Incorporation of Agouti-Related Protein (AgRP) Human Single Nucleotide Polymorphisms (SNPs) in the AGRP-Derived Macrocyclic Scaffold c[Pro-Arg-Phe-Phe-Asn-Ala-Phe-DPro] Decreases Melanocortin-4 Receptor Antagonist Potency and Results in the Discovery of Melanocortin-5 Receptor Antagonists

Abstract

While the melanocortin receptors (MCRs) are known to be involved in numerous biological pathways, the potential roles of the MC5R have not been clearly elucidated in humans. AgRP, an MC3R/MC4R antagonist and MC4R inverse agonist, contains an exposed ß-hairpin loop composed of six residues (Arg-Phe-Phe-Asn-Ala-Phe) that is imperative for binding and function. Within this active-loop of AgRP, four human missense polymorphisms were deposited into the NIH Variation Viewer database. These polymorphisms, Arg111Cys, Arg111His, Phe112Tyr, and Ala115Val (AgRP full-length numbering) were incorporated into the peptide macrocycles c[Pro¹-Arg²-Phe³-Phe⁴-Xaa⁵-Ala⁶-Phe⁷-DPro⁸], where Xaa was Dap⁵ or Asn⁵, to explore the functional effects of these naturally-occurring substitutions in a simplified AgRP scaffold. All peptides lowered potency at least 10-fold in a cAMP accumulation assay compared to the parent sequences at the MC4Rs. Compounds MDE 6-82-3c, ZMK 2-82, MDE 6-82-1c, ZMK 2-85, and ZMK 2-112 are also the first AgRP-based chemotypes that antagonize the MC5R.

Graphical Abstract

Introduction

Five unique melanocortin receptors (MCRs) have been identified to date.^1-8 These receptors are members of the family A of G protein-coupled receptors (GPCRs). The MCRs have been shown to be involved in several different physiological functions such as skin pigmentation (MC1R),^1,2 steroidogenesis (MC2R),^1,9 and energy homeostasis (MC3/4R).^10-13 While many biological activities have been attributed to the MC1–4Rs, the physiological role(s) of the MC5R have not been fully identified and characterized for this most ubiquitously expressed melanocortin receptor. Based upon the phenotype of the MC5RKO mice, the MC5R has been reported to be involved in sebaceous gland function.¹⁴ The melanocortin receptor family is stimulated by several endogenous peptides including α-, ß-, and γ- melanocyte stimulating hormone (MSH) and adrenocorticotropin hormone (ACTH), all processed from the pro-opiomelanocortin (POMC) gene transcript.¹⁵ In addition to these agonists, there are two known naturally occurring antagonists for the melanocortin receptor family: agouti-signaling protein (ASP)^16,17 and agouti-related protein (AgRP).^18,19 These antagonist peptides both contain an Arg-Phe-Phe motif in the carboxyl terminal region, as reviewed,²⁰ which has been proposed to be the sequence responsible for receptor affinity and antagonist activity.^21,22 AgRP is expressed centrally in the arcuate nucleus of the hypothalamus¹⁹ and antagonizes the MC3R and MC4R,^4,19,23 effectively increasing feeding behavior.^13,24-26 Additionally, AgRP functions as an inverse agonist at the MC4R, decreasing the amount of cyclic adenosine monophosphate (cAMP) produced by the cell in the absence of the agonist ligand stimulation^27,28 Feeding studies reported that central administration of melanocortin agonists such as NDP-MSH (a synthetic potent analogue of α-MSH) or melanotan II (MTII, a potent cyclic analogue of α-MSH) decreased food intake and prevented weight gain in mice.^29-33 Contrarily, central injections of AgRP antagonist prompted long-acting increases in feeding behavior in fasted or sated animals.^19,32,34-36

In addition to the reported activity of AgRP at the MC3R and MC4R, there have been mixed reports about AgRP at the MC5R. Pooled fractions of recombinant AgRP using a baculovirus expression system in insect cells resulted in a rightward potency shift of α-MSH in HEK293 cells expressing the hMC5R at 100 nM concentrations.¹⁹ Another report using recombinant AGRP expressed from COS-7 cells indicated that AgRP was unable to displace radiolabeled NDP-MSH at the hMC5R expressed in L cells at up to 40 nM concentrations, although functional activity was not determined.¹⁸ A third report using chemically synthetized AgRP ligands [either the hAGRP(87-132) C-terminal domain, or a fusion of the mAGRP(21-85) to the hAGRP(87-132) generating the (Leu127Pro)AGRP] indicated that AgRP could functionally antagonize α-MSH at the MC5R at 100 nM concentrations, and could displace radiolabeled NDP-MSH (IC₅₀ = 310 nM) and radiolabeled AgRP (IC₅₀ = 26 nM) at the MC5R.³⁷ The differing reported activities could be due to the activity ranges assayed, varying degrees of AgRP purity based upon the methods used to generate the peptide, or cellular assay conditions between the different reports. Previous reports of AgRP derived ligands have not reported MC5R antagonist activity, although inverse agonist efficacy has previously been observed.^38,39

Melanocortin ligands like MTII were originally developed before cloning of the receptors and to function as MC1R agonists in frog and lizard skin pigmentation assay. ^30,40 After cloning of the receptors, these compounds were developed for the clinic to prompt melanogenesis (tanning of the skin), aide in melanoma detection, and as melanoma chemotherapeutics.^41-44 Since the early 2000’s the MC4R, and to a lesser extent the MC3R, have been targeted due to their role in energy homeostasis and clinical indications in obesity and anorexia.^45-49 Obesity has grown to be an epidemic, with projections estimating that over 50% of Americans will be overweight or obese by 2030.⁵⁰ Approximately 5% of these cases are monogenic and linked to a disrupted central melanocortin system, where patients’ obesity manifests in early childhood.^51-53 Diseases of negative energy imbalance, causing failure to thrive in young children, anorexia, and cachexia (disease-associated wasting) are also concerning because a decreased drive to eat can stunt growth in children and delay the body’s response to healing.^47-49,54,55 Manipulation of the melanocortin system in either direction could offer a viable solution to combating these diseases and their associated comorbidities.

Activation of the MC4R leads to a decrease in food intake, but compounds that have made it into the clinic have ultimately failed due to unwanted side effects like increases in blood pressure,⁵⁶ darkening of the skin,^57,58 and increased erectile function,⁵⁹ as reviewed by Ericson et al. ⁶⁰ One compound, RM-493 (Setmelanotide), manufactured by Rhythm Pharmaceuticals, has seen clinical success in obese patients that do not produce the POMC gene transcript by decreasing body weight without the increase in blood pressure seen in the clinic with other melanocortin compounds.⁶¹ However, one patient with POMC deficiency was reported to have a decrease in diastolic blood pressure while being treated with RM-493 that was present during the extension phase of the study as well.⁵⁷ RM-493 has also been investigated in a Phase II trial in patients with rare genetic obesities (POMC deficiency, leptin/leptin receptor deficiency, Prader-Willi syndrome, Bardet-Biedl, etc.) clinical trial: NCT03013543. No drugs aimed towards increasing appetite through the melanocortin cascade have made it to the clinic to date.

Within the melanocortin system, many single nucleotide polymorphisms (SNPs) have been identified in the receptors,^51,62-67 agonists,^68,69 and antagonists,^48,70-74 which may be linked to a predisposition for an obese or lean phenotype in humans. Receptor polymorphisms can alter protein conformation or structure thereby decreasing the affinity of the normally processed ligands.^51,62-66 Mutations within genes that code for the peptide ligands can prevent the peptides from being synthesized or processed properly, alter ligand affinity or activity at the receptor, disrupting the natural signaling cascade and ultimately altering the phenotype of the patient.^68,69,72-76

This study focused on exploring four single nucleotide polymorphisms that result in missense mutations in the putative active loop of AgRP, containing the Arg-Phe-Phe amino acid motif and the three following residues Asn, Ala, and Phe. This sequence is located on an exposed ß-hairpin loop in the structure of AgRP (Figure 1). ^77,78 Truncation studies have demonstrated that the cysteine-rich carboxyl terminus of AgRP is sufficient to modulate receptor activity at the MC3R and MC4R, and possibly the MC5R as discussed above.^19,77,79 In the native sequence the loop is exposed between a disulfide bridge, Figure 1.

Figure 1.

Illustration of the proposed binding loop of AgRP, with the critical residues Arg-Phe-Phe in dark purple. PDB ID: 1HYK ⁷²

The AgRP antagonist, in its native form, is 132 amino acids in length, with five disulfide bonds throughout the carboxyl terminus; however, biologically active AgRP is attributed to residues 87-132.^19,37 Replicating this protein in its entirety or in its active form, to study multiple polymorphisms is costly and labor-intensive, therefore driving the discovery of smaller, more synthetically amenable probes and scaffolds. In 2015, we reported the discovery that the macrocyclic octapeptide AgRP mimetic scaffold, c[Pro-Arg-Phe-Phe-Asn-Ala-Phe-DPro], was reported to have nanomolar potency at the MC4R.⁸⁰ This scaffold uses a DPro-Pro head-to-tail cyclization to create a macrocycle ligand postulated to orient the key pharmacophore AGRP based side-chain groups in a similar conformation to that of the native AGRP endogenous antagonist. This macrocycle is smaller by 37 amino acids from AgRP (87-132), and retains nanomolar MC4R antagonist potency. This macrocycle is amenable to substitutions, allowing for the investigation of multiple potentially deleterious mutations within the putative AGRP active loop. The SNPs were accessed from the NIH Variation Viewer in November 2017 (https://www.ncbi.nlm.nih.gov/variation/view/). The mutations are R111H (rs199927717), R111C (rs1012110755), F112Y (rs200972106), and A115V (rs773319622) based upon the full length AGRP amino acid sequence, Figure 2.⁸¹

Figure 2.

SNPs in the purported binding loop of AgRP, deposited into the NIH Variation Viewer as of November 2017, (https://www.ncbi.nlm.nih.gov/variation/view/).

These mutations were systematically incorporated and assayed in the macrocyclic octapeptide scaffold. The polymorphisms were also incorporated into the scaffold with the endogenous Asn⁵ to diaminopropionic acid (Dap) substitution, as this modification was previously reported to be equipotent to AgRP(87-132) at the MC4R.⁸⁰ Screening deposited mutations from a publicly available database versus using the genome sequenced from a small subset of an afflicted population, which has previously been the source of polymorphisms,^{51,57,58,63,68,69} allows for a broader understanding of the melanocortinergic control of energy homeostasis through the central receptors, and potentially other physiologically relevant functions through the peripheral receptors. In addition, this broad sourcing allows for a greater understanding of the contribution of polymorphisms in monogenic diseases. To fully characterize these SNPs, they would need to be synthesized in the full length of AgRP. It is hypothesized that the octapeptide scaffold can be utilized as a novel tool to screen polymorphisms that may be disruptive to AgRP signaling at the MCRs and provide new chemical probes to investigate MCR physiology in vivo. Using the two previously mentioned scaffolds, an eight-compound library was assayed for activity at the mouse MC1R, MC3R, MC4R, and MC5R and the human MC4R. Compounds were not tested at the MC2R as it has previously only been activated by full-length ACTH.^1,9

Results

All peptides were synthesized using a microwave-assisted synthesizer and standard fluorenylmethoxycarbonyl (Fmoc) techniques.^82,83 Peptides were cleaved from the resin and cyclized using BOP and HOBt, with their side chain protecting groups intact as previously described.^38,39,80,84 Peptides were purified to >95% purity using semi-preparative reverse-phase high performance liquid chromatography (RP-HPLC) after removing the protecting groups from the residues. Purity was determined using an analytical RP-HPLC, using two distinct solvent systems, Table 1. Molecular weights were assessed using either MALDI-TOF/TOF or ESI-TOF/MS (University of Minnesota Mass Spectrometry Laboratory), Table 1. The potencies of these compounds in live cell cAMP accumulation assay are reported in Tables 2 and 3. The binding affinities in live cell ¹²⁵I-radiolabeled NDP-MSH binding assay are reported in Tables 4 and 5.

Table 1.

Analytical data for peptides synthesized in this study.

Peptide	Sequence	Retention Time (min)a		M (calc)	mass spectral analysis (M+1)	Purity %
		System 1	System 2
ZMK 2-82	c[Pro-Arg-Phe-Phe-Asn-Val-Phe-DPro]	19.5	31.1	1004.5	1005.6	>98%
ZMK 2-85	c[Pro-Arg-Phe-Phe-Dap-Val-Phe-DPro]	19.7	31.3	976.5	977.6	>95%
ZMK 2-96	c[Pro-His-Phe-Phe-Asn-Ala-Phe-DPro]	17.8	28.2	957.4	958.4	>98%
ZMK 2-99	c[Pro-His-Phe-Phe-Dap-Ala-Phe-DPro]	18.0	29.7	929.5	930.4	>95%
ZMK 2-110	c[Pro-Arg-Tyr-Phe-Asn-Ala-Phe-DPro]	15.4	24.6	992.5	993.5	>99%
ZMK 2-112	c[Pro-Arg-Tyr-Phe-Dap-Ala-Phe-DPro]	15.4	25.6	964.5	965.5	>98%
ZMK 3-18	c[Pro-Cys-Phe-Phe-Asn-Ala-Phe-DPro]	20.9	31.8	924.1	924.8	>97%
ZMK 3-20	c[Pro-Cys-Phe-Phe-Dap-Ala-Phe-DPro]	21.9	32.8	896.1	896.8	>95%

^aPeptide retention times (min) are reported for solvent system 1 (10% acetonitrile in 0.1% trifluoroacetic acid/water and a gradient to 90% acetonitrile over 35 min) and solvent system 2 (10% methanol in 0.1% trifluoroacetic acid/water and a gradient to 90% methanol over 35 min). An analytical Vydac C18 column (Vydac 218TP104) was used with a flow rate of 1.5 mL/min. The peptide purity was determined by HPLC at a wavelength of 214 nm. Molecular mass was determined by MALDI-MS or ESI-TOF/MS (Applied Biosystems-Sciex 5800 MALDI/TOF/TOF-MS, Bruker BioTOF II ESI-TOF/MS, University of Minnesota Mass Spectrometry Lab).

Table 2.

Summary of the antagonist pA₂ and inverse agonist pharmacology of AgRP macrocyclic analogues incorporating both single nucleotide polymorphism mutations and Dap residues at the human melanocortin 4 receptor (hMC4R).^a

	Peptide	Sequence	hMC4R	Observed Inverse Agonism?
			pA2
	AgRP		8.7 ± 0.1	Yes
ASN 5	1* (MDE 6-82-3c)	c[Pro-Arg-Phe-Phe-Asn-Ala-Phe-DPro]	8.0 ± 0.2	Yes
	ZMK 2-82	c[Pro-Arg-Phe-Phe-Asn-Val-Phe-DPro]	7.8 ± 0.03	Yes
	ZMK 2-96	c[Pro-His-Phe-Phe-Asn-Ala-Phe-DPro]	5.8 ± 0.2	Yes
	ZMK 2-110	c[Pro-Arg-Tyr-Phe-Asn-Ala-Phe-DPro]	6.5 ± 0.2	Yes
	ZMK 3-18	c[Pro-Cys-Phe-Phe-Asn-Ala-Phe-DPro]	5.8 ± 0.2	No
DAP 5	2* (MDE 6-82-1c)	c[Pro-Arg-Phe-Phe-Dap-Ala-Phe-DPro]	8.6 ± 0.2	Yes
	ZMK 2-85	c[Pro-Arg-Phe-Phe-Dap-Val-Phe-DPro]	7.6 ± 0.06	Yes
	ZMK 2-99	c[Pro-His-Phe-Phe-Dap-Ala-Phe-DPro]	6.4 ± 0.09	No
	ZMK 2-112	c[Pro-Arg-Tyr-Phe-Dap-Ala-Phe-DPro]	7.5 ± 0.2	Yes
	ZMK 3-20	c[Pro-Cys-Phe-Phe-Dap-Ala-Phe-DPro]	<5.0	No

^aThe indicated errors represent the standard error of the mean (SEM) determined from at least three independent experiments. The antagonistic pA₂ values were determined using the Schild analysis and the agonist NDP-MSH. >100,000 indicates that the compound was examined but lacked agonist activity at up to 100 μM concentrations. <5.0 indicates that no antagonist potency was observed in the highest concentration ranged assayed (10,000, 5,000, 1,000, and 500 nM). Inverse agonist indicates that inverse agonist pharmacology was observed.

^*The pharmacology for 1 and 2 has previously been reported.⁸⁰

Table 3.

Summary of the cAMP based agonist EC₅₀ and antagonist pA₂ pharmacology values of AgRP macrocyclic analogues incorporating both single nucleotide polymorphism mutations and Dap residues at the mouse melanocortin receptors.^a

	Peptide	Sequence	mMC1R	mMC3R		mMC4R		mMC5R
			EC50 (nM)	EC50 (nM)	pA2	EC50 (nM)	pA2	EC50 (nM)	pA2
	NDP-MSH		0.010±0.003	0.07±0.03		0.9±0.6		0.13±0.03
ASN 5	1* (MDE 6-82-3c)	c[Pro-Arg-Phe-Phe-Asn-Ala-Phe-DPro]	25% @ 100 μM	>100,000	6.3 ± 0.1	>100,000	8.2 ± 0.1	Inverse Agonist −10%	6.4 ± 0.2
	ZMK 2-82	c[Pro-Arg-Phe-Phe-Asn-Val-Phe-DPro]	23% @ 100 μM	>100,000	5.9 ± 0.1	16% @ 100 μM	7.2 ± 0.1	>100,000	6.1 ± 0.09
	ZMK 2-96	c[Pro-His-Phe-Phe-Asn-Ala-Phe-DPro]	23% @ 100 μM	>100,000	< 5.0	>100,000	5.3 ± 0.1	>100,000	N/A
	ZMK 2-110	c[Pro-Arg-Tyr-Phe-Asn-Ala-Phe-DPro]	>100,000	>100,000	5.8 ± 0.2	>100,000	6.1 ± 0.03	Inverse Agonist −35% @ 100 μM	N/A
	ZMK 3-18	c[Pro-Cys-Phe-Phe-Asn-Ala-Phe-DPro]	46% @ 100 μM	>100,000	< 5.0	>100,000	5.9 ± 0.1	Inverse Agonist −32% @ 100 μM	<5.0
DAP 5	2* (MDE 6-82-1c)	c[Pro-Arg-Phe-Phe-Dap-Ala-Phe-DPro]	30% @ 100 μM	>100,000	6.5 ± 0.09	>100,000	8.7 ± 0.1	Inverse Agonist −15%	7.3 ± 0.03
	ZMK 2-85	c[Pro-Arg-Phe-Phe-Dap-Val-Phe-DPro]	79% @ 100 μM	27% @ 100 μM	6.1± 0.03	37% @ 100 μM	7.5 ± 0.1	>100,000	6.3 ± 0.03
	ZMK 2-99	c[Pro-His-Phe-Phe-Dap-Ala-Phe-DPro]	56% @ 100 μM	>100,000	5.4 ± 0.1	>100,000	6.0 ± 0.1	>100,000	N/A
	ZMK 2-112	c[Pro-Arg-Tyr-Phe-Dap-Ala-Phe-DPro]	48% @ 100 μM	>100,000	6.5 ± 0.1	>100,000	7.3 ± 0.1	>100,000	6.5 ± 0.2
	ZMK 3-20	c[Pro-Cys-Phe-Phe-Dap-Ala-Phe-DPro]	4000 ± 2000	29% @ 100 μM	< 5.0	42% @ 100 μM	< 5.0	>100,000	N/A

^aThe indicated errors represent the standard error of the mean (SEM) determined from at least three independent experiments. The antagonistic pA₂ values were determined using the Schild analysis and the agonist NDP-MSH. >100,000 indicates that the compound was examined but lacked agonist activity at up to 100 μM concentrations. N/A indicates that the compound was not examined. A percentage denotes the percent maximal stimulatory response observed at 100 μM concentrations but not enough stimulation was observed to determine an EC₅₀ value. <5.0 indicates that no antagonist potency was observed in the highest concentration ranged assayed (10,000, 5,000, 1,000, and 500 nM). Inverse agonist indicates that inverse agonist pharmacology was observed with the percent decrease from basal indicated. For inverse agonists, a decrease in cAMP signal was observed without a sigmoidal dose-response curve, the percent change from basal at 100 μM concentrations is indicated.

^*The pharmacology for 1 and 2 has previously been reported, with the exception of the antagonistic pA₂ values at the mMC5R which are reported herein for the first time.

Table 4.

Summary of binding IC₅₀ affinities using the ¹²⁵I-NDP-MSH orthosteric binding ligand, of AgRP macrocyclic analogues incorporating both single nucleotide polymorphism mutations and Dap residues at the mouse melanocortin receptors (mMCRs).^a

	Peptide	Sequence	mMC1R		mMC3R		mMC4R		mMC5R
			IC50 (nM)	Fold Diff	IC50 (nM)	Fold Diff	IC50 (nM)	Fold Diff	IC50 (nM)	Fold Diff
	NDP-MSH		0.29 ± 0.09		8.2 ± 0.7		1.30 ± 0.01		140 ± 50
ASN 5	1 (MDE 6-82-3c)	c[Pro-Arg-Phe-Phe-Asn-Ala-Phe-DPro]	9000 ± 2000	1	17000 ±5500	1	70 ± 20	1	12300 ± 800	1
	ZMK 2-82	c[Pro-Arg-Phe-Phe-Asn-Val-Phe-DPro]	15000 ± 8000	2	53000 ± 11000	3	38 ± 2	2	29000 ±3000	2
	ZMK 2-96	c[Pro-His-Phe-Phe-Asn-Ala-Phe-DPro]	>100,000		>100,000		6700 ± 600	300	>100,000
	ZMK 2-110	c[Pro-Arg-Tyr-Phe-Asn-Ala-Phe-DPro]	>100,000		>100,000		2000 ± 100	100	>100,000
	ZMK 3-18	c[Pro-Cys-Phe-Phe-Asn-Ala-Phe-DPro]	24600 ± 100	3	>100,000		1900 ± 500	90	>100,000
DAP 5	2 (MDE 6-82-1c)	c[Pro-Arg-Phe-Phe-Dap-Ala-Phe-DPro]	5200 ± 300	1	3200 ± 600	1	1.9 ± 0.6	1	1700 ± 300	1
	ZMK 2-85	c[Pro-Arg-Phe-Phe-Dap-Val-Phe-DPro]	8000 ± 5000	2	10000 ± 100	3	8.2 ± 0.8	4	2700 ± 100	2
	ZMK 2-99	c[Pro-His-Phe-Phe-Dap-Ala-Phe-DPro]	25000 ± 6000	5	15000 ± 3000	5	1000 ± 300	500	12000 ± 6000	7
	ZMK 2-112	c[Pro-Arg-Tyr-Phe-Dap-Ala-Phe-DPro]	18000 ± 1000	4	28000 ± 12000	9	134 ± 7	70	36000 ± 2000	20
	ZMK 3-20	c[Pro-Cys-Phe-Phe-Dap-Ala-Phe-DPro]	33000 ± 2000	6	49000 ± 3000	15	10000 ± 2000	5200	>100,000

^aThe indicated errors represent the standard deviation (SD) determined from at least two independent experiments, each consisting of duplicate replicates. >100,000 indicates that the compound was examined but lacked binding affinity at up to 100 μM concentrations. In addition, NDP-MSH, c[Pro-Arg-Phe-Phe-Asn-Ala-Phe-DPro], and c[Pro-Arg-Phe-Phe-Dap-Ala-Phe-DPro] are included as experimental controls and as reference controls. The fold difference (fold diff) is determined between the peptide as compared to the c[Pro-Arg-Phe-Phe-Asn-Ala-Phe-DPro] or c[Pro-Arg-Phe-Phe-Dap-Ala-Phe-DPro] controls respectively.

Table 5.

Summary of binding affinity IC₅₀ values using the ¹²⁵I-NDP-MSH orthosteric agonist ligand to competitively displace the AgRP based macrocyclic analogues incorporating both single nucleotide polymorphism mutations and Dap residues at the hMC4R.^a

	Peptide	Sequence	hMC4R
			IC50 (nM)	Fold Diff
	NDP-MSH		10 ± 2
ASN 5	1 (MDE 6-82-3c)	c[Pro-Arg-Phe-Phe-Asn-Ala-Phe-DPro]	70 ± 20	1
	ZMK 2-82	c[Pro-Arg-Phe-Phe-Asn-Val-Phe-DPro]	250 ± 40	3
	ZMK 2-96	c[Pro-His-Phe-Phe-Asn-Ala-Phe-DPro]	22000 ± 2000	300
	ZMK 2-110	c[Pro-Arg-Tyr-Phe-Asn-Ala-Phe-DPro]	14300 ± 1200	2000
	ZMK 3-18	c[Pro-Cys-Phe-Phe-Asn-Ala-Phe-DPro]	19300 ± 1400	300
DAP 5	2 (MDE 6-82-1c)	c[Pro-Arg-Phe-Phe-Dap-Ala-Phe-DPro]	7.5 ± 0.4	1
	ZMK 2-85	c[Pro-Arg-Phe-Phe-Dap-Val-Phe-DPro]	32.2 ± 0.5	4
	ZMK 2-99	c[Pro-His-Phe-Phe-Dap-Ala-Phe-DPro]	2700 ± 600	400
	ZMK 2-112	c[Pro-Arg-Tyr-Phe-Dap-Ala-Phe-DPro]	1016 ± 9	100
	ZMK 3-20	c[Pro-Cys-Phe-Phe-Dap-Ala-Phe-DPro]	15700 ± 1400	2100

^aThe indicated errors represent the standard deviation determined from at least two independent experiments, each consisting of duplicate replicates. >100,000 indicates that the compound was examined but lacked binding affinity at up to 100 μM concentrations. NDP-MSH, c[Pro-Arg-Phe-Phe-Asn-Ala-Phe-DPro], and c[Pro-Arg-Phe-Phe-Dap-Ala-Phe-DPro] are included as experimental controls and reference controls. The fold difference (fold diff) is determined between the peptide as compared to the c[Pro-Arg-Phe-Phe-Asn-Ala-Phe-DPro] or c[Pro-Arg-Phe-Phe-Dap-Ala-Phe-DPro] controls respectively.

SNP Incorporations, Asn-Substituted Scaffold:

Compound 1 (MDE 6-82-3c), which contains the endogenous sequence of AgRP from residues 111-116 (Arg-Phe-Phe-Asn-Ala-Phe), cyclized through a DPro-Pro motif, was previously reported to have antagonist activity at the mMC3R and mMC4R (pA₂= 6.3 and 8.2 respectively), inverse agonist activity at the mMC5R, and partially stimulated the mMC1R (25% of NDP-MSH signal at 100 μM).⁸⁰ In the present study, this compound was additionally found to possess antagonist activity at the hMC4R (pA₂= 8.0) and mMC5R (pA₂= 6.4), as well as inverse agonist activity at the hMC4R (Tables 2 and 3).

This scaffold was designed to recreate the active loop of AgRP cyclized through a DPro-Pro motif that had previously been demonstrated to mimic β-hairpin conformations in other macrocyclic peptides.⁸⁵ Therefore, this octapeptide scaffold was used to investigate four naturally occurring AgRP SNPs, without synthesizing the full-length peptide [46 residue AgRP(87-132)]. Substitution of the Ala115 with Val (rs773319622) resulted in ZMK 2-82. This peptide (ZMK 2-82) was able to partially stimulate the mMC1R (23% NDP-MSH signal at 100 μM), and was unable to stimulate the mMC3R, mMC4R, and mMC5R at the concentrations assayed. When assayed as an antagonist at the mMC3R, mMC4R, and mMC5R, ZMK 2-82 exhibited antagonist activity (pA₂= 5.9, 7.2, and 6.1 respectively) and a pA₂ of 7.7 at the hMC4R. Replacing the Arg111 to a His (rs199927717) resulted in the peptide ZMK 2-96, which partially stimulated the mMC1R (23% at 100 μM) and did not stimulate the mMC3R, mMC4R, or the mMC5R. ZMK 2-96 did not possess antagonist activity at the mMC3R (pA₂ < 5.0) and was a micromolar antagonist at the mMC4R (pA₂= 5.3) and hMC4R (pA₂= 5.8). Substituting the Phe112 with Tyr (rs200972106) resulted in ZMK 2-110, which was unable to stimulate the mMC1R, mMC3R, or mMC4R at the concentrations assayed, but possessed inverse agonist activity at the mMC5R (−35% at 100 μM), a pharmacology previously observed using this scaffold.^38,39 ZMK 2-110 was a micromolar to sub-micromolar antagonist at the mMC3R, the mMC4R, and the hMC4R (pA₂= 5.8, 6.1, and 6.5 respectively). The last peptide synthesized using this scaffold was ZMK 3-18, which contained an Arg111 to Cys (rs1012110755) substitution. This peptide (ZMK 3-18) partially stimulated the mMC1R (46% at 100 μM), was unable to stimulate the mMC3R or mMC4R, and possessed inverse agonist activity at the mMC5R (−32% at 100 μM). ZMK 3-18 did not have antagonist activity at the mMC3R or mMC5R (pA₂ < 5.0) but was a micromolar antagonist at the mMC4R (pA₂=5.9) and at the hMC4R (pA₂=5.8). Binding affinities reflected the functional activity data and ranged from micro to nanomolar affinities, depending on the SNP incorporated, at all receptors assayed. Binding data is summarized in Tables 4 and 5.

SNP Incorporations, Dap-Substituted Scaffold:

Compound 2 (MDE 6-82-1c) incorporates the basic amino acid diaminopropionic acid (Dap) at the Asn⁵ position in the sequence c[-Pro-Arg-Phe-Phe-Asn⁵-Ala-Phe-DPro] and has been previously reported to be an equipotent antagonist to endogenous AgRP antagonist at the MC4R.⁷⁵ Due to the increased potency of 2, as compared to 1, it was hypothesized that the potential losses in potency from the SNPs might be better detected in compound 2 as it and AgRP have similar staring potencies than the less potent compound 1. This macrocycle (2) was able to partially stimulate the mMC1R (30%), was unable to stimulate the mMC3R and mMC4R within the assayed concentrations, and possessed inverse agonist activity at the mMC5R and hMC4R. Compound 2 possessed antagonist activity at the mMC3R, mMC4R, mMC5R, and hMC4R with pA₂ values of 6.5, 8.7, 7.3, and 8.6 respectively. Replacement of Ala115 (AGRP numbering) with Val resulted in the compound ZMK 2-85. This peptide (ZMK 2-85) was able to partially stimulate the mMC1R (79% of maximal NDP-MSH signal at 100 μM), the mMC3R (27%), and the mMC4R (37%), and was unable to stimulate the mMC5R. ZMK 2-85 also possessed antagonist activity at the mMC3R, mMC4R, mMC5R, and hMC4R (pA₂= 6.1, 7.5, 6.3, and 7.6 respectively). Substitution of the Arg111 (AGRP numbering) with a His residue produced ZMK 2-99, which partially activated the mMC1R (56% at 100 μM) and was unable to stimulate the mMC3R, mMC4R, or the mMC5R. ZMK 2-99 was a micromolar antagonist at the mMC3R (pA₂= 5.4) and mMC4R (pA₂= 6.0) and a sub-micromolar antagonist at the hMC4R (pA₂= 6.4). Converting the Phe at the 112 position (AGRP numbering) to Tyr, coupled with the Dap substitution resulted in ZMK 2-112. This peptide (ZMK 2-112) partially stimulated the mMC1R (48% @ 100μM), was unable to stimulate any other receptor assayed, and possessed sub-micromolar antagonist potency at the mMC3R, mMC4R, mMC5R, and hMC4R (pA₂= 6.5, 7.3, 6.5, and 7.5 respectively). The last peptide contained an Arg111 (AGRP numbering) to Cys substitution (ZMK 3-20). ZMK 3-20 was the only peptide that acted as a full agonist at the mMC1R, stimulating the receptor to 98% of the NDP-MSH control, with an EC₅₀ of 4000 ± 2000 nM. ZMK 3-20 partially stimulated the mMC3R (29% @ 100μM) and mMC4R (42% @ 100μM) but was unable to stimulate the mMC5R or the hMC4R up to 100 μM concentrations. This peptide did not have any apparent antagonist activity (pA₂ <5.0) at the mMC3R, the mMC4R, or the hMC4R at the highest concentrations of antagonists assayed (10μM). All Dap-containing peptides, in general, possessed more potent binding IC₅₀ affinities at the receptors assayed compared to their Asn-containing counterparts, which mirrored the cAMP functional data. Binding IC₅₀ affinities are reported in Tables 4 and 5.

Discussion and Conclusions

This work, for the first time, explored human single nucleotide polymorphisms resulting in missense mutations in the purported binding loop of the endogenous AgRP antagonist that were identified from deposited information into the NIH Variation Viewer database. The protein AgRP antagonizes the MC3R and the MC4R in the central nervous system, in addition to functioning as an inverse agonist at the MC4R. These functions prevent the endogenous agonists from binding in the orthosteric receptor domain, which prompts an increase in food intake when administered in vivo. AgRP has been reported to have varied pharmacology at the MC5R,^18,19,37 but any potential physiological roles of these data have not been reported. The database SNPs examined herein have not been previously explored to investigate their potential for altering cellular signaling and functional relevance at the MCRs. To examine the potential impact these SNPs have on cAMP signaling at the MCRs, they were incorporated into the octapeptide macrocyclic scaffold c[Pro-Arg-Phe-Phe-Xaa-Ala-Phe-DPro], where Xaa is either Asn or Dap. This scaffold was previously reported to possess nanomolar potency (Asn) or equipotent potency (Dap) to AgRP (87-132) at the mMC4R.⁸⁰ This scaffold was chosen for this study to incorporate the four polymorphisms due to its potency and synthetic amenability.

All compounds, except ZMK 2-110, were able to partially or fully stimulate the mMC1R, consistent with prior work published utilizing this scaffold.^80,84 All compounds tested resulted in at least a 10-fold decreased antagonist potency at the mMC4R and hMC4R, as compared to the respective control compounds 1 and 2 (Tables 2 and 3), suggesting that SNPs may impact AgRP function. The greatest loss of activity was seen in compounds with substitutions at the Arg position, decreasing potency by at least a 150-fold relative to the parent control compounds. The Arg position has been previously studied in the Arg-Phe-Phe pharmacophore of AgRP.²¹ In a structure activity relationship study published on the homologous agouti (ASP) protein, which shares the Arg-Phe-Phe motif, Kiefer et al. reported the conclusion that charge contributed to activity at the Arg position, as binding increased as more positive substitutions were used Lys>His>Gln>Ala.²² The His substitution may maintain a potential positive charge (depending upon local environmental pH), resulting in decreased potency. In the Arg to Cys polymorphisms, the Cys substitution may be disrupting the native disulfide bonds by introducing new disulfide bonds. Ten cysteine residues are found naturally in the carboxyl end of AgRP, which form five bonds, and create a “cysteine knot” structure.⁷⁸ Disruption of this structure could potentially alter the conformation of the protein and decrease its ability to bind properly, although the current scaffold does not permit the examination of this hypothesis, since it does not replicate the cysteines of the C-terminal domain. The substitution of Tyr at the Phe112 position exhibited a 110-fold loss in the Asn scaffold and a 25-fold loss in the Dap scaffold, as compared to their respective parent compounds. These findings are consistent with previously reported structure-activity studies using this scaffold, where either a complete loss in antagonist potency or a decrease in potency from the native sequence at the MC4R was described at the Phe112 position when unnatural aromatic amino acids were substituted, with the exception of a Nal(1’) substitution, which maintained the potency of the parent compound.⁸⁰ The findings presented herein recapitulate the hypothesis in the field that the Arg-Phe-Phe sequence is critical for orthosteric binding and activity as previously reported.^20-22 Lastly, the Ala115 containing macrocyclic peptides resulted in 31- and 15-fold decreased antagonist potency. This position appears to be more amenable to substitutions, albeit resulting in a decreased response relative to control compounds. It has been postulated, based upon GPCR homology molecular modeling, that the ligand Ala side chain is juxtaposed a histidine receptor residue of the MC4R (H264, TM6), when AgRP is bound.⁸⁶ Previously reported ligand structure-activity studies have postulated that hydrophobic residues and polar residues (specifically Ser) substituted at the Ala115 position maintain at least micromolar antagonist potency (Ser exhibiting nanomolar potency) compared to the parent compound, but acidic residues Glu and Asp ablated antagonist activity at the MC4R.³⁸ Modeling these SNP containing peptides herein, illustrate the apparent loss of two critical interactions between the Arg wild-type residue within AgRP and Glu100 and Asp126 residues in the MC4R, Figure 3. See Experimental for full description of modeling and ligand docking.

Figure 3.

Illustration of the Dap macrocyclic peptide scaffold containing the SNPs examined herein, docked into a GPCR homology molecular model of the human (hMC4R PDB ID:4z35-based model). SNPs are shown in green, while wild-type residues are shown in blue, interactions with hMC4R are shown in yellow, atom type is indicated with standard colors (O is red, N is blue, S is yellow). A) The wild-type Arg, with polar contacts shown in yellow dashes between Arg and Glu100/Asp126. B) The Arg to Cys SNP that possessed the greatest shift in potency. C) The His to Arg SNP containing peptide. D) The Phe to Tyr SNP containing peptide. E) The Ala to Val SNP containing peptide.

Many of the compounds were inactive when assayed as agonists at the mMC5R, apart from ZMK 2-110 and ZMK 3-18, which demonstrated inverse agonist activity, similar to the parent compounds as previously reported.⁸⁰ Based upon the results from the binding studies (Table 4), demonstrating compounds that bound to the MC5R with micromolar binding IC₅₀ values, the 1* (MDE 6-82-3c), ZMK 2-82, ZMK 3-18, 2* (MDE 6-82-1c), ZMK 2-85, and ZMK 2-112 ligands were tested for MC5R antagonist pharmacology. The ZMK 2-82, ZMK2-85, and ZMK2-112 peptides possessed micromolar antagonist potencies at the mMC5R (pA₂= 6.1, 6.3, and 6.5, respectively), suggesting that if AgRP has functional activity at the MC5R, these SNP may lead to altered MC5R signaling, Figure 4.

Figure 4.

Illustrations of the in vitro antagonist pharmacology of select orthosteric compounds at the mMC5R. The data has normalized to basal and NDP-MSH values in the cAMP accumulation assay. A Schild antagonist experimental design was applied and the agonist NDP-MSH was utilized in these experiments. Data points represent values from at least three independent experiments and the error bars represent the standard error of the mean (SEM).

This scaffold may therefore also serve as a lead in the further development of selective and potent MC5R antagonist ligands, which are lacking in the field to date. While the exact physiological role(s) of the ubiquitously expressed MC5R has not been elucidated, it has been linked to exocrine gland function in mice.¹⁴ Additionally, the MC5R has been investigated for its role in muscle glucose uptake⁸⁷ and as a treatment for acne (reviewed by Zhang et al.).⁸⁸ Non-selective MC5R antagonists have previously been reported, including HS024 that inhibits α-MSH-mediated cAMP stimulation at the MC1R, MC3R, MC4R, and MC5R when administered at 100 nM concentrations.⁸⁹ Many ligands have been reported to selectively bind to the MC5R,^90,91 although functional antagonist studies were not reported for these compounds. A peptidomimetic macrocycle (PG20N) was reported to be an MC5R antagonist (pA₂= 8.3) that did not activate the MC1R and MC4R at concentrations up to 10 μM, and was reported to possess partial agonist activity at the MC3R (EC₅₀ = 50 nM; 21% the effect of MTII).⁹² PG20N was derived from the synthetic melanocortin agonist MTII and antagonist SHU9119, while the library in the current study was based upon an AgRP template. Utilizing a different starting scaffold may allow the development of more potent and selective MC5R antagonist ligands that can be used to help determine the physiological roles of the MC5R in vivo.

The two compounds that were the most active at the mMC1R (Figure 5), ZMK 2-85 and ZMK 3-20, were also able to partially stimulate the mMC3R and mMC4R (27/37% and 29/42% of the maximal NDP-MSH signal respectively).ZMK 2-85 possessed antagonist pharmacology at the mMC3R, mMC4R, and mMC5R (pA₂= 6.1, 7.5, and 6.3 respectively). Compounds with this activity profile have been previously reported for the melanocortin system.⁹³ Interestingly, ZMK 3-20 had no antagonist activity (pA₂ < 5.0) at the micromolar concentrations assayed at either the mMC3R or the mMC4R or the hMC4R but was a full micromolar agonist (4 ± 2 μM) at the mMC1R and partially stimulated the mMC4R (42% of NDP-MSH at 100 μM). The Arg to Cys mutation, when incorporated with the basic residue Dap, produced a melanocortin ligand with unique pharmacology as compared to the rest of the library examined herein, which may be an important lead for studying the nuances of each receptor subtype and generating completely selective ligands.

Figure 5.

Illustration of the agonist pharmacology observed for ZMK2-85, ZMK3-18, and ZMK3-20 at the mMC1R, as compared to the control agonist NDP-MSH. Data points represent values from three independent experiments and error bars represent the standard error of the mean (SEM).

From the findings in this study, it may be speculated that individuals with these polymorphisms express a form of AgRP that results in decreased MC4R potency, which may be ineffective at stimulating hunger, potentially resulting in a lean phenotype; however, the extent of any physiology consequences is not currently known and would need to be further experimentally verified. The scaffold used was an AgRP mimetic, therefore full-length AgRP incorporating the SNPs would need to be synthesized before concluding any physiologically based hypotheses relating the human SNPs and molecular function. There are many redundant mechanisms that nature has established to maintain energy homeostasis and satiety. For example, it has been reported that animals with genetic modifications preventing them from producing AgRP, NPY, or both, still maintain a normal weight, postulated to a result of compensatory pathways involved in the regulation of feeding and satiety homeostasis.⁹⁴ Gropp et al. proposed this was likely due to the maintenance of the AgRP/NPY producing neurons in these animals, allowing for alternate compensatory pathways. When the neurons were ablated using diphtheria toxin, animals exhibited the expected phenotype within 24 hours, demonstrating that the AgRP/NPY neurons are mandatory for feeding behavior.⁹⁵ However, these experiments represent situations where the signal (AgRP) is absent through neuronal ablation or genetic knockdown. In the case of the SNPs, the signal (AgRP) can transmit to the recipient (MC4R/MC3R), but at a potentially decreased potency. It is unknown what the human phenotype of this circumstance would be. It was apparent from the studies performed herein, that these polymorphisms altered the melanocortin receptor pharmacology from the endogenous sequence when tested in a live cell cAMP based functional assays and binding affinity assays and therefore should be further investigated in patients.

This study reports the first incorporation of deposited human SNPs of the binding loop of AgRP into a functional macrocyclic scaffold to determine the physiological consequences of the polymorphisms, as well as describes the first AgRP-derived mMC5R peptide antagonists. All compounds assayed demonstrated a decrease in antagonist potency at the MC4R compared to parent ligands of the wild-type sequence or the more potent Asn⁵Dap substitution scaffolds examined. These polymorphisms may represent a functional change when incorporated into the full-length AgRP, potentially altering AgRP signaling in individuals possessing these SNPs. At the MC1R, one compound did not possess agonist activity (ZMK 2-110), one ligand was a full agonist (ZMK 3-20), and the others partially stimulated it at 100 μM concentrations. Incorporation of the Dap residue at the 114 position (AGRP sequence numbering) did not universally increase MC4R potency (see ZMK 3-20) like previously reported.⁸⁴ Additionally, both SNPs that change Arg111 (to either His or Cys) saw the greatest decrease in potency, followed by the Phe112Tyr substitution, recapitulating the hypothesis in the field that the Arg-Phe-Phe(111-113 AGRP numbering) motif is critical for activity. Further phenotypic characterization of individuals possessing these SNPs will be important in determining the relevance of AgRP polymorphisms in the broader context of modulated energy homeostasis.

Experimental

Peptide Synthesis:

Peptides were synthesized using standard Fmoc chemistry.^82,83 Coupling reagents 2-(1-H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate (HBTU), benzotriazol-1-yl-oxy-tris(dimethylamino) phosphonium hexafluorophosphate (BOP), and 1-hydroxybenzotriazole (HOBt), the H-Pro-2-chlorotrityl resin, and Fmoc-protected amino acids were purchased from Peptides International (Louisville, KY). Dichloromethane (DCM), methanol (MeOH), acetonitrile (ACN), dimethylformamide (DMF) and anhydrous ethyl ether were purchased from Fisher (Fairlawn, NJ). Trifluoroacetic acid (TFA), dimethyl sulfoxide (DMSO), piperidine, triisopropylsilane (TIS), and N,N-diisopropylethylamine (DIEA) were purchased from Sigma-Aldrich (St. Louis, MO). All reagents and chemicals were ACS grade or better and were used without further purification.

Peptides were synthesized on a 0.05 mmol scale using H-Pro-2-chlorotrityl resin (0.76 meq/g substitution) using a manual microwave synthesizer (CEM Discover SPS, Matthews, NC). Amino acids were added from carboxyl to amino terminal with two repeated steps consisting of Fmoc removal and amino acid coupling. Fmoc groups were removed with 20% piperidine in DMF for 2 min at room temperature and additionally for 4 min using microwave irradiation at 75°C at 30W. Incoming amino acids (3.1 eq) were installed with HBTU (3 eq) and DIEA (5 eq) with microwave irradiation at either 50°C (His or Cys) or 75°C for 5 min at 30W. Arginine couplings required 5.1 eq of amino acid, 5 eq HBTU, and 7 eq of DIEA and an additional 5 minutes of irradiation time (10 min total). Deprotection and coupling steps were separated by at least four DMF washes and the ninhydrin⁹⁶ or chloranil (for Proline )⁹⁷ colorimetric assays were used to monitor reaction progress. After synthesis, peptides were cleaved from the resin, maintaining side-chain protecting groups using a 99:1 solution of DCM: TFA for a total of 6 min. The cleavage solution was concentrated under N₂ (g) and peptides were precipitated using ice-cold ethyl-ether. Side-chained protected peptides were cyclized head to tail forming an amide bond through the Arg and Pro residues, in DCM using BOP (3 eq) and HOBt (3 eq) at a peptide concentration of 0.33 mg/mL overnight. The DCM was subsequently removed under vacuum. Without further purification, side chain protecting groups were removed using a solution of a 95:2.5:2.5 solution of TFA:H₂O:TIS for 2 h. Peptide solution was concentrated under N₂ (g) and cyclized peptide was precipitated using ice-cold ethyl-ether. Peptides were purified using RP-HPLC (Shimadzu) configured with a UV detector and a semi-preparative C18 bonded silica column (Vydac 218TP1010, 1 × 25 cm). Peptides were determined to have >95% purity before being assayed. Purity was determined by analytical RP-HPLC (Shimazdu) configured with a photodiode array detector and an analytical C18 silica column (Vydac 218TP104, 0.46 × 25 cm) in two different solvent systems. Molecular mass was determined by MALDI-MS or ESI-TOF/MS (Applied Biosystems-Sciex 5800 MALDI/TOF/TOF-MS, Bruker BioTOF II ESI-TOF/MS, University of Minnesota Mass Spectrometry Lab).

cAMP AlphaScreen® Bioassay:

Compounds were dissolved in DMSO, apart from NDP-MSH, which was dissolved in H₂O, to a stock concentration of 10⁻² M. The compounds were characterized in HEK293 cells stably expressing the mMC1R, mMC3R, mMC4R, mMC5R, and hMC4R using a cAMP AlphaScreen® assay (Perkin Elmer), following the manufacturers’ instructions and as previously described.^80,98 As the MC2R can only be stimulated by ACTH, it was excluded from this study.^1,9 Peptides were assayed first as agonists at the four mouse receptors. Compounds that did not elicit a full agonist response were then assayed as antagonists at the mouse MC3R, MC4R, and MC5R, and the human MC4R. Experiments were performed three independent times performed in duplicate.

To summarize, cells that had reached 75-85% confluency were removed with Versene (Gibco®) and plated at a concentration of 10,000 cells per well in a 384-well plate (Optiplate™), with 10 μL solution of prepared stimulation buffer (1X HBSS, 5 mM HEPES, 0.5 mM IBMX, 0.1% BSA, pH= 7.4) and 0.5 μg of anti-cAMP acceptor beads. 5 μL of peptide containing solutions (concentrations ranging from 10⁻⁴ to 10⁻¹³, ligand potency dependent) was added to the wells to stimulate the cells and cells were incubated for 2 hours in the dark at room temperature. To determine the antagonist activity, cells were incubated with both the potent MCR agonist NDP-MSH (concentrations ranging from 10⁻⁶ to 10⁻¹²) and peptide (concentrations of 10,000 nM, 5,000 nM, 1,000 nM, and 500 nM), using a Schild paradigm.⁹⁹

Following the incubation period, a 10 μL solution containing: 0.5 μg streptavidin coated donor beads, 0.62 μmol biotinylated-cAMP tracers, and lysis buffer (5 mM HEPES, 0.3% Tween-20, 0.1% BSA, pH= 7.4) was added under green-light and incubated for an additional 2 h in the dark and at room temperature. Plates were read on an EnSpire® (Perkin Elmer) using the 384-Alpha plate pre-normalized assay protocol established by the manufacturer.

¹²⁵I-NDP-MSH Competitive Binding Assay:

Peptide macrocycles were dissolved in DMSO and assayed using stably transfected HEK293 cells at the mMCRs and the hMC4R. The MC2R was excluded from this study. The resulting dose-response binding curves were used to determine the reported IC₅₀ values, using a protocol previously described.¹⁰⁰

Briefly, cells were grown to >90% confluency in a 12-well polystyrene plate (Corning Life Sciences). Culture media was removed, and cells were incubated for 1 hour at 37°C, 5% CO₂ with 500 μL solution of 0.1% BSA in Dulbecco’s Modified Eagle Medium (DMEM) containing the peptide macrocycles (concentrations ranging from 10⁻⁴ to 10⁻¹⁰ M) and a constant of 100,000 cpm/well ¹²⁵I-NDP-MSH. Post incubation, the media was removed, and the cells were washed and lysed with 500 μL of 0.1 M NaOH and 500 μL of 1% Triton X-100. Cells were incubated with the lysis solution for 10 minutes at room temperature before being transferred to 12mm x 75 mm polystyrene tubes. Radioactivity was quantified using a WIZARD2 automatic gamma counter (PerkinElmer). Specific binding was determined using nonradioactive NDP-MSH as a positive control, and the specific binding for compounds was normalized to 100% relative to NDP-MSH. Experiments were performed twice with duplicate wells in each experiment.

Data Analysis:

The EC₅₀ and pA₂ values obtained from the AlphaScreen™ are representative of the means of duplicates assayed in at least three independent experiments. Compounds that did not possess agonist activity at the mMC3R, mMC4R, or mMC5R were screened as antagonists at these receptors and at the hMC4R. The EC₅₀ and pA₂ means and standard errors of the means (SEM) were determined using a non-linear least-squares analysis using PRISM software (v4.0, GraphPad Inc.) fit to the collected data. All peptides were assayed as TFA salts and not corrected for peptide weight. Inverse agonism was determined as a percent change from basal compared to the maximal signal of NDP-MSH at 100 μM. The IC₅₀ values obtained from the ¹²⁵I-NDP-MSH competitive binding assays are representative of the means of duplicates assayed in two independent experiments. The IC₅₀ means and standard deviations (SD) were determined using a non-linear least-squares analysis using PRISM software (v4.0, GraphPad Inc).

Homology models of human MC4R with peptide-based antagonists:

In the absence of crystal structures of melanocortin receptors (MCRs), homology modeling remains a viable alternative for structural analysis of MCR-ligand complexes. In 2005 the first homology model was generated of the inactive conformation of hMC4R in complex with the C-terminal region of hAGRP (PDB ID:2iqv, 2iqr) ⁸⁶ using distance geometry calculations by DIANA. ¹⁰¹ The model was based on crystal structures of bovine rhodopsin (PDB ID:1gzm) ¹⁰², which has only ~15% sequence similarity to hMC4R, the NMR structure of hAGRP (PDB ID: 1hyk, residues 87-132) ¹⁰³, and experimentally-derived structural restraints.⁸⁶ Our subsequent modeling by distance geometry calculations in 2009 ¹⁰⁴ was based on the beta-2 adrenergic receptor structure (PDB ID: 2rh1) that has higher (~30%) sequence similarity to hMC4R. Both 1gzm- and 2rh1-based models were rather similar with rmsd 2.2 Å for 240 Cα-atoms, though they demonstrated some helix shift and deviations in intracellular (IL) and extracellular (EL) loop conformations. Structures of ILs generally followed the corresponding templates, EL2 represented just a short interhelical link in extended conformation, EL3 was restricted by two disulfide bridges connecting EL3 to TM6 (C271-C277) and to N-terminus (C40-C279) ⁸⁶, while EL1 was either unfolded (in 2rh1-based model) or folded into a β-hairpin (in 1gzm-based model).

At present, these models can be substantially updated due to the progress in structural determination of GPCRs and advances in computational methods for homology modeling. Currently, ~50 crystal structures of GPCRs are available and can be used for homology modeling by automated servers, such as SWISS-MODEL ¹⁰⁵, Phyre2 ¹⁰⁶, RaptorX ¹⁰⁷, M4T ¹⁰⁸, and IntFOLD. ¹⁰⁹ Comparison of results of automated modeling of hMC4R by these servers demonstrated that crystal structures of human lysophospatidic acid receptor 1 (PDB ID: 4z35) ¹¹⁰ and of the human sphingosine 1-phosphate receptor 1 (PDB ID: 3v2y) ¹¹¹ likely represent the optimal templates, as they are evolutionary closer to MCRs with the highest (~34%) sequence similarity to hMC4R (27-29% sequence identity) and, unlike other GPCR structures, lack the conserved Pro in the TM5, which is also absent of MCRs. This evolutionarily conserved Pro from TM5 produces a helical aneurism (an α-bulge with 5 residues per turn), which affects the position of the extracellular part of the TM5 and the conformation of the short EL2.

The latest modeling of the hMC4R (UniProtKB ID: P32245, residues 40-317) in the inactive conformation was based on the refinement of the 4z35-based model of hMC4R generated by SWISS-MODEL server. ¹⁰⁵ The particular focus was made on the reformation of ELs, as these loops enclose the ligand binding pocket and likely participate in receptor-ligand interactions.^112,113 The procedure included remodeling of EL3 to satisfy disulfide bridge restraints ⁸⁶ and of the EL1 to exclude helical bulges and interference with N-terminus attached to EL3 via disulfide, as well as the adjustment of side chains to produces their allowed conformations that lack mutual hindrances. The obtained model was refined by energy minimization (100 steps) with CHARMm force field implemented in QUANTA (Accerlys) using a dielectric constant (ε) of 10 and the adopted-basis Newton-Raphson method. After minimization the rmsd between hMC4R model and its structural template (4z35) was 1.6 Å for 208 equivalent Cα-atoms. The 4z35-based model slightly deviated from previous models: rmsd with 1gzm- and 2rh1-based models were 2.2 and 1.8 Å for 240 Cα-atoms, respectively.

Ligand docking:

The NMR structure of hAGRP (87-132) fragment ¹⁰³ was used for docking into the new 4z35-based model of hMC4R. The structure of the octapeptide c[Pro-Arg-Phe-Phe-Dap-Ala-Phe-DPro] was produced from the central β-hairpin loop of hAGRP (¹¹⁰CRFFNAFC¹¹⁷) by modifying C110 to L-Pro, C117 to D-Pro, and N114 to Dap, followed by Pro-DPro cyclization. The model of the octapeptide was refined by energy minimization with QUANTA/CHARMm. The docking of hAGRP (87-132) and compound 2 octapeptide into 4z35-based model of hMC4R was done by superposition of the β-hairpin central loop with its position in previous models followed by manual adjustment of ligand position and side chain orientations to avoid substantial hindrances with receptor residues and to satisfy experimentally-derived ligand-receptor interactions.^86,112-114 The ligand docking pose was subsequently refined using the solid docking module of QUANTA. Finally, the receptor-ligand complexes were shortly minimized using QUANTA/CHARMm (50 steps, using ε of 10 and the adopted-basis Newton-Raphson method).

Ligand docking mode:

The central loop (110-117) of hAGRP (87-132) and loop-derived octapeptide contain key residues Arg111, Phe112, and Phe113, which are responsible for antagonist properties of these ligands.^80,115 Therefore, positions and interactions of these residues with receptor are critical for the efficient binding of AGRP-derived ligands and locking the receptor in the inactive conformation. In the proposed hMC4R-ligand models, Phe-Phe motif of the ligand display trans side chain configuration, which agrees with impaired activity of Tic-substituted analogues of these residues restricted in the gauche configuration of the aromatic ring.⁸⁰ Observed interactions of the central loop residues are in perfect agreement with experimental data ^86,114,116: Arg111 forms ionic interactions with receptor D126 and E100; Phe112 form hydrophobic interactions with I129, F261, L265, and F284; Phe113 is in contact with F184; Phe116 forms hydrophobic interactions with F284 and M281. Remarkably, Asn114 from the central loop of AGRP is spatially close to D189 from the EL2, while corresponding Dap of octapeptide forms H-bond and ionic interactions between its amine group and carboxyl group of D189, Figure 6. Besides, patch of positively charged residues (Arg120, Lys121, Arg131 ) from the C-terminal loop are spatially close and may form ionic interactions with EL1 (D111, D113) in compliance with experimental data ¹¹³, while residues from the N-terminal loop forms multiple interactions with EL2 and EL3, supporting experimental observations. ¹¹²

Figure 6.

Models of complexes of hMC4R (PDB ID: 4z35) with hAGRP(87-132) (A, B) and compound 2 (MDE 6-82-1c) octapeptide (C). (A) Ionic interactions between positively charged residues (shown by blue sticks with dots) from central and C-terminal loops of hAGRP(87-132) with patch of negatively charged residues (shown by red sticks) from TM2-EL1-TM3 of hMC4R; (B, C) Binging pocket residues of hMC4R (shown by yellow sticks with O-atoms colored red) that interacts with residues from central loop of AGRP (shown by blue sticks) in accordance with mutagenesis data ^{86, 112-114}. TM helices and extracellular loops (ELs) of hMC4R is shown by cartoons colored grey, ligands are shown by cartoon and sticks colored blue, five disulfide bridges in hAGRP are shown by blue sticks with S-atoms colored yellow.

Abbreviations Used:

ACTH

Adrenocorticotropin Hormone

AgRP

Agouti-Related Protein

ASP

Agouti Signaling Protein

GPCR

G-protein Coupled Receptor

MCR

Melanocortin Receptor

MC1R

Melanocortin 1 Receptor

MC3R

Melanocortin 3 Receptor

MC4R

Melanocortin 4 Receptor

MC5R

Melanocortin 5 Receptor

m

Mouse

h

Human

Fmoc

9-fluorenylmethoxycarbonyl

cAMP

Cyclic 5’-Adenosine Monophosphate

MSH

Melanocyte Stimulating Hormone

POMC

Proopiomelanocortin

NDP-MSH

(4-Norleucine-7-D-Phenylalanine)-Ac-Ser-Tyr-Ser-Nle-Glu-His-DPhe-Arg-Trp-Gly-Lys-Pro-Val-NH₂

Dap

Diaminopropionic acid

RP-HPLC

Reverse Phase- High Performance Liquid Chromatography

DMEM

Dulbecco’s Modified Eagle Medium; SEM, Standard Error of the Mean; SD, Standard Deviation