Isolation, structural characterization, and bioactivity of a novel neuromedin U analog from the defensive skin secretion of the Australasian tree frog, Litoria caerulea.

We report the isolation of a novel bioactive peptide, neuromedin U-23 (NmU-23), from the defensive skin secretion of the Australasian tree frog, Litoria caerulea. The primary structure of the peptide was established by a combination of microsequencing, mass spectroscopy and site-directed antiserum immunoreactivity as SDEEVQVPGGVISNGYFLFRPRN-amide (M(r) 2580.6). A synthetic replicate of frog NmU-23 displaced monoradioiodinated rat NmU-23 from uterine membranes in a dose-dependent fashion indistinguishable from nonisotopically labeled rat NmU-23. In a rat uterine smooth muscle strip preparation, synthetic frog NmU-23 produced dose-dependent contractions identical to porcine NmU-25. However, in a preparation of human urinary bladder muscle strip, the synthetic frog peptide was more potent than porcine NmU-25 in eliciting contraction and produced desensitization of the preparation to the latter peptide. This report demonstrates that the defensive skin secretion of a frog contains a novel peptide exhibiting a high degree of primary structural similarity to the endogenous vertebrate peptide, NmU, and that this frog skin analog displays biological activity in mammalian tissues.

Amphibian skin secretions have long been known as a rich source of biologically active peptides (1). Currently, in excess of 100 peptides have been structurally characterized, and these have been classified into several families based upon primary structural similarities (2). Many of these peptides have primary structures that are either identical to their endogenous mammalian counterparts (e.g. bradykinin from the skin of the European common frog, Rana temporaria (3)) or possess discrete regions of identity with the active sites of such (e.g. bombesin from the skin of the European frog, Bombina bombina (4)). A number of approaches have been adopted to identify and isolate bioactive peptides from amphibian skin on the basis of physiological tests or bioassay. For example, contractile activity of smooth muscle preparations has enabled several peptides to be identified from a large number of amphibian species (5)(6)(7)(8). Peptide tyrosine tyrosine, synthesized primarily in the endocrine cells of the gut and in brain stem neurons (9), was isolated from a skin extract of the South American leaf frog, Phyllomedusa bicolor, on the basis of its antifungal activity (10). The glandular secretion of the green tree frog, Litoria caerulea, found in the northern territories of Australia and in New Guinea, is known to contain large quantities of peptides, only a few of which have been structurally characterized. These include a group of homologous peptides, the caerins (11), certain of which have antibacterial and antiviral properties (12); the caeridins, whose bioactivity is unknown (13); and caerulein, with physiological effects indistinguishable from CCK-8 (14). To date, the identity of other peptides within the skin secretion remains to be established.
The neuromedin U (NmU) 1 group of peptides exhibit limited sequence similarity with other peptide families. The prototype peptides, NmU-8 and NmU-25, the former representing the C-terminal octapeptide amide of the latter, were isolated from extracts of porcine spinal cord tissues (15), and both were found to be myoactive. NmU immunoreactivity has been found throughout the peripheral and central nervous systems, with the highest concentrations in the pituitary gland, gastrointestinal tract, and nucleus accumbens (16). Primary structural information currently exists on six other members of this peptide family from the rat (17,18), frog (19), guinea pig (20), rabbit (21), dog (22), and chicken (23). The primary structure of a putative human NmU-25 has been deduced from cDNA encoding the human NmU precursor (24). NmU peptides have been shown to be potent stimulators of uterine smooth muscle contraction in vitro (15) and to exert a marked hypertensive effect in rats when administered systemically (25). NmU-8 and NmU-25 cause potent and selective reduction of splanchnic blood flow in the dog and rat (26,27), and NmU-8 alters ion transport in porcine jejunum by a noncholinergic, neuronal mechanism (28). A variety of motor responses have been reported to be mediated by NmU in the gastrointestinal tract of several different species (28,29) and between longitudinal and circular muscle of particular tissues (30). NmU may also exert a direct and indirect effect on the regulation of growth, structure, and function of the adrenal cortex (31)(32)(33). We report here, for the first time, the isolation and structural character-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ization of a 23-amino acid residue NmU analog from the defensive skin secretion of an amphibian. We further demonstrate that a synthetic replicate of this peptide is biologically active in smooth muscle tissues of rat and man and that it can effectively compete with its endogenous mammalian counterparts for specific high affinity binding sites on membrane preparations from such tissues.

MATERIALS AND METHODS
Synthetic NmU-8 and NmU-25 and Lys 0 -NmU-8 were purchased from Peninsula Laboratories (St. Helens, United Kingdom); NmU-8 tracer was monoradioiodinated within our laboratory. Carbamylcholine chloride (Carbachol) was supplied by Sigma (Dorset, UK). All other reagents were of analytical or HPLC grade.
NmU-8 Radioimmunoassay-Guinea pigs were immunized with porcine Lys 0 -NmU-8 conjugated to ovalbumin using glutaraldehyde. Animals were given primary subcutaneous immunizations of 50 g of coupled peptide dispersed in 1 ml of Freund's complete adjuvant and monthly boosters of 10 g of coupled peptide dispersed in 1 ml of incomplete Freund's adjuvant. After two booster injections, an antiserum with appropriate affinity for radioimmunoassay was obtained. All of the reactants for the NmU radioimmunoassay were diluted in 0.04 M sodium phosphate buffer (pH 7.2) containing 0.14 M sodium chloride and 2% (v/v) horse serum. The assay system consisted of 100 l of diluted NmU-8 antiserum (reference code GP 9320 (1:38,000)), 100 l of monoradioiodinated NmU-8 tracer (100 Bq, 2 pg), and 100 l of porcine NmU-8 standard (0 -500 pg/assay tube) or sample. Antiserum and NmU-8 standards (or samples) were incubated for 24 h at 4°C, with further incubation of the assay system for 24 h at 4°C following the addition of monoradioiodinated NmU-8 tracer. Separation of bound from free counts was achieved by the addition of 1 ml of 0.05% (w/v) dextran-coated charcoal, followed by centrifugation at 1100 ϫ g for 30 min at 4°C. Charcoal pellets, containing free counts, were counted using a Nuclear Enterprises NE 1600 ␥-counter. Under these conditions, the sensitivity of the assay was 1.5 pg of NmU-8/assay tube. Cross-reactivity of the antiserum with substance P, calcitonin generelated peptide, pancreatic polypeptide, vasoactive intestinal peptide, neuropeptide tyrosine, peptide tyrosine tyrosine, and neuropeptide phenylalanine was assessed.
Collection of Skin Secretions-Young adult, captive bred specimens (n ϭ 4, snout to vent length of 8 cm) of L. caerulea were obtained from a local supplier. The frogs were maintained in terraria at a temperature of 24°C with a 12 h/12 h light/dark cycle and were fed crickets twice weekly. Following a 4-week period of acclimatization, skin secretions were obtained from the frogs by gentle electrical stimulation (4-ms pulse width, 50 Hz, 5 V) using platinum electrodes placed over the paired parotoid glands. Secretions were visible after a few seconds, and these were washed into a glass beaker with deionized water. The resultant secretions were snap frozen using liquid nitrogen prior to lyophilization. Approximately 50 mg, dry weight, of skin secretion could be obtained from the four frogs on a monthly basis without obvious deleterious effects on the frogs over a period of 2 years.
Gel Permeation Chromatography-Lyophilized skin secretions (approximately 50 mg, dry weight) were reconstituted in 3 ml of 2 M acetic acid and applied to a precalibrated 90 ϫ 1.6-cm chromatographic column packed with Sephadex G50 (medium) gel, equilibrated in 2 M acetic acid. The column was eluted at a flow rate of 10 ml/h, and 2.5-ml fractions were collected.
Sequence Analysis and Mass Spectrometry-The purified NmU immunoreactive peptide was subjected to automated Edman degradation using an Applied Biosystems Procise 491 microsequencer. The limit for detection of phenylthiohydantoin amino acids was 0.2 pmol. The correct mass of the purified peptide was verified by using matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS) on a linear timeof-flight Biflex mass spectrometer (Bruker Instruments), with a positive detection mode and ␣-cyano-4-hydroxycinnamic acid as the matrix.
Internal mass calibration of the instrument with known standards established the accuracy of mass determination as Ϯ0.1%.
The correct mass of the purified peptide was verified by matrixassisted laser desorption/ionization mass spectrometry on a linear timeof-flight MALDI II mass spectrometer (Kratos, Manchester, UK) using the positive detection mode and a sinapinic acid matrix. The spectra were obtained by summing over 50 laser pulses (337 nm). The [M ϩ H] ϩ peak of bovine insulin (m/z 5735) was used for internal mass calibration.
Rat Uterus Preparation-Virgin female Wistar rats (200 -250 g) were killed by cervical dislocation, and both uterine horns were removed en bloc and placed in ice-cold De Jalon solution (154 mM NaCl, 5.95 mM NaHCO 3 , 5.63 mM KCl, 0.54 mM CaCl 2 ⅐2H 2 O, 2.78 mM glucose) equilibrated with a carbogen mixture (95% O 2 , 5% CO 2 ). Each uterine horn was halved, mounted in a 2-ml organ bath containing De Jalon solution, and maintained at 30°C. NmU-8, NmU-25, and frog NmU-23 were dissolved in fresh De Jalon solution each day and added directly to the bath. Contractions were measured isometrically using UFI force displacement transducers (Pioden Controls Ltd., Canterbury, UK). Concentration response curves were constructed by the addition of a single concentration of an agonist, followed by two 5-min washes of the tissue subsequent to a maximum contraction being achieved; the process was then repeated with a higher concentration of the agonist. Tissue responses were normalized by expression in grams of tension produced per milligram of tissue (wet weight) (g/mg). Drug potencies were expressed as EC 50 values, determined by the method of De Lean et al. (34). Results were expressed as mean values Ϯ S.E., and significance (p value Ͻ 0.05) was tested using analysis of variance.
Human Bladder Preparation-Full thickness specimens were taken from the dome of the bladder from two cadaver donors (with consent) and immediately placed in ice-cold Krebs solution (120 mM NaCl, 5.9 mM KCl, 15.4 mM NaHCO 3 , 1.2 mM NaH 2 PO 4 , 2.5 mM CaCl 2 , 1.2 mM MgCl, 11.5 mM glucose) previously equilibrated with a carbogen mixture (97% O 2 , 3% CO 2 ) to give a stable pH of 7.4. Muscle strips (two 2 ϫ 10-mm strips) comprising parallel muscle bundles were dissected from the specimen using a dissection microscope and fine silk (5/0) ligatures tied to each end of the strips. Strips were mounted in 2-ml organ baths containing Krebs solution and maintained at 37 Ϯ 0.5°C with constant bubbling (97% O 2 , 3% CO 2 ). After an initial application of 1 g of tension, tissue was allowed to equilibrate for 1 h, during which time the bath was constantly perfused with Krebs solution (37°C at 1 ml/min). Strips were initially challenged with carbachol (100 M) in order to test the integrity of the muscle and obtain a measure of the maximum contraction. NmU-8, NmU-25, and frog NmU-23 were dissolved fresh each day in Krebs solution and added directly to the bath. Agonists were applied in a noncumulative manner; however, after exposure to an agonist, the organ bath was drained and washed (2 ϫ 5 min) as described above before being perfused at 1 ml/min with fresh, gas-equilibrated Krebs (37°C) for 20 min. Measurement of contraction was performed as described above.
Synthetic rat-NmU-23 was iodinated by the chloramine T method as described previously (35). Peak fractions were assayed for binding to uterine membranes, and the average specific activity, as determined by a specific NmU radioimmunoassay (37), was 24 Bq/fmol. The assay was performed as described by Nandha et al. (35). Briefly, incubations were carried out at 4°C for 60 min in a final volume of 0.5 ml of binding buffer containing 50 mM Tris-HCl (pH 7.4) and aprotinin (30 g/ml). The presence of constant amounts of membrane protein (200 g/ml) and radiolabeled peptide (2000 Bq; 0.17 nM) with the concentration of the unlabeled synthetic peptide varied from 0 to 1 M. The receptor-[ 125 I]rat NmU complex was separated from free tracer by centrifugation at 15,600 ϫ g for 2 min. The pellet was washed once with ice-cold binding buffer (0.5 ml) and counted in a ␥-counter. Nonspecific binding was determined in the presence of 1 M unlabeled rat NmU-23. Specific binding is defined as total binding minus nonspecific binding. Binding data were analyzed by nonlinear regression to determine the affinity (K D ) of binding sites using Receptor-Fit programs (Lundon Software, Cleveland, OH).

RESULTS
Peptide Isolation-NmU immunoreactivity was detected in the gel permeation chromatographic fractions of lyophilized L. caerulea skin secretion eluting in fractions suggestive of a molecular mass of less than 3 kDa (data not shown). Radioimmunoassay indicated the presence of a least 4.5 nmol of NmU immunoreactive peptide in the sample. Following semipreparative HPLC fractionation of 50% of the pooled NmU immunoreactive gel permeation chromatographic fractions, immunoreactivity was localized to a single peak spanning three fractions (Fig.  1A). Further purification of these fractions by diphenyl reverse phase HPLC produced a single symmetrical peak of UV absorbance coincident with NmU immunoreactivity (Fig. 1, B and C). This purified peptide was subjected to structural analysis.
Sequence Analysis and Mass Spectrometry-The primary structure of the purified peptide was established unequivocally as SDEEVQVPGGVISNGYFLFRPRN, and the observed mass of 2580.6 Da corresponded precisely with the theoretically predicted mass of the peptide with an amidated C-terminal residue (2580.8 Da). The full molar cross-reactivity of the frog NmU-23 with the amide requiring NmU antiserum employed in a radioimmunoassay confirmed the presence of this common post-translational modification. Sequence alignment of the newly identified peptide with those of the Swiss-Prot and Gen-Bank TM peptide/protein sequence data bases revealed a very high degree of sequence similarity with peptides of the NmU family (Table I). Approximately 87% of the NmU immunoreactivity applied to the original chromatographic column was purified to homogeneity. Rat Uterus Preparation-The frog NmU-23, along with porcine NmU-25 and NmU-8 produced dose-dependent contractions in rat uterine smooth muscle preparations (Fig. 2), with EC 50 values of 1.1 Ϯ 0.1, 1.8 Ϯ 0.1, and 12 Ϯ 0.6 nM, respectively. The potencies of porcine NmU-25 and frog NmU-23 were not significantly different; however, both peptides were significantly more potent than porcine NmU-8 (NmU-25 was 6.7 times and NmU-23 was 10.9 times more potent than NmU-8). The maximum contraction elicited by each peptide was not significantly different.
Human Bladder Preparation-Application of frog NmU induced contractions of human detrusor smooth muscle strips in a dose-dependent manner (n ϭ 4) (Fig. 3). It was apparent that 10 nM frog NmU-23 produced stronger contraction, suggesting a greater potency when compared with an equimolar concentration of porcine NmU-25 (Fig. 4). Furthermore, upon reapplication of the same concentration of porcine NmU-25 following frog NmU-23, a 35.3 Ϯ 12.1% decrease was apparent when compared with the previously evoked response (n ϭ 4).
Receptor Binding Assay-Synthetic frog NmU-23 displaced binding of 125 I-labeled rat NmU-23 from rat uterine membranes in a dose-dependent fashion indistinguishable from nonisotopically labeled rat NmU-23 (Fig. 5). Scatchard analysis of saturation binding data produced dissociation constants (K d ) for rat NmU-23 and frog NmU-23 of 5.10 Ϯ 2.83 and 12.72 Ϯ 5.29 nM, respectively (n ϭ 3). , v/v); linear gradient 15-70% B in 55 min, at a flow rate of 1 ml/min. NmU-LI was identified in one absorbance peak (arrow). C, final purification of the peptide was performed using a 15 ϫ 0.46 cm Kromasil C18 column; mobile phases A and B and the linear gradient were identical to the previous purification step, above. NmU-LI was isolated to one purified peak (arrow).  (18). The amino acid sequence for rabbit neuromedin U can be accessed in the Swiss-Prot Database under accession no. P34965 (20). The amino acid sequence for human neuromedin U can be accessed in the Swiss-Prot Database under accession no. P48645 (23). The amino acid sequence for rat neuromedin U can be accessed in the Swiss-Prot Database under accession no. P12760 (42). The amino acid sequence for chicken neuromedin U can be accessed in the Swiss-Prot Database under accession no. P34963 (22). The amino acid sequence for pig neuromedin U can be accessed in the Swiss-Prot Database under accession no. P34964 (14). The amino acid sequence for dog neuromedin U can be accessed in the Swiss-Prot Database under accession no. P34962 (21). The amino acid sequence for guinea pig neuromedin U can be accessed in the Swiss-Prot Database under accession no. P34966 (19).

DISCUSSION
The defensive skin secretions of frogs are known to contain a plethora of biologically active peptides, many of which share common primary structural features with endogenous vertebrate regulatory peptides. Here, we report the identification and structural characterization of a NmU analog from the skin secretion of a frog for the first time. This finding thus extends the list of endogenous vertebrate regulatory peptides with frog skin counterparts and vice versa. Table I illustrates the structural similarity of frog NmU-23 to other members of the NmU family, and from this, it is clear that the highest degree of similarity (70%) is shared with the C-terminal 23 amino acid residues of intestinal NmU-25 from the frog, Rana temporaria (19). Frog skin NmU-23 has the same number of amino acid residues within its sequence as rat NmU-23; however, the two analogs differ in that the frog peptide is attenuated by two N-terminal amino acid residues, whereas two amino acid deletions have apparently occurred within the central domain of the rat peptide (17,18). Most members of the NmU family have a putative dibasic or monobasic amino acid residue cleavage site at positions 16 and 17, which in some species leads to the endogenous generation of an N-terminally truncated molecular variant containing the C-terminal active core of the peptide (15,22,38). Frog NmU-23 and rat NmU-23 are unique in that they lack this putative site, and therefore it is probable that they exist only as the elongated form. Conservation of Glu 5 and Gln 8 in the N-terminal region of porcine NmU-25 is found in the frog NmU-23, and the highly conserved Pro 10 is present in all other species except chicken (23). The N-terminal region is thought to increase the potency and prolong the activity of the peptide (15), and preservation of these residues is likely to be essential for biological activity.  The quantity of NmU-23 in L. caerulea skin secretion (several nmol) is comparatively lower than previously reported levels of other peptides from this frog, such as caerulein, which is reported in quantities as great as 1.33 micromol/g of dried skin (39). This factor, however, does not diminish its significance, since biological potency is perhaps more important than absolute quantity. The concentration of this peptide in secretions is, however, significantly higher than the levels of NmU found in neuroendocrine source tissues from other species (rat spinal cord, 61.0 pmol/g (40); chicken intestine, 97.5 pmol/g (23); rabbit small intestine, 22 pmol/g (21); and pig spinal cord, 26.6 pmol/g (40)). These levels are somewhat lower, however, than those found in frog gut, which are reported to be as much as 10-fold higher but still considerably less than levels present in L. caerulea skin secretion (19). NmU immunoreactivity was not detected previously in the skin of R. temporaria (19), and we have confirmed this observation using our radioimmunoassay. It is possible that NmU, like other previously identified vertebrate regulatory peptide analogs, is of restricted distribution within frog taxa, since we have failed to detect this peptide in skin secretions from ranid, discoglossid, leptodactylid, and neotropical hylid frogs. However, further studies on larger samples of species are required to address this question in a meaningful manner.
In studies designed to determine myotropic activity, frog skin NmU-23 produced a similar contractile response to porcine NmU-25 in a rat uterine smooth muscle preparation. The frog peptide was approximately 10.9 times more potent on a molar basis when compared with porcine NmU-8. The latter peptide produced similar EC 50 values to those reported previously (16 Ϯ 5 nM (22); 46 Ϯ 8 nM (38)). In addition, the IC 50 value obtained for inhibition of monoradioiodinated rat NmU binding to rat uterine membrane preparations (60 nM) was of a similar magnitude to that previously reported (35)). However, in the absence of suitable NmU receptor antagonists, it is not possible to state unequivocally whether both peptides activate or indeed bind to the same receptor. With such qualifications in mind, the synthetic replicate of the natural frog skin peptide did displace monoradioiodinated NmU from uterine smooth muscle membranes in a dose-dependent fashion indistinguishable from nonisotopically labeled NmU-23. Previously, porcine NmU-8 has been shown to exert a concentration-dependent contractile effect on human detrusor smooth muscle strips (40), and this effect can now be extended to include both NmU-25 and frog NmU-23. In addition, the synthetic frog peptide appeared to exert almost twice the contractile effect of porcine NmU-25. The reduction in response to 10 nM porcine NmU-25, after exposure to 10 nM frog NmU-23, may be due to an increased tachyphylactic affect of frog NmU after a relatively short exposure time. It may be that frog NmU has a comparably longer half-life than porcine NmU-25, due to the absence of the dibasic processing site where cleavage to the attenuated and more labile porcine NmU-8 form is likely to reduce expression of the bioactivity.
The discovery of a novel member of the NmU family in frog skin secretions demonstrates that this molecule is of important biological function, although its actions are likely to be specific for certain tissues. For many years, a relationship between similar peptides from the gut, brain, and skin of amphibians has been known, due to the embryological origin of all of these tissues from the neural crest (41). It may be that a regulatory loop exists between the pituitary and the skin in amphibians, as suggested previously (10).
Virtually nothing is known of the physiological roles that skin peptides play in amphibian survival; however, they may be involved in antipredation, modes of reproduction, and mating behavior or involved in control in osmoregulation (42). Establishment of these functions is difficult, due of the large variety of endogenous skin compounds, which could interact together to generate complex effects not produced by a single compound alone. Subsequently, the in vitro effects of NmU may not represent the actual physiological function. Previous studies have demonstrated that the secretion of the African clawed frog, Xenopus laevis, promotes yawning and other orofacial movements in several predatory snakes, which slowed ingestion rate and facilitated escape of the frog (43,44). Cerulein, as well as the myoactive and neuromodulatory peptide, xenopsin, were reported to be responsible for some of these antipredator actions (42). In the arboreal L. caerulea, the high content of cerulein (13), in conjunction with NmU, may elicit peripheral or neurocentral effects to the predator that protect this species from being consumed. An alternative strategy for NmU may be to act systemically on the frog itself, to promote hypertension following injury. This would benefit the frog by preventing death by rapid blood loss. These actions would be facilitated by cerulein, which supports cardiovascular function following massive hemorrhage (45,46). It is clear that skin NmU represents a further peptide whose specific role for the amphibian has yet to be elucidated.
This report demonstrates that amphibian skin is a very valuable resource for the discovery of novel and functionally significant bioactive peptides, many of which have mammalian counterparts that themselves await discovery.