Article pubs.acs.org/jmc
Minimum Active Structure of Insulin-like Peptide 5 Alessia Belgi,†,‡ Ross A. D. Bathgate,†,‡,§,▽ Martina Kocan,⊥,▽ Nitin Patil,†,∥ Suode Zhang,† Geoffrey W. Tregear,†,‡ John D. Wade,*,†,§,∥ and Mohammed Akhter Hossain*,†,§,∥ †
Florey Institute of Neuroscience and Mental Health, ‡Department of Biochemistry and Molecular Biology, §Florey Department of Neuroscience and Mental Health, and ∥School of Chemistry, The University of Melbourne, Victoria 3010, Australia ⊥ Monash Institute of Pharmaceutical Sciences, Monash University, 399 Royal Parade, Parkville, Victoria, 3052, Australia S Supporting Information *
ABSTRACT: Insulin-like peptide 5 (INSL5) is a complex two-chain peptide hormone constrained by three disulfide bonds in a pattern identical to insulin. High expression of INSL5 in the colon suggests roles in activation of colon motility and appetite control. A more recent study indicates it may have significant roles in the regulation of insulin secretion and β-cell homeostasis. This peptide thus has considerable potential for the treatment of eating disorders, obesity, and/or diabetes. However, the synthesis of INSL5 is extremely challenging either by chemical or recombinant means. The A-chain is very poorly soluble and the B-chain is highly aggregating in nature which, together, makes their postsynthesis handling and purification very difficult. Given these difficulties, we have developed a highly active INSL5 analogue that has a much simpler structure with two disulfide bonds and is thus easier to assemble compared to native INSL5. This minimized peptide represents an attractive new mimetic for investigating the functional role of INSL5.
■
INTRODUCTION
between interacting partners and therefore strongly indicates INSL5 is the native ligand for RXFP4.4 Following the identification of the receptor for INSL5, several studies were undertaken in an attempt to elucidate the physiological function(s) of the INSL5/RXFP4 pair. It has been proposed that INSL5 is involved in appetite stimulation and activation of colon motility.9 In addition, the results of more recent studies suggest that INSL5 is also involved in the regulation of insulin secretion and β-cell homeostasis.10 More specifically, morphometric and immunohistological analyses revealed that the INSL5 knockout mice had markedly reduced average islets area and β-cell numbers that caused reduced insulin production leading to increased blood glucose levels. INSL5 agonists therefore have the potential to be therapeutics for the management of eating disorders and/or diabetes. Conversely, an INSL5 antagonist may modulate appetite and feeding behavior leading to reduced weight gain in an environment with access to high caloric and palatable food. An understanding of the interaction between a peptide and its receptor(s) is crucial for the potential development of effective therapeutics. Chemical peptide synthesis has long played a major role in such structure−activity and drug development studies. We recently undertook the preparation of human INSL5 (hINSL5; Table 1) via an established sequential disulfide bond formation protocol.8 The synthesis was
Insulin-like peptide 5 (INSL5) is a peptide hormone that belongs to the insulin superfamily of peptides that includes insulin, insulin-like growth factors I and II (IGFI, IGFII), relaxins 1−3, and insulin-like peptides 3−6 (INSL3−6).1 Like insulin, it has a complex structure with two chains (A and B) and three disulfide bonds. It was first identified in 1999 while screening Expressed Sequence Tags (EST) databases searching for the Cys motif that is conserved within the B-chain of the peptides in the insulin superfamily.2 This peptide showed 40% homology with relaxin, 34% with INSL3, 30% with insulin, 29% with INSL6 (also identified in 1999 by the same research group3) and IGFII, 28% with IGFI, and 24% with INSL4.2 INSL5 is predicted to consist of a 21 residue A-chain and 24 residue B-chain held together by 3 disulfide bonds in a manner similar to that of insulin and other members of the family with the exception of IGFI and -II. In 2005, Liu et al. identified INSL5 as a ligand for the G-protein-coupled receptor GPCR1424 which is now known as relaxin family peptide receptor 4 (RXFP4).5 A related peptide, relaxin-3, also binds to RXFP4 although GPCR135 (RXFP3) is its native receptor.4,6 Both recombinant7 and synthetic INSL58 have been demonstrated to have high affinity and potency at RXFP4. Additionally, INSL5 and RXFP4 are expressed in an overlapping pattern in the colon in contrast to relaxin-3 which is expressed in the brain together with its receptor RXFP3. This suggests a coevolution of INSL5 and RXFP4 which is typical © 2013 American Chemical Society
Received: June 20, 2013 Published: November 4, 2013 9509
dx.doi.org/10.1021/jm400924p | J. Med. Chem. 2013, 56, 9509−9516
Journal of Medicinal Chemistry
Article
Table 1. Receptor Binding Affinity (pIC50) and Activation (pEC50) on the RXFP4 Receptora
a
pIC50 is the negative logarithm of the concentration of peptide needed to displace 50% of the labeled peptide in one-site competition binding assays; pEC50 is the concentration of peptide that produces a response which is 50% of its maximal effect, expressed as negative logarithm. Emax (%): efficacy maximum of the peptide in the cAMP assay relative to forskolin as 100%. NA = not active in the concentration range tested on receptor RXFP4, ND = not determined; *p < 0.0001 vs mouse INSL5; †p < 0.05 vs human INSL5; **p > 0.05 vs mouse INSL5; §p > 0.05 vs human INSL5. Z = pyroglutamic acid.
unexpectedly challenging, as both chains (A and B) were resistant to standard synthesis protocols and required highly optimized conditions. In particular, the use of a strong tertiary amine base (DBU) to deprotect Fmoc was required for successful assembly of the B-chain, and a pseudoproline was required at the C-terminus for successful assembly of the Achain. In addition, the A-chain was very poorly soluble and the B-chain was highly aggregating in solution which resulted in a very low yield of target peptide as previously described.8 For this reason, attention was diverted to the mouse insulin-like peptide, mINSL5 (Table 1),11 which, because of fewer hydrophobic residues, was predicted to be easier to assemble compared with hINSL5 and which also can be readily tested in vivo in mice.11 However, it was also found to be exceptionally difficult to prepare and required optimization of the synthetic protocol including the use of stronger deprotection and coupling reagents as well as a polycationic solubilizing tag.11
Given these difficulties, it would be highly desirable if a simplified INSL5 analogue consisting of either a single or two shorter chains could be developed and which retained in vitro and in vivo biological activity. INSL5 is closely related to relaxin-3 as they both belong to the same branch in the phylogenetic tree of the relaxin family. In addition, the native receptors of INSL5 (RXFP4) and relaxin-3 (RXFP3) belong to the same class of G-proteincoupled receptors.5,12 On the basis of such a close relationship between ligands (relaxin-3 and INSL5) and receptors (RXFP3 and RXFP4), it is most likely that INSL5 interacts with RXFP4 in a similar way to relaxin-3 interacting with RXFP313 where the latter interaction has been extensively characterized.14 Thus, we have designed and synthesized all INSL5 analogues (single chain or modified two chains) in this study based upon our recent findings on the related peptide, human relaxin-3 (H3 relaxin, Table 1).14 The modifications have been primarily 9510
dx.doi.org/10.1021/jm400924p | J. Med. Chem. 2013, 56, 9509−9516
Journal of Medicinal Chemistry
Article
It is recognized that ERK is a cellular pathway known to be activated through GPCRs. However, there was no report that RXFP4 can signal through ERK. Here we first tested our native hINSL5 on human RXFP4 receptor in the ERK 1/2 phosphorylation assay18 and showed, for the first time, that RXFP4 is able to signal through ERK signaling pathway (Figure 1). Then we tested all four mINSL5 single chain analogues (1−
applied to mINSL5 peptide, as it was shown to be more active on RXFP4 compared to hINSL5.8,11 Here we have shown that the A-chain of INSL5 can be truncated significantly without altering its biological activity and that significant activity is retained following further simplification to two chains bearing two, instead of three, disulfide bonds in the native peptide.
■
RESULTS
The amino acids at the C-termini of both A- and B-chains of native INSL5 contain free carboxyl groups (COOH). Recently we chemically prepared INSL5 analogues with amidated (CONH2) C-termini.11 The bioassay results showed that INSL5 analogues having amidated C-termini were less active compared to native INSL5 with free C-termini. Therefore, all the analogues in this study were made with free carboxylic acid. As we have discussed previously, the close relationship between both the peptides H3 relaxin and INSL5 and their cognate receptors RXFP3 and RXFP4 leads to the hypothesis that INSL5 interacts with RXFP4 in a fashion similar to the better characterized relaxin-3/RXFP3.5,12,13 The B-chain of H3 relaxin alone is an agonist of both RXFP3 and RXFP4, although with greatly reduced affinity and potency compared to native H3 relaxin.6,15 Moreover, the residues identified to date as being involved in binding and activation of the receptors are all contained within the B-chain of H3 relaxin.16 Thus, it is logical to determine whether the B-chain alone of INSL5 interacts with RXFP4. We therefore synthesized an analogue of the B-chain of mINSL5, 1 (Table 1), where the two interchain cysteine residues were replaced with alanine. Additionally, we have recently developed a single chain RXFP3 antagonist, R3B1-22R14a (Table 1). In this current study, the corresponding truncation was applied to mINSL5 B-chain, and analogue 2 (Table 1) was designed and synthesized to determine if, for INSL5, a B-chain truncated analogue can also behave as an antagonist on RXFP4. Competition binding and cAMP activity assays (details are in Supporting Information) were performed for analogues 1 and 2. The peptides showed neither affinity nor activity for the RXFP4 receptor (Table 1). Because of their inactivity, attention was then directed toward the A-chain alone. Previous studies on H3 relaxin showed that the A-chain is important primarily for maintaining the overall structure of H3 relaxin.17 Those earlier studies also showed that the A-chain can be substituted with the A-chain from other members of the relaxin family with no dramatic effect on H3 relaxin interaction with both RXFP3 and RXFP4.15 As INSL5 was assumed to interact with RXFP4 in a fashion similar to that of H3 relaxin, it is unlikely that the Achain will bind to RXFP4. Nevertheless, there is no evidence or prior data showing that, in the case of INSL5, the A-chain is also not involved in either binding or activation or both. Moreover, an extensive structure−function study of the A-chain of relaxin-like peptides has highlighted that the role of A-chain is peptide- and receptor-dependent. 17 Therefore, as a preliminary screening of the possible activity of INSL5 Achain alone, two mINSL5 A-chain analogues, 3 and 4 (Table 1), were designed, synthesized, and tested. The side chain of cysteine residues (SH group) were capped either with tBu, Acm, or acetamide, which are very small in size (Table 1 and in Supporting Information), in analogues 3 and 4. The analogue 4 was a cyclic form of analogue 3. The A-chain analogues (3 and 4) showed no affinity/activity on CHO-K1/RXFP4 cells (Table 1 and in Supporting Information).
Figure 1. ERK1/2 phosphorylation assay on CHO-K1/RXFP4 cells of analogues 1−4, hINSL5 acid, and mINSL5 acid. The data are the result of n = 3−6 independent experiments and are expressed as the mean ± SEM.
4) together with native mINSL5. We observed that native mINSL5 is active in the ERK 1/2 phosphorylation assay, and in fact it is more potent compared with hINSL5. Such difference in the activity is consistent with our previous observation where we showed that mINSL5 caused greater forskolin-stimulated cAMP inhibition compared to hINSL5.11 We also confirmed that, like in the cAMP assay, all single chain mINSL5 analogues (1−4) are inactive in this ERK phosphorylation assay (Figure 1). As neither the A- nor B-chain alone is reasonably active on RXFP4, both chains of INSL5 are probably essential for activity. Therefore, mINSL5 analogue 5 (Table 1) was prepared in which the two chains remain intact without any mutation or truncation. However, the peptide was simplified by removing the intra-A-chain disulfide bridge. A similar strategy was recently used to successfully produce a high affinity RXFP3 agonist.14b This analogue, with two disulfide bridges, is simpler and thus considerably easier to assemble than the native peptide which has three disulfide bridges (Table 1). The minimized analogue 5 was then tested in competition binding assay on RXFP4-expressing CHO-K1 cells, and it showed strong affinity for the receptor with pIC50 values of 7.32 ± 0.10 (Figure 2A, Table 1). It showed strong ability to activate the RXFP4 receptor measured via its ability to inhibit forskolinstimulated cAMP production in CHO-K1/RXFP4 cells. Consistent with the affinity data, its potency is also very strong with pEC50 8.38 ± 0.20. (Figure 2B, Table 1). One additional analogue (mouse analogue 6, Table 1) was synthesized. It had a truncation at the N-terminus of the Achain up to Cys8, which is involved in the interchain disulfide bridge. Analogue 6 showed strong affinity for RXFP4 receptor with pIC50 values of 7.06 ± 0.10 (Figure 2A, Table 1). Consistent with the affinity value, its potency was also strong with pEC50 8.04 ± 0.11 (Figure 2B, Table 1). Interestingly, the most minimized analogue 6 was still as active as analogue 5 as the dose−response curves for mINSL5 analogues 5 and 6 almost overlap (Figure 2), and there is no significant difference in their pEC50 values (Table 1). 9511
dx.doi.org/10.1021/jm400924p | J. Med. Chem. 2013, 56, 9509−9516
Journal of Medicinal Chemistry
Article
Figure 2. In vitro RXFP4 receptor bioassays: (A) Competition binding curve of analogues 5 and 6 and mINSL5 acid competing with 5 nM of Eu-(A) mINSL5; (B) cAMP inhibition assay of analogue 5, analogue 6, and mINSL5 acid. The data are the result of n = 3 independent experiments and are expressed as the mean ± SEM.
Figure 3. In vitro RXFP4 receptor bioassays: (A) Competition binding curve of analogue 7 and hINSL5 acid competing with 5 nM of Eu-(A) mINSL5; (B) cAMP inhibition assay of analogue 7 and hINSL5 acid. The data are the result of n = 3 independent experiments and are expressed as the mean ± SEM.
Analogue 7 (Table 1), the human equivalent of the most minimized analogue 6, was finally designed and synthesized. It showed strong affinity for the RXFP4 receptor with a pIC50 value of 6.28 ± 0.16 (Figure 3A, Table 1). Consistent with the affinity value, its potency was also strong, with a pEC50 of 7.40 ± 0.19 (Figure 3B, Table 1). Importantly this peptide showed a similar affinity and efficacy shift in comparison to native hINSL5 as the mINSL5 equivalent did to native mINSL5 (Figure 2, Table 1). To further verify whether the minimized analogues follow the same signaling pathways as native INSL5, we compared mouse INSL5 and its two minimized analogues 5 and 6 in the ERK1/2 phosphorylation assay (Figure 4). The results showed that, like in the cAMP assay (Figure 2), these minimized analogues are very potent in the ERK assay. However, these analogues are less potent as compared to native mINSL5, which correlates with the cAMP results. Importantly, mouse analogue 5 activates ERK (Figure 4) with efficacy identical to that of native hINSL5 (Figure 1), and this efficacy is about 30% higher than for analogue 6 (Figure 4). To get some structural insights, the circular dichroism (CD) spectra of synthetic mINSL5 together with two minimized analogues 5 and 6 were recorded (Figure 5). The studies revealed that all three peptides possess a varied degree of αhelical conformation (with a pronounced double minimum at around 208 and 222 nm). The α-helical content of mINSL5 calculated from the mean residual ellipticity (MRE) at 222 nm, [θ]222, was found to be 38% ([θ]222 = −13 613), which is consistent with the recently reported result on native
Figure 4. ERK1/2 phosphorylation assay on CHO-K1/RXFP4 cells of analogues 5 and 6 with mINSL5 acid. The data are the result of n = 3 independent experiments and are expressed as the mean ± SEM.
Figure 5. CD spectrum of synthetic mINSL5 analogues 5 and 6 compared with synthetic native mINSL5. CD was performed in 20 mM PBS buffer at pH 7.4. 9512
dx.doi.org/10.1021/jm400924p | J. Med. Chem. 2013, 56, 9509−9516
Journal of Medicinal Chemistry
Article
mINSL5.11 The α-helical content of analogue 5, calculated in the same way ([θ]222 = −6031, α-helix content
