Conformation of the Core Sequence in Melanocortin Peptides Directs Selectivity for the Melanocortin MC3 and MC4 Receptors*

Melanocortin peptides regulate a variety of physiological processes. Five melanocortin receptors (MC-R) have been cloned and the MC3R and MC4R are the main brain MC receptors. The aim of this study was to identify structural requirements in both ligand and receptor that determine γ-melanocyte-stimulating hormone (MSH) selectivity for the MC3R versus the MC4R. Substitution of Asp10 in [Nle4]Lys-γ2-MSH for Gly10 from [Nle4]α-MSH, increased both activity and affinity for the MC4R while the MC3R remained unaffected. Analysis of chimeric MC3R/MC4Rs and mutant MC4Rs showed that Tyr268 of the MC4R mainly determined the low affinity for [Nle4]Lys-γ2-MSH. The data demonstrate that Asp10 determines selectivity for the MC3R, however, not through direct side chain interactions, but probably by influencing how the melanocortin core sequence is presented to the receptor-binding pocket. This is supported by mutagenesis of Tyr268 to Ile in the MC4R which increased affinity and activity for [Nle4]Lys-γ2-MSH, but decreased affinity for two peptides with constrained cyclic structure of the melanocortin core sequence, MT-II and [d-Tyr4]MT-II, that also displayed lower affinity for the MC3R. This study provides a general concept for peptide receptor selectivity, in which the major determinant for a selective receptor interaction is the conformational presentation of the core sequence in related peptides to the receptor-binding pocket.

The development of selective ligands for the MC receptors has been hampered by the absence of detailed knowledge about the structural requirements of peptide ligands for selective MC receptor binding and activation. Nevertheless, it has been demonstrated that HFRW (MSH (6 -9)) forms the core sequence of melanocortins, which is necessary to bind to all MC receptors (21)(22)(23). Ligand selectivity may therefore be determined by residues outside the core region either through a selective interaction with different receptor subtypes, by altering folding of the core sequence, or by a combination of both.
Although the MC3R and MC4R both recognize ␣-MSH, the affinity of ␥-MSH is 50-fold higher for the MC3R. Of the three forms of ␥-MSH, ␥ 1 -and ␥ 2 -MSH (11 and 12 amino acid residues, respectively) are most related, while ␥ 3 -MSH has an extended C terminus. In vivo amidation of the C-terminal Gly 12 residue of ␥ 2 -MSH results in the formation of ␥ 1 -MSH with an C-terminal Phe 11 -amide. In mammals, the natural forms of ␥-MSH contain an additional N-terminal Lys residue (3). ␣-MSH and Lys-␥ 2 -MSH both contain the core sequence, HFRW, a Tyr residue at position 2 and a Met residue at position 4, while N-and C-terminal residues and the residue at position 5 differ (Fig. 1). Recently it was shown that Asp 10 in Lys-␥ 2 -MSH determined MC3R selective activation (24). However, it is not clear whether replacement of Asp 10 in Lys-␥ 2 -MSH increases binding affinity or only increases efficacy for the MC4R. Moreover, it remains to be determined whether there exists a direct selective interaction of Asp 10 with the MC3R, or whether Asp 10 induces a peptide structure that is favorable for the MC3R. To solve this problem a more detailed analysis is required regarding the contribution of each individual amino acid in the ligand to receptor binding and activation. Therefore, the aim of this study was to gain insight into the molecular mechanism of the selectivity of ␥ 2 -MSH for the MC3R versus the MC4R. Using a gain of function approach, we tested [Nle 4 ]␣-MSH and [Nle 4 ]Lys-␥ 2 -MSH and derivatives with exchanged amino acid residues on in vitro binding and activation of wild type and chimeric MC3R/MC4R and mutant MC4Rs. We demonstrate here that the selectivity of ␥ 2 -MSH for the MC3R versus the MC4R is determined by a single amino acid residue in the ligand and, to a large extend, a single residue in the receptor. A new concept is proposed in which MC3R/MC4R selectivity is determined by how the melanocor-tin core sequence is presented to the receptor-binding pocket.  4 ,Gly 10 ,Pro 12 ]Lys-␥ 2 -MSH were synthesized using solid phase N-(9-fluorenyl)methoxycarbonyl chemistry and purified as described in Schaaper et al. (25). The products were analyzed using liquid chromatography mass spectrometry. Ion spray mass spectrometry performed on a Micromass Quattro sq confirmed the expected molecular weights. All other peptides mentioned in Table II were synthesized using pepscan and their concentration and purity were verified by high pressure liquid chromatography (26). All peptides in Table II were N-terminal acetylated and C-terminal amidated. Peptides were dissolved in 1 mM hydrogen chloride and diluted in phosphate-buffered saline or binding buffer (see below).

Peptides
Radioiodination of NDP-␣-MSH-Iodination was performed as described by Tatro et al. (27). In short, 4 g of NDP-␣-MSH was mixed with 1.2 IU of bovine lactoperoxidase (Calbiochem) and 1 mCi of Na 125 I (ICN) in a final volume of 100 l of 0.05 M phosphate buffer (pH 6.5). Then 5 l of 0.003% H 2 O 2 was added every 60 s. After 4 min 50 l of 1 mM dithiothreitol was added to stop the reaction. Then, the sample was high pressure liquid chromatography purified with a Bondapak C18 column 3.9 ϫ 300 mm (Waters, Division of Millipore) by elution with a 22-52% acetonitrile gradient in 10 mM ammonium acetate (pH 5.5) in 40 min. Specific activity of the [ 125 I]NDP-␣-MSH was determined in the ␤-galactosidase activation assay and was approximately 0.4 pmol/ 100,000 cpm.
Construction of Chimeric and Mutant Receptors-Chimeric receptors were made with polymerase chain reaction using overlapping primers directed against the region coding for identical residues in the rat MC3R and human MC4R in TM1, the junction of TM3-IC2, and in TM5 (Fig. 2). Thus, within the junction sites of the chimeric receptor the amino acid sequence was not changed. Both chimeric and mutant receptors were made with a similar stategy. In the first polymerase chain reaction (Cloned Pfu DNA Polymerase, Stratagene) separate fragments were obtained with primer T7 combined with the 3Ј mutagenic/chimeric primer and with SP6 combined with the 5Ј mutagenic/ chimeric primer (complete overlap with the 3Ј primer). Fragments were isolated and purified from agarose gel (QIAEX Gel Extraction Kit, QIAGEN) and a second polymerase chain reaction (Taq DNA polymerase, Perkin-Elmer) was performed with the two purified bands as template for primers SP6 and T7. The polymerase chain reaction products were digested with BamHI and XhoI, separated through agarose gel electrophoresis, and purified. Chimera 3B, MC4(267-282 MC3), MC4(267-273 MC3), MC4(278 -282 MC3), and MC4R mutants, F267L/ Y268I (double mutant), F267L, Y268I, S270T, Q273T, M281T were cloned into pcDNA3.1. Chimera 3AB, 3C, 3D, and 4D were cloned into pcDNA1/Neo). The inserts of the clones were sequenced completely (T7 sequencing kit, Pharmacia Biotech) and only mutations introduced by the mutagenic/chimeric primers were found.
Cells and Transfection-HEK 293 cells were grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal calf serum. Optimal amounts of receptor cDNA used for transfection of cells for the activation assay were determined empirically. Approximately 7 ϫ 10 6 cells were transiently transfected with 15-200 ng of the rat MC3R, the human MC4R, chimeric or mutant receptor cDNA constructs, combined with 7 g of the pCRELacZ for the ␤-galactosidase activation assay using the calcium phosphate precipitation method. For the receptor binding assay, cells were transfected with 7 g of the receptor cDNA constructs.
␤-Galactosidase Activation Assay-The assay makes use of the ␤-galactosidase (lacZ) gene fused to five copies of the cyclic AMP response element (CRE) to detect the activation of CRE-binding protein resulting from increased intracellular cAMP and Ca 2ϩ (28). Twenty-four hours after transfection 293 HEK cells were distributed into 96-well plates (Primaria). The next day the cells were treated for 6 h with various concentrations of peptides in Dulbecco's modified Eagle's medium supplemented with 0.5% bovine serum albumin, 25 mM Hepes (pH 7.4), and 50 g/ml (150 KIU/ml) aprotinin (Sigma). After treatment the cells were lysed, frozen, thawed, and assayed for ␤-galactosidase activity. For each peptide 12 data points were measured in quadruplicate. EC 50 values were then calculated with a 95% confidence interval using GraphPad Prism software (sigmoidal dose-response curve fitting, variable slope). Experiments were repeated at least twice with the same results.
Adenylate Cyclase Assay-293 HEK cells stably expressing human MC4R, rat MC3R, mutant MC4(Y268I), and chimera MC4(267-282 MC3) were grown in poly-L-lysine-coated (Sigma) 24-well Costar plates. Agonist stimulated adenylate cyclase activity was measured as described by Salomon (29,30). In short, after prelabeling with 500 l of [ 3 H]adenine (NEN Life Science Products Inc.) in a concentration of 2 Ci/ml, the 293 cells were incubated for 20 min at 37°C in phosphatebuffered saline containing 0.1 mM isobutylmethylxanthine (IBMX), 1 M forskolin, and agonist in a concentration ranging from 10 Ϫ11 to 10 Ϫ5 M. The cells were harvested and [ 3 H]cAMP formation was determined. For each peptide 12 duplicate data points were measured. EC 50 values were then calculated with a 95% confidence interval using GraphPad Prism software (sigmoidal dose-response curve fitting, variable slope).
Receptor Binding Assay-Transfected HEK 293 cells were grown in poly-L-lysine-coated 24-well Costar plates. Two days after transfection, the cells were incubated with 100,000 cpm of [ 125 I]NDP-␣-MSH and various concentrations of peptides diluted in binding buffer consisting of Ham's F-10 medium (Life Technologies, Inc.) (pH 7.4) containing 2.5 mM calcium chloride, 0.25% bovine serum albumin, 10 mM Hepes, and 50 g/ml (150 KIU/ml) aprotinin. After incubation for 30 min at room temperature, the cells were washed twice with ice-cold Tris-buffered saline containing 2.5 mM calcium chloride and lysed in 1 M sodium hydroxide. Radioactivity of the lysates was counted in a Packard Cobra ␥-counter. Competition curves were fitted from 11 duplicate data points with GraphPad Prism software, nonlinear regression, one site competition. K i values were calculated using the Cheng and Prusoff equation with 95% confidence interval. Experiments were repeated at least twice with the same results.

RESULTS
Characterization of the Reference Peptides-In order to delineate ␣/␥-MSH selectivity we first excluded influence of peptide length, N-and C-terminal modifications, and oxidation of the Met residue, and therefore used synthetic [Nle 4 ]␣-MSH and [Nle 4 ]Ac-Lys-␥ 2 -MSH-NH 2 as reference peptides (the latter will be referred to as [Nle 4 ]Lys-␥ 2 -MSH). Table I Table II). The only exception was [Nle 4 ,Phe 12 ]␣-MSH, which exhibited a 2-fold decrease in activity for the MC4R and a 2-fold increase in affinity for the MC3R. However, when Gly 10 or Lys 11 of [Nle 4 ]␣-MSH were substituted for Asp 10 or Arg 11 a clear loss of affinity and activity was observed for both the MC3R and MC4R. When the Glu 5 to Gly 5 substitution was combined with the Gly 10 to Asp 10 substitution in [Nle 4 ]␣-MSH, the affinity and activity for the MC4R decreased more than 3-fold, whereas the affinity for the MC3R remained the same as for [Nle 4 ]␣-MSH.
Next, each amino acid residue in [Nle 4 ]Lys-␥ 2 -MSH was substituted for corresponding residues of [Nle 4 ]␣-MSH (lower panel of Table II). When Lys 1 was substituted for Ser 1 a slight decrease in affinity was observed for both MC3R and MC4R but the activity remained unaffected. Substitution of Arg 11 for Lys 11 slightly increased activity for both receptors. Substitution of Val 3 for Ser 3 or Gly 13 for Val 13 decreased affinity for only the MC4R, while the activity remained unaffected. With respect to binding affinity, [Val 13 ]Lys-␥ 2 -MSH displayed the largest difference between MC3 and MC4 as did [Ser 3 ]Lys-␥ 2 -MSH in activation. Interestingly, when Asp 10 was substituted for Gly 10 , a 5-fold increase in both affinity and activity for the MC4R was observed, while the MC3R was unaffected.
[Nle 4 ,Gly 10 ]Lys-␥ 2 -MSH had an even higher affinity for the MC4R than [Nle 4 ]␣-MSH. The Phe 12 for Pro 12 substitution also gave an increase in affinity and activity of more than 3-fold for the MC4R, but not for the MC3R. Strikingly, the Asp 10 to Gly 10 substitution combined with the Phe 12 to Pro 12 substitution further increased the affinity to almost 13-fold for the MC4R, but there was no additive effect of these two substitutions on MC4R activation.

Receptor Domains Involved in [Nle 4 ]Lys-␥ 2 -MSH Selective
Binding-The MC3R and the MC4R share 58% overall amino acid identity and 76% similarity. The transmembrane regions (TM) show the highest degree of homology while the intra-and extracellular loops (IC and EC) have lower homology (Fig. 2). To identify regions of the MC3R responsible for [Nle 4 ]Lys-␥ 2 -MSH selectivity a series of MC3R and MC4R chimeric receptors were generated with boundaries in stretches of amino acid residues with complete homology in order to minimize effects on receptor folding. These chimeric receptors were designed to determine the contribution of each of the extracellular domains to ligand recognition. Thus, the first extracellular loop with or without the N-terminal domain (named 3B and 3AB, respectively), the second (3C) and third (3D) extracellular loop of the MC4R were swapped individually with the corresponding domain of the MC3R. These chimeric receptors were transfected into 293 HEK cells and the affinities of NDP-␣-MSH, [Nle 4 ]␣-MSH, [Nle 4 ]Lys-␥ 2 -MSH, and [Nle 4 ,Gly 5 ,Asp 10 ]␣-MSH were determined (the latter two peptides displayed MC3R selectivity). Fig. 3 summarizes ligand affinity for the MC3R, MC4R, and five chimeric receptors. All chimera bound [ 125 I]NDP-␣-MSH, demonstrating that they were all expressed on the plasma membrane. In general, the affinities of MSH peptides for MC3R and chimera 3AB, 3B, and 4D were higher than for MC4R and chimeras 3C and 3D. For example, the affinity of NDP-␣-MSH on these chimera was 6 -20-fold higher than for the MC4R, but the same as for the MC3R. Gray circles indicate residues identical in the rat MC3R and the human MC4R.
Black circles indicate regions with complete homology that were used as boundaries for construction of the chimeric receptors. These were named after the extracellular domain of the MC3R that was placed into the MC4R. Thus, chimera 3AB is the MC4R containing the N terminus through TM3 (third transmembrane domain) of the MC3R, chimera 3B is the MC4R containing the IC1 (first intracellular loop) through TM3 of the MC3R, chimera 3C is the MC4R containing IC2, TM4, EC2 (second extracellular loop) and part of TM5 of the MC3R, chimera 3D is the MC4R containing part of TM5 through the C terminus of the MC3R and vice versa, chimera 4D is the MC3R containing part of TM5 through the C terminus of the MC4R. The radioligand receptor binding assay was used to further analyze the chimeric and mutant receptors described below. In order to determine in more detail which residues are important for selectivity we first substituted residues 267 to 282 of the MC4R (containing EC3), hereafter named MC4(267-282 MC3), for corresponding residues of the MC3R (Fig. 4). Fig. 5A (Fig. 5B). Therefore, F267L and the Y268I were also mutated separately. The affin-  Independent substitutions of Gly 10 and Lys 11 in [Nle 4 ]␣-MSH for Asp 10 and Arg 11 , respectively, led to a significant decrease in activity and affinity for both MC3R and MC4R. This indicates that introduction of these residues in [Nle 4 ]␣-MSH prohibited interaction with both receptors. In a previous study (32) the affinity of [Asp 10 ]␣-MSH also appeared to be much lower than that of ␣-MSH for MC3R and MC1R. Still, Gly 10 , and also Lys 11 , of [Nle 4 ]␣-MSH are probably not important for a direct interaction with the receptor, since Ala substitution on positions 10 and 11 of ␣-MSH did not affect affinity on the MC3R nor the MC1R (23).
Substitution of Pro 12 of [Nle 4 ]␣-MSH for Phe 12 slightly increased MC3R affinity while decreasing activity, but not affinity, for the MC4R. This is in agreement with the observation that substitution of Pro 12 for Ala in ␣-MSH did not affect binding to the MC3R (22). This implies that the residue at this position is not essential for receptor binding but may be important for MC4R activation only. Our data and others (33)    Interestingly, it was found that all chimera containing the 3B segment (TM2, EC1, and TM3) of the MC3R displayed a high affinity for MSH peptides, similar to MC3R. In this same region of the MC1R, Yang et al. (35) suggested an important role for residues Glu 94 in TM2 and Asp 117 or Asp 121 in TM3 of the MC1R in forming an ionic binding pocket for the Arg 8 residue of NDP-␣-MSH, ␣-MSH, ␥ 2 -MSH (Arg 7 ), and MT-II (35). Although these residues are conserved in all MC receptors, other, non-conserved residues may influence the strength of the receptor-ligand interaction, explaining the overall lower affinity of MSH peptides, used in this study, for the MC4R versus the MC3R.
A selective receptor-ligand interaction could occur through a mechanism of exclusion rather than a specific recognition. In that respect, Tyr 268 of the MC4R may hinder [Nle 4 ]Lys-␥ 2 -MSH interaction and the following model may apply. His 264 of the MC4R is the equivalent of the His 260 in the MC1R and may, as was suggested for the MC1R (36,37), also be important for interaction with ␥-MSH. His 264 has been demonstrated to be essential for melanocortin peptide activation of MC4R as well (38). His 264 of the MC4R is located only four residues lower in TM6 than Tyr 268 . Therefore, Tyr 268 may mask His 264 and thereby exclude an interaction with [Nle 4 ]Lys-␥ 2 -MSH (Fig. 7).
Using NDP-␣-MSH containing Asp 10 , Arg 11 , and Phe 12 of ␥-MSH, Schioth et al. (39) suggested that the ␥-MSH C terminus hinders MC4R binding. However, this peptide showed an equal loss of affinity for the MC3R. Nevertheless, this loss of affinity agrees with our data for [Nle 4 ,Asp 10 ]␣-MSH. This suggests that the presence of two acidic amino acid residues on positions 5 and 10 (Glu and Asp, respectively), lead to a significant decrease in MC3R and MC4R affinity, possibly because of repulsion of these two negatively charged residues. This is in agreement with the high affinity of cyclic lactam-(cyclization between residues 5 and 10) and disulfide-bridged (cyclization between residues 4 and 10) melanocortin derivatives (40 -44). Taken together, these data suggest that in the active conformation the residues in positions 4/5 and 10 are in close proximity. We here propose that the presence of Asp 10 in [Nle 4 ]Lys-␥ 2 -MSH decreased the chance of the core sequence to be in the optimal conformation necessary to bind to the MC4R.
To investigate whether presentation of a constraint core sequence determined MC4R selectivity, the affinity of cyclic melanocortin peptides with a structurally constrained core sequence was tested. MT-II displayed marked increased and [D-Tyr 4 ]MT-II equal affinity as compared with NDP-␣-MSH for the MC4R, but not for the MC3R and MC4(Y268I). These differences in ligand activity appeared to be represented by differences in affinity. Thus, mutation of Tyr 268 in MC4R toward the Ile residue present on the corresponding position in MC3R increased both affinity and activity of [Nle 4 ]Lys-␥ 2 -MSH and decreased affinity and activity for the high affinity MC4R agonist, [D-Tyr 4 ]MT-II. Thus, a single amino acid residue determined MC3/MC4R selectivity for different ligands and MC4R binding and activation increases when the core sequence is presented in a constraint conformation.
Recently, it was shown that cyclic peptides have higher affinity than ␣-MSH on all MC-Rs (45). Strikingly, for the MC4R, all cyclic peptides displayed higher or similar affinity than for MC3R (45)(46)(47), while linear peptides always seemed to exhibit lower affinity on the MC4R than on the MC3R (this study). Therefore, for design of new MC-R selective ligands, cyclization may be appropriate for the MC4R. In contrast, modification in linear MSH peptides may be more valuable for MC3R selectivity, as was suggested by Haskell-Luevano et al. (41). A similar model was proposed for the opioid receptors in which linear dynorphin A analogues were generally more selective for the -opioid receptor while cyclic constrained dynorphin A peptides demonstrated slight selectivity forversus ␦-opioid receptors or were nonselective (48). Taken together, these data suggests that the message sequence residues do not interact with conserved neuropeptide receptor residues in the same manner. Indeed, it has been shown that the HFRW pharmacophore interacts differently with all MC-R (49).
Here, we propose a general concept for selective receptorligand interaction that may apply to all peptide receptors. Ligand residues outside the peptide core sequence direct the conformation of the receptor interacting core sequence presented to the receptor-binding pocket, and thereby determine selectivity. There are several examples that emphasize the critical role of residues positioned outside the core of true contact residues, in determining selectivity of ligands to, for instance, opioid (36) and neuropeptide Y receptors (50).
This study provides, for the first time, a detailed analysis of the structural requirements for selective MC3R versus MC4R recognition. Using a gain of function approach we have demonstrated that the Asp 10