NMR Insights into the Structure-Function Relationships in the Binding of Melanocortin Analogues to the MC1R Receptor

Linear and cyclic analogues of the α-melanocyte stimulating hormone (α-MSH) targeting the human melanocortin receptor 1 (MC1R) are of pharmacological interest for detecting and treating melanoma. The central sequence of α-MSH (His-Phe-Arg-Trp) has been identified as being essential for receptor binding. To deepen current knowledge on the molecular basis for α-MSH bioactivity, we aimed to understand the effect of cycle size on receptor binding. To that end, we synthesised two macrocyclic isomeric α-MSH analogues, c[NH-NO2-C6H3-CO-His-DPhe-Arg-Trp-Lys]-Lys-NH2 (CycN-K6) and c[NH-NO2-C6H3-CO-His-DPhe-Arg-Trp-Lys-Lys]-NH2 (CycN-K7). Their affinities to MC1R receptor were determined by competitive binding assays, and their structures were analysed by 1H and 13C NMR. These results were compared to those of the previously reported analogue c[S-NO2-C6H3-CO-His-DPhe-Arg-Trp-Cys]-Lys-NH2 (CycS-C6). The MC1R binding affinity of the 22-membered macrocyclic peptide CycN-K6 (IC50 = 155 ± 16 nM) is higher than that found for the 25-membered macrocyclic analogue CycN-K7 (IC50 = 495 ± 101 nM), which, in turn, is higher than that observed for the 19-membered cyclic analogue CycS-C6 (IC50 = 1770 ± 480 nM). NMR structural study indicated that macrocycle size leads to changes in the relative dispositions of the side chains, particularly in the packing of the Arg side chain relative to the aromatic rings. In contrast to the other analogues, the 22-membered cycle’s side chains are favorably positioned for receptor interaction.


Introduction
The melanocortin receptor 1 (MC1R) belongs to a five-member subfamily of G protein-coupled receptors (GPCR), the melanocortin receptors (MCRs) family that also includes MC2-, MC3-, MC4-and MC5-receptors [1][2][3][4][5]. The human gene coding the MC1R was first cloned in 1992 [1,6]. Since then, genetic analysis, phenotypic association and structure-function studies opened new perspectives for the assignment of its function as a key regulator of skin biological events. It is involved in the regulation of skin pigmentation, animal coat coloration, and melanocyte function [5,7]. Polymorphisms in the receptor gene have been attributed to the red hair phenotype, melanoma and non-melanoma skin cancer [7][8][9]. MC1R is expressed in melanocytes, melanoma cells, macrophages, and brain, as 1 Data for PG10N and for CycS-C6 were taken from [19] and [20], respectively.

Affinity of Peptides CycN-K6 and CycN-K7 to MCR1
The binding affinities of the peptides CycN-K6 and CycN-K7 to MC1R were tested in a competitive binding assay using [ 125 I]-[Nle 4 ,DPhe 7 ]-α-MSH ([ 125 I]-NDP-α-MSH) as radioligand and murine melanoma B16F1 cells ( Figure S1, Supplementary Materials), as previously described for other α-MSH analogues [20]. Table 1 lists the IC50 values obtained for those two peptides, and for the 19-membered macrocyclic peptides PG10N and CycS-C6 [19,20]. Strikingly, the 25-membered peptide CycN-K7 turned out to have less affinity for the receptor than the 22-membered peptide CycN-K6, which, however, exhibits more affinity than the 19-membered peptide CycS-C6. This indicates that the relationship between cycle size and receptor affinity is not straightforward. Therefore, a detailed structural comparison of these three peptides is required to understand the molecular basis for their binding to receptor MC1R.

NMR Characterization of CycN-K6 and CycN-K7 and Comparison to CycS-C6
Peptides CycN-K6 and CycN-K7 are structural isomers, which differ in the Lys side chain involved in cyclization ( Figure 1). Once their chemical shifts (δ) were assigned (see Section 4 and Tables S1 and S2, Supplementary Materials), the identity of the Lys residue involved in cyclization could be confirmed unambiguously by 1 H and 13 C δ values of the Lys Cεε'H2 groups (Tables S1 and  S2), and by the Nuclear Overhauser effects (NOEs) between the p-NO2-benzoic acid ring and the Lys side chain ( Figure 2 and Table S3, Supplementary Materials).
As previously found in peptide CycS-C6 [20], these two macrocyclic peptides show many δ values that differ strongly from the random coil δ values. For instance, the chemical shift deviations (Δδ = δobserved − δrandom coil, ppm) obtained for 1 Hα, 13 Cα and 13 Cβ atoms lie clearly outside the random coil range ( Figure 3 and Figure S2, Supplementary Materials), which shows that peptides CycN-K6 and CycN-K7 form ordered conformations in aqueous solution. Moreover, the sign of the observed deviations, which are negative in ΔδHα and ΔδCβ, and positive in ΔδCα for most residues (Figure 3 and Figure S2), is indicative of turn-like conformations. Concerning side chain protons, the up-field shifts exhibited by Arg (Table 2) indicate that these side chain protons experience anisotropic effects from   [19,20], respectively.

Affinity of Peptides CycN-K6 and CycN-K7 to MCR1
The binding affinities of the peptides CycN-K6 and CycN-K7 to MC1R were tested in a competitive binding assay using [ 125 I]-[Nle 4 ,DPhe 7 ]-α-MSH ([ 125 I]-NDP-α-MSH) as radioligand and murine melanoma B16F1 cells ( Figure S1, Supplementary Materials), as previously described for other α-MSH analogues [20]. Table 1 lists the IC 50 values obtained for those two peptides, and for the 19-membered macrocyclic peptides PG10N and CycS-C6 [19,20]. Strikingly, the 25-membered peptide CycN-K7 turned out to have less affinity for the receptor than the 22-membered peptide CycN-K6, which, however, exhibits more affinity than the 19-membered peptide CycS-C6. This indicates that the relationship between cycle size and receptor affinity is not straightforward. Therefore, a detailed structural comparison of these three peptides is required to understand the molecular basis for their binding to receptor MC1R.

NMR Characterization of CycN-K6 and CycN-K7 and Comparison to CycS-C6
Peptides CycN-K6 and CycN-K7 are structural isomers, which differ in the Lys side chain involved in cyclization ( Figure 1). Once their chemical shifts (δ) were assigned (see Section 4 and Tables S1 and S2, Supplementary Materials), the identity of the Lys residue involved in cyclization could be confirmed unambiguously by 1 H and 13 C δ values of the Lys C εε' H 2 groups (Tables S1 and S2), and by the Nuclear Overhauser effects (NOEs) between the p-NO 2 -benzoic acid ring and the Lys side chain ( Figure 2 and Table S3, Supplementary Materials).
As previously found in peptide CycS-C6 [20], these two macrocyclic peptides show many δ values that differ strongly from the random coil δ values. For instance, the chemical shift deviations (∆δ = δ observed − δ random coil , ppm) obtained for 1 H α , 13 C α and 13 C β atoms lie clearly outside the random coil range ( Figure 3 and Figure S2, Supplementary Materials), which shows that peptides CycN-K6 and CycN-K7 form ordered conformations in aqueous solution. Moreover, the sign of the observed deviations, which are negative in ∆δ Hα and ∆δ Cβ , and positive in ∆δ Cα for most residues (Figure 3 and Figure S2), is indicative of turn-like conformations. Concerning side chain protons, the up-field shifts exhibited by Arg (Table 2) indicate that these side chain protons experience anisotropic effects from the ring currents of one or more aromatic residues, as was also observed in peptide CycS-C6 [20]. Despite these common general characteristics, the detailed examination of NMR parameters highlights the existence of important differences between peptides CycN-K6 and CycN-K7, and also among these peptides and the previously reported CycS-C6 [20]. The differences are observed both at the cycle backbone and at certain side chains. The most remarkable differences at cycle backbone are observed for residues His and Arg, for which peptide CycN-K6 exhibits ∆δ Hα values of opposite sign relative to those observed in peptides CycS-C6 and CycN-K7 ( Figure 3 and Table 2). In addition, 3 J αN coupling constants, which have a well-established relationship with φ angles, displayed differences among the three peptides. DPhe residue shows a small 3 J αN value (<5 Hz) in peptide CycS-C6 [20], which is typical of the i + 1 residue in β-turns, but values about 6-7 Hz in peptides CycN-K6 and CycN-K7 (Table 2). In contrast, the 3 J αN value is small for the His residue in peptide CycN-K6, and displays a large value, which is characteristic of extended chains, in peptide CycS-C6 (Table 2). This suggests that the residues forming a β-turn structure differ between peptide CycS-C6 and CycN-K6. No residue shows a small value in the case of peptide CycN-K7 (Table S4), probably because its larger macrocyclic size imposes less restrictions on mobility. It seems obvious that the differences in macrocycle conformations should translate into changes in the relative dispositions of the side chains. Thus, the magnitudes of the Arg up-field shifted 1 H δ differ significantly among the three peptides, and follow the decreasing order CycS-C6 > CycN-K7 > CycN-K6 (Table 2). These differences must arise from changes in the packing of the Arg side chain relative to the aromatic rings, as confirmed in the calculated structures (see below). All these conformational variations account for the differences in affinity to receptor MC1R (Table 1). the ring currents of one or more aromatic residues, as was also observed in peptide CycS-C6 [20]. Despite these common general characteristics, the detailed examination of NMR parameters highlights the existence of important differences between peptides CycN-K6 and CycN-K7, and also among these peptides and the previously reported CycS-C6 [20]. The differences are observed both at the cycle backbone and at certain side chains. The most remarkable differences at cycle backbone are observed for residues His and Arg, for which peptide CycN-K6 exhibits ΔδHα values of opposite sign relative to those observed in peptides CycS-C6 and CycN-K7 ( Figure 3 and Table 2). In addition, 3 JαN coupling constants, which have a well-established relationship with φ angles, displayed differences among the three peptides. DPhe residue shows a small 3 JαN value (<5 Hz) in peptide CycS-C6 [20], which is typical of the i + 1 residue in β-turns, but values about 6-7 Hz in peptides CycN-K6 and CycN-K7 (Table 2). In contrast, the 3 JαN value is small for the His residue in peptide CycN-K6, and displays a large value, which is characteristic of extended chains, in peptide CycS-C6 (Table 2). This suggests that the residues forming a β-turn structure differ between peptide CycS-C6 and CycN-K6. No residue shows a small value in the case of peptide CycN-K7 (Table S4), probably because its larger macrocyclic size imposes less restrictions on mobility. It seems obvious that the differences in macrocycle conformations should translate into changes in the relative dispositions of the side chains. Thus, the magnitudes of the Arg up-field shifted 1 H δ differ significantly among the three peptides, and follow the decreasing order CycS-C6 > CycN-K7 > CycN-K6 (Table 2). These differences must arise from changes in the packing of the Arg side chain relative to the aromatic rings, as confirmed in the calculated structures (see below). All these conformational variations account for the differences in affinity to receptor MC1R (Table 1).    [20]. 3 Values taken from [21]. 4 Experimental errors in 3 J αN are ± 0.4 Hz. 5 Range reported for non-structured model pentapeptides [22].

NMR Solution Structures of Analogues CycN-K6 and CycN-K7
According to the qualitative analysis of the NMR data, peptides CycS-C6 [20], CycN-K6 and CycN-K7 present some conformational differences. The size of the macrocycle seems to affect the cycle conformation, which, in turn, induces changes in the relative positions of aromatic and positively-charged side chains. To achieve a detailed comparison of the three-dimensional arrangement of the side chains in the three peptides, we proceeded to calculate the structures of peptides CycN-K6 and CycN-K7 from the experimental NMR parameters (see Materials and methods). The solution structure of CycS-C6 had been previously determined following the same calculation protocol [13]. The structural ensembles of the three peptides are represented in Figure 4. Peptide CycS-C6, which has the smallest cycle size (19 atoms), has the lowest RMSD value (0.4 Å). Thus, as expected, the smaller macrocycle restricts the most the available conformations. However, CycN-K7, which has the biggest cycle size (25 atoms), has a slightly lower RMSD (0.7 Å) than peptide

NMR Solution Structures of Analogues CycN-K6 and CycN-K7
According to the qualitative analysis of the NMR data, peptides CycS-C6 [20], CycN-K6 and CycN-K7 present some conformational differences. The size of the macrocycle seems to affect the cycle conformation, which, in turn, induces changes in the relative positions of aromatic and positively-charged side chains. To achieve a detailed comparison of the three-dimensional arrangement of the side chains in the three peptides, we proceeded to calculate the structures of peptides CycN-K6 and CycN-K7 from the experimental NMR parameters (see Materials and methods). The solution structure of CycS-C6 had been previously determined following the same calculation protocol [13]. The structural ensembles of the three peptides are represented in Figure 4. Peptide CycS-C6, which has the smallest cycle size (19 atoms), has the lowest RMSD value (0.4 Å). Thus, as expected, the smaller macrocycle restricts the most the available conformations. However, CycN-K7, which has the biggest cycle size (25 atoms), has a slightly lower RMSD (0.7 Å) than peptide CycN-K6 (0.8 Å), whose cycle size (22 atoms) is intermediate between CycS-C6 and CycN-K7. Although the difference in RMSD value between CycN-K6 and CycN-K7 is very small, it is interesting to note that increasing or decreasing the cycle size in the antimicrobial peptide Gramicidin S results in a well-defined antiparallel β-sheet or a less-defined distorted structure depending on the number of cycle residues [23]. Apart from this, the main differences among the structures lie in the relative position of the side chains (Figure 4), in particular the Trp side chain shows quite a different orientation in the three peptides ( Figure S3, Supplementary Materials). These differences should account for the distinct affinities to MC1R receptor exhibited by the three peptides.

atoms) is intermediate between CycS-C6 and CycN-K7.
Although the difference in RMSD value between CycN-K6 and CycN-K7 is very small, it is interesting to note that increasing or decreasing the cycle size in the antimicrobial peptide Gramicidin S results in a well-defined antiparallel β-sheet or a less-defined distorted structure depending on the number of cycle residues [23]. Apart from this, the main differences among the structures lie in the relative position of the side chains (Figure 4), in particular the Trp side chain shows quite a different orientation in the three peptides ( Figure S3, Supplementary Materials). These differences should account for the distinct affinities to MC1R receptor exhibited by the three peptides.

Discussion
Based on sequence homology, the structure of the MC1R receptor consists of seven transmembrane helices, TM1 to TM7 [12,25], which are very probably arranged in the helical bundle characteristic of the GPCR superfamily. In the proposed model for the complex between the human MC1R receptor and a rigid cyclic melanocyte-stimulating hormone core peptide c[His-DPhe-Arg-Trp-Gly], the ligand lies into an inner channel of the helical bundle, and close to the extracellular loops. In that model [24], the guanidinium group of the Arg residue interacts with a cluster of negatively-charged residues and polar residues at TM1 and TM7 helices; the DPhe aromatic ring lies

Discussion
Based on sequence homology, the structure of the MC1R receptor consists of seven transmembrane helices, TM1 to TM7 [12,25], which are very probably arranged in the helical bundle characteristic of the GPCR superfamily. In the proposed model for the complex between the human MC1R receptor and a rigid cyclic melanocyte-stimulating hormone core peptide c[His-DPhe-Arg-Trp-Gly], the ligand lies into an inner channel of the helical bundle, and close to the extracellular loops. In that model [24], the guanidinium group of the Arg residue interacts with a cluster of negatively-charged residues and polar residues at TM1 and TM7 helices; the DPhe aromatic ring lies in a hydrophobic cavity between TM1 and TM2 helices; the Trp indole moiety interacts with a hydrophobic cluster formed by residues from TM3, TM6, and TM7 helices, and it might be hydrogen-bonded to an Asp at TM3 helix; and the His imidazole group points towards the extracellular loops and contacts with residues at TM2 helix. Helices TM4 and TM5 do not participate in the binding [24,26].
Considering this complex network of interactions, the structural differences among the three peptides ( Figure 4 and Figure S3, Supplementary Materials) can explain their distinct affinity to MC1R ( Table 1). As we previously reported [20], the extra-cyclic Lys present in CycS-C6 strongly decreases affinity to MC1R (Table 1). Likely, this additional long and positively-charged Lys side chain impedes suitable accommodation of the side chains of the cycle residues at the MC1R binding pocket. The extra-cyclic Lys might interfere either by steric hindrance or by competing with the Arg binding site. Strikingly, the MC1R affinity of CycN-K6, which also has an extra-cyclic Lys, but a larger cycle, is about ten-fold higher than in CycS-C6 (Table 1). Indeed, this result prompted us to characterize the structure of peptide CycN-K6 (see Introduction). As we hypothesized [20], the differences in the size of the cycle give rise to different relative arrangements of all the side chains. These differences can be appreciated for the His/DPhe, DPhe/Arg, Arg/Trp, and DPhe/Trp pairs in Figure 4. For example, the aromatic rings of DPhe3 and Trp5 extend outwards from the same macrocycle face in CycS-C6 (61 ± 24 • ) but from different sides in CycN-K6 (99 ± 38 • ). Considering that DPhe and Trp aromatic rings bind to different sub-sites in the MC1R receptor [24], it seems likely that these rings' position on different cycle faces in CycN-K6 will predispose them to interact favorably with MC1R. In contrast, the alignment of these aromatic rings on the same cycle face in CycS-C6 predisposes them unfavorably for binding, and it can be expected that part of the CycS-C6/MC1R binding energy will have to be "spent" to change their orientation. These expectations are supported by the higher MC1R affinity of CycN-K6 compared to CycS-C6 (Table 1). In the case of peptide CycN-K7, which shows an intermediate affinity value between the other two peptides, the orientation of the DPhe and Trp rings (69 ± 20 • ) is more similar to CycS-C6 than CycN-K6 (see also Figure S3, Supplementary Materials).
Analogously, the other residue pairs are probably better predisposed in CycN-K6 to interact with the MC1R receptor. For instance, it can be seen in Figure 4 that the Arg/DPhe and DPhe/His pairs display quite different arrangements in the three peptides. Since the DPhe aromatic ring interacts with residues at the TM1 and TM2 helices, Arg with residues at TM1, and His with residues at TM2, positioning these three side chains in a conformation that resembles their orientation when bound to the MC1R receptor should facilitate complex formation. Based on this line of reasoning, we can infer that the melanocortin core residues (His, DPhe, Arg, and Trp) in CycN-K6 are favorably arranged to bind the MC1R receptor.
Evidently, the conformation of the free and MC1R-bound peptides could differ. Then, considering peptides whose side chains are not favorably predisposed to bind the receptor, we can expect that flexible peptides will bind better than rigid ones. This is because less binding energy will be "spent" forcing a flexible peptide to adopt the proper orientation as compared to a rigid one. Indeed, some cases in which the most flexible variant of a peptide is more active than those more structurally restricted have been reported [27,28]. Although RMSD value is not a dynamic parameter, in general, low RMSD values correspond to well-defined, and usually mostly rigid structures, whereas high RMSD values are found for less-defined, and more-flexible conformers. Accordingly, the structure of the 19-membered macrocycle CycS-C6 is better defined, and therefore is likely to be more rigid, than those of the 22-and 25-macrocycles CycN-K6, and CycN-K7, respectively. Hence, if receptor binding requires structural re-arrangements, these two peptides will adapt more easily than CysS-C6. This explains the lower affinity for MC1R by CysS-C6 in comparison with CycN-K7, even though the side chains of these two peptides dispose differently from CycN-K6, and more similarly between them.
Coming back to the Lys residue, which is present in the three peptides, and strongly decreases affinity to MC1R (see above CysS-C6 vs. PG10N; Table 1); the distribution of the positively-charged residues (Arg4 and Lys6/7; Figure 4) differs among the three peptides. Thus, the fact that the detrimental effect of Lys is not so drastic in CycN-K6, and CycN-K7 could be explained by its disfavoring effect being offset by other favorable factors, such as a suitable pre-arrangement of the side chains in CycN-K6; and by the higher flexibility of these larger macrocycles, which makes feasible for the Lys side chain to be placed so that it does not disturb the favorable key interactions of the conserved melanocortin core residues. The conformational space available for Lys6 in CycN-K7, which is within the cycle, must be more restricted than for Lys7 in CycN-K6, which is outside the cycle (Figures 1 and 4), and likely contributes to the lower MC1R affinity of CycN-K7 relative to CycN-K6 (Table 1).
Taken together, our results have shown that appropriate side chain orientations, as proposed by Cho and coworkers [29], and conformational flexibility play an important role in the binding affinity to melanocortin receptors. Upon the increase of the cycle size, the side chain orientations and the cycle flexibility are changed, and as a consequence binding affinity is affected.

Competitive Binding Assays
NDP-α-MSH was radioiodinated as described [20]. To determine the inhibitory concentration of 50 % (IC 50 ) values for the α-MSH analogues, competitive binding assays with [ 125 I]NDP-α-MSH in B16F1 melanoma cells were carried out in triplicate following the previously described procedure [20]. IC 50 values were calculated from the plots of the percentage of [ 125 I]NDP-α-MSH bound to the cells vs. the concentrations of displacing peptides by using GraphPad Prism software (GraphPad Software Inc., La Jolla, CA, USA).

NMR Sample Preparation
Samples for acquisition of NMR spectra at approximately 1 mM peptide concentration were prepared by solving the lyophilized peptides (~0.5 mg) in 0.5 ml H 2 O/D 2 O (9:1 v/v).
Sodium-2,2-dimethyl-2-silapentane-5-sulfonate (DSS) was added as an internal reference. A glass microelectrode was used to measure pH, which was adjusted to 2.5 by adding minimal amounts of NaOD or DCl. The pH readings were not corrected for isotope effects.

Acquisition of NMR Spectra
A Bruker (Bruker Española S.A, Madrid, Spain) AV-600 spectrometer operating at a proton frequency of 600.13 MHz and equipped with a cryoprobe was used to acquire NMR spectra. The temperature of the NMR probe was calibrated using a methanol sample. 1D and 2D spectra, i.e., 1 H, 1 H-COSY (phase-sensitive correlated spectroscopy), 1 H, 1 H-TOCSY (total correlated spectroscopy), and 1 H, 1 H-ROESY (rotating frame nuclear Overhauser spectroscopy) were recorded and processed as previously reported [20,30]. 1 H-13 C and 1 H-15 N heteronuclear single quantum coherence spectra (HSQC) were acquired at natural abundance of the heteronuclei. 13 C and 15 N δ-values were indirectly referenced by multiplying the spectrometer frequency that corresponds to 0 ppm in the 1 H spectrum, assigned to internal DSS reference, by 0.25144953 and 0.101329118, respectively [31].

NMR Spectral Assignment
1 H resonances were assigned by analysis of 2D 1 H, 1 H-COSY, 1 H, 1 H-TOCSY, and 1 H, 1 H-ROESY spectra using the program Sparky [32] and following standard sequential assignment methods [33,34]. Next, the 13 C and 15 N resonances were straightforwardly assigned from the cross-correlations observed in the corresponding HSQC spectra between the proton and the bound carbon or nitrogen, respectively. Supplementary Tables S1 and S2 list all chemical shifts for peptides CycN-K6 and CycN-K7. The 3 J αN coupling constants were measured at the doublet signals observed for the HN amide protons in the 1D 1 H-NMR spectra acquired in aqueous solution at pH 2.5 and 25 • C. They are listed at Supplementary Table S3.

Structure Calculation
The three dimensional structures were calculated from distance restraints derived from the cross-peaks present in 1 H, 1 H-ROESY spectra of the peptides CycN-K6 and CycN-K7 ( Figure S4) and using torsion angle dynamics as implemented in the CYANA program (P. Guntert's group, Institute of Biophysical Chemistry, Goethe-University, Frankfurt am Main, Germany) [35,36]. The peak lists for 1 H, 1 H-ROESY spectra were generated by interactive peak picking, and the peak volumes obtained by the automatic integration function of Sparky [32]. The CYANA protocol consisted of seven iterative cycles of automatic assignment of distance restraints and structure calculation, followed by a final standard structure calculation. In addition, a negative value for the torsion angle phi for all amino acids, except for DPhe, plus the distance restraints needed for cyclization between Lys6 or Lys7 and the p-NO 2 -benzoic acid linker were used during the NOE assignment/structure calculation cycles. In each cycle, the structure calculation started from 100 randomized conformers and the standard CYANA-simulated annealing schedule was used with 10,000 torsion angle dynamics steps per conformer. The final structures, which are the ensemble of the 20 structures with the lowest target functions, were visualised and analysed using the MOLMOL program [37].