Relaxin-like bioactivity of ovine Insulin 3 (INSL3) analogues Antonia A. Claasz 1 , Courtney P. Bond 2 , Ross A. Bathgate 1 , Laszlo Otvos Jr 3 , Nicola F. Dawson 1 , Roger J. Summers 2 , Geoffrey W. Tregear 1 and John D. Wade 1 1 Howard Florey Institute, University of Melbourne, Victoria, Australia; 2 Department of Pharmacology, Monash University, Victoria, Australia; 3 The Wistar Institute, Philadelphia, USA Relaxin is an insulin-like peptide consisting of two separate chains (A and B) joined by two inter- and one intrachain disulfide bonds. Binding to its receptor requires an Arg–X– X–X–Arg–X–X–Ile motif in the B-chain. A related member of the insulin superfamily, INSL3, has a tertiary structure that is predicted to be similar to relaxin. It also possesses an Arg–X–X–X–Arg motif within its B-chain, although this is displaced by four amino acids towards the C-terminus from the corresponding position within relaxin. We have previ- ously shown that synthetic INSL3 itself does not display relaxin-like activity although analogue (Analogue A) with an introduced arginine residue in the B-chain giving it an Arg cassette in the exact relaxin position does possess weak activity. In order to identify further the structural features that impart relaxin function, solid phase peptide synthesis was used to prepare three additional analogues for bioassay. Each of these contained point substitutions within the arginine cassette. Analogue D contained the full human relaxin binding cassette, Analogue G consisted of the native INSL3 sequence containing an Arg to Ala substitution, and Analogue E was a further modification of Analogue A, with the same substitution. Each analogue was fully chemically characterized by a number of criteria. Detailed circular dichrosim spectroscopy analyses showed that the changes caused little alteration of secondary structure and, hence, overall conformation. However, each analogue displayed only weak relaxin-like activity. These results indicate that while the arginine cassette is vital for relaxin-like activity, there are additional, as yet unidentified structural require- ments for relaxin binding. Keywords: cAMP; ligand binding; receptor; solid phase peptide synthesis; THP-1 cells. Relaxin is a 6-kDa peptide with two chains linked by one intra- and two interchain disulfide bonds which is charac- teristic of the insulin superfamily of peptide hormones. Relaxin displays a wide variety of biological effects, including connective tissue remodelling in the female reproductive tract [1], stimulation of chronotropic and inotropic responses in heart atria [2], and regulation of fluid balance [3]. It has also been shown to be a potent vasodilator [4,5] as well as an antifibrotic agent [6] suggesting that it may have potential therapeutic applica- tions. Extensive structure–function studies undertaken over the years have led to the identification of the key structural elementsthatareinvolvedintheinteractionofrelaxinwith its receptor [7,8]. In particular, a motif in the B-chain, consisting of two arginine residues in positions B13 and B17 [9,10], forming an arginine cassette (Arg–X–X–X–Arg), has been demonstrated to be essential for receptor binding. More recently it has been found that the B20 isoleucine (IleB20) residue is also crucial for receptor binding [11]. When it is replaced with another residue, such as Ala or Thr, the ability of relaxin to bind to its receptor is greatly reduced [11]. One exception to this effect is when Ile is replaced with Val. This substitution results in very similar bioactivity to its IleB20 counterpart; indeed, some native relaxins contain a Val in this position [12]. This B20 Ile/Val residue, along with the previously mentioned Arg cassette, forms an extended Arg–X–X–X–Arg–X–X–Ile/Val cassette, which is now regarded as the principal relaxin receptor binding motif. The primary structure of INSL3 (also known as relaxin- like factor and Leydig cell insulin-like peptide), a recently identified member of the insulin-like superfamily, possesses striking similarity to relaxin with a predicted A-chain of 26 residues and a B-chain of 32 residues. Intriguingly, the INSL3 B-chain contains an Arg–X–X–X–Arg binding cassette, however, it is displaced towards the C-terminus by four residues. Although INSL3 displays its own distinct physiology with important roles in testis descent and ovarian function [13], previous work indicated that INSL3 may interact with relaxin receptors [14]. Studies from our laboratory using synthetic ovine INSL3 showed that this peptide has no relaxin-like activity in the rat atrial bioassay [15], and did not cause a significant reduction in the response to relaxin when the two peptides were assayed together. Modelling of bombyxin, an insulin-like protein from the silkworm, together with the known tertiary structures of insulin and relaxin show a distinctive insulin- like structure, consisting of two a-helices in the A-chain and three disulfide bonds [16]. INSL3 is predicted to possess a similar insulin-like structure, so its lack of relaxin-like activity is probably a consequence of the absence of specific amino acids required for the correct relaxin-like binding Correspondence to J. D. Wade, Howard Florey Institute of Experi- mental Physiology and Medicine, The University of Melbourne, Victoria 3010, Australia. Fax: 61 3 9348 1707, Tel.: 61 3 8344 7285, E-mail: j.wade@hfi.unimelb.edu.au Abbreviations: INSL3, insulin-like peptide 3; LGR7/8, leucine-rich repeat G-protein coupled receptor 7/8. (Received 19 June 2002, revised 1 September 2002, accepted 5 November 2002) Eur. J. Biochem. 269, 6287–6293 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.03348.x motif in the correct position. An analogue with an Arg residue replacing the His in position B12, thus creating an Arg–X–X–X–Arg cassette in the same position as in relaxin (as aligned by cysteine residues) was then chemically synthesized (Analogue A). It produced weak relaxin-like activity in the isolated rat atria [15], clearly reinforcing the importance of the Arg binding cassette in the relaxin conformation. Thus, INSL3 and the effect of this single amino acid substitution, provides us with a useful molecular template with which to study any further structural requirements for full relaxin binding. However, as the relaxin-like activity produced by this analogue was com- paratively weak, it is clear that there are further require- ments besides the Arg cassette for full interaction of relaxin with its receptor. This prompted us to examine further modifications of the INSL3 B-chain in comparison with relaxin. Firstly an analogue was produced with a full human relaxin Arg cassette (RELVR) placed into the B-chain (Analogue D, Fig. 1) in the relaxin position. Secondly, as mentioned above, IleB20 is also a vital residue for relaxin- like bioactivity and this corresponds to a Val residue (B19) in INSL3. However, an additional Arg residue is located in the B20 position of INSL3, adjacent to this Val and may have the effect of hindering the interaction of the Val residue with the relaxin receptor. To further investigate this, we undertook the chemical synthesis of two new INSL3 analogues where the ArgB20 residue was substituted for a less bulky Ala residue (Analogues E and G; Fig. 1). Removal of the ArgB20 might be expected to have the effect of unmasking the ValB19 residue, allowing it to interact with the receptor. Each of these newly designed analogues (D, E and G), together with the previously prepared analogue (A), were assayed in the established cAMP bioassay [17], and relaxin receptor binding assay. Both assays are based on the THP-1 human monocytic cell line [18], which expresses the human relaxin receptor. MATERIALS AND METHODS Peptide synthesis For each INSL3 analogue, both the A- and B-chains were separately synthesized by the continuous flow Fmoc solid-phase method on a 0.15-mmol scale. Each A-chain was assembled manually on a CRB Pepsynthesiser, using Fmoc-His(Trt)-Novasyn PA 500 (Novabiochem, Switzer- land) as support. Amino acid acylation was performed with HOBt-catalyzed Fmoc-amino acid pentaflourophenyl esters (Auspep, Melbourne, Australia), with the exception of arginine whichwas activated withO-(7-aza-benztriazol-1-yl)- N,N,N,N¢-tetramethylammonium hexaflourophosphate (HATU) and DIEA in dimethylformamide. B-chain syn- theses were carried out using an automated MilliGen 9050 synthesiser on PAC-PEG-PS support. Amino acid acylation was performed with HOBt in dimethylformamide with the addition of 1,3-diisopropylcarbodiimide. For both chains, N a -Fmoc deprotection was with 20% piperidine in dimethyl- formamide. All amino acid couplings were of 30 min duration. Cleavage and purification Protected A- and B-chain resins were each treated with 82.5% trifluoroacetic acid/5% phenol/5% H 2 O/5% thioan- isole/2.5% ethanedithiol (v/v/v/v) plus three drops of triethylsilane for 3 h. The trifluoroacetic acid solution was then removed under a stream of nitrogen, followed by diethyl ether extraction twice. The resulting peptide pellet was then resuspended in 0.1% (v/v) aqueous trifluoroacetic acid and lyophilized. Crude peptides were purified using a Waters HPLC system using a Vydac C 18 reverse-phase (RP) column (10 · 250 mm 218TP), with a solvent system of 0.1% (v/v) aqueous trifluoroacetic acid (buffer A) and 0.1% (v/v) trifluoroacetic acid in acetonitrile (buffer B) in linear gradient mode. Fractions were collected and lyophilized. Chain combination Disulfide bond formation between purified chains was performed in 0.15 M 3-(cyclohexamino)propanesulfonic acid (CAPS), pH 10.5, 1 M guanidine hydrochloride, 10% (v/v) methanol, 2.5 m M dithiothreitol buffer, vigorously stirring at 4 °C. The reaction was monitored by RP-HPLC at hourly intervals, and terminated by addition of trifluoro- acetic acid when no starting B-chain was remaining (after approximately 24 h). Preparative RP-HPLC, as described above, was then used to isolate and purify the product. Characterization Purity of both peptides was analyzed using MALDITOF MS, performed in the linear mode on a Bruker BIFLEX instrument, analytical RP- and cation-exchange HPLC. Cation-exchange HPLC was performed using a Poly CAT column and a 0–70% gradient (Buffer A: 25 m M KH 2 PO 4 , pH 7, and Buffer B: 0.5 M KCl + 25 m M KH 2 PO 4 ,pH7). Peptide quantitation was performed by amino acid analysis on a GBC automatic amino acid analyzer (Melbourne, Australia). Circular dichroism spectroscopy Circular dichroism (CD) spectroscopy was carried out using a Jasco J-720 spectropolarimeter using a 0.2-mm path length cell in the 180–260 nm range. The peptides were dissolved in double-distilled water at room temperature. Four scans were averaged, and the resulting crude spectra were smoothed by the algorithm provided by Jasco. Mean residue ellipticity is expressed in deg cm 2 Ædmole )1 with using a mean residue mass of 110. Alpha helicity was calculated from the ellipticity values at 208 nm, according to the algorithm of Greenfield and Fasman [19]. The peptide Fig. 1. Sequence of INSL3 B chain, analogue B-chain sequences, and relaxin B-chain. Underlined residues in the relaxin B-chain indicate the relaxin binding motif. Underlined residues in INSL3 and analogue B-chains indicate the displaced Arg cassette. High- lighted residues indicate amino acid substitutions in comparison to INSL3. 6288 A. A. Claasz et al.(Eur. J. Biochem. 269) Ó FEBS 2002 concentrations were 0.17–1.38 mgÆmL )1 , as determined by amino acid analysis (above). cAMP bioassay Peptides were assayed for relaxin-like activity in THP-1 cells, using a cAMP ELISA (Biotrak), as previously described [17]. In short, THP-1 cells were cultured in suspension, and plated out in a 96-well plate, at a density of 40 000 cells per well. Cells were then incubated with 1 l M forskolin and 50 l M IBMX (in 200 lL RPMI 1640, containing 10% foetal calf serum, penicillin/streptomycin and 2 m ML -glutamine), plus the peptide of interest for 30 min at 37 °C. Analogues were assayed alongside 5 n M H2(B29) relaxin and medium with IBMX and forskolin only (used as control). The 96-well plate was then centri- fuged and the supernatant immediately aspirated. Cells were lysed in 200 lL lysis buffer (from kit) per well, and 100 lL of the lysate per well used in the ELISA to measure cAMP levels. Each analogue was tested in at least three experiments in quadruplicate and the data analyzed by one-way ANOVA followed by Newman–Keuls multiple comparison test. Binding assay THP-1 cells were grown in modified RPMI 1640 medium, spun down and resuspended in binding buffer [20 m M Hepes, 50 m M NaCl, 1.5 m M CaCl 2 ,1%(w/v)BSA, 0.1 mgÆmL )1 lysine, 0.01% NaN 4 ,pH7.5][14]togive 2 · 10 6 cellsÆwell )1 in a 96-well plate. The cells were incubated in binding buffer with 100 p M 33 P-labeled human gene 2 (B33) relaxin [H2 (B33)], [20], at 25 °Cfor90minin the absence or presence of increasing concentrations of unlabelled human gene 2 (B29) relaxin [H2(B29)]; 100 p M to 30 n M ), INSL3 (30 n M to 30 l M ), Analogue D and Analogue E (10 n M to 300 l M ). Nonspecific binding was definedwithH2(B29)(1l M ). Cells were harvested using a Packard 96-well plate cell harvester and Whatman GF/C glass fibre filters treated with 0.5% poly(ethylenimine). The filters were washed three times with modified binding buffer (20 m M Hepes, 50 m M NaCl, 1.5 m M CaCl 2 ),driedina 37 °C oven, 30 lL of scintillant (Microscint O, Packard) added and the radioactivity counted by liquid scintillation spectrometry (TopCount TM ,Packard). RESULTS The chemical synthesis of the individual chains proceeded without difficulty, and following cleavage and purification, were obtained in good overall yield. Each of the A- and B-chains were subjected to detailed chemical characteriza- tion prior to their combination in solution. Analogues E and G were obtained in modest yield due to the difficulty in preventing B-chain dimerization during the process of chain combination. Analysis of both peptides by MALDITOF MS showed principal products with the expected molecular weights, but also the trace presence of A-chain (not shown). No other contaminants were detected. Analytical cation-exchange HPLC was used to further assess the purity of the analogues, as free A-chain coeluted with the combined A- and B-chain on reverse-phase HPLC. Cation exchange HPLC produced a single peak, and no free A-chain was detected (data not shown) indicating that only trace amounts of free A-chain were present. Mass spectral analysis of the main peak for each analogue contained the target disulfide bonded A- and B-chain product (not shown). Secondary structure analysis by CD spectroscopy showed the presence of a positive band for all three INSL3 peptides at 187 nm and a negative band at 205 nm. An additional negative shoulder was seen at around 222 nm (Fig. 2). The helical content of these peptides were around 25% and 11% for Analogues E and G, respectively, although these figures need to be treated with caution due to the low peptide concentration which may influence the accuracy of the results. However, there were clearly no major spectral differences found compared to unmodified INSL3, but all three ovine sequences exhibited significantly reduced alpha- helicity compared to the human relaxin peptide. While the CD spectrum of relaxin showed bands characteristic for alpha helices (192 nm positive, and 208 nm and 223 nm negative bands), the helix for INSL3 analogs were signifi- cantly less alpha helical, as indicated by the blueshift of the pi-pistar bands to 187 nm and 205 nm. These CD spectra were remarkably similar to those of the N-methyl- D -aspartate-antagonist peptide conantokin G [21], and featured characteristics of 3–10 helices, which can be considered as repeating b turns [22]. Indeed, high resolution NMR spectra of conantokin G verified the 3–10 helix structure [23]. Hence, the intensity differences between the INSL3 peptide spectra may reflect differences in the turn- forming potential. However, the very similar spectral characteristics argue for simple inaccuracies in concentra- tion determination. In this regard it is interesting to note that the amino acid changes in the INSL3 to more closely Fig. 2. CD spectra for human relaxin (solid line), sheep INSL3 (dashed line), Analogue E (dots and dashes) and Analogue G (dotted line) in water. Ó FEBS 2002 Relaxin activity of B chain analogues of ovine INSL3 (Eur. J. Biochem. 269) 6289 mimic the relaxin structure were not accompanied by more alpha helical spectral transitions in water (Fig. 2). Analogues D, E and G, along with previously synthesized Analogue A were subjected to a cAMP bioassay using THP-1 cells. These cells have been shown to express the relaxin receptor at a density of  275 receptors per cell [17]. A preliminary competitive whole-cell binding assay, using 33 P-labelled H2(B33) relaxin with increasing concentrations of unlabelled H2 (B29) relaxin, was performed in order to verify the receptor number in our batch of cells, before these were used to assay any peptides. Scatchard analysis showed that receptors were expressed at a density of  265 receptors/cell (data not shown), with a dissociation constant (K d ¼ 0.13 n M ), which is similar to what has previously been found [17]. THP-1 cells responded to increased concentrations of H2(B29) relaxin with increased cAMP production, with an approximate EC50 of 0.1 n M . Maxi- mum cAMP stimulation was achieved at 5 n M (Fig. 3), and hence was taken as the 100% value used for comparison with the analogues. Each peptide was assayed alongside varying concentra- tions of H2(B29) relaxin, as an internal control in each assay, and also for the purpose of comparison. All analogues were assayed at a concentration of 100 and 500 n M , and at additional concentrations where possible, alone or in the presence of 5 n M H2(B29) relaxin to ascertain whether these analogues could augment or antagonize the response to relaxin. INSL3 itself was also assayed at 500 n M , on its own as well as in the presence of H2(B29) relaxin. Free A-chain was also assayed both in isolation and in the presence of H2(B29) relaxin to ensure that it did not have any effect on its own, in order to control for the presence of free A-chain detected by mass spectro- metry as a contaminant in trace amounts in both analogues E and G. The THP-1 cAMP assay was responsive to H2(B29) relaxin and analogues A, D, and E. However, INSL3 produced no significant response, nor did it have any effect on the response to relaxin (Fig. 4A). Analogue A (Fig. 4B) produced approximately 60% of the maximum relaxin response at 100 n M , and around 75% maximum response at 500 n M , while a 55% maximum response at 100 n M and a 90% maximum response at 500 n M was produced by Analogue D (Fig. 4C). Analogue E displayed a response at 50% of maximum relaxin response at 100 n M , and approximately 80% of maximum relaxin response at 500 n M (Fig. 4D). Furthermore, a  40% response was observed at 50 n M , and no significant response was observed at 10 n M . Therefore, while Analogue E did display some degree of bioactivity the activity of the previous analogues was not improved upon. Neither Analogue G (Fig. 4E) nor free A chain (not shown) produced any response. Interestingly, Analogue A produced a significant reduc- tion in the response to relaxin when the two were assayed in combination (P < 0.05, by ANOVA followed by New- man–Keuls multiple comparison test). Analogues D and E also produced a slight reduction in the relaxin response, however, these results did not reach statistical significance. Due to the limited amounts of peptide available, higher doses could not be assayed for antagonist activity. The results obtained from the THP-1 binding assay (Fig. 5) showed that INSL3 had low affinity for the relaxin receptor (pK i 6.0 ± 0.32, n ¼ 4) compared to H2(B29) relaxin (pK i 8.7 ± 0.11, n ¼ 11). Incorporation of the relaxin Arg motif into INSL3 (Analogue D) increased the affinity of the peptide fivefold for the relaxin receptor (pK i 6.7 ± 0.18, n ¼ 4) but it was still 100-fold weaker than H2(B29) relaxin. Removal of both Arg motifs from INSL3 (Analogue G) further reduced affinity for the receptor (pK i 5.4 ± 0.20, n ¼ 4). Due to the limited amounts of available peptide, analogues A and E could not be assessed in the THP-1 binding assay. DISCUSSION While species variation in amino acid sequences of relaxin is significant, and may be as great as 60%, relaxin peptides will generally interact with relaxin receptors between species [24,25]. The common elements to relaxin peptides between species are the disulfide bond patterns, and the Arg–X–X– X–Arg–X–X–Ile motif in the B-chain. Given that INSL3 has the former and a portion of the latter, we sought to make it more relaxin-like by successively introducing all of the Arg cassette components. Such a study would help provide a clear indication of the role of this cassette for relaxin function. As earlier studies showed that human, ovine (unpublished data) and rat [26] INSL3, all show no activity in the THP-1 cell bioassay and, in our hands, the ovine peptide is slightly easier to chemically synthesize, we chose to use the ovine sequence as the template for further study. The four INSL3 analogues that were selected for preparation allowed examination of the amino acid residues required for interaction of relaxin with its receptor. It was previously shown that while native INSL3 is not a relaxin- like hormone at physiological concentrations, replacement of B-chain HisB12 with Arg did result in some relaxin-like activity in isolated rat atria [15]. The interaction of Analogue A with the receptor indicated that INSL3 must have a similar secondary structure to relaxin. Although active, Analogue A was still 100-fold less potent than H2(B29) relaxin, clearly indicating that the arginine cassette is probably only one of a number of requirements for full relaxin-like bioactivity. This has been highlighted by the recent discovery that IleB20 is crucial for binding of relaxin to its receptor [11], which corresponds to ValB19 in INSL3 Fig. 3. Standard response to relaxin in the cAMP/THP-1 cell assay. Data is expressed as percentage maximum response. pEC50 ¼ 10 (0.1 n M ). Maximum response observed at  5n M . 6290 A. A. Claasz et al.(Eur. J. Biochem. 269) Ó FEBS 2002 (when sequence is aligned according to the cysteine residues). It was shown that when IleB20 was replaced with an Ala residue, it had much less affinity for the relaxin receptor in mouse brain. However, when IleB20 was replaced with a Val residue, binding was not affected. Val is also present in this position in some native relaxins, such as porcine relaxin [12]. This is of particular importance when comparing the B-chains of relaxin and INSL3. In INSL3, the corresponding residue to the relaxin IleB20 residue is a Val. This means that yet another residue of importance for relaxin-like activity is present in INSL3, however, the adjacent ArgB20 residue may well sterically hinder the interaction of this residue with the relaxin receptor. It was speculated that activity of Analogue A [15] could be improved upon by removing this additional Arg residue found in INSL3 (ArgB20). INSL3 demonstrated no significant relaxin-like activity in the THP-1 cAMP assay. It also did not have any significant effect on the cAMP response to H2(B29) relaxin when the two were assayed in combination. In a different assay system, the rat atrial bioassay, INSL3 also has no significant effect on chronotropic or inotropic responses [15]. At the time of writing this paper, the sequences of the relaxin and INSL3 receptors became available, named LGR7 and LGR8, respectively [27]. The two receptors are G-protein coupled receptors, and share approximately 60% sequence homology. Interestingly, both receptors bind relaxin. Clearly aspects of the ligand-binding domain of both receptors must share some similarity, while the ligands themselves have subtle differences which affords them their specificity. A better understanding of the binding require- ments of these individual ligands may ultimately allow us to design specific analogues to avoid cross-reactivity between the two receptors. We have since confirmed, using RT-PCR, that THP-1 cells only express LGR7 (data not shown). Analogue A and D both displayed relaxin-like activity in the THP-1 cAMP assay. Analogue A produced  75% of the maximum response at 500 n M , while Analogue D at the same concentration produced  80% of the maximum response. In the rat atrial bioassay Analogue A at 1 l M produced 42% and 45% of the maximum response to H2(B29) relaxin in the chronotropic and inotropic assay, respectively [15]. Analogue E contains a human relaxin binding motif, as well as having the Arg21 residue replaced by Ala21. The replacement of the Arg residue failed to increase relaxin activity. Analogue E produced 80% of the maximum H2(B29) relaxin cAMP response at 500 n M and was therefore equipotent to Analogues A and D. In Analogue G ArgB20 was replaced by an Ala residue. Analogue G would thus serve as a negative control, in case this residue replacement was sufficient to induce relaxin-like activity. Analogue G failed to have activity in the relaxin Fig. 4. cAMP response to sheep INSL3, Analogues A, D, E and G. (A) cAMP response in THP-1 cells to sheep INSL3, expressed as a per- centage of maximum relaxin response. (B) cAMP response to Ana- logue A, expressed as a percentage of maximum relaxin response. * P < 0.05 vs. relaxin 5 n M . (C) cAMP response to Analogue D, expressed as a percentage of maximum relaxin response. (D) cAMP response to Analogue E, expressed as a percentage of maximum relaxin response. (E) cAMP response to Analogue G, expressed as a percentage of maximum relaxin response. Each analogue was tested in at least three experiments in quadruplicate and the data analyzed by one-way ANOVA followed by Newman–Keuls multiple comparison test. Ó FEBS 2002 Relaxin activity of B chain analogues of ovine INSL3 (Eur. J. Biochem. 269) 6291 bioassay, and furthermore showed a lower affinity for the relaxin receptor in the binding assay, reinforcing the importance of the Arg–X–X–X–Arg cassette B13–17 posi- tion for relaxin-like bioactivity. The circular dichroism spectroscopy studies performed on these peptides showed that both INSL3 and analogues E and G were less a-helical than human relaxin. As the primary receptor binding site of relaxin is located in the a-helical region of the B-chain, any changes in the helicity in this region could well change the orientation of the binding cassette and hence have a considerable effect on the receptor–ligand interaction. This could therefore be a major contributing factor to the reduced bioactivity of these analogues. Of further interest is the reduction in relaxin response seen when H2(B29) relaxin and Analogue A were assayed together. This is in contrast to the rat atrial bioassay [15], where no antagonist activity was observed for this analogue. Analogues D and E show some potential for antagonist activity, although this was not significant at the concentra- tions assayed. A possible explanation for the antagonist activity of Analogue A could be the presence of the additional ArgB20 residue in the B-chain along with the lack of a human relaxin Arg cassette (this analogue had the RHFVR motif). In comparison to Analogue D, which has the human relaxin Arg (RELVR) cassette, Analogue A shows a trend towards causing a lesser relaxin-like response in this assay. The complete RELVR sequence may cause a slight change in the overall structure of this particular region of the B-chain, allowing a better interaction of Analogue D with the receptor compared to Analogue A, where the Arg cassette has the sequence RHFVR. Furthermore, the GluB13 residue in the complete Arg cassette, as in Analogue D, has a negative charge, while the HisB13 residue in Analogue A has a positive charge. This may also influence the interaction of these analogues with the relaxin receptor. This unfavourable effect on the structure of the B-chain in Analogue A coupled with the additional ArgB20 residue may be sufficient to give Analogue A weak antagonist activity. Even if Analogue D still has the additional Arg, the more favourable interaction of the human Arg cassette with the receptor may serve to counteract any antagonist activity of this analogue. Analogue E, while not having the complete human relaxin Arg cassette (RHFVR, as in Analogue A), does not possess the additional ArgB20 residue, reducing its potential for antagonist activity. Our data from this study has re-emphasized the import- ance of an Arg cassette in the correct position for relaxin- like activity. However, it has also clearly demonstrated that there are other necessary factors for full relaxin-like bio- activity. The overall structure of the INSL3 B-chain may have subtle, yet vital differences compared to relaxin, which necessarily prevents it from fully interacting with the relaxin receptor. To investigate this aspect, other residues in the INSL3 B-chain could be substituted, to give rise to further INSL3 analogues. Furthermore, the INSL3 A-chain could also display structural differences to the relaxin A-chain. It has previously been shown that the relaxin A-chain is essential for full bioactivity [28], most likely by acting as a scaffold to hold the B-chain in its correct conformation. Our study also suggests the distinct possibility of produ- cing a relaxin antagonist by changing the structure of INSL3. However, for this to be achieved, full elucidation of the INSL3 tertiary structure will ultimately be required for a better understanding of the essential differences between relaxin and INSL3. Nevertheless, INSL3 is clearly not only a useful tool for understanding the requirements for agonist activity at the relaxin receptor, but it also has the potential to assist us in designing and preparing a relaxin antagonist, which will provide a useful molecular probe of relaxin action. Given the recent confirmation of the potent vaso- active effect of relaxin and the fact that vasodilator pathways are key targets for new therapeutics, relaxin is an attractive peptide for further development into both agonists and antagonists. With the recent cloning of the relaxin and INSL3 receptors [27] and the discovery that both receptors bind relaxin, the subtle differences between the two ligands is of even more importance, particularly in the design of specific therapeutics. ACKNOWLEDGEMENTS The work described herein was supported by an Institute Block Grant Reg Key Number 983001 from the National Health and Medical Research Council of Australia. We thank Mare Cudic (Wistar Institute) for performing the CD spectroscopy. We also thank Kathryn Smith of the Howard Florey Institute for performing amino acid analysis. REFERENCES 1. Steinetz, B.G., O’Byrne, E. & Kroc, R.L. (1980) The role of relaxin in cervical softening in mammals. In Dilation of the Uterine Cervix (Naftolin, F.S.P., ed), pp. 157–177. Raven Press, New York, USA. 2. Kakouris, H., Eddie, L.W. & Summers, R.J. (1992) Cardiac effects ofrelaxininrats.Lancet 339, 1076–1078. 3. Weisinger, R.S., Burns, P., Eddie, L.W. & Wintour, E.M. (1993) Relaxin alters the plasma osmolality-arginine vasopressin relationship in the rat. J. Endocrinol. 137, 505–510. 4. Novak, J., Ramierez, R.J., Gandley, R.E., Sherwood, O.D. & Conrad, K.P. (2002) Myogenic activity is reduced in small renal arteries isolated from relaxin-treated rats. Am. J. Physiol. Reg- ulatory Integrative Comp. Physiol. 283, R349–R355. 5. Dschietzig, T., Richter, C., Bartsch, C., Laule, M., Armbruster, F.P., Baumann, G. & Stangl, K. (2001) The pregnancy hormone relaxin is a player in human heart failure. FASEB J. 15, 2187– 2195. 6. Garber, S.L., Mirochnik, Y., Brecklin, C.S., Unemori, E.N., Singh, A.K., Slobodskoy, L., Grove, B.H., Arruda, J.A. & Dunea, Fig. 5. Competition for 33 P-labeled H2(B33) binding by INSL3 (n ¼ 4), analogues D (INSL3-D; n ¼ 4) and G (INSL3-G; n ¼ 4) in comparison with H2(B29) relaxin (n ¼ 11). 6292 A. A. Claasz et al.(Eur. J. Biochem. 269) Ó FEBS 2002 G. (2001) Relaxin decreases renal interstitial fibrosis and slows progression of renal disease. Kidney Int. 59, 876–882. 7. Bullesbach, E.E. & Schwabe, C. (1995) Structural contribution of the A-chain loop in relaxin. Int. J. Pept. Protein Res. 46, 238–243. 8. Schwabe, C. & Bullesbach, E.E. (1994) Relaxin: structures, func- tions, promises, and nonevolution. Faseb J. 8, 1152–1160. 9. Bullesbach, E.E. & Schwabe, C. (1988) On the receptor binding site of relaxins. Int. J. Pept. Protein Res. 32, 361–367. 10. Bullesbach, E.E., Yang, S. & Schwabe, C. (1992) The receptor- binding site of human relaxin II. A dual prong-binding mechan- ism. J. Biol. Chem. 267, 22957–22960. 11. Bullesbach, E.E. & Schwabe, C. (2000) The relaxin receptor- binding site geometry suggests a novel gripping mode of interac- tion. J. Biol. Chem. 275, 35276–35280. 12. Georges, D. & Schwabe, C. (1999) Porcine relaxin, a 500 million- year-old hormone? the tunicate Ciona intestinalis has porcine relaxin. Faseb J. 13, 1269–1275. 13. Adham, I.M., Emmen, J.M. & Engel, W. (2000) The role of the testicular factor INSL3 in establishing the gonadal position. Mol. Cell Endocrinol. 160, 11–16. 14. Bullesbach,E.E.&Schwabe,C.(1995)AnovelLeydigcellcDNA- derived protein is a relaxin-like factor. J. Biol. Chem. 270, 16011– 16015. 15. Dawson, N.F., Tan, Y.Y., Macris, M., Otvos, L., Summers, R.J., Tregear, G.W. & Wade, J.D. (1999) Solid-phase synthesis of ovine Leydig cell insulin-like peptide – a putative ovine relaxin? J. Pept. Res. 53, 542–547. 16. Nagata, K., Hatanaka, H., Kohda, D., Kataoka, H., Nagasawa, H., Isogai, A., Ishizaki, H., Suzuki, A. & Inagaki, F. (1995) Three- dimensional solution structure of bombyxin-II an insulin-like peptide of the silkmoth Bombyx mori: structural comparison with insulin and relaxin. J. Mol. Biol. 253, 749–758. 17. Parsell, D.A., Mak, J.Y., Amento, E.P. & Unemori, E.N. (1996) Relaxin binds to and elicits a response from cells of the human monocytic cell line, THP-1. J. Biol. Chem. 271, 27936–27941. 18. Tsuchiya, S., Yamabe, M., Yamaguchi, Y., Kobayashi, Y., Konno, T. & Tada, K. (1980) Establishment and characterization of a human acute monocytic leukemia cell line (THP-1). Int. J. Cancer 26, 171–176. 19. Greenfield, N. & Fasman, G.D. (1969) Computed circular dichroism spectra for the evaluation of protein conformation. Biochemistry 8, 4108–4116. 20. Tan, Y.Y., Wade, J.D., Tregear, G.W. & Summers, R.J. (1999) Quantitative autoradiographic studies of relaxin binding in rat atria, uterus and cerebral cortex: characterization and effects of oestrogen treatment. Br.J.Pharmacol.127, 91–98. 21. Zhou, L.M., Szendrei, G.I., Fossom, L.H., Maccecchini, M.L., Skolnick, P. & Otvos, L. Jr (1996) Synthetic analogues of con- antokin-G: NMDA antagonists acting through a novel poly- amine-coupled site. J. Neurochem. 66, 620–628. 22. Smith, J.A. & Pease, L.G. (1980) Reverse turns in peptides and proteins. CRC Crit. Rev. Biochem. 8, 315–399. 23. Skjaerbaek, N., Nielsen, K.J., Lewis, R.J., Alewood, P. & Craik, D.J. (1997) Determination of the solution structures of con- antokin-G and conantokin-T by CD and NMR spectroscopy. J. Biol. Chem. 272, 2291–2299. 24. Sherwood, O.D. (1994) Relaxin. In The Physiology of Reproduc- tion (Knobil, E. & Neill, J.D., eds), pp. 861–1009. Raven Press Ltd, New York, USA. 25. Bathgate,R.A.D.,Siebel,A.,Tovote,P.,Claasz,A.A.,Macris, M., Tregear, G.W. & Parry, L.J. (2002) Purification and char- acterization of relaxin from the tammar wallaby (Macropus eugenii): Bioactivity and expression in the corpus luteum. Biol. Reprod. 67, 293–300. 26. Smith,K.J.,Wade,J.D.,Claasz,A.A.,Otvos,L.Jr,Temelcos,C., Kubota, Y., Hutson, J.M., Tregear, G.W. & Bathgate, R.A. (2001) Chemical synthesis and biological activity of rat INSL3. J. Pept. Sci. 7, 495–501. 27. Hsu, S.Y., Nakabayashi, K., Nishi, S., Kumagai, J., Kudo, M., Sherwood, O.D. & Hsueh, A.J. (2002) Activation of orphan re- ceptors by the hormone relaxin. Science 295, 671–674. 28. Bullesbach, E.E. & Schwabe, C. (1994) Functional importance of the A-chain loop in relaxin and insulin. J. Biol. Chem. 269, 13124– 13128. Ó FEBS 2002 Relaxin activity of B chain analogues of ovine INSL3 (Eur. J. Biochem. 269) 6293 . Relaxin-like bioactivity of ovine Insulin 3 (INSL3) analogues Antonia A. Claasz 1 , Courtney P. Bond 2 , Ross A. Bathgate 1 , Laszlo Otvos Jr 3 , Nicola F. Dawson 1 , Roger J. Summers 2 , Geoffrey. data analyzed by one-way ANOVA followed by Newman–Keuls multiple comparison test. Ó FEBS 2002 Relaxin activity of B chain analogues of ovine INSL3 (Eur. J. Biochem. 269) 6291 bioassay, and furthermore. display some degree of bioactivity the activity of the previous analogues was not improved upon. Neither Analogue G (Fig. 4E) nor free A chain (not shown) produced any response. Interestingly, Analogue