Discovery of Ghrelin(1–8) Analogues with Improved Stability and Functional Activity for PET Imaging

Abstract

The highest affinity ghrelin-based analogue for fluorine-18 positron emission tomography, [Inp¹,Dpr³(6-FN),1Nal⁴,Thr⁸]ghrelin(1–8) amide (1), has remarkable subnanomolar receptor affinity (IC₅₀ = 0.11 nM) toward the growth hormone secretagogue receptor 1a (GHSR). However, initial in vivo PET imaging and biodistribution of [¹⁸F]1 in mice demonstrated an unfavorable pharmacokinetic profile with rapid clearance and accumulation in liver and intestinal tissue, prompting concerns about the metabolic stability of this probe. The aims of the present study were to examine the proteolytic stability of ghrelin analogue 1 in the presence of blood and liver enzymes, structurally modify the peptide to improve stability without impeding the strong binding affinity, and measure the presently unknown functional activity of ghrelin(1–8) analogues. The in vitro stability and metabolite formation of 1 in human serum and liver S9 fraction revealed a metabolic soft spot between amino acids Leu⁵ and Ser⁶ in the peptide sequence. A focused library of ghrelin(1–8) analogues was synthesized and evaluated in a structure–activity–stability relationship study to further understand the structural importance of the residues at these positions in the context of stability and receptor affinity. The critical nature of l-stereochemistry at position 5 was identified and substitution of Ser⁶ with l-2,3-diaminopropionic acid led to a novel ligand with substantially improved in vitro stability while maintaining subnanomolar GHSR affinity. Despite the highly modified nature of these analogues compared to human ghrelin, ghrelin(1–8) analogues were found to recruit all G protein subtypes (Gα_{q/11/13/i1/oB}) known to associate with GHSR as well as β-arrestins with low micromolar to nanomolar potencies. The study of these analogues demonstrates the ability to balance desirable ligand properties, including affinity, stability, and potency to produce well-rounded candidate molecules for further in vivo evaluation.

Peptides are firmly established as medicinally relevant biomolecules owing to their moderate molecular mass, high target affinity, selectivity, sufficiently rapid clearance from blood and non-target tissues, and ease of synthesis.¹ As such, peptides have been utilized in the treatment and diagnosis of various diseases including a multitude of cancer types.²⁻⁴ Molecular imaging affords the ability to visualize and track disease states in a minimally invasive manner. Over the past 20 years, peptides have been developed into imaging agents through incorporation of radionuclei for imaging and targeted therapy. The high target affinity and specificity of peptides as radiopharmaceuticals allows rapid accumulation of the tracer at the target site and provides particular utility toward the imaging and treatment of tumors that express high levels of peptide-binding receptors.^5,6 Despite these promising attributes of peptides, there are obstacles that limit their utility, particularly in their pharmacokinetic profile. Proteolysis is the predominant reason why natural peptides often demonstrate low metabolic stability resulting in a short plasma half-life, indicating rapid clearance from the bloodstream.^1,7 While the presence of proteases is ubiquitous throughout the body, degradation of peptides primarily occurs in tissues including the blood, liver, kidneys, and gastrointestinal tract.⁸ The metabolic rate of peptides is highly dependent on factors that stem from the nature of its chemical structure including size, charge, and lipophilicity.⁹ Most natural peptides tend to be highly hydrophilic and therefore have a tendency toward high kidney uptake and a strong preference for renal clearance. However, it is also possible for smaller (<1 kDa), more lipophilic peptides to be passively taken up in the liver, metabolized, and excreted through hepatobiliary elimination.⁹ Regrettably, these pitfalls can be detrimental to the pharmacokinetic profile and utility of peptide-based radiopharmaceuticals.^5,6

The growth hormone secretagogue receptor 1a (GHSR), also referred to as the ghrelin receptor, is one of several G protein-coupled receptors found to be differentially expressed in various diseases compared to healthy tissue.¹⁰⁻¹² When activated by its endogenous ligand, ghrelin, a 28 amino acid peptide, GHSR regulates several physiological processes, including appetite, growth hormone secretion, and energy homeostasis.¹³⁻¹⁵ Ghrelin action through GHSR has complex signaling pharmacology, through G protein dependent (Gα_q/11, Gα_12/13, and Gα_i/o) and independent (β-arrestin) mechanisms, and high basal constitutive activity.¹⁶ Due to this complexity of downstream signaling and the multitude of physiological functions, it is believed that specific signaling cascades and accessory regulatory proteins may modulate GHSR-regulated functional responses. For example, GHSR-coupled Gα_q/11-mediated calcium signaling is mainly related to growth hormone secretion, whereas β-arrestin signaling is linked with mitogenic activity and intracellular lipid storage.^17,18

The ghrelin–GHSR axis has been linked to numerous metabolic disorders, such as obesity, diabetes, and cachexia, as well as cancer.¹⁹⁻²³ The strong affinity and specificity of ghrelin for its target made it a valuable starting point for the development of imaging probes targeting the GHSR. However, like many natural peptides, ghrelin suffers from poor metabolic stability when administered in vivo.²⁴ Consequently, the development of ghrelin-based GHSR imaging agents not only needs to modify ghrelin to include a traceable imaging moiety, but also to improve its pharmacokinetic behavior and promote delivery to the targeted receptor. As such, it is prudent to investigate how structural modifications affect metabolic stability in addition to target affinity. Several early reports studied the structure–activity relationship (SAR) between ghrelin and the GHSR laying the groundwork for the development of ghrelin analogues as imaging probes.²⁵⁻²⁷ Since that time, several ghrelin mimetics have been developed as candidates for optical and nuclear imaging of this receptor target.²⁸⁻³⁸ In previous work, we reported a high-affinity fluorine-containing ghrelin analogue, truncated to only the first eight amino acids for ¹⁸F-PET imaging. The peptide [Inp¹,Dpr³(6-FN),1Nal⁴,Thr⁸]ghrelin(1–8) amide (1) displayed subnanomolar GHSR affinity (IC₅₀ = 0.11 nM), making it the strongest affinity ghrelin analogue reported to date.³⁶

The study that resulted in peptide 1 focused its efforts on investigating the SAR of ghrelin(1–8) at positions 1, 4, and 8 as well as the side-chain at position 3. Other positions in the peptide sequence remained untouched. Common strategies known to improve metabolic stability without hindering receptor affinity were employed, including amidation of the C-terminus and the linkage between Ser³ and the acyl side chain.^25,26 Positions 1 and 4 were modified to include unnatural amino acids, isonipecotic acid (Inp) and 1-naphthylalanine (1Nal), respectively, likely improving the stability further, particularly at the N-terminus, which would otherwise be exposed to enzymatic degradation from exopeptidases. Embedded in the hydrophobic side chain at position 3, a key location for receptor binding, is fluorine, which may be incorporated as fluorine-18 for PET imaging. Subsequent in vivo evaluation of [¹⁸F]1 in normal mice revealed rapid blood clearance and high tracer uptake in the liver and intestines.³⁹ A possible explanation for this observation is poor metabolic stability of the imaging probe. While peptide 1 offers substantially improved binding affinity over natural ghrelin, there has been little investigation into the proteolytic stability of this analogue. Despite the presence of multiple unnatural amino acids in peptide 1, other areas of the molecule remain susceptible to proteolytic cleavage. Such areas are referred to as metabolic soft spots, and they can be addressed through targeted structural modification. Metabolic stability of other peptidic GHSR ligands has been studied in vitro through incubations in various human media including serum, liver/kidney microsomes, and the liver/kidney S9 fraction for the purpose of sports drug testing applications.^40,41 Due to the rapid blood clearance observed for this probe and its propensity to be taken up in the liver, one objective of the present study was to examine the proteolytic stability of ghrelin analogue 1 in the presence of blood and liver metabolic enzymes to identify the presence of metabolic soft spots. As a result of this study, a metabolic soft spot between Leu⁵ and Ser⁶ was identified and served as the basis for the development of a focused library of novel ghrelin(1–8) analogues based on 1 for targeted improvement of proteolytic stability, while maintaining a strong binding affinity toward the GHSR.

Since GHSR activation can trigger multiple signaling pathways that contribute to various physiological functions, we further wished to investigate whether the novel ghrelin analogues retained functional activity. We also investigated functional selectivity of the ghrelin analogues by comparing their signaling output to that triggered by natural human ghrelin. Together, these studies aimed to obtain novel GHSR ligands with strong affinity, stability, and functional activity.

Results and Discussion

Metabolic Soft Spot in Ghrelin(1–8) Analogue 1

One of the drawbacks of using peptides as pharmaceutical agents is their tendency toward rapid blood clearance due to low proteolytic stability.⁹ The in vitro stability of peptide 1 in human serum was recently reported, revealing substantial proteolytic degradation and a half-life of 4.7 h.³⁹ In the present study, as an additional comparison, a control sequence of ghrelin(1–8), with minimal modification compared to the native sequence, was also evaluated by the same method. The peptide [Dpr³(n-octanoyl)]ghrelin(1–8) (2) (Figure 1) was synthesized, and its serum stability was analyzed. Peptide 2 was completely degraded after 6 h with a half-life of 1.8 h, confirming that our previous work modifying positions 1, 4, and 8, which give rise to the backbone of peptide 1, already positively impacted in vitro stability and did not warrant further investigation (Figure 2).³⁶ The use of an amide linkage at position 3 was maintained since it is well known that cleavage of the natural ester at this position occurs rapidly in the presence of proteases, which would complicate the analysis of observing metabolites resulting from peptide backbone fragmentation.²⁴ As a control, the chemical stability of both peptides 1 and 2 was evaluated and both were found to be chemically stable when incubated in buffer at 37 °C over 24 h (Figures S1 and S2), indicating that all degradation in the serum experiments was enzyme-mediated.

Figure 1.

Structures of ghrelin(1–8) analogues 1 and 2.

Figure 2.

Stability of peptides (A) 1 and (B) 2 in human serum over 24 h. (Part A is adapted from adapted from ref (39) with permission from The Royal Society of Chemistry)

The metabolic soft spot that led to proteolysis of peptide 1 was identified through metabolite analysis following a 24 h incubation of peptide 1 in human serum. Over the course of the experiment, four peaks corresponding to metabolites 3–6 were observed asynchronously by UHPLC–MS (Figure 3A) and identified by their mass-to-charge ratio (m/z) (Figure 4). Metabolite 3 was observed after 30 min and steadily increased over the next 5.5 h, after which its growth slowed (Figure 3B). The rate of formation of the other three metabolites 4–6 started to increase later in the experiment with metabolite 6 only observed in substantial quantities after 24 h. The pattern in which the metabolites arose suggests the possibility that metabolites 4–6 were formed from further proteolytic degradation of metabolite 3 instead of solely from the parent peptide 1.

Figure 3.

(A) Stacked UHPLC-UV chromatograms of peptide 1 metabolism in human serum at select time points. Asterisk (*) indicates parent peptide 1. UV peaks for metabolites 3–6 indicated with their m/z values obtained via mass spectrometry. (B) Quantification of metabolites 3–6 over time. AUC = Area under curve from selected ion chromatogram.

Figure 4.

Structures of identified metabolites produced from peptide 1 in human serum.

To confirm the identity of metabolites 3–6, each metabolite was independently synthesized and characterized via UHPLC–MS. The retention time for each synthesized metabolite matched that of the observed metabolites from the serum assay data (Figure S3). Additionally, the mass and UV absorbance data were also a match in all cases. This thereby confirmed the identity of the serum metabolites and identified the amide bond between Leu⁵ and Ser⁶ in peptide 1 as a critical metabolic soft spot.

In order to investigate the hepatic stability of peptide 1, an in vitro stability assay in the human liver S9 fraction was performed. Samples of both peptides 1 and 2 were incubated with S9 fraction media at a low concentration over the course of 4 h, and the amount of intact peptide was measured by UHPLC–MS at various time points. Peptide 1 was completely degraded after 4 h with a half-life of 1.1 h, which is nearly double the half-life of 0.57 h for peptide 2 (Figure 5A). Three proposed metabolites identified based on their m/z values indicate possible degradation from the C-terminus of peptide 1 with cleavage of amino acids Thr⁸ (7), Pro⁷ (8), and Ser⁶ (3). The amount present of metabolites 7 and 8 (Figure S4) appeared to decrease after the first hour of incubation, indicating that these metabolites were themselves being metabolized. Interestingly, one of these metabolites, which was present in the largest quantity by the end of the experiment (Figure 5B), was metabolite 3, also observed as a product of serum metabolism. This further supports the presence of a metabolic soft spot between Leu⁵ and Ser⁶ and that the enzymes that metabolize this peptide are not solely expressed in one specific organ.

Figure 5.

(A) Stability of peptides 1 and 2 in human liver S9 fraction over 4 h. (B) Quantification of metabolites 3, 7, and 8 over time from peptide 1. AUC = Area under curve from selected ion chromatogram.

Another consideration is the fact that an amide bond was used to conjugate the fluorine-bearing aryl group to the peptide backbone at position 3. Since this amide bond is also susceptible to hydrolysis in the same manner as peptide bonds, it is worthwhile to note that no cleavage of the fluorine-containing aromatic group was observed in our in vitro stability studies.⁵ This suggests that the [¹⁸F]6-fluoro-2-napthyl (6-FN) moiety in this class of ghrelin(1–8) analogues would likely not be lost during in vivo PET imaging.

Structure–Activity–Stability Study of Ghrelin(1–8) Analogues

Following identification of Leu⁵ and Ser⁶ in peptide 1 as a metabolic soft spot through both serum and hepatic stability investigation, a targeted library of ghrelin(1–8) analogues was developed in an effort to improve the in vitro stability while maintaining the strong binding affinity of this peptide. Currently, there is limited information describing the SAR between positions 5 and 6 in ghrelin and the targeted receptor. An alanine scan of ghrelin(1–14) showed negligible changes in IC₅₀ when alanine was substituted for Leu⁵ or Ser⁶ compared to the unmodified peptide sequence suggesting that modifications at these two positions may be well tolerated.²⁷ Analogues of peptide 1 were synthesized via standard Fmoc solid-phase peptide synthesis, purified by reverse-phase HPLC-MS, and characterized by UHPLC–MS.

The binding affinities of all analogues toward the GHSR were evaluated in a competitive radioligand-displacement binding assay against [¹²⁵I]-ghrelin(1–28) using HEK293 cells transiently transfected with GHSR-eYFP (Table 1). As controls, acylated human ghrelin(1–28) and peptide 1 were evaluated and found to possess binding affinity values of 2.48 and 0.09 nM, respectively, which are in agreement with published literature values.^36,39 Additionally, all analogues were evaluated in vitro for proteolytic stability in human serum and human liver S9 fraction, the results of which are summarized in Table 1.

Table 1. In Vitro Binding Affinity and Proteolytic Stability of Ghrelin(1–8) Analogues^b.

#	sequence	IC50 (nM)	Serum t1/2 (h)	S9 t1/2 (h)
	human ghrelin(1–28)	2.48	n.d	n.d
1	Inp-S-Dpr(6-FN)-1Nal-LSPT-NH2	0.09a	4.7a	1.08
2	GS-Dpr(n-octanoyl)-FLSPE-OH	n.d	2.0	0.57
9	Inp-S-Dpr(6-FN)-1Nal-lSPT-NH2	56.5 ± 6.5	>24	>4
10	Inp-S-Dpr(6-FN)-1Nal-LsPT-NH2	3.55 ± 0.25	>24	>4
11	Inp-S-Dpr(6-FN)-1Nal-lsPT-NH2	11.5 ± 0.3	>24	>4
12	Inp-S-Dpr(6-FN)-1Nal-LtPT-NH2	2.52 ± 0.32	>24	>4
13	Inp-S-Dpr(6-FN)-1Nal-Nle-SPT-NH2	1.50 ± 0.27	1.7	0.55
14	Inp-S-Dpr(6-FN)-1Nal-Cha-SPT-NH2	1.88 ± 0.67	2.4	1.18
15	Inp-S-Dpr(6-FN)-1Nal-LCPT-NH2	7.15 ± 0.18	<0.5	0.95
16	Inp-S-Dpr(6-FN)-1Nal-LTPT-NH2	0.62 ± 0.10	>24	1.25
17	Inp-S-Dpr(6-FN)-1Nal-L-Hse-PT-NH2	1.43 ± 0.08	>24	1.40
18	Inp-S-Dpr(6-FN)-1Nal-LNPT-NH2	1.90 ± 0.56	>24	>4
19	Inp-S-Dpr(6-FN)-1Nal-LQPT-NH2	1.21 ± 0.12	>24	2.03
20	Inp-S-Dpr(6-FN)-1Nal-L-Abu-PT-NH2	2.54 ± 0.65	>24	0.46
21	Inp-S-Dpr(6-FN)-1Nal-L-Dpr-PT-NH2	0.32 ± 0.03	>24	>4
22	Inp-S-Dpr(6-FN)-1Nal-L-S(N-Me)-PT-NH2	1.32 ± 0.58	>24	>4
23	Inp-S-Dpr(6-FN)-1Nal-β-homoL-SPT-NH2	2.11 ± 0.21	>24	1.48

^aRef (39).

^bn.d – not determined.

A common practice to improve metabolic stability of peptides through structural modification is by inverting the stereochemistry at specific sites within the peptide sequence. The use of d-amino acids in place of l-amino acids serves as an effective method to disrupt the ability for proteases to recognize and bind the ligand. Analogue 9, which substituted Leu⁵ with d-Leu, was completely stable in both stability assays. However, this modification led to a dramatically negative impact on binding affinity, producing an IC₅₀ value of 56.5 nM. Therefore, it appears that the stereochemistry at position 5 is critical for favorable binding toward the GHSR. A recent analysis of a cryo-electron microscopy structure of ghrelin bound to a GHSR-G_q complex showed a H-bond between the ghrelin main chain at Leu⁵ and Q302 of the receptor.⁴² It would not be unreasonable to then suggest that d-stereochemistry at this position may cause steric hinderance and negatively impact this interaction. In contrast, substitution of Ser⁶ to d-Ser (10) did not cause such a drastic reduction in binding affinity, indicating that stereochemistry at this position is not as vital. Rather, compound 10 resulted in an IC₅₀ value of 3.55 nM, which is comparable to that of natural ghrelin(1–28). Predictably, an amalgamation of the alterations made in analogues 9 and 10 resulted in analogue 11 with a binding affinity value between those two compounds (IC₅₀ = 11.5 nM). Furthermore, the use of d-Thr⁶ in place of serine (12) also resulted in a reduction in binding affinity similar to that of compound 10, indicating that stereochemistry at positions 5 and 6 within this peptide sequence should remain in the l configuration to preserve binding affinity despite the improved stability of the d-stereoisomers. Interestingly, modification at positions 5 and/or 6 with d-amino acids also resulted in the suppression of metabolites 7 and 8, which formed during incubation with liver S9 fraction. This indicates that structural modification at one position in the molecule may lead to enhanced stability of other nearby areas.

In maintaining appropriate l stereochemistry at position 5, two non-canonical amino acids, norleucine (Nle) and cyclohexylalanine (Cha), which both possess hydrophobic alkyl side chains, were substituted in place of Leu⁵ in an effort to maintain similar physicochemical properties. Both analogues 13 and 14 led to a mild reduction in binding affinity with IC₅₀ values of 1.50 and 1.88 nM, respectively. However, the surprising result was that both analogues also led to a reduction in both serum and hepatic metabolic stability despite their unnatural nature. It is possible that the side chains of Nle and Cha were similar enough to Leu allowing for sufficient or likely improved recognition by metabolizing proteases. Therefore, the focus of further optimization was narrowed down to position 6 and the peptide backbone in that area of the peptide.

To explore the structure–activity–stability relationship (SASR) at position 6, Ser⁶ was substituted with other polar, uncharged amino acids including cysteine (15), threonine (16), homoserine (Hse) (17), asparagine (18), and glutamine (19). Predictably, due to the propensity for cysteine to be oxidized in a biological environment, analogue 15 was found to possess a half-life in human serum of less than 30 min. However, the IC₅₀ value for this analogue was also substantially increased. Analogues 16–19 explored the chemical space occupied by the residue in position 6. Peptides 17–19 demonstrated IC₅₀ values in the low nanomolar range, which is still a reduction in binding affinity compared to the parent peptide 1. However, analogue 16 with Thr⁶ was found to possess subnanomolar affinity with an IC₅₀ value of 0.62 nM, which is only a slight reduction compared to peptide 1. Interestingly, all four of the analogues 16–19 resulted in substantial improvement in serum stability with half-lives over 24 h. This was surprising since modifications in these analogues involved natural amino acids, with the exception of 17. However, stability in the liver S9 fraction was not substantially improved in peptides 16 and 17. Analogue 19 demonstrated a half-life nearly twice as long as peptide 1 in the liver S9 fraction media and analogue 18 was confirmed to have a half-life of over 4 h. Unlike modifications using d-amino acids, some of these modifications at position 6 were unable to also suppress hydrolysis at the C-terminus in order to improve hepatic stability.

In addition to evaluating the impact of expanding the local chemical space at position 6, we also sought to explore the necessity of the serine hydroxyl group in this position. As such, analogue 20 substituted Ser⁶ with l-α-aminobutyric acid (Abu), which replaced the polar hydroxyl group with a non-polar methyl group. Removal of polarity on the position 6 residue resulted in a reduction in binding affinity with an IC₅₀ value of 2.54 nM, indicating that the serine hydroxyl plays at least a minor role in binding to the GHSR. In fact, this finding supports the cryo-electron microscopy structure of GHSR-bound ghrelin that showed Ser⁶ contributes to binding through a H-bond between the serine hydroxyl and E197 in ECL2.⁴² In contrast to removing polarity from this position, analogue 21 replaced Ser⁶ with Dpr⁶ (Figure 6), which placed a formal positive charge on this residue under physiological conditions. Remarkably, the binding affinity of this analogue was very similar to the parent peptide with an IC₅₀ value of 0.32 nM. Additionally, analogue 21 demonstrated full stability over 24 h in human serum, signifying that hydrolysis of peptide 1 in human serum is likely dependent on the presence of serine in position 6. Furthermore, the half-life of over 4 h in the human liver S9 fraction demonstrates the ability of this single modification at position 6 to also substantially suppress degradation from the C-terminus.

Figure 6.

Structure and in vitro data for ghrelin(1–8) analogue 21. (A) Structure; (B) IC₅₀ curve (IC₅₀ = 0.32 ± 0.03 nM) (n = 2); (C) human serum stability (t_1/2 > 24 h); (D) human liver S9 fraction (t_1/2 > 4 h).

In addition to modifying the side chain residues, modification to the peptide backbone can also result in improved proteolytic stability. To that end, N-methylation of the amide nitrogen between positions 5 and 6 resulted in analogue 22. While a reduction in binding affinity was observed compared to peptide 1, this analogue was also stable in both serum and liver S9 fraction media. A recent report suggested a weak interaction between this amide proton and the GHSR backbone, which may explain the reduction in affinity due to its removal.⁴³ Substitution of Leu⁵ to l-β-homoleucine (23) expanded the peptide backbone by an additional carbon, thus adding additional flexibility to the peptide. However, this negatively impacted the binding affinity and while serum stability was improved substantially, liver S9 fraction stability was only marginally improved.

Functional Characterization of Ghrelin(1–8) Analogues

The impact of highly modified ghrelin(1–8) analogues on the functional activation of GHSR is currently unknown. Since the SASR study revealed position 6 to be the most promising location for modification resulting in improved stability and strong binding affinity, we focused our investigation specifically on the potential impact of changes here on the recruitment of different G proteins and β-arrestins. Human ghrelin and peptide 1 as well as new compounds 20, 21, and 22 were evaluated to provide insight on the structural influence of this position on GHSR activation. These compounds were selected based on the different chemical properties at position 6. Peptides 1, 20, and 21 highlight examples of different chemical properties, polar uncharged, non-polar, and polar charged, respectively, at the position 6 side chain. Methylation of the backbone amide in peptide 22 provided insight on the impact caused by removing a potential hydrogen bond at this location.

Calcium signaling experiments were performed on GHSR expressing HEK293 cells to evaluate activation of the Gα_q/11 signaling pathway. Ghrelin and the peptides (1, 20, 21, and 22) tested were found to activate calcium signaling, exhibiting EC₅₀ values in the low nanomolar range (Figure 7). Peptides 1 (EC₅₀ = 8 nM) and 21 (EC₅₀ = 2 nM) showed improved potency over natural human ghrelin (EC₅₀ = 23 nM), albeit with lower efficacy. However, removal of polarity from the position 6 side chain (20) did cause a decrease in potency compared to peptide 1, resulting in an EC₅₀ value of 30 nM, similar to natural ghrelin (EC₅₀ = 23 nM). These findings are generally consistent with the observed trend in binding affinity for these compounds with one notable exception; methylation of the amide nitrogen at position 6 (22) resulted in a substantial drop in potency (EC₅₀ = 119 nM) compared to ghrelin and the other truncated ghrelin(1–8) analogues despite the fact that the binding affinity IC₅₀ of this analogue is similar to those of natural ghrelin and peptide 20. This suggests that peptide 22 may be a partial agonist. It appears that maintaining side chain polarity and a free amide backbone is important for potent calcium signaling.

Figure 7.

Ghrelin(1–8) analogues activate Gα_q/11 coupled calcium signaling in GHSR expressing HEK293 cells. Data are mean ± SEM of three independent experiments performed in duplicate (n = 3).

Recruitment of other G proteins (Gα_q, Gα₁₃, Gα_i1, Gα_oB) as well as β-arrestins 1 and 2, which are known partners for GHSR, was also evaluated (Figure 8) using a BRET assay on GHSR and G proteins or β-arrestins co-expressing HEK293 cells. All peptides were able to trigger recruitment of all G protein subunits and β-arrestins tested with no substantial difference compared to natural ghrelin. The obtained EC₅₀ values were in the low micromolar range for recruitment of G proteins (Figure 8A–D) and in the nanomolar range for β-arrestins (Figure 8E,F). It is well-established that GHSR has high basal constitutive activity regardless of the presence of its endogenous ligand ghrelin.⁴⁴⁻⁴⁸ The constitutive GHSR activity and change in the receptor conformation could potentially affect the exogenous ligand-induced G protein recruitment and thus the observed EC₅₀ values. Overall, here we show that ghrelin(1–8) analogues recruit all G-proteins and β-arrestins that could lead to unbiased downstream signaling and functional activity of GHSR with similar potency as ghrelin.

Figure 8.

Ghrelin(1–8) analogues trigger recruitment of G proteins and β-arrestins to GHSR expressed on HEK293 cells. (A) Gα_q, (B) Gα₁₃, (C) Gα_i1, (D) Gα_oB, (E) β-arrestin-1, and (F) β-arrestin-2. Data are mean ± SEM of at least three independent experiments performed in duplicate or triplicate (n ≥ 3).

In conclusion, the GHSR has been implicated as a potential biomarker for multiple disease states resulting in the development of several molecular probes derived from ghrelin. However, one of the key barriers to developing peptides as pharmaceutical agents is their propensity to undergo rapid enzymatic degradation through proteolysis. To better understand and improve the stability of a high-affinity GHSR ligand, peptide 1, we evaluated the in vitro stability of this ligand in human serum and human liver S9 fraction. The metabolites identified through these experiments revealed a novel metabolic soft spot between Leu⁵ and Ser⁶ in this peptide sequence. This metabolic weak point was investigated through a focused library of ghrelin(1–8) analogues, which possessed modifications to residues 5 and 6, as well as to the peptide backbone. Insight into the structural importance of the residues at these positions was elucidated. In particular, the critical nature of l-stereochemistry at position 5 and, to a lesser extent, position 6 was revealed. Additional exploration of the local chemical space at residue 6 concluded that this residue needed to remain small in size and contain a polar group in order to preserve the subnanomolar binding affinity. Peptide 21, which substituted Ser⁶ with Dpr, demonstrated substantial improvement in proteolytic stability in both human serum and human liver S9 fraction compared to parent peptide 1, which demonstrated the capability of a modification at position 6 to influence stability at the C-terminus of this octapeptide. The influence of these modifications at residue 6 on functional GHSR activation was also investigated. Coupling of GHSR to G protein α-subunits or to β-arrestins was assessed using BRET assays and activation of Gα_q/11 coupled calcium signaling was monitored. All peptides showed similar functional activity compared to natural ghrelin and did not show any considerable bias toward any of the pathways assessed here. These findings suggest that these truncated, highly modified ghrelin(1–8) analogues with increased metabolic stability may be useful tools to study the ghrelin receptor in vivo and mimic natural ghrelin activity. The study of these ghrelin(1–8) analogues demonstrates the ability to balance desirable ligand properties, including affinity, stability, and potency, to produce well-rounded candidate molecules for future in vivo evaluation as diagnostic or therapeutic agents.

Methods

General Information

All common solvents were obtained from Fisher Scientific. Amino acids, resins, and coupling agents were obtained from Chem-Impex and Aapptec. Human [¹²⁵I]-ghrelin (NEX388010UC) was purchased from Perkin Elmer. All other reagents were purchased from Sigma-Aldrich, Fisher Scientific, or Oakwood Chemicals and were used as received. Analytical and preparative reverse-phase LC–MS was performed on a system consisting of a Waters 600 controller, Waters prep degasser, and Waters MassLynx software. The UV absorbance was detected using a Waters 2998 Photodiode array detector. A preparative (Agilent Zorbax PrepHT SB-C18 Column 21.2 × 150 mm, 5 μm) or analytical column (Agilent Zorbax SB-C18 column 4.6 × 150 mm, 5 μm) was used. The solvent system ran gradients of solvent A, 0.1% trifluoroacetic acid (TFA) in acetonitrile (ACN), and solvent B, 0.1% TFA in MilliQ (18.2 mΩ·cm conductivity) water, over 10 min with a 5 min wash. A flow rate of 20 mL/min or 1.50 mL/min was used for preparative and analytical runs, respectively. Analytical LC–MS was performed using a Waters Inc. Acquity UPLC H-class instrument in combination with a Xevo QToF mass spectrometer. A Waters Acquity analytical column (UPLC BEH C18 2.1 × 50 mm, 1.7 μm) was used and the UV absorbance was detected using a Waters Acquity PDA detector. The solvent system ran gradients of solvent A, 0.1% formic acid in ACN (Optima grade, Fisher Scientific), and solvent B, 0.1% formic acid in MilliQ water (18.2 mΩ·cm conductivity), at a flow rate of 0.6 mL/min over 3 min followed by a wash over 1 min. Solution phase reactions were monitored by thin layer chromatography (TLC) using aluminum-backed silica gel TLC plates. High-resolution mass spectra were determined in positive ion mode using an electrospray ionization (ESI) ion source on either a Waters Xevo QToF or a Bruker micrOToF II mass spectrometer.

General Fmoc Solid Phase Peptide Synthesis

Peptides were synthesized by standard Fmoc solid-phase peptide synthesis on a Biotage Syrowave automated peptide synthesizer. Briefly, peptides were synthesized at a 0.1 mmol scale on Rink amide MBHA resin (0.39 mmol/g). The resin was initially swelled in dichloromethane (DCM, 4 mL) followed by Fmoc deprotection using 40% piperidine in N,N′-dimethylformamide (DMF) (1.2 mL) for two cycles (3 min, 12 min). Amino acid coupling was completed by adding the appropriate Fmoc-protected amino acid (4 equiv) in DMF, HCTU (4 equiv) in DMF, and N,N′-diisopropylethylamine (DIPEA) (8 equiv) in N-methyl-2-pyrrolidone (NMP) to the resin and reacting for 40 min. The cycle of Fmoc deprotection followed by amino acid coupling was repeated until all amino acids were coupled to the resin. All further synthetic modifications were performed manually.

Allyloxycarbonyl deprotection of the Dpr³ side chain was performed manually under an inert N₂ atmosphere by mixing the peptide resin with phenylsilane (296 μL, 2.40 mmol) in DCM (4 mL) for 5 min followed by the addition of tetrakis(triphenylphosphine)palladium(0) (Pd(PPh₃)₄) (17 mg, 0.015 mmol) and reacting for 10 min. Coupling of 6-fluoro-2-naphthoic acid to the free side chain amine on Dpr³ was then performed manually by reacting the resin with a solution of 6-fluoro-2-naphthoic acid (3 equiv), HCTU (3 equiv), and DIPEA (6 equiv) in DMF (2 mL) twice (2 h, 16 h).

Global deprotection and resin cleavage of the peptides were performed by reacting the peptide resin in a solution of 95% TFA, 2.5% triisopropylsilane (TIPS), and 2.5% H₂O (2 mL) for 5 h. The cleaved peptide was precipitated from ice-cold tert-butyl methyl ether (TBME) and centrifuged at 3000 rpm for 10 min. The supernatant was then decanted, and the resulting peptide pellet was re-dissolved in 20% ACN in H₂O, frozen, and lyophilized until dry. The crude peptides were purified using preparative LC–MS, and collected fractions were combined, frozen, and lyophilized until dry. Purity was determined using analytical UHPLC–MS and is summarized in Table S1.

Small Molecule Synthesis

3-(1-Naphthyl)-l-alanine (6)

Fmoc-3-(1-naphthyl)-l-alanine (0.0781 g, 0.1 mmol) was suspended in DCM (3 mL). Piperidine was added dropwise until the white, opaque solution turned transparent. The mixture was then stirred for 20 min at room temperature before removal of excess solvent under reduced pressure to obtain a fine, white, crude powder. Trituration of the crude product with a small amount of cold DCM yielded pure metabolite 6 (0.010 g, 27%). UHPLC–MS method: 05–95% ACN (0.1% TFA) in H₂O (0.1% TFA), 3 min run; t_R (min) 0.92. HRMS (ESI⁺): [M + H]⁺ calculated = 216.1019; [M + H]⁺ observed = 216.1051.

Competitive Radioligand Binding Assay (IC₅₀)

The binding affinities of ghrelin analogues toward the GHSR were determined using a competitive radioligand-displacement binding assay using human [¹²⁵I]-ghrelin(1–28) (PerkinElmer Inc., cat. NEX388010UC) as the radioligand. Human Embryonic Kidney 293 (HEK293) cells were transiently transfected with the GHSR-eYFP pcDNA3.1(+) plasmid using an XtremeGENE 9 DNA transfection agent (Sigma-Aldrich, cat. 06365787001) according to the manufacturer’s protocol over 48 h prior to being frozen at 2 million cells per vial in 10% DMSO in fetal bovine serum (FBS). Human ghrelin(1–28) (Cedarlane Labs, cat. HOR-297) was tested as a reference to ensure the integrity of the results. A frozen aliquot of cells was thawed and centrifuged (3000 rpm, 10 min, room temperature), and the cell pellet was resuspended in binding buffer (HEPES (25 mM), MgCl₂·6H₂O (5 mM), EDTA (2.5 mM), CaCl₂ (1 mM), and 0.4% bovine serum albumin (BSA), pH = 7.4). The radioligand (15 pM) and cells (100,000 cells per assay tube) were added to solutions of the individual ghrelin(1–8) analogues in binding buffer (in triplicate at different concentrations of 10^–6 to 10^–12 M) at a final volume of 300 μL and incubated at 37 °C for 20 min with agitation. The tubes were then centrifuged at 13,000g for 5 min and unbound [¹²⁵I]-ghrelin was removed in the supernatant. The cell pellets were then washed with 500 μL of ice-cold TRIS–HCl buffer (50 mM, pH = 7.4) and centrifuged again at 13,000g for 5 min. The quantity of [¹²⁵I]-ghrelin bound to the cell membranes was measured by a gamma counter. Non-linear regression analysis to fit a 4-parameter dose response curve using GraphPad Prism 8 provided IC₅₀ values.

Chemical Stability

The chemical stabilities of peptides 1 and 2 were assessed by incubating each peptide (1 mM) in phosphate-buffered saline (PBS, pH = 7.4) at 37 °C. The peptide was allowed to equilibrate in solution for 10 min prior to the collection of sample aliquots. Aliquots (15 μL) of the peptide solution were removed in triplicate at 0, 0.5, 1, 2, 4, 6, and 24 h, and mixed with 0.1% TFA in ACN (60 μL). The samples were then analyzed by UHPLC–MS (5 μL volume injected). The selected ion chromatogram corresponding to the m/z value of the peptide was obtained and the resulting peak was integrated to quantify the amount of intact peptide remaining at each time-point. Both peptides were 100% intact after 24 h.

Human Serum Stability

The stability of ghrelin(1–8) analogues in human serum from human male AB plasma (Sigma-Aldrich, cat. H4522) was assessed by incubating each peptide (1 mM) at 37 °C in a 25% serum solution containing PBS at pH 7.4. The peptide was allowed to equilibrate in solution for 10 min prior to the collection of sample aliquots. Aliquots (15 μL) of the peptide solution were removed in triplicate at 0, 0.5, 1, 2, 4, 6, and 24 h. The reaction was quenched with 4% NH₄OH(aq) (60 μL), and salts and proteins were removed using a Waters Oasis HLB microextraction plate (30 μm) (Waters, cat. 186001828BA). The cartridge was first activated with methanol (MeOH, 200 μL) followed by water (200 μL). The sample was loaded onto the cartridge and the column was washed with 5% MeOH in water (200 μL). The peptide was then eluted using 2% formic acid in MeOH (2 × 30 μL). An aliquot (20 μL) of the eluted peptide was diluted with a solution of 0.1% TFA in H₂O (80 μL) and subsequently analyzed by UHPLC–MS (5 μL volume injected). The selected ion chromatogram corresponding to the m/z value of the peptide was obtained and the resulting peak was integrated to quantify the amount of intact peptide remaining at each time point.

Human Liver S9 Fraction Stability

The stability of ghrelin(1–8) analogues in the pooled human liver S9 fraction (Fisher Scientific, cat. HMS9PL) was assessed by incubating each peptide (3 μM) with the S9 fraction (1 mg/mL), MgCl₂·6H₂O (3 mM), and potassium phosphate buffer (0.1 M, pH = 7.4) at 37 °C with gentle agitation at 300 rpm. Aliquots (50 μL) of the reaction mixture were removed in triplicate at 0, 0.5, 1, 2, 3, and 4 h. The reaction was quenched with ice-cold 25% MeOH in ACN (200 μL), vortexed, and placed on ice for 20 min, which caused the S9 fraction material to precipitate. The mixture was then centrifuged at 14,000g for 10 min. An aliquot (100 μL) of the supernatant was then mixed with H₂O (100 μL) and subsequently analyzed by UHPLC–MS (10 μL volume injected). The selected ion chromatogram corresponding to the m/z value of the peptide was obtained and the resulting peak was integrated to quantify the amount of intact peptide remaining at each time point.

Calcium Signaling

The GHSR-activated change in intracellular calcium ([Ca²⁺]_i) was measured as described previously with some modifications.⁴⁹ HEK293 cells were seeded in a poly-d-lysine coated 96-well black cell culture microplate (Clear Bottom, Polystyrene, Thermo Scientific) at a density of 4 × 10⁴ cells per well. The cells were transiently transfected with GHSR-SmBit-pcDNA3.1(+) plasmids (30 ng/well) using an X-tremeGENE 9 transfection mixture (ratio 3:1 μL/μg DNA) as per the manufacturer’s instruction and cultured for 48 h in complete Dulbecco’s modified Eagle’s medium (DMEM, Gibco). The cells were rinsed with PBS (2 × 100 μL per well) and incubated with calcium indicator Fluo-4 NW (50 μL per well, F36206, Thermo Scientific) at 37 °C for 30 min and for further 15 min at room temperature in the dark. The change in [Ca²⁺]_i was measured as a change in fluorescence on a FlexStation3 (Molecular Devices) microplate reader. The real-time fluorescence spectra were recorded using the Flex Fluorescence Bottom-read mode with excitation and emission wavelengths of 494 and 515 nm, respectively. The total run time of each spectrum was 180 s, with the baseline recorded for 20 s followed by the addition of agonist (ghrelin, 1, 20, 21, 22) at different concentrations (10^–6 to 10^–11 M). A calcium ionophore (A23187, 6 μM, Sigma-Aldrich) was used as a control in all experiments to determine the maximum possible calcium response per well. The measured changes in [Ca²⁺]_i responses are presented as a percentage of maximum ghrelin (1000 nM) response. The concentration-effect curves were fitted and analyzed using the non-linear regression dose–response stimulation three-parameter model, and the logEC₅₀ values were obtained on GraphPad Prism 8.

G Protein and β-Arrestin Recruitment Assays

Recruitment of Gα_q, Gα₁₃, Gα_i1, or Gα_oB and β-arrestin-1 or β-arrestin-2 to GHSR was measured using a bioluminescence resonance energy transfer (BRET) assay.^50,51 To measure Gα_q, Gα₁₃, Gα_i1 or Gα_oB recruitment, HEK293 cells (4 × 10⁴ cells per well) were transiently transfected with plasmids encoding enhanced yellow fluorescent protein (eYFP) tagged GHSR (GHSR-eYFP) and Renilla luciferase (Rluc8) tagged Gα subunits (Rluc8:eYFP, 2:20 ng/well) using XtremeGENE 9 in an opaque white 96-well plate for 48 h. pcDNA5/FRT/TO-Gα_q/13/i1/oB-Rluc8 plasmids were a gift from Dr. Bryan Roth [Addgene: 140982 (Gα_q-Rluc8); 140986 (Gα₁₃- Rluc8); 140973 (Gα_i1- Rluc8); 140977 (Gα_oB- Rluc8)].⁵² The culture medium was replaced with 70 μL Hanks’ Balanced Salt Solution (HBSS, Invitrogen) and 10 μL of coelenterazine-h substrate (50 μM, Nanolight Technology). eYFP (530 nm) and Rluc8 (470 nm) emissions were measured at room temperature immediately after the addition of 30 μL of natural human ghrelin or ghrelin analogues 1, 20, 21, and 22 in HBSS at different concentrations (10^–5 to 10^–11 M) using the integrated automatic pipettor on a FlexStation3 plate reader.

To measure β-arrestin-1 or β-arrestin-2 recruitment, HEK293 cells were transiently transfected as described above with plasmids encoding GHSR-eYFP and Rluc tagged β-arrestin-1 or β-arrestin-2 (Rluc: eYFP, 3:30 ng/well). β-arrestin-1 and β-arrestin-2 plasmids were a kind gift from Dr. Michel Bouvier. Post-transfection, the culture medium was replaced with different concentrations (10^–5 to 10^–12 M) of natural human ghrelin or ghrelin analogues 1, 20, 21, and 22 in HBSS (70 μL) and incubated for a total of 20 min at 37 °C. 10 μL of coelenterazine-h substrate (50 μM) was added to each well 10 min before recording eYFP and Rluc signals on a Mithras LB 940 plate reader (Berthold Technologies). The data are presented as a BRET ratio (eYFP/ Rluc). The concentration–effect curves were fitted and analyzed using the non-linear regression dose–response stimulation three-parameter model, and the logEC₅₀ values were obtained on GraphPad Prism 8.

Glossary

ABBREVIATIONS

6-FN

6-fluoro-2-naphthyl

ACN

acetonitrile

DCM

dichloromethane

DIPEA

N,N-diisopropylethylamine

DMF

N,N-dimethylformamide

Dpr

l-2,3-diaminopropionic acid

eYFP

enhanced yellow fluorescent protein

Fmoc

fluorenylmethoxycarbonyl

GHSR

growth hormone secretagogue receptor 1a

HCTU

2-(6-chloro-1H-benzotriazole-1-yl)-1,1,3,3-tetramethylaminium hexafluorophosphate

HEK293

human embryonic kidney 293 cells

HEPES

4-(2-hydroxyethyl)piperazine-1-ethanesulfonic acid

MBHA

4-methylbenzhydrylamine

MeOH

methanol

NMP

N-methyl-2-pyrrolidone

PET

positron emission tomography

SAR

structure–activity relationship

TBME

tert-butyl methyl ether

TFA

trifluoroacetic Acid

THPTA

tris-hydroxypropyltriazolylmethylamine

TIPS

triisopropylsilane

TLC

thin layer chromatography

UHPLC

ultra high-performance liquid chromatography

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsptsci.3c00088.

• Characterization data as well as IC₅₀, stability, and functional activity curves for the library of ghrelin(1–8) analogues (PDF)

Author Present Address

Marina D. Childs – Lawson Health Research Institute, 750 Base Line Road East, London, Ontario, N6C 2R5

Author Present Address

Derian Hodgson – Hiram Walker & Sons Limited, 2072 Riverside Drive East, Windsor, Ontario, N8Y 4S5.

Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

This work was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC) and Canadian Institutes for Health Research (CIHR) grant # 376560.

The authors declare no competing financial interest.

Special Issue

Published as part of the ACS Pharmacology & Translational Science virtual special issue “Diagnostic and Therapeutic Radiopharmaceuticals”.