of ECE to big ET-1.     

Characterization of phosphoramidon-sensitive metalloproteinases with endothelin-converting enzyme activity in porcine lung membrane

Endothelin-1 (ET-1), a 21 amino-acid potent vasoconstrictor peptide, is produced from the biologically inactive intermediate big ET-1 via an endoproteolytic cleavage between Trp-21 and Val-22 by endothelin converting enzyme (ECE). cDNA sequence analysis predicts that the two other members of the endothelin family, ET-2 and ET-3, are also generated from the corresponding intermediates called big ET-2 and big ET-3, respectively. The metalloproteinase inhibitor phosphoramidon inhibited the conversion of big ET-1 into mature ET-1 both in vivo and in cultured endothelial cells, suggesting that ECE may be a neutral metalloproteinase. In this study, we solubilized and partially purified ECE from the membrane fraction of porcine lung. Using gel filtration chromatography, we separated two distinct ECE activities, designated M1 (apparent molecular mass approx. 300 kDa) and M2 (approx. 65 kDa). Optimum pH for the cleavage of big ET-1 by M1 and M2 was 7.0 and 7.5, respectively. M1 efficiently converted human big ET-1(1–38) to ET-1, but not human big ET-2(1–37) or human big ET-3(1–41)-amide. In contrast, M2 converted both big ET-1 and big ET-2, but not big ET-3. M1 was inhibited by phosphoramidon (IC₅₀ approx. 1 μM) but not by thiorphan or bacitracin. In contrast, M2 was inhibited by much lower concentrations of phosphoramidon (IC₅₀ approx. 0.3 nM), as well as by thiorphan and bacitracin. ECE activity in M1 was able to bind to a concanavalin A-agarose column and was eluted by α-methyl-d-glucoside, indicating that the ECE is glycosylated. From these results, M1 and M2 from the porcine lung membrane are similar to the candidate of ECE in endothelial cells and neutral endopeptidase in kidney (EC 3.4.24.11), respectively. Taken in conjunction with the previous finding that neither thiorphan nor bacitracin affected the conversion of endogenously synthesized big ET-1 in cultured endothelial cells, we conclude that physiologically relevant ECE found in the endothelial cells is more similar to M1 than to M2.     

Endothelin-1 induces prostacyclin release from bovine aortic endothelial cells

The effects of endothelin-1 (ET-1) on the release of prostacyclin from cultured bovine aortic endothelial cells were studied. ET-1 induced a time- and dose-dependent release of 6-keto PGF1 alpha, the stable metabolite of prostacyclin, with an apparent EC50 value of 3.0 +/- 0.9 nM (n = 6). ET-1 up to a concentration of 500 nM did not affect cellular integrity. Preincubation of the cells for 30 min with 10 microM indomethacin inhibited ET-1 (100 nM) - induced prostacyclin release by 90%. These findings indicate that ET-1 can directly stimulate prostacyclin release from endothelial cells probably through a receptor mediated mechanism.     

Synthesis and antineoplastic properties of an ether glycerophosphonocholine, and analog of ET-18-OCH3-GPC.

A glycerophosphonocholine analog of the ether-linked lipid, rac-1-O-octadecyl-2-O-methyl-glycero-3-phosphocholine (ET-18-OCH3-GPC), was synthesized in which the head group is nonhydrolyzable by phospholipase C. The phosphonate analog used in this study is rac-3-octadecyloxy-2-methoxy-propyl-phosphonocholine, C18H37OCH2CH(OCH3)CH2P(O)(O)OCH2CH2N+(CH3)3. The activity of the synthetic phosphonate was tested in the human leukemic cell line, HL-60, and the human undifferentiated cervical carcinoma, C-41. The glycerophosphonocholine inhibited [3H]thymidine uptake by HL-60 cells with an EC50 value of 5-7 microM. The glycerophosphate ET-18-OCH3-GPC had an EC50 value of approximately 2 microM against HL-60 cells. The EC50 values estimated from cell viability experiments were similar to that for [3H]thymidine uptake. The EC50 value for C-41 cells was about 10-15 microM. The data demonstrate that the glycerophosphonocholine is a promising anti-cancer drug for the treatment of both leukemia and solid tumors. Furthermore, the data demonstrate that phospholipase C-catalyzed hydrolysis of ET-18-OCH3-GPC does not play an important role in the cytotoxic action of the ether-linked glycerolipids.     

N-ethylmaleimide differentiates endothelin converting activity by two types of metalloproteinases derived from vascular endothelial cells

We have recently found that cultured vascular endothelial cells (ECs) contain two types of metalloproteinases which convert big endothelin-1 (big ET-1) to endothelin-1 (ET-1) via a single cleavage between Trp21 and Val22. In the present study, two enzymes were clearly differentiated by using sulfhydryl blocking reagents and anion-exchange HPLC. As reported, the converting activity of the membrane fraction of ECs was specifically inhibited by phosphoramidon. N-ethylmaleimide (NEM) markedly enhanced the apparent converting activity of the membrane fraction. This enhancement was not due to the direct action on the converting enzyme, but rather to inhibition of the degradation of big ET-1 and/or ET-1. In contrast, the converting activity of the cytosolic fraction was abolished by NEM treatment. Effects of phosphoramidon and NEM on converting activities of both fractions were confirmed after anion-exchange HPLC of each fraction, using a COSMOGEL QA column. Our results provide new information on two types of metalloproteinases which convert big ET-1 to ET-1, in vascular ECs.     

NO-induced modulation of calcium-oscillations in pulmonary vascular smooth muscle

The effect of the nitric oxide (NO) donor sodium nitroprusside (SNP) on both [Ca(2+)](i)and mechanical activity was studied in the rat isolated pulmonary artery (RPA). In freshly isolated myocytes loaded with 1 microM indo-lacetoxymethyl ester for 30 min, short (40-60 s) application of ATP (100 microM) or ET-1 (0.1 microM) induced 3-6 cyclic rises in [Ca(2+)](i)(Ca-oscillations) of decreasing amplitude. Preincubation of cells with SNP (10-250 microM) for 10 min had no effect on the resting [Ca(2+)](i)value, but progressively abolished the oscillations. A similar effect was obtained with 8-bromo-cGMP (100-500 microM). SNP (0.001-100 microM) concentration-dependently relaxed ATP (10 mM, n = 4) and ET-1 (0.1 microM, n = 4)-precontracted RPA. 1H-[1,2,4]oxadiazolol [4,3,-a]quinoxalin-1-one (ODQ, 10 microM), a potent inhibitor of the cytosolic guanylyl cyclase, fully reversed the effect of SNP on ATP-induced [Ca(2+)](i)oscillations as well as on ATP-precontracted RPA. In contrast, N-[2-(methylamino)ethyl]-5-isoquinolinesulfonamide (H8, 10 microM), a potent inhibitor of cGMP-dependent protein kinase (PKG), did not alter the effect of SNP. Caffeine (5 mM) induced only one transient [Ca(2+)](i)-increase (n = 24), the amplitude of which was altered neither by SNP nor by 8-bromo-cGMP. Our results show that the relaxing effect of NO in RPA is related, at least in part, to its action on the Ca-signalling pathway. NO interacts with inositol trisphosphate pathway without interacting with the ryanodine-sensitive receptor. Finally, the effect of NO involves an increase in cGMP but appears independent of activation of PKG.     

Phosphoramidon-sensitive endothelin-converting enzyme in vascular endothelial cells converts big endothelin-1 and big endothelin-3 to their mature form.

Incubation of big endothelin-3 (big ET-3(1-41)) with the membrane fraction obtained from cultured endothelial cells (ECs) resulted in an increase in immunoreactive-ET (IR-ET). This increasing activity was markedly suppressed by phosphoramidon, which is known to inhibit the conversion of big ET-1(1-39) to ET-1(1-21). Reverse-phase HPLC of the incubation mixture of the membrane fraction with big ET-3 revealed one major IR-ET component corresponding to the elution position of synthetic ET-3(1-21). When the cultured ECs were incubated with big ET-3, a conversion to the mature ET-3, as well as an endogenous ET-1 generation, was observed. Both responses were markedly suppressed by phosphoramidon. By the gel filtration of 0.5% CHAPS-solubilized fraction of membrane pellets of ECs, the molecular mass of the proteinase which converts big ET-1 and big ET-3 to their mature form was estimated to be 300-350 kDa. Phosphoramidon almost completely abolished both converting activities of the proteinase. We conclude that the above type of phosphoramidon-sensitive metalloproteinase functions as an ET-converting enzyme to generate the mature form from big ET-1 and big ET-3 in ECs.     

Maturation depresses cGMP-mediated decreases in [Ca2+]i and Ca2+ sensitivity in ovine cranial arteries

Because cerebrovascular cGMP levels vary significantly during maturation, we examined the hypothesis that the ability of cGMP to relax cerebral arteries also changes during maturation. In concentration-response experiments, potassium-induced tone in basilar arteries was significantly more sensitive to a nonmetabolizable cell-permeant cGMP analogue 8-(p-chlorophenylthio)-cGMP (8-pCPT-cGMP) in term fetal [-log one-half maximal concentration (EC(50)) = 4.4 +/- 0.1 M] than in adult (-log EC(50) = 4.0 +/- 0.1 M) ovine basilar arteries. Serotonin-induced tone also revealed significantly greater sensitivity to the cGMP analogue in fetal (-log EC(50) = 4.9 +/- 0.1 M) than in adult (-log EC(50) = 4.7 +/- 0.1 M) basilars. In fura 2-loaded preparations, 8-pCPT-cGMP had no significant effect on cytosolic calcium concentrations in potassium-contracted arteries but at 6 microM significantly reduced calcium only in fetal basilars (Delta = 33 +/- 8%). Higher 8-pCPT-cGMP concentrations reduced cytosolic calcium in both fetal and adult basilars. Similarly, in both potassium- and 5-hydroxytryptamine (5-HT)-contracted preparations, low concentrations of 8-pCPT-cGMP reduced myofilament calcium sensitivity only in fetal basilars (Delta = 29 +/- 6 and Delta = 42 +/- 10%, respectively), whereas higher concentrations reduced calcium sensitivity in both fetal and adult arteries. In beta-escin-permeabilized arteries, equivalent reductions in basal and agonist-enhanced myofilament calcium sensitivity were produced by much lower 8-pCPT-cGMP concentrations in fetal (172 and 61 microM, respectively) than in adult (410 and 231 microM, respectively) basilars. The mechanisms mediating cGMP-induced vasorelaxation appear similar in fetal and adult arteries, with the exception that they are much more sensitive to cGMP in fetal than adult arteries. These age-related differences in the sensitivity of cytosolic calcium concentration, basal, and agonist-enhanced myofilament calcium sensitivity to cGMP can easily explain why both potassium- and 5-HT-induced tone are more sensitive to cGMP in fetal than adult cerebral arteries.     

Endothelin-1 stimulates glial cell line-derived neurotrophic factor expression in cultured rat astrocytes

Effects of endothelin-1 (ET-1) on glial cell line-derived neurotrophic factor (GDNF) production in cultured astrocytes were examined. Treatment of cultured astrocytes with ET-1 (100 nM) increased mRNA levels of GDNF in 1-6h. The effect of ET-1 was inhibited by BQ788, an ET(B) receptor antagonist, but not by FR139317, an ET(A) receptor antagonist. ET-1 stimulated release of GDNF into culture medium. Dexamethasone (1 microM) and pyrrolidine dithiocarbamate (PDTC, 100 microM), which inhibit activation of NFkappaB, prevented the increases in GDNF mRNA by H(2)O(2). In contrast, the effect of ET-1 was not affected by dexamethasone and PDTC. The increase of astrocytic GDNF mRNA by ET-1 was inhibited by BAPTA/AM (30 microM) and PD98059 (50 microM), but not by calphostin C, staurosporine, and cyclosporine A. These results suggest that ET-1 stimulated expression of astrocytic GDNF through ET(B) receptor-mediated increases in cytosolic Ca(2+) and ERK activation.     

Elasmobranch color change: A short review and novel data on hormone regulation.

Skins of Potamotrygon reticulatus are light in color in vitro, exhibiting punctate melanophores. Alpha-Melanocyte stimulating hormone (EC(50) = 4.58 x 10(-9) M) and prolactin (EC(50) = 1.44 x 10(-9) M) darken the skins in a dose-dependent manner. The endothelins ET-1, ET-2 and ET-3, and the purines, ATP, and uracil triphosphate (UTP) were not able to induce either skin lightening or darkening. Forskolin and the calcium ionophore A23187 promoted a dose-dependent darkening response, whereas N(2), 2'-O-dibutyryl guanosine 3'-5'-cyclic monophosphate (db cyclic GMP), phorbol-12-myristate-13-acetate (TPA), and 1-oleoyl-2-acetyl-sn-glycerol (OAG) were ineffective. The maximal response obtained with the calcium ionophore A23187 was only 76% of maximal darkening. These results indicate that the cyclic adenosine 3'-5'-monophosphate (cAMP) pathway is probably involved in the pigment dispersion of P. reticulatus melanophores. Other experiments should be done to further investigate how cytosolic calcium may be physiologically increased, and the existence of a putative cross-talk between calcium and cAMP signals. In conclusion, the only hormones effective on P. reticulatus melanophores were prolactin and alpha-MSH. No aggregating agent has been shown to antagonize these actions. Prolactin effect on elasmobranch melanophores adds a novel physiological role to this ancient hormone. J. Exp. Zool. 284:485-491, 1999.