Signaling Modulation via Minimal C-Terminal Modifications of Apelin-13

Abstract

Apelin is an endogenous peptide that is involved in many diseases such as cardiovascular diseases, obesity, and cancer, which has made it an attractive target for drug discovery. Herein, we explore the penultimate and final sequence positions of [Pyr¹]-apelin-13 (Ape13) via C-terminal N_α-alkylated amide bonds and the introduction of positive charges, potentially targeting the allosteric sodium pocket, by assessing the binding affinity and signaling profiles at the apelin receptor (APJ). Synthetic analogues modified within this segment of Ape13 showed high affinity (K_i 0.12–0.17 nM vs Ape13 K_i 0.7 nM), potent Gα_i1 activation (EC₅₀ Gα_i1 0.4–0.9 nM vs Ape13 EC₅₀ 1.1 nM), partial agonist behavior disfavoring β-arrestin 2 recruitment for positively charged ligands (e.g., 49 (SBL-AP-058), EC₅₀ β-arr2 275 nM, E_max 54%) and high plasma stability for N-alkyl ligands (t_1/2 > 7 h vs Ape13 t_1/2 0.5 h). Combining the benefits of the N_α-alkylated amide bond with the guanidino substitution in a constrained ligand led to 63 (SBL-AP-049), which displayed increased plasma stability (t_1/2 5.3 h) and strong reduction of β-arrestin 2 signaling with partial maximal efficacy (EC₅₀ β-arr 864 nM, E_max 48%), significantly reducing the hypotensive effect in vivo.

The apelin receptor (APJ) was discovered in 1993 and deorphanized in 1998 by Tatemoto and co-workers when they isolated its first endogenous ligand, the apelin peptide.^1,2 The second endogenous ligand, ELABELA, was reported 15 years later by two independent groups while studying the implications of peptides in the embryogenesis process.^3,4 However, these two endogenous ligands have distinct signaling profiles and recent evidence also suggests that they have different mechanisms of action, such as in the cardiovascular system.^5,6 Of the two ligands, apelin is studied more extensively for historical reasons, but also because it is widely distributed in many tissue types and is involved in multiple physiological processes.⁷ Several isoforms of the apelin peptide exist after post-translational processing of prepro-apelin (77-aa long) into apelin-36, apelin-17, apelin-13, apelin-12, and [Pyr¹]-apelin-13 (I, Figure 1). The latter is the predominant isoform in human plasma since the N-terminal glutamine cyclizes to pyroglutamic acid (Pyr).⁸⁻¹¹ [Pyr¹]-apelin-13 (Ape13) is a full agonist and has been extensively studied in diabetes,¹²⁻¹⁴ obesity,¹² cancer,¹⁵⁻¹⁸ and cardiovascular diseases.¹⁹⁻²³ Recent reviews focusing on biological applications and signaling are available,²⁴⁻²⁷ but it is noteworthy that several compounds targeting the apelin receptor were clinically tested in phase I or II to treat cardiovascular complications, such as heart failure.²⁸⁻³⁰ The apelin-13 peptide has a very short in vivo half-life (<5 min) and is known to be degraded at the C-terminus between Pro¹²-Phe¹³ by the metalloproteases angiotensin converting enzyme II (ACE2) and proline carboxy peptidase (PRCP) (Figure 1).^31,32 More recently, neprilysin (NEP) has been identified, which has Arg⁴-Leu⁵ and Leu⁵-Ser⁶ as cleavage sites (Figure 1).^33,34

Figure 1.

Linear and macrocyclic peptide-based apelin analogues in literature. NEP: neprilysin; ACE2: angiotensin-converting enzyme 2; PRCP: prolyl carboxypeptidase. IV–VI are modifications of [Pyr¹]-apelin-13 at positions 12 and 13, and these analogues have Nle instead of Met¹¹ to avoid oxidation of the Met side chain.

As a member of class A GPCRs, the apelin receptor triggers the G protein and β-arrestin signaling pathways upon activation.³⁵ Two of the G proteins involved are Gα_i, which inhibits adenylate cyclase and reduces cyclic AMP production,³⁶ and Gα₁₂, which leads to RhoGEF recruitment and ROCK activation.^37,38 β-arrestin is also recruited to the cell membrane after receptor activation.³⁹ This protein is mainly associated with receptor desensitization and endocytosis, but also activates other downstream signaling pathways and effectors, such as ERK1/2 and Raf.³⁹ Altogether, the above signaling pathways trigger physiological responses such as vasodilation, cardiac contraction, and angiogenesis.^35,40,41

Related to the present work, the literature has demonstrated that apelin administration is beneficial in heart failure models and has cardioprotective properties. However, the activation of the β-arrestin signaling pathway seems to cause cardiomyocyte hypertrophy.⁶ Therefore, APJ ligands that favor G protein activation and block β-arrestin signaling appear promising for patients with heart failure and other cardiovascular diseases. Interestingly, Li and co-workers recently demonstrated that APJ transgenic mutant mice (APJ I107A), which are unable to recruit β-arrestins after receptor activation do not seem to exhibit impairment of cardiac function after exercise or in response to pressure overload-induced cardiac hypertrophy.⁴² Altogether, these results suggest that intracellular signaling pathways may be subtly altered under pathological conditions, thus emphasizing the need for a comprehensive analysis of ligands with unique or unbalanced signaling profiles.⁴³

Although many advances have been made in apelin research, it remains unclear why certain modifications of [Pyr¹]-apelin-13 allow the triggering of different signaling profiles following APJ activation, a phenomenon called functional selectivity or biased agonism. Biased agonism is now accepted as representing a handle to prevent receptor desensitization, due to β-arrestin recruitment, and thus prolong the drug’s therapeutic effect.^44,45 Brame et al. reported a macrocyclic analogue of apelin-13 (II, Figure 1) which showed biased G protein activation over β-arrestin recruitment.⁴⁶ More recently, we have also reported the design and synthesis of novel macrocyclic analogues of apelin-13.⁴⁷ These structures (e.g., III, Figure 1), containing a C-terminal macrocycle, showed potent bias toward G protein activation.⁴⁷⁻⁴⁹ Additionally, we showed that conformationally constrained or unnatural amino acids (e.g., IV, V, VI, Figure 1) at the 12th (Pro) or 13th (Phe) position had beneficial effects, both in terms of affinity and stability, compared to the apelin-13 peptide.^50,51 Unfortunately, no biased ligand was obtained in these latter efforts. Recently, Vederas’ group has worked on modifications of the NEP cleavage site. They obtained agonists with good affinity and stability, but without studying the potential for signaling bias.⁵² In addition, many efforts have been made in recent years to develop small molecule agonists of the APJ receptor.^28,53−60 Nevertheless, the development of peptide analogues allows an easy and rapid approach for structure–activity relationship (SAR) studies and functional selectivity, which is still poorly understood.

Only a handful of publications exist describing the rational approach to design a biased GPCR ligand. Among them, Roth and co-workers postulated that modulating the interaction of ligands with the transmembrane domain 5 (TM5) and extracellular loop 2 (ECL2) of the GPCR could impact signaling, as successfully demonstrated by the development of β-arrestin biased agonists of the dopamine D2 receptor.⁶¹ The interaction of orthosteric ligands with ECL2 has also been used to explain the bias for other GPCRs, in casu the serotonin receptor 5-HT2R.⁶² Although the X-ray structure of an apelin analogue in complex with a mutant form of the receptor has been solved and critical interactions of Ape13 binding to APJ have been suggested for agonist activity, it remains speculative that analogous interactions could be applied to produce a similar result.⁶³ The C-terminus of Ape13 occupies the binding site between the TM1, TM2, TM6, and TM7 of the receptor and its SAR has been extensively studied in the past, in which certain modifications, especially on Phe¹³, resulted in high affinity and biased ligands.^51,64 In contrast, modifications at the N_α-amide bond between residues Pro¹²-Phe¹³ have been scarcely investigated, although they may help to establish new interactions and signaling profiles, and potentially increase proteolytic stability.^50,65

Herein, we describe the synthesis of a series of [Pyr¹]-apelin-13 analogues in which the amide bond at the native Pro¹²-Phe¹³ peptide bond was substituted with different alkyl groups. The Nle¹¹-1-Nal¹²-Phe¹³ segment (V, Figure 1) was taken as a starting point given its recently reported increased affinity and signaling potency for Gα₁₂.⁵⁰ A wide variety of substituents varying in length, nature (aliphatic vs aromatic), polarity, and steric bulk was applied. These substituents induce conformational shifts by steric interactions, potentially influencing any eventual signaling. On the other hand, unnatural amino acids bearing a positive charge were also designed and introduced into the C-terminus of [Pyr¹]-apelin-13 to test their ability to modulate the signaling profile. The insertion of a positive charge was inspired by the work of the Yokoyama group, which demonstrated an inverse agonist mechanism on the leukotriene receptor BLT1 (class A GPCR) by targeting the allosteric sodium ion binding site with positively charged analogues.⁶⁶ Other studies have also shown the potential of the sodium pocket for a pronounced signaling bias for class A GPCRs, including the angiotensin 1 receptor (AT1R) which has the highest homology to APJ.^1,67−69 Several class A GPCRs share the same amino acid sequence for the allosteric sodium pocket, and the APJ receptor holds this sequence, suggesting the presence of this pocket.⁶⁷

In contrast to the previously reported macrocyclic analogues, the ligands in this study only encompass minimal backbone modifications compared to the endogenous ligand and are intended to provide additional insights into the SAR of APJ and apelin. The affinity of the modified peptides for APJ was measured by a competitive binding assay against [¹²⁵I][Nle⁷⁵,Tyr⁷⁷]Pyr-apelin-13, and their ability to activate the canonical signaling pathways Gα_i1 and β-arrestin 2 was evaluated using bioluminescence resonance energy transfer (BRET)-based biosensors in HEK293 cells overexpressing APJ. Proteolytic stability was assessed for all analogues in rat plasma. Finally, synthetic analogues that showed beneficial pharmacological properties in vitro were selected for continuous in vivo blood pressure measurements.

Results and Discussion

Recently published results by our group have shown that insertion of sterically or covalently constrained unnatural amino acids produces metabolically stable [Pyr¹]-apelin-13 analogues, but without bias in signaling.^50,51 Hence, we further studied C-terminal modifications by adding alkyl groups ranging from a methyl to bulky benzyl ethers in the 1-Nal¹²-Phe¹³ dipeptide segment. The introduction of the additional substituent further inhibits protease access and is expected to increase hydrophobic interactions with the receptor. To avoid difficult resin couplings and to ensure easy peptide purifications, all N_α-alkylated dipeptides and phenylalanine derivatives were synthesized in solution prior to assembly by solid-phase peptide synthesis (SPPS).

Synthesis of N-Substituted Dipeptides

C-Terminal dipeptides of type 3 were synthesized by first monoalkylating the N_α of phenylalanine methyl ester 1 according to three different procedures from the literature (Scheme 1).^70,71 For aromatic aldehydes (Conditions A, Scheme 1), phenylalanine methyl ester 1 was first dissolved in dry DCM under an inert atmosphere, to which aldehyde was added. After imine formation, the hydride donor NaBH₄ was added, which allowed complete reduction to the desired secondary amine. For aliphatic aldehydes, a one-step reductive amination gave the best results. Here, NaBH(OAc)₃ was added to a solution of phenylalanine and aldehyde (Conditions B, Scheme 1). Upon insertion of the 4-nitrobenzyl substituent, a third route was applied, in which mono-N_α alkylation was achieved using 4-nitrobenzyl bromide and Cs₂CO₃ in N,N-dimethylformamide (DMF) (Conditions C, Scheme 1). This alternative route was followed due to the observed epimerization of Phe when performing reductive amination conditions with 4-nitrobenzaldehyde on phenylalanine. All final products were concentrated in vacuo after extraction and precipitated in a solution of 2 N HCl in Et₂O to afford the white HCl salt. Next, to prevent prolonged or repeated coupling using expensive coupling reagent mixtures, an acyl chloride-based coupling was used. Commercially available Fmoc-1-Nal-OH was treated with SOCl₂ in dry DLM at room temperature. Full conversion was achieved after overnight reaction. This procedure enabled the synthesis of the targeted dipeptides (Conditions D, Scheme 1). However, in selected cases, the HATU reagent was sufficiently reactive to allow coupling between Fmoc-1-Nal-OH and alkylated phenylalanine bearing an aliphatic chain such as methyl, ethyl, n-propyl, etc. (Conditions E, Scheme 1).

Scheme 1. Synthesis of Fmoc-1-Nal-(N-R)-Phe-OH 3.

Conditions: A) 1) RCHO, Et₃N, dry DCM, rt, 2h, 2) NaBH₄, 3) 2 N HCl Et₂O; B) 1) RCHO, Et₃N, NaBH(OAc)₃, dry DCM, rt, 2 h, 2) 2 N HCl Et₂O; C) 1) 4-NO₂-benzyl-bromide, Cs₂CO₃, DMF, rt, 16 h, 2) 2 N HCl Et₂O; D) Fmoc-1-Nal-Cl, DIPEA, dry DCM, rt, 16 h; E) Fmoc-1-Nal-OH, HATU, DIPEA, dry DMF, rt, 3 h; F) SnMe₃(OH), DCE, 60 °C, 16 h; G) 1 N HCl/1,4-dioxane (1:1), 90 °C, 16 h.

After dipeptide synthesis, orthogonal hydrolysis was needed to obtain the free C-terminus in compounds of type 3. Standard aqueous acidic conditions (e.g., 1 N HCl/acetone (1:1 v/v) or 1 N HCl/dioxane (1:1 v/v)) at reflux temperature resulted in good conversion for aliphatic substitutions, but poor conversion and compound degradation for most aromatic substitutions. Shifting to basic conditions (LiOH in methanol at elevated temperature) improved the conversion significantly, but with epimerization as a byproduct. However, the applying conditions reported by Nicolaou et al. afforded good conversion and low degradation (Conditions F, Scheme 1).⁷² All synthetic details can be found in the Supporting Information.

Synthesis of Substituted Phenylalanine Derivatives

Scheme 2 depicts the synthetic route toward the 4-amino-methylated compound 11, starting from natural phenylalanine amino acid 6. First, commercially available amide 4 was treated with formalin, followed by acyl iminium formation from 5 in H₂SO₄, and allowing substitution at the para-position of 6. Protection of amine 7 with Boc₂O resulted in 8 which was orthogonally deprotected with a 20% NaOH solution (in H₂O/EtOH) to afford 9. Alloc protection, giving compound 10, followed by Boc-to-Fmoc interconversion, resulted in the final unnatural amino acid 11.

Scheme 2. Synthetic Pathway towards Fmoc-(4-CH₂NHAlloc)-Phe-OH 11.

Previous work involving the Tyr(OBn) substitution of Phe¹³ led to a substantial increase in [Pyr¹]-apelin-13 binding to the pM level and still has, to our knowledge, the highest reported affinity for the APJ-receptor.⁵¹ For this reason, benzylated tyrosine was used as a scaffold to implement an additional, positively charged functional group. Two types of residues (15a and 15b; Scheme 3) were envisioned, with charged benzyloxy and alkoxy extensions built into the Tyr residue. The unnatural amino acid 14a was prepared from commercially available 4-cyanobenzyl bromide 12. First, the bromide was converted to the alcohol, using BaCO₃ in water under reflux conditions, followed by nitrile reduction applying standard Pd/C under H₂-atmosphere conditions.⁷³ Compound 13, obtained after Alloc-protection, underwent a Mitsunobu reaction with Boc-L-Tyr-OMe to cleanly deliver 14a in good yield. Simultaneous Boc-deprotection and ester hydrolysis resulted in the unprotected amino acid whose amine was Fmoc protected to yield 15a. Second, Alloc protection of commercially available 3-aminopropyl-bromide 16 resulted in 17, which was used to alkylate the phenol group of Boc-L-Tyr-OMe in DMF with Cs₂CO₃ in excellent yield, to present compound 14b. The protective group switch, as described above, resulted in 15b. The Alloc-protected building blocks eventually served as precursors for the aminated and guanidinylated peptide ligands.

Scheme 3. Synthetic Pathway towards the Substituted and SPPS-Compatible Tyrosine Residues.

Solid-Phase Synthesis of [Pyr¹]-Apelin-13 Analogues

Once available, the (un)natural amino acids and prepared Fmoc-1-Nal-(N_α-R)-Phe-OH dipeptide analogues were used to synthesize the envisaged analogues of [Pyr¹]-apelin-13 using standard Fmoc-based SPPS. Dipeptide anchoring and elongation were performed on 2-chlorotrityl chloride resin to prevent the formation of diketopiperazines (DKPs), which is a common problem in the synthesis of N-alkylated peptides by Fmoc-SPPS (Scheme 4, Conditions A).⁷⁴ Resin loadings (ca. 0.4–0.7 mmol/g resin) were determined by UV absorbance measurements of dibenzofulvene after Fmoc cleavage of the first anchored residue (see Supporting Information). Peptide elongation was performed using standard Fmoc-based SPPS with DMF/HATU/DIPEA as the coupling mixture. To prevent oxidation side products due to the presence of Met,¹¹ this residue was replaced by Nle for all peptides. This unnatural amino acid does not affect the affinity of the apelin-13 analogues, as previously reported.^51,75 To access amine- or guanidine-functionalized peptides, the Alloc side chain protecting group in compound 18b was removed by using a catalytic amount of Pd(PPh₃)₄ in the presence of PhSiH₃ as a scavenger. Thereupon, to obtain guanylated analogues, the free amine was converted to guanidine with the guanylating reagent N,N′-bis-Boc-1H-pyrazole-1-carboxamidine on solid support (Scheme 4, Conditions C). As a result, the impact of amine and guanidine functional groups on affinity and signaling could be studied.

Scheme 4. SPPS Assembly of [Pyr¹]-Apelin-13 Analogues.

Conditions: A) Standard deprotection and cleavage; B) Deprotection and cleavage strategy for [Pyr¹]-apelin-13 analogues with an amine on the C-terminal amino acid; C) Deprotection, guanylation and cleavage strategy for [Pyr¹]-apelin-13 analogues with a guanidine on the C-terminal amino acid.

Binding Affinity and APJ Signaling of the N-Alkylated [Pyr¹]-Apelin-13 Analogues

Impact of Aliphatic N_α-Substituents on Affinity and Signaling

First, aliphatic substituents were introduced, but this type of modification showed no significant effect on ligand binding compared to the unalkylated reference (Table 1). Insertion of a methyl, ethyl, and propyl group provided good binding affinity (19 (SBL-AP-022), K_i 0.15 nM; 20 (SBL-AP-023), K_i 0.12 nM; 21 (SBL-AP-024), K_i 0.17 nM vs ref with K_i 0.12 nM) and longer alkyl groups, such as isobutyl, cyclohexylmethyl and n-heptyl, were also well tolerated (22 (SBL-AP-025), K_i 0.26 nM; 23 (SBL-AP-027), K_i 0.6 nM; 24 (SBL-AP-026), K_i 0.5 nM vs ref with K_i 0.12 nM). Interestingly, most of these alkyl groups resulted in a slight increase in Gα_i1 activation, compared to the reference compound and the native ligand. Isobutyl alkylation delivered one of the most potent Gα_i1 activators in our study (22, EC₅₀ Gα_i1 0.4 nM vs ref, EC₅₀ Gα_i1 4.4 nM; Ape13, EC₅₀ Gα_i1 1.1 nM). Only the cyclohexylmethyl substitution resulted in a slight loss of Gα_i1 activation compared to the reference sequence and a 6-fold loss compared to Ape13 (23, EC₅₀ Gα_i1 7 nM vs ref, EC₅₀ Gα_i1 4.4 nM; Ape13, EC₅₀ Gα_i1 1.1 nM). Moreover, among the analogues with aliphatic alkyl substituents, insertion of the cyclohexylmethyl group was the only case in which a significant decrease in β-arrestin 2 recruitment was observed, compared to the reference and the endogenous ligand (23 (SBL-AP-027), EC₅₀ β-arr2 184 nM vs ref, EC₅₀ β-arr2 41 nM; Ape13, EC₅₀ β-arr2 40 nM). Altogether, although some differences in signaling were observed, no significant impact was noted for this series when compared with Ape13.

Table 1. Chemical Structures of the Alkylated Substituents (1-Nal¹²-N(R)-Phe¹³), Affinity and Functional Activities of C-Terminally Modified [Pyr¹]-Apelin-13 Analogues.

compound no.	compound code	peptide sequence	Ki binding (nM)a	Gαi1 (nM)b	β-arrestin 2 (nM) (Emax%)c	rat plasma t1/2 (h)d
Ape13		Pyr-R-P-R-L-S-H-K-G-P-Nle-Pro-Phe-OH	0.7 ± 0.1	1.1 ± 0.1	40 ± 4	0.5 ± 0.1e
ref	KT04-39	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-F-OH	0.12 ± 0.01	4.4 ± 0.6	41 ± 4 (113%)	0.4 ± 0.1e
19	SBL-AP-022	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(N-Me)Phe-OH	0.15 ± 0.01	1.8 ± 0.7	28 ± 7 (103%)	>7
20	SBL-AP-023	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(N-Et)Phe-OH	0.12 ± 0.02	0.7 ± 0.3	38 ± 7 (105%)	>7
21	SBL-AP-024	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(N-nPr)Phe-OH	0.17 ± 0.02	1.0 ± 0.4	38 ± 8 (103%)	>7
22	SBL-AP-025	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(N-iBu)Phe-OH	0.26 ± 0.06	0.4 ± 0.1	52 ± 17 (101%)	>7
23	SBL-AP-027	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(N-cyclohexylmethyl)Phe-OH	0.6 ± 0.2	7 ± 2	184 ± 55 (88%)	>7
24	SBL-AP-026	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(N-nheptyl)Phe-OH	0.50 ± 0.07	1.4 ± 0.3	62 ± 7 (94%)	>7
25	SBL-AP-028	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(N-Bn)Phe-OH	0.25 ± 0.03	3 ± 1	61 ± 14 (101%)	>7
26	SBL-AP-029	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(N-PhEt)Phe-OH	0.4 ± 0.1	0.9 ± 0.2	70 ± 34 (91%)	>7
27	SBL-AP-030	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(N-Me-1′-naph)Phe-OH	0.57 ± 0.08	5 ± 2	115 ± 35 (97%)	>7
28	SBL-AP-031	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(N-Me-2′-naph)Phe-OH	1.1 ± 0.3	7 ± 3	139 ± 52 (87%)	>7
29	SBL-AP-036	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(N-(4′-Et)Bn)Phe-OH	0.5 ± 0.2	4 ± 2	89 ± 16 (94%)	>7
30	SBL-AP-035	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(N-(4′-tBu)Bn)Phe-OH	7 ± 2	7 ± 3	312 ± 93 (87%)	>7
31	SBL-AP-032	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(N-(4′-I)Bn)Phe-OH	0.3 ± 0.1	0.75 ± 0.09	82 ± 45 (83%)	>7
32	SBL-AP-043	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(N-(4′-NO2)Bn)Phe-OH	0.17 ± 0.03	0.9 ± 0.1	34 ± 3 (96%)	>7
33	SBL-AP-064	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(N-(4′-CF3)Bn)Phe-OH	0.9 ± 0.2	0.6 ± 0.2	22 ± 3 (83%)	>7
34	SBL-AP-034	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(N-(4′-OH)Bn)Phe-OH	0.14 ± 0.09	5 ± 3	108 ± 52 (92%)	>7
35	SBL-AP-033	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(N-(4′-MeO)Bn)Phe-OH	0.27 ± 0.03	5 ± 3	97 ± 34 (96%)	>7
36	SBL-AP-041	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(N-MediOBn)Phe-OH	0.21 ± 0.08	5 ± 1	114 ± 55 (93%)	>7
37	SBL-AP-037	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(N-(4′-BnO)Bn)Phe-OH	1.4 ± 0.3	7 ± 1	119 ± 33 (92%)
38	SBL-AP-038	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(N-(3′-BnO)Bn)Phe-OH	0.9 ± 0.3	11 ± 7	258 ± 74 (95%)	>7
39	SBL-AP-039	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(N-(4′-PhO)Bn)Phe-OH	1.3 ± 0.7	9 ± 2	252 ± 118 (84%)	>7
40	SBL-AP-040	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(N-(3′-PhO)Bn)Phe-OH	3 ± 2	24 ± 11	365 ± 130 (87%)	>24
41	SBL-AP-042	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(N-methyldibenzofurane)Phe-OH	4.8 ± 0.4	0.8 ± 0.2	121 ± 29 (89%)	>24

^aK_i was calculated from experimental IC₅₀ values (the concentration of ligand that displaces 50% of radiolabeled [¹²⁵I][Nle⁷⁵,Tyr⁷⁷]Pyr-apelin-13) using the Cheng–Prusoff equation.⁷⁸ Values represent the mean ± SEM of two or three experiments, each performed in duplicate.

^bEC₅₀ corresponds to the concentration of ligand that produces 50% dissociation of Gα_i1 from the Gβγ subunits. All compounds showed E_max Gα_i1 > 96%.

^cEC₅₀ is the concentration of ligand that produces 50% recruitment of β-arrestin 2 to APJ. EC₅₀ values represent the mean ± SEM of three experiments, each performed in triplicate. All E_max values can be found in Table S2 in the Supporting Information.

^dValues represent the mean ± SEM of three experiments. “>7” means between 7 and 24 h.

^eReported in ref (50).

Impact of Aromatic N_α-Substituents on Affinity and Signaling

To further explore the effect of the substituent, we synthesized analogues with increased bulk by implementing benzyl, phenethyl, 1/2-naphthyl and 4-ethyl/4-tbutyl-benzyl groups (i.e., compounds 25 to 30; Table 1). Overall, the additional bulk did not significantly impact the binding affinity of these ligands. Only analogue 30 (SBL-AP-035) led to a significant decrease in affinity (30 (SBL-AP-035), K_i 7 nM vs ref, K_i 0.12 nM). This suggests that a tert-butyl group represents the steric hindrance threshold to impact affinity at the para-position of benzyl. Phenethyl alkylation showed a 4-fold increase in Gα_i1 activation (26 (SBL-AP-029), EC₅₀ Gα_i1 0.9 nM vs ref, EC₅₀ Gα_i1 4.4 nM), but did not impact β-arrestin 2 recruitment. Hence, the series of aromatic N_α-substituents was extended even further to potentially modulate the signaling profile.

Addition of polar groups (e.g., hydroxyl or nitro substituents; Table 1) on the para-position of the benzyl moiety resulted in a maintained binding affinity (32 (SBL-AP-043), K_i 0.17 nM; 34 (SBL-AP-034), K_i 0.14 nM) compared to the unsubstituted benzyl group (25 (SBL-AP-028), K_i 0.25 nM. Remarkably, an electron-withdrawing group on the benzyl moiety (as in 32) favors Gα_i1 activation, unlike an electron-donating group (34-36), which has no significant effect (32 EC₅₀ Gα_i1 0.9 nM; vs 34–36 EC₅₀ Gα_i1 5 nM). However, the recruitment of β-arrestin 2 seems to be independent of these electronic effects. The addition of iodine (31 (SBL-AP-032)) had the same effect as the nitro group on affinity and signaling. To reduce potential toxicity risks, the nitro group, known to be toxicophoric, was replaced with an electron-withdrawing trifluoromethyl bioisostere, resulting in slightly reduced affinity (32 (SBL-AP-043), K_i 0.17 nM vs 33 (SBL-AP-064), K_i 0.9 nM) and similar signaling potency.⁷⁶ Finally, electron-donating alkoxy groups (35 (SBL-AP-033) and 36 (SBL-AP-041)) induced no improvement.

Additional substitution on the benzyl group via extensions with a benzyloxy and phenoxy unit resulted in a slight loss of affinity compared to the reference and unsubstituted benzyl group, but values similar to those of Ape13 were noted (Table 1). A decrease in signaling potency was observed for Gα_i1 (37 (SBL-AP-037)–40 (SBL-AP-040), EC₅₀ Gα_i1 7–24 nM), but it was even more pronounced for β-arrestin 2 recruitment (37 (SBL-AP-037)–40 (SBL-AP-040), EC₅₀ β-arr2 119–365 nM), compared to the reference (ref, EC₅₀ Gα_i1 4.4 nM and EC₅₀ β-arr2 41 nM). As expected, the most bulky and rigid compound with the methyldibenzofuran extension presented the lowest measured affinity (41 (SBL-AP-042), K_i 4.8 nM vs ref, K_i 0.12 nM). In contrast, Gα_i1 signaling increased, compared to the reference peptide (41 (SBL-AP-042), EC₅₀ Gα_i1 0.8 nM vs ref, EC₅₀ Gα_i1 4.4 nM).

It is noteworthy that all analogues behave as full agonists of the Gα_i1 pathway (E_max Gα_i1 > 96%). Interestingly, some compounds are partial agonists of β-arrestin 2 recruitment, but their efficacy remained high (E_max β-arr2 > 83%). A trend is observable between the size of the N_α-substituents and E_max: the larger the substituent, the lower the E_max.

To summarize, the results showed that the binding pocket displays a high tolerance to a wide variety of bulky substituents. Our observations are in agreement with recent X-ray structures of APJ complexed with an apelin-17 mimetic peptide that revealed that Pro¹² and Phe¹³ are surrounded in the orthosteric binding pocket by hydrophobic residues with plenty of space available to diversify.⁶³ In previous modifications of the Pro¹² position of linear Ape13 analogues, we only noticed improvements in affinity and stability, but signaling remained unbiased.^61,62,77 In this subseries, we did not obtain biased analogues, but a decrease in potency and maximum efficacy of β-arrestin 2 recruitment was observed.

Binding Affinity and APJ Signaling of the Positively Charged [Pyr¹]-Apelin-13 Analogues

Impact of Positively Charged Residues at the C-Terminus on Affinity and Signaling

The above results showed that adding purely apolar aliphatic and aromatic groups at the C-terminus of [Pyr¹]-apelin-13 analogues had no significant effect on affinity and signaling. However, the addition of polar and electron-withdrawing functionality had a significantly impact on signaling patterns. Hence, substitution of the para-position of Phe¹³ with a cyano group was carried out first. Compound 42 (SBL-AP-018) provided no significant change in binding affinity compared to the reference ligand (42, K_i 0.13 nM vs ref, K_i 0.12 nM), but a 10- and 3-fold increase was obtained in Gα_i1 activation and β-arrestin 2 recruitment, respectively, compared to the reference (Table 2). Moreover, analogue 42 was found to be the best β-arrestin 2 recruiter in this study 42, (EC₅₀ β-arr2 14 nM). Introduction of a positively charged aminomethyl group via a nitrile reduction (43 (SBL-AP-019)) resulted in a 20-fold loss of binding affinity compared to the reference peptide, a 6-fold loss of Gα_i1 activation, and a significant reduction in β-arrestin 2 recruitment was observed, compared to the nitrile analogue 42. Guanylating the primary amine (44 (SBL-AP-020); Table 2) led not only to a further loss of affinity but also a loss of signaling efficacy (K_i 4.6 nM; EC₅₀ Gα_i1 10 nM; EC₅₀ β-arr2 423 nM with E_max of 59%). The guadinylated analogue of Phe¹³ (45 (SBL-AP-051); Table 2) further decreased affinity, but Gα_i1 signaling remained nevertheless in the low nM range, while the β-arrestin 2 pathway was attenuated (with E_max of 67%) to a similar extent of the guanidinomethyl analogue 44 (SBL-AP-020). Interestingly, the positively charged compounds are the only ones in this series that are partial agonists in β-arrestin 2 recruitment. Thus, the presence of guanidine moieties seemed promising, and we therefore altered the position of this positive charge relative to the peptide backbone via extension of the side chain (see entries 48 (SBL-AP-057) to 51 (SBL-AP-048); Table 2).

Table 2. Chemical Structures of Ligands Bearing a Positive Charge, Affinity and Functional Activities of C-Terminally Modified [Pyr¹]-Apelin-13.

compound no.	compound code	peptide sequence	Ki binding (nM)a	Gαi1 (nM)b	β-arrestin 2 (nM) (Emax%)c	rat plasma t1/2 (h)d
Ape13		Pyr-R-P-R-L-S-H-K-G-P-Nle-Pro-Phe-OH	0.7 ± 0.1	1.1 ± 0.1	40 ± 4	0.5 ± 0.1e
ref	KT04-39	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-F-OH	0.12 ± 0.01	4.4 ± 0.6	41 ± 4	0.4 ± 0.1e
42	SBL-AP-018	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-Phe(4-CN)-OH	0.13 ± 0.02	0.4 ± 0.2	14 ± 5 (100%)	<1
43	SBL-AP-019	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-Phe(4-CH2-NH2)-OH	3.4 ± 1.0	2.4 ± 0.9	105 ± 39 (80%)	1.0 ± 0.01
44	SBL-AP-020	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-Phe(4-CH2-Gn)-OH	4.6 ± 0.1	10 ± 4	423 ± 58 (59%)	<1
45	SBL-AP-051	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-Phe(4-Gn)-OH	7 ± 2	12 ± 5	303 ± 130 (67%)	<1
46	SBL-AP-045	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-Tyr(O-4-NO2-Bn)-OH	0.15 ± 0.004	0.9 ± 0.4	60 ± 5 (99%)	5.4 ± 0.9
47	SBL-AP-046	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-Tyr(O-4-CN-Bn)-OH	0.12 ± 0.03	0.4 ± 0.2	28 ± 5 (99%)	3.0 ± 0.4
48	SBL-AP-057	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-Tyr(O-4-Bn-CH2-NH2)-OH	3.0 ± 0.4	10 ± 3	324 ± 49 (80%)	1.3 ± 0.1
49	SBL-AP-058	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-Tyr(O-4-Bn-CH2-Gn)-OH	1.1 ± 0.4	11 ± 4	275 ± 14 (54%)	1.1 ± 0.02
50	SBL-AP-047	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-Tyr(O-4-npropyl-NH2)-OH	0.30 ± 0.08	1.2 ± 0.5	53 ± 2 (97%)	<1
51	SBL-AP-048	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-Tyr(O-4-npropyl-Gn)-OH	3 ± 1	1.4 ± 0.5	197 ± 18 (70%)	<1
52	SBL-AP-021	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-Arg-OH	1.2 ± 0.3	6 ± 2	116 ± 19 (76%)	<1
53	SBL-AP-054	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-d-Arg-OH	25 ± 5	10 ± 2	906 ± 229 (66%)	2.8 ± 0.1
54	SBL-AP-052	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-Lys-OH	7.8 ± 0.4	14 ± 4	708 ± 185 (67%)	<1
55	SBL-AP-053	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-hArg-OH	3.2 ± 0.2	6 ± 3	412 ± 89 (72%)	<1

^aK_i was calculated from experimental IC₅₀ values (the concentration of ligand that displaces 50% of radiolabeled [¹²⁵I][Nle⁷⁵,Tyr⁷⁷]Pyr-apelin-13) using the Cheng–Prusoff equation.⁷⁸ Values represent the mean ± SEM of two or three experiments, each performed in duplicate.

^bEC₅₀ corresponds to the concentration of ligand that produces 50% dissociation of Gα_i1 from the Gβ_γ subunits. All compounds showed an E_max Gα_i1 > 99%.

^cEC₅₀ is the concentration of ligand that produces 50% recruitment of β-arrestin 2 to the apelin receptor. EC₅₀ values represent the mean ± SEM of three experiments, each performed in triplicate. All E_max values can be found in Table S2 in the Supporting Information.

^dValues represent the mean ± SEM of three experiments.

^eReported in ref (50).

The introduction of 4′-substituted benzyl groups on Tyr, replacing Phe,¹³ was performed. Replacement of Phe¹³ with 4-nitro- and 4-cyanobenzyl-Tyr (46 (SBL-AP-045) and 47 (SBL-AP-046), respectively) led to similar binding affinity and β-arrestin 2 recruitment to the reference ligand (Table 2). However, higher Gα_i1 potencies were observed (46, EC₅₀ Gα_i1 0.9 nM; 47, EC₅₀ Gα_i1 0.4 nM vs ref, EC₅₀ Gα_i1 4.4 nM). Reduction of nitrile toward aminomethyl (48 (SBL-AP-057)) provided very similar data compared to the previous aminomethylated analogue (43, K_i 3.4 nM vs 48, K_i 3.0 nM; with β-arr2 E_max of 80% for both), again showing the benefit of positive charges in finding biased ligands. Additionally, the corresponding guanylated analogue 49 (SBL-AP-058) recovered some affinity (49, K_i 1.1 nM), compared with the aminomethyl derivative 48, but more importantly, a maximum β-arrestin 2 recruitment efficacy of only 54% could be reached.

In a next step, we expanded the series with different side chain lengths (50-55; Table 2) to identify more precisely where the positive charge could be placed relative to the backbone to optimize the signaling bias. Substitutions on Tyr¹³ with a linear alkyl chain bearing amine or guanidine functionality were investigated, alongside the use of more standard basic amino acids, such as Arg and Lys, or variants thereof. Altogether, this subset of analogues provided ligands with high affinity (e.g., 50 (SBL-AP-047) and 52 (SBL-AP-021)), high Gα_i1 signaling potency (e.g., 50 and 51 (SBL-AP-048)) and some with partial agonism, strongly disfavoring β-arrestin 2 recruitment (with a range between 66% and 76%). A reduction of the β-arrestin 2 signaling was observed, especially for the analogues possessing d-Arg¹³ and l-Lys¹³ (53 (SBL-AP-054), EC₅₀ β-arr2 906 nM, E_max 66%; 54 (SBL-AP-052), β-arr2 708 nM, E_max 67%). Therefore, this suggests that, in addition to a positive charge, a shorter aliphatic side chain seems to decrease β-arrestin 2 recruitment and its maximum efficacy.

Synthesis of Constrained Analogues Bearing a Positive Charge

In an attempt to further optimize ligand binding, bias, and stability as well as to pinpoint more precisely in which direction the charged group should be oriented, we combined the N-alkyl substituent and the positive charge in this series of constrained analogues. We first opted to design tetrahydroisoquinoline-based constraint type 59 (Scheme 5), which correspond to more rigid versions of derivatives 50 (SBL-AP-047) and 51 (SBL-AP-048) (Table 2). Such custom residues not only allow χ-space screening but also represent tools for increasing the affinity and metabolic stability of more flexible peptide analogues.^50,79

Scheme 5. Synthetic Pathway Towards Constrained Amino Acids and Dipeptides Bearing a Positive Charge, 59 and 61.

Panels: (A) Synthetic pathway toward (Fmoc-Tic(6-O-propyl-NHAlloc)-OH 59; (B) Synthetic pathway toward Fmoc-L-Aia-Orn(Alloc)-OH 61.

To obtain scaffold 59, treatment of meta-tyrosine 56 under Pictet–Spengler conditions gave the expected 6-hydroxy-tetrahydroxyisoquinoline Tic(6-OH) in excellent yield (Scheme 5, panel A). Next, the amine and acid were protected to deliver 57. The free phenol was engaged in a substitution reaction with previously synthesized Alloc-protected amino-propyl-bromide 17, to afford 58, which was Boc-deprotected and saponified in a one-pot protocol. Finally, Fmoc protection resulted in 59. Additionally, constrained building block 61 was prepared from Fmoc-2-formyl-L-Trp-OH 60, following a literature procedure (Scheme 5, panel B).⁸⁰ Compound 61 served as a precursor for the Aia-Arg-bearing sequence 64.

Binding Affinity and APJ Signaling of the Constrainded and Positively Charged [Pyr1]-Apelin-13 Analogues

Impact of Constrained Analogues Bearing a Positive Charge at the C-Terminus on Binding Affinity and Receptor Signaling

With the combination of both modifications, we observed for the analogue carrying Tic amino acid (62 (SBL-AP-059)) native-like affinity and Gα_i1 signaling, whereas β-arrestin 2 recruitment is decreased 4-fold, with the maximum efficacy of 87% (Table 3, Figure 2). Replacing 1-Nal¹² with the amino-indoloazepinone (Aia) template and introducing a C-terminal Orn (63 (SBL-AP-049); Table 3, Figure 2) produced a decrease in receptor binding affinity and Gα_i1 signaling, but more importantly in β-arrestin 2 recruitment (63, K_i 24 nM; EC₅₀ Gα_i1 22 nM; EC₅₀ β-arr2 864 nM). However, replacement of 1-Nal¹² in 52 with the constrained Aia residue resulted in a 4-fold loss of binding affinity (52, K_i 1.2 nM vs 64 (SBL-AP-050), K_i 4.7 nM), a 2-fold loss in Gα_i1 activation and a 9-fold loss in β-arrestin 2 recruitment (Table 3, Figure 2). Interestingly, 63 and 64 behaved as partial agonists on the β-arrestin 2 pathway with E_max values of 48% and 82%, respectively. These data suggest that the substituted Tic analogue (62) does not provide optimal positive charge orientation to optimize signaling bias, but that the Aia moiety, used as a 1-Nal surrogate, provides one of the lowest maximal β-arrestin 2 recruitment efficacies. Noteworthily, both types of constraints improved proteolytic stability (vide infra).

Table 3. Chemical Structures of Constrained Ligands Bearing a Positive Charge and Affinity, Functional Activities of C-Terminally Modified [Pyr¹]-Apelin-13 Analogues.

compound no.	compound code	peptide sequence	Ki binding (nM)a	Gαi1 (nM)b	β-arrestin 2 (nM) (Emax%)c	rat plasma t1/2 (h)d
Ape13		Pyr-R-P-R-L-S-H-K-G-P-Nle-Pro-Phe-OH	0.7 ± 0.1	1.1 ± 0.1	40 ± 4	0.5 ± 0.1e
ref	KT04-39	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-F-OH	0.12 ± 0.01	4.4 ± 0.6	41 ± 4	0.4 ± 0.1e
62	SBL-AP-059	Pyr-R-P-R-L-S-H-K-G-P-Nle-1-Nal-(O-6-npropyl-Gn)Tic-OH	0.74 ± 0.01	1.3 ± 0.5	110 ± 28 (87%)	>7
63	SBL-AP-049	Pyr-R-P-R-L-S-H-K-G-P-Nle-Aia-Orn-OH	24 ± 4	22 ± 13	864 ± 321 (48%)	5.3 ± 0.1
64	SBL-AP-050	Pyr-R-P-R-L-S-H-K-G-P-Nle-Aia-Arg-OH	4.7 ± 0.1	13 ± 6	993 ± 616 (82%)	4.8 ± 0.4

^aKi was calculated from experimental IC₅₀ values (the concentration of ligand that displaces 50% of radiolabeled [¹²⁵I][Nle⁷⁵,Tyr⁷⁷]Pyr-apelin-13) using the Cheng–Prusoff equation.⁷⁸ Values represent the mean ± SEM of two or three experiments, each performed in duplicate.

^bEC₅₀ corresponds to the concentration of ligand that produces 50% dissociation of Gα_i1 from the Gβγ subunits. All compounds showed an E_max Gα_i1 > 97%.

^cEC₅₀ is the concentration of ligand that produces 50% recruitment of β-arrestin 2 to the apelin receptor. EC₅₀ values represent the mean ± SEM of three experiments, each performed in triplicate. All E_max values can be found in Table S2 in the Supporting Information.

^dValues represent the mean ± SEM of three experiments. > 7 h means between 7 and 24 h.

^eReported in ref (50).

Figure 2.

Functional assay and plasma stability of 62, 63, and 64 compared to Ape13. BRET concentration–response curves monitoring β-arrestin 2 recruitment (A) and degradation curves in rat plasma (B). Data are represented as the mean ± SEM of at least three separated experiments.

In Vitro Stability in Rat Plasma

The [Pyr¹]-Apelin-13 analogues in this study were evaluated for in vitro proteolytic stability in rat plasma to test the impact of amide N_α-substituents and insertion of various phenylalanine derivates on degradation. The N_α-substitution led to a substantial increase in plasma stability (19-41, t_1/2 > 7 h, Table 1), compared with the reference peptide and Ape13 (ref, t_1/2 0.4 h; Ape13, t_1/2 0.5 h). It is well-known that ACE2 is responsible for the main degradation of Ape13 analogues by hydrolysis between the last two C-terminal amino acids (Figure 1). Therefore, as expected, the addition of substituents on the backbone amide bond between 1-Nal¹² and Phe¹³ significantly increased plasma stability, presumably by blocking cleavage by ACE2. Furthermore, incorporation of small polar substituents on the para-position of Phe¹³ (42–45) did not significantly impact plasma stability (Table 2). The same observation was made for the charged aliphatic substituents in compounds 50-51. In contrast, the addition of benzyloxy derivates of Phe¹³ (46-49), creating much bulkier side chains, increased plasma stability up to 10-fold in comparison with the reference and Ape13 (46, t_1/2 5.4 h vs ref, t_1/2 0.4 h; Ape13, t_1/2 0.5 h). Additionally, aliphatic side chains at the 13th position (as in 52, 54, 55, Table 2) did not elicit any change in plasma stability, with the exception of analogue 53 bearing D-Arg¹³ (t_1/2 2.8 h, Table 2) which exhibits inverted stereochemistry, compared to the other analogues (with an L-configuration). Moreover, constrained analogues (62-64) with backbone-anchored side chains increased plasma stability (62, t_1/2 > 7 h; 63, t_1/2 5.3 h; 64, t_1/2 4.8 h, Table 3). Overall, the alkylated peptide bonds, increased side chain size and inverted stereochemistry at the C-terminus of [Pyr¹]-Apelin-13 analogues decreased degradation in rat plasma. It is noteworthy that the positive charge had no influence on the stability of the compounds.

In Vivo Blood Pressure Measurement

To complete this study, we investigated the impact of applied chemical modifications on blood pressure in anesthetized rats. Apelin analogues were administered to rats via a single intravenous (i.v.) injection at a dose of 19.6 nmol/kg, while the mean arterial blood pressure (MABP) was continuously monitored. As recently reported by our group, the drop in blood pressure reaches the maximum hypotensive effect when [Pyr¹]-Apelin-13 is injected at this bolus dose.⁸¹ Moreover, we have previously shown that the β-arrestin recruitment upon APJ activation correlates with the physiological hypotension effect observed in rats.⁴⁰ Therefore, different analogues harboring an N_α-substituent, a positive charge at the C-terminal, or both, with maximal and reduced β-arrestin 2 recruitment potencies and efficacies were selected for in vivo blood pressure measurements. Analogues 19 and 47 produced a drop in blood pressure comparable to that of Ape13 (19, Max ΔMABP −37 mmHg; 47, Max ΔMABP −34 mmHg vs Ape13, Max ΔMABP −36 mmHg), which could be rationalized by their similarity in terms of β-arrestin 2 recruitment. Analogue 40, the N-alkylated compound with the lowest β-arrestin 2 recruitment potency (EC₅₀ β-arr2 365 nM, E_max 87%), produced a 2-fold lower hypotensive effect than the endogenous ligand (40, Max ΔMABP −17 mmHg). The conformationally constrained and positively charged analogues 63 and 64 were the two most promising compounds in this study due to their good affinity for APJ, strong activation of the Gα_i1 pathway, greatly increased proteolytic stability and limited recruitment of β-arrestin 2, with partial maximum efficacy (63, EC₅₀ β-arr2 864 nM, E_max 48%; 64, EC₅₀ β-arr2 993 nM, E_max 82%). Analogue 64 induced a 2-fold smaller pressure drop than Ape13 (64, Max ΔMABP −18 mmHg vs Ape13, Max ΔMABP −36 mmHg) and, interestingly, analogue 63 did not even affect blood pressure (63, Max ΔMABP −3 mmHg vs Saline, Max ΔMABP −2 mmHg). At a very higher dose (65.0 nmol/kg), however, analogue 63 demonstrated a blood pressure lowering effect that was nevertheless less significant than for Ape13 at this dose (63, Max ΔMABP −22 mmHg vs Ape13, Max ΔMABP −36 mmHg). Both analogues 63 and 64 showed the benefit of partial agonism in the β-arrestin 2 recruitment pathway. All analogues tested exhibit a longer plasma half-life than Ape13, yet no difference was observed in the duration of blood pressure effects.

Conclusion

In this study, 40 analogues of [Pyr¹]-apelin-13, including backbone-alkylated dipeptides as well as modified amino acids, were synthesized. The general observation of this study is that the addition of various alkyl groups (e.g., Et, 4-nitrobenzyl, 4-benzyloxy) onto the N_α of Phe¹³ is well tolerated and produces excellent APJ ligands with affinities in the very low nM range (0.12–0.9 nM). Moreover, potent Gα_i1 activators were obtained (EC₅₀ Gα_i1 0.4–0.9 nM). Interestingly, the introduction of longer and bulkier substituents (e.g., N-(3′-BnO)Bn (38) and N-(3′-PhO)Bn (40)) resulted in compounds with similar binding affinities to Ape13, but significant losses were observed in the signaling potency for the β-arrestin 2 pathway (EC₅₀ β-arr2 258–365 nM) with a small decrease in the maximum efficacy (E_max 87–895%). These modifications also drastically increased stability in rat plasma with half-lives greater than 7 h. IWhen tested in vivo, the compound characterized by a reduced ability to recruit β-arrestin 2 also induced a decreased hypotensive effect (cfr. 19 vs 40 in Figure 3). Incorporation of a positive charge on the side chain of the C-terminal residue (herein a modified Phe or Tyr) provided (near-)native binding affinities, but at the same time unique signaling profiles where partial agonism for β-arrestin 2 recruitment was obtained (e.g., 44, 45, 49, 51–53, 63, 64). Among these, the constrained and positively charged C-terminal dipeptide Aia-Orn in analogue 63 gave rise to a promising analogue with high affinity (K_i 24 nM), high Gα_i1 activation (EC₅₀ Gα_i1 22 nM), low β-arrestin 2 recruitment potency with partial maximum efficacy (EC₅₀ β-arr2 864 nM, E_max 48%), improved stability (t_1/2 5.3 h) and a greatly decreased hypotensive effect. In conclusion, we report the useful design and synthesis of apelin-13 analogues with high binding affinities and improved plasma stability by means of minimal structural modifications at the apelin C-terminus. These bioactive molecules represent novel pharmacological tools that may help shed new light on apelin-related pathologies.

Figure 3.

Effects of Ape13 and selected analogues on blood pressure in anesthetized rats. Recorded mean arterial blood pressure (ΔMABP) after intravenous injection of (A) Ape13 and selected analogues at 19.6 nmol/kg and (B) Ape13 and 63 at 65.0 nmol/kg. Maximal reduction of mean arterial pressure (Max ΔMAP) of (C) Ape13 and selected analogues at 19.6 nmol/kg and (D) Ape13 and 63 at 65.0 nmol/kg. Data are represented as the mean ± SEM (n = 5). Statistical analyses were performed with a one-way ANOVA followed by a Dunnett’s multiple comparisons test. **, p < 0.01, ***, p < 0.001, and ****, p < 0.0001 vs Ape13.

Experimental Section

General Information and Compound Characterization

Unless stated otherwise, all commercial chemicals were used without further purification. ¹H and ¹³C NMR spectra were recorded using a Bruker Avance II 500 MHz at ambient temperature with the frequencies of ¹H and ¹³C set, respectively, at 500.03 and 125.74 MHz. The chemical shifts were reported in delta (δ) units in parts per million (ppm) relative to the signal of the deuterated solvent. For CDCl₃, the singlet in ¹H NMR was calibrated at 7.26 ppm and the central line of the triplet in ¹³C NMR at 77.0 ppm. When recording in MeOD-d₄ or DMSO-d₆, the calibration was performed, respectively, at 3.31 and 2.50 ppm for ¹H NMR and 49.0 and 39.5 ppm for ¹³C NMR. Assignments were made using one-dimensional (1D) ¹H and ¹³C spectra and two-dimensional (2D) HSQC, HMBC, and COSY spectra. Multiplicities were described as singlet (s), doublet (d), triplet (t), quartet (q), multiplet (m), and broad (br), or a combination thereof. The corresponding coupling constants (J values) were reported in Hertz (Hz). Analytical RP-HPLC was performed on a VWR-Hitachi Chromaster HPLC with a Chromolith HighResolution RP-18C column from Merck (150 × 4.6 mm, 1.1 μm, 150 Å). The flow rate was 3 mL/min and UV detection occurred at 214 nm. The solvent system used consisted of 0.1% TFA in ultrapure water (A) and 0.1% TFA in acetonitrile (B) with a gradient from 3% B to 100% B over a 6 min runtime. For LC/MS analysis, the HPLC unit used was a Waters 600 system combined with a Waters 2487 UV detector at 215 nm and as stationary phase an EC 150/2 NUCLEODUR 300-5 C18 ec-column (150 × 2.1 mm², 3 μm, 300 Å). The solvent system used was 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B) with a gradient going from 3% to 100% B over 20 min with a flow rate of 0.3 mL/min. The MS unit, coupled to the HPLC system, was a Micromass QTOF-micro system. For the high-resolution mass spectroscopy, the same MS system was used with reserpine (2.10–3 mg/mL) solution in H₂O:CH₃CN (1:1) as reference. Microwave syntheses were performed using a Biotage Initiator⁺ SP Wave. Automated flash chromatography was performed using Grace Reveleris X2 system equipped with an ELSD and UV detector (254 or 280 nm). The normal phase columns used for the systems were Grace Silica Flash Cartridges of 40 g with a flow rate of 40 mL/min unless stated otherwise. Semipreparative RP-HPLC-purifications were done using a Gilson HPLC system with Gilson 322 pump equipped with a INTERCHIM Vydac 150HC C18 (250 × 22 mm², 10 μm) column and Waters UV/vis-156 detector at 215 nm. The same solvent system was used as applied for the analytical RP-HPLC with a flow rate of 20 mL/min.

Binding Experiments

Binding experiments were carried out on the cell membranes of HEK293 stably expressing the YFP-tagged human APJ receptor. Cells were frozen at −80 °C for storage and only thawed right before the experiments. For membrane extraction, cells were suspended in EDTA solution (1 mM EDTA and 10 mM Tris-HCl), transferred to a falcon tube, and centrifuged at 1600 g for 15 min at 4 °C. The precipitate was suspended in binding buffer (50 mM Tris-HCl, 0.2% BSA, pH 7.4). Binding assay was carried out in 96-well plates: 15 μg of membrane proteins was incubated with 0.2 nM radiolabeled [¹²⁵I][Nle⁷⁵,Tyr⁷⁷]Pyr-apelin-13³⁵ (820 Ci/mmol) and the test ligand with a range of concentrations from 10^–5 to 10^–11 M in a total volume of 200 μL for 1 h at room temperature. The incubation mixtures were filtered through a glass fiber filter (Millipore, preabsorbed of PEI 0.5% for 2 h at 4 °C) to remove unbound ligands, and the filtered membranes were washed three times with 150 μL of cold binding buffer (4 °C). The γ emission was measured using a γ-counter 1470 Wizard from PerkinElmer (Waltham) (80% efficiency). Nonspecific binding did not exceed over 5% of the total signal (determined by incubation with 10^–5 M unlabeled [Pyr¹]-apelin-13). IC₅₀ values, determined from those results using GraphPad Prism 9, represent the concentration of the tested ligand displacing 50% of the radiolabeled ligand from the receptor. The K_D of [Pyr¹]-apelin-13 is 1.8 nM, determined by the saturation binding assay. The dissociation constant K_i value was calculated from IC₅₀ using the Cheng–Prusoff equation, and results were displayed as mean ± SEM of two to three independent experiments, each done in duplicate.⁷⁸

BRET Assays for Gαi1 Activation and β-Arrestin 2 Recruitment

HEK293 cells seeded in T175 flasks were allowed to grow in high-glucose DMEM supplemented with 10% FBS, 100 U/mL penicillin/streptomycin at 37 °C in a humidified chamber at 5% CO2. All transfections were carried out with PEI. After 24 h, cells were transfected with the plasmids coding for hAPJ, Gαi1-RlucII(91), GFP10-G_γ2, and G_β1 (for the BRET-based Gα_i1 activation assay) or coding for hAPJ, CAAX-GFP10 and RlucII-βarrestin 2.^51,82,83 To perform BRET assays, cells were transferred into white 96-well plates BD Bioscience (Mississauga, Canada) at a concentration of 50 000 cells/well 24 h after transfection and incubated at 37 °C overnight. Cells were then washed with phosphate-buffered saline (PBS), and 90 μL of Hank’s balanced salt solution (HBSS) was added to each well. Then, cells were stimulated with analogues at concentrations ranging from 10^–5 to 10^–11 M for 5 min at 37 °C (Gα_i1) or for 30 min at room temperature (β-arrestin 2). After stimulation, 5 μM coelanterazine 400a was added to each well and the plate was read using the BRET2 monochromatic filter set of a MithrasLB943 plate reader (Berthold Technologies, Germany). The BRET ratio was determined as GFP10em/RlucIIem. Data were plotted, and EC₅₀ values were determined using GraphPad Prism 9. Each data point represents the mean ± SEM of at least three different experiments each done in triplicate.

Rat Plasma Stability

In a 96-well plate, 6 μL of compound (1 mM aqueous solution) was incubated with 27 μL of plasma (from male Sprague–Dawley rat) at 37 °C for 0, 1, 2, 4, 7, and 24 h. Degradation was quenched by adding 140 μL of 1% formic acid in acetonitrile-ethanol (1:1) solution containing N,N-dimethylbenzamide 0.25 mM (internal standard). This mixture was transferred to a filter plate Impact Protein Precipitation (Phenomenex, California), and another 96-well UPLC plate was placed at the bottom to collect filtrates. Both plates were centrifuged at 800 g for 10 min at 4 °C. The filtrates were diluted with 30 μL of distilled water and analyzed using an Acquity UPLC-MS system class H (column Acquity UPLC CSH C18 (2.1 mm × 50 mm), 1.7 μm particles with pores 130 Å). The degradation was calculated by comparing the mass spectrum of each point with those at 0 min (plasma-inactivated before adding the compound). Peptide half-life was calculated from the degradation curve using the exponential one-phase decay function in GraphPad Prism 9. Data were presented as mean ± SEM when the data are >1 h and <7 h of at least three different experiments each done in simplicate.

In Vivo Blood Pressure Measurement

Adult male Sprague–Dawley rats, 8–10 weeks of age (Charles River Laboratories, St-Constant, Quebec, Canada), were kept on a 12 h light/12 h dark cycle with access to food and water ad libitum. The animal experimental protocols in this study were approved by the Animal Care Committee of Université de Sherbrooke and complied with policies and directives of the Canadian Council on Animal Care.

Male Sprague–Dawley rats (250–300 g) anesthetized with ketamine/xylazine injection (87/13 mg/kg intramuscular (i.m.)) were placed in supine position on a thermostatic pad. A blood pressure monitoring catheter (PE 50 filled with heparinized saline) was inserted into the right carotid artery and connected to a Micro-Med transducer (model TDX-300, Calabasas) and a MicroMed blood pressure analyzer (model BPA-100c). Bolus injection of vehicle (isotonic saline) followed 5 min later by the injection of Ape13, saline, or modified analogues, 19, 40, 47, 63, or 64, (given at 19.6 or 65 nmol/kg; volume of 0.25 mL over 10 s) was carried out through another catheter (PE10) inserted into the left jugular vein. To remove residual injected substances, this i.v. catheter was flushed with saline (0.2 mL) immediately after each injection.

Glossary

Abbreviations

1-Nal

3-(1-naphthyl)-l-alanine

AA

amino acid

ACE2

angiotensin-converting enzyme 2

ACN

acetonitrile

Aia

amino-indoloazepinone

AMP

adenosine monophosphate

Ape13

[Pyr¹]-apelin-13

Arg

arginine

Atm

atmosphere

Bn

benzyl

Boc

tert-butyloxycarbonyl

BRET

bioluminescence resonance energy transfer

BSA

bovine serum albumin

CSH

charged surface hybrid

DCM

dichloromethane

DIAD

diisopropyl azodicarboxylate

DIPEA

N,N-diisopropylethylamine

DKP

diketopiperazine

DMEM

Dulbecco’s modified eagle medium

DMF

N,N-dimethylformamide

ECL2

extracellular loop

EDC

1-ethyl3-(3-(dimethylamino)propyl)carbodiimide

EDTA

ethylenediaminetetraacetic acid

ERK

extracellular-signal-regulated kinase

Fmoc

9-fluorenylmethoxycarbonyl

GFP

green fluorescent protein

Gn

guanidine

GPCR

G-protein-coupled receptor

hAPJ

human APJ

HATU

1-[bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxide hexafluorophosphate

HBSS

Hank’s balanced salt solution

HEK

human embryonic kidney

HRMS

high-resolution mass spectrometry

Lys

lysine

MABP

mean arterial blood pressure

Met

methionine

MS

mass spectrometry

NEP

neprilysin

Nle

norleucine

NMM

N-methylmorpholine

Orn

ornithine

PBS

phosphate-buffered saline

PEI

polyethylenimine

Ph

phenyl

Phe

phenylalanine

PRCP

proline carboxypeptidase

Pro

proline

Pyr

pyroglutamic acid

Raf

rapidly accelerated fibrosarcoma

RhoGEF

Rho-guanine nucleotide exchange factor

RlucII

Renilla luciferase

rt

room temperature

SAR

structure–activity relationship

SEM

standard error of the mean

SPPS

solid-phase peptide synthesis

TFA

trifluoroacetic acid

THF

tetrahydrofuran

Tic

tetrahydroisoquinoline carboxylic acid

TIPS

triisopropylsilane

TM

transmembrane domain

Tris

tris(hydroxymethyl)aminomethane

Trp

tryptophan

Tyr

tyrosine

UPLC

ultrahigh-performance liquid chromatography

YFP

yellow fluorescent protein

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsptsci.2c00219.

• Synthetic procedures for unnatural amino acids and compound characterization; SPPS procedures; HRMS spectra; analytical UPLC-MS spectra; structures and molecular formula strings of the investigated compounds; competitive binding curves, saturation BRET curves monitoring Gα_i1 activation and β-arrestin 2 recruitment; degradation curves in rat plasma (PDF)

Author Contributions

^∥ L.T. and R.V.D.H. contributed equally to this study. S.B., P.-L.B. and P.S. contributed equally to directing this study.

Financial support from Université de Sherbrooke, the Canadian Institutes of Health Research, the Canada Foundation for Innovation, the Fonds de la Recherche du Québec en Santé (FRQS) is gratefully acknowledged. The Canadian Francophonie Scholarship Program (PCBF) is also acknowledged for scholarship grants to K.T. P.S. is the recipient of the Canada Research Chair in Neurophysiopharmacology of Chronic Pain. E.M. and P.L.B. are members of the FRQNT-funded Proteo Network. R.V.D.H. and S.B. thank the Strategic Research Program (SRP50) of the Research Council of VUB for the financial support.

The authors declare no competing financial interest.

Special Issue

Published as part of the ACS Pharmacology & Translational Science virtual special issue “GPCR Signaling”.