Selective homodimerization of unprotected peptides using hybrid hydroxydimethylsilane derivatives

Abstract

We developed a simple and straightforward way to dimerize unprotected peptide sequences that relies on a chemoselective condensation of hybrid peptides bearing a hydroxydimethylsilyl group at a chosen position (either C-ter, N-ter or side-chain linked) to generate siloxane bonds upon freeze-drying. Interestingly, the siloxane bond sensitivity to hydrolysis is strongly pH-dependent. Thus, we investigated the stability of siloxane dimers in different experimental conditions. For that purpose, ²⁹Si, ¹³C and ¹H NMR spectra were recorded to accurately quantify the ratio of dimer/monomer. More interestingly, we showed that ¹H resonances of the methylene and methyl groups connected to the Si can be used as sensitive probes to monitor siloxane hydrolysis and to determine the half-lives of the dimers. Importantly, we showed that the dimers were rather stable at pH 7.4 (t_1/2 ≈ 400 h) and we applied the dimerization strategy to bioactive sequences. Once optimized, three dimers of the growth hormone releasing hexapeptide (GHRP-6) were prepared. Interestingly, their pharmacological evaluation revealed that the activity of the dimeric ligands could be switched from agonist to inverse agonist depending on the position of dimerization.

---

Introduction

Homodimerization of bioactive molecules is a well-known strategy to improve the pharmacological profile of the corresponding monomer. Dimerization was first applied to increase the local concentration of a ligand in the neighborhood of a receptor. In some reported cases, such a strategy leads to a large increase of the dimer affinity for its receptor. There are three hypotheses to explain such a phenomenon.¹ First, a dimer may connect two binding sites in close proximity (i.e. two receptors or an additional binding subsite). Alternatively, a dimer may first bind to a single receptor and help in the recruitment of another receptor. This is particularly relevant when signaling is initiated by receptor oligomerization.² At last, a “statistical binding effect” has to be taken into account: binding of a dimeric ligand is favored by the higher local concentration of binding elements at the neighborhood of the receptor. The dimerization may also improve the stability towards proteases particularly in the case of peptide ligands and may affect the selectivity³ and the biological activity of the resulting compounds. Dimerization strategy was applied to numerous peptide hormones including serotonin,^4,5 enkephalins,⁶ bradykinin,⁷ TGF-β.⁸ More recently, dimerization was also applied to the design of protein–protein interaction inhibitors.⁹ Several chemical strategies, in solution or on solid support, were used for peptide dimerization using specific functional groups and spacer arms. As examples of peptide dimer syntheses, we can cite the anchorage of a bifunctionalized spacer on a solid support and the simultaneous synthesis of the two ligand peptides from C to N-terminus,¹⁰ the coupling of two identical protected peptides which bear a single free reactive function on a bifunctionalized spacer.⁷

In this study, we focused our efforts on an alternate dimerization process based on hydroxydimethylsilane peptides. To our knowledge, most of the existing strategies require orthogonal protecting groups and multistep procedures to obtain homodimeric peptides at the exception of dimerization of unprotected peptide sequence containing cysteine residue through the formation of a disulfide bond.¹¹ Advantageously, the use of dimethylhydroxysilane peptides allows the dimerization of any peptides, even those containing a Cys residue and greatly simplifies and speeds up the process. Indeed, dimerization occurs chemoselectively in the presence of unprotected side chains.

Moreover, it is of particular interest that hybrid dimethylhydroxysilane monomers are prepared by classical peptide synthesis and can be purified before dimerization. We recently applied this strategy to the synthesis of linear and branched peptide-polymers and tripeptide dimers.¹² In this work, we aimed at demonstrating that the strategy is not only applicable to the synthesis of N-ter dimers but also to C-ter and internal bioactive peptide sequence dimerization. Interestingly we proved here that siloxane formation simply occurred during lyophilization after preparative RP-HPLC of the hybrid monomer. The ²⁹Si NMR signals were different in siloxane bonds and silanols but we demonstrated that, ¹H NMR could be used advantageously to measure the SiOH/SiOSi ratio. Thanks to quick and sensitive ¹H NMR analyses, we monitored the stability of siloxane bonds in different solvents but more importantly in water at various pHs. At last, we synthesized three types of dimers based on the sequence of the growth hormone releasing hexapeptide (GHRP-6). The dimerization strategy was applied at three different positions of the peptide sequence: C-terminal, N-terminal and intra-sequence through a modified lysine residue (Scheme 1). Then, the binding and signaling properties of these dimers were evaluated on intact HEK293T cells expressing the growth hormone secretagogue receptor 1a (GHS-R1a) involved in important physiological pathways.¹³

Scheme 1 Synthesis of GHRP-6 peptide homodimers based on dimethylhydroxysilyl peptides. 5: N-terminal dimerization; 7: C-terminal dimerization; 9: internal residue dimerization.

Materials and methods

All solvents and reagents were used as supplied. Solvents used for LC/MS were of HPLC grade. Dichloromethane (DCM); N,N-dimethylformamide (DMF) were obtained from Carlo Erba. Fmoc amino acid derivatives, [benzotriazol-1-yloxy(dimethylamino)methylidene]dimethylazanium hexafluorophosphate (HBTU), Fmoc-Rink amide CM resin, 2-chloro chlorotrityl PS resin, were purchased from Iris Biotech (Marktredwitz, Germany). Diisopropylcarbodiimide (DIC), diisopropylethylamine (DIEA), trifluoroacetic acid (TFA) were obtained from Sigma-Aldrich (St. Louis, MO, USA). Hexafluoroisopropanol (HFIP) and triisopropylsilane (TIS) were obtained from Alfa Aeser and Acros respectively. 3-Isocyanatopropyldimethylchlorosilane was purchased from Sikemia (France).

The following abbreviations were used: ICPDMCS – isocyanatopropyldimethylchlorosilane. Other abbreviations used are recommended by the IUPAC-IUB Commission.

Peptide synthesis

Microwave assisted peptide syntheses were performed using Fmoc/tBu SPPS strategy on a Liberty™ Microwave Peptide Synthesizer (CEM Corporation, Matthews, NC) providing MW irradiation at 2450 MHz. Temperature consign was set up at 70 °C as previously described.^14,15

Ethane-1,2-diamine anchoring on chloro-2-chlorotrityl PS resin

The chloro-2-chlorotrityl chloride PS resin (4.00 g, 8.8 mmol, initial theoretical loading = 2.2 mmol g⁻¹) was conditioned for 15 min in DCM. The resin was filtered and a solution of ethane-1,2-diamine (1.175 mL, 17.6 mmol, 2 eq.) in DMF (40 mL) containing DIEA (3.060 mL, 17.6 mmol, 2 eq.) was added. The resin was stirred overnight and washed with DMF (2×), DCM/MeOH 1/1 v/v (2×), DMF (2×), DCM (2×) and dried overnight in vacuo. The experimental loading of the resin was determined after the first amino acid coupling. Optical density of dibenzofulvene-piperidine adduct formed during Fmoc deprotection with a piperidine/DMF 2/8 v/v solution was measured at 299 nm and the loading was calculated according to a previously reported method.¹⁵

Coupling step

On a 0.1 mmol resin scale (either H-Rink amide ChemMatrix resin 1.2 mmol g⁻¹ or H₂N(CH₂)₂NH-chlorotrityl-PS resin 0.8 mmol g⁻¹), coupling reactions were performed with 5 eq. of amino acid (0.5 mmol, 2.5 mL of 0.2 M solution in DMF), 5 eq. of HBTU (0.5 mmol, 2.5 mL of 0.2 M solution in DMF), and 10 eq. of DIEA (1 mmol, 0.5 mL of 2 M NMP solution). Resin was stirred for 90 minutes at room temperature or heated at 70 °C under microwave irradiation (50 W) for 10 minutes. The resin was filtered and then washed with DMF (2×) and DCM (2×).

Fmoc deprotection step

Fmoc removal was carried out in DMF/pip 8/2 v/v solution for 3 minutes under stirring, at room temperature. The resin was filtered and a DMF/pip solution was added again and stirred for 20 minutes. When MW irradiation was used (50 W, 70 °C), two cycles of 30 s and 3 minutes were used. In both cases, the resin was filtered and then washed with DMF (2×) and DCM (2×).

Alloc deprotection

The Alloc lysine side-chain protecting group of the peptide anchored on a Rink amide resin was removed selectively. Phenylsilane (20 eq.) was added to a solution of Pd(PPh₃)₄ (0.1 eq.) in DCM (4 mL g⁻¹ of resin). The resulting solution was poured on the peptidyl-resin (1 eq.) in an open vessel. The resin was stirred for 2 h and then washed with DCM (2×) and DMF (2×). A solution of potassium O-ethyl dithiocarbonate in DMF (50 mg in 10 mL) was added and the resin was stirred for 20 minutes before final washings with DMF (2×) and DCM (2×).

HFIP-mediated cleavage of protected peptide from 2-chlorotrityl resin

The peptidyl-resin was stirred in a solution of hexafluoroisopropanol/DCM 1/4 v/v (40 mL g⁻¹ of resin) for 30 min. The resin was filtered. The solution was concentrated under vacuum to yield the protected peptide.

Silylation in solution

The peptide to be silylated was solubilized in anhydrous DMF under argon. DIEA (4 eq.) and 3-isocyanatopropylchlorodimethylsilane (1 eq.) were added. The released hydrogen chloride released was removed under an argon flow. The mixture was stirred for 2 h and then concentrated under vacuum. The hybrid peptide was precipitated by addition of diethyl ether and dried under vacuum.

Side-chain deprotection in solution

Side-chain amino acid protecting groups were removed in TFA/triisopropylsilane/H₂O 95/2.5/2.5 v/v/v. After 30 minutes under stirring, the solution was concentrated under vacuum. The unprotected dimethylhydroxysilyl peptide was precipitated in diethyl ether and dried under vacuum.

Silylation on solid support

3-Isocyanatopropylchlorodimethylsilane (3 eq.) was added in a solution of DIEA (3 eq.) in anhydrous DMF (10 mL g⁻¹ of resin). The resulting solution was poured on the peptidyl resin and the resin was stirred overnight before being filtered and washed with DMF (2×) and DCM (2×).

Cleavage from Rink amide ChemMatrix resin along with side-chain deprotection

The peptide was cleaved from Rink amide ChemMatrix resin in TFA/triisopropylsilane/H₂O 95/2.5/2.5 v/v/v. The resin was stirred in the cleavage mixture for 2 × 2 h. The resin was filtered and the solution was concentrated under vacuum. The unprotected peptide was precipitated by addition of diethyl ether.

Purification

Hybrid hydroxydimethylsilyl-peptides were purified by preparative chromatography. Purifications were performed using a Waters HPLC 4000 instrument, equipped with a UV detector 486 and a Waters Delta-Pack 40 × 100 mm, 100 Å, 15 μm, C18 reversed-phase column using a flow rate of 50 mL min⁻¹. Solvents employed were water/0.1% TFA and ACN/0.1% TFA. UV detection was performed at 214 nm. Only the hybrid monomers were recovered. Condensation into dimers occurred after purification, during lyophilization.

NMR analysis

NMR samples contained 0.06 mmol (3-aminopropyl)dimethylsilyl compounds dissolved in DMSO-d₆ or D₂O (600 μl). The ²⁹Si, ¹³C and ¹H spectra were collected on a Bruker AVANCE III 600 MHz spectrometer at 298 K. Peptide dimers (4 μmol) were dissolved in DMSO-d₆ and spectra were collected on a Bruker AVANCE III HD 400 MHz spectrometer at 298 K. Both spectrometers were equipped with a 5 mm BroadBand Observation BBFO probe. The deuterated solvent was used as an internal deuterium lock. Chemical shifts are given in δ ppm calibrated with residual protic solvent (e.g., DMSO: 2.50 ppm). A quantitative inverse-gated experiment was used for ²⁹Si NMR using a recovery delay of 60 s (128 scans with a spectral width of 5950 Hz). The same sequence was used for quantitative ¹³C NMR with the same recovery delay (64 scans with a spectral width of 15 120 Hz). For determination of silicon chemical shifts, a DEPT sequence was used to benefit from the polarization transfer of hydrogens near the silicon atom (256 scans, spectral width of 11 900 Hz and recovery delay of 2 s). ¹H NMR spectra were typically recorded using 16 scans with a spectral width of 6000 Hz and a recovery delay of 2 s. Spectra were processed and visualized either with Topspin 3.2 (Bruker Biospin) on a Linux station or with MestReNova 6.0.2 (Mestrelab research).

Ligand binding assay

K_i values were determined from binding competition experiments performed on intact HEK293T cells expressing the GHS-R1a using a Homogenous Time Resolved Fluorescence (HTRF) assay as previously described.¹⁶ The HTRF signal was collected in a PHERAstar microplate reader (BMG LABTECH). K_i values were obtained from binding curves using GraphPad Prism software (GraphPad Software, Inc., San Diego, CA).

Inositol phosphate assay

Inositol phosphate accumulation assay was carried out 48 h after transfection on adherent cells in 96-well plates at a density of 50 000 cells per well. IP1 production was measured using the IP-One HTRF kit (Cisbio Bioassays) as previously described.¹⁶ Briefly, cells were stimulated for 30 min at 37 °C with the ligand to be tested in 70 μl of IP1 stimulation buffer. An anti-IP1 antibody labelled with Lumi4-Tb (15 μl) and an IP1-d2 derivative (15 μl), were added on cells. The medium was incubated for 1 h at room temperature. Signals at 665 nm and 620 nm were detected using a PHERAstar (BMG LABTECH) fluorescence reader. Values are expressed as ΔF. ΔF corresponded to: (ratio 665 nm/620 nm of the assay − ratio 665 nm/620 nm of the negative control)/ratio 665 nm/620 nm of the negative control. The negative control corresponded to the Lumi4-Tb blank and was used as an internal assay control. Inositol phosphate accumulation was expressed as the percentage of the maximal ghrelin response using the formula: (ΔF mock cells − ΔF receptor transfected cells)/(ΔF mock cells − ΔF maximal ghrelin stimulation for receptor transfected cells). EC₅₀ values were determined from dose–response curves using GraphPad Prism software (GraphPad Software, Inc., San Diego, CA).

Results and discussion

In the frame of this study, characterization of the dimeric siloxane species was of first importance. In vitro biological assays mostly require small quantities of bioactive compounds and thus, peptide analogues to be tested are mostly synthesized at a small scale (<0.1 mmol). In this context, sensitive and non-destructive methods have to be privileged. In a general manner, NMR is the method of choice considering nuclei having detectable isotopes of high abundance. The distinction between dimethylsiloxane (dimer) or hydroxydimethylsilane (monomer) could be achieved by ²⁹Si NMR; chemical shifts being different between a monomer and a dimer. However, due to the low natural abundance and the rather weak detection sensitivity of ²⁹Si isotope, ²⁹Si NMR analyses was not sufficient for a convenient characterization of hybrid peptides, their corresponding dimers and the study of the dimers stability. Thus, despite a narrow spectral width and the numerous peptides resonances in the ¹H dimension, we first showed that quick ¹H NMR analyses allowed us to readily quantify the relative proportions of monomers and dimers both in D₂O and DMSO-d₆ and to measure the half-lives of the dimers in different conditions. Indeed, the ¹H resonances of methylene and methyl groups connected to the Si allow to discriminate the dimeric from the monomeric peptide state.

²⁹Si, ¹³C and ¹H NMR chemical shifts assignment of (3-aminopropyl)dimethylsilyl compounds as model for the hybrid peptide characterization

First, we assigned the NMR chemical shifts of the simple model 1,3-bis(3-aminopropyl)tetramethyl disiloxane (Fig. 1) and its monomer, which possessed the same dimerization arm as the hybrid peptides of the study.

Fig. 1 Model compounds for NMR determination of ¹H, ¹³C and ²⁹Si shifts. Representative atoms of interest for the identification of both dimer and monomer are marked in bold.

Compound 1a was obtained from commercially available 1,3-bis(3-aminopropyl)tetramethyl disiloxane poured into hydrogen chloride ether solution. On the contrary of its neutral counterpart, 1a was soluble in water enabling the NMR analysis in aqueous solutions. Compound 1a was solubilized either in DMSO-d₆ or D₂O and analyzed by ¹H, ¹³C and ²⁹Si NMR. Results are reported in Table 1 (first column, dimer 1a). As expected, when compound 1a was left in D₂O for a prolonged time, monomer 2a appeared upon hydrolysis. This latter was analyzed by NMR (Table 1). Hydrochloric acid was added to dimer 1a in DMSO-d₆ to force hydrolysis and subsequently identify the chemical shifts of monomer 2a (Table 1). Since all the peptides were purified in trifluoroacetic acid-containing solvents and therefore recovered as trifluoroacetate salts, dimer 1b was also prepared. It was prepared from 1,3-bis(3-aminopropyl)tetramethyl disiloxane poured into trifluoroacetic acid ether solution and solubilized in DMSO-d₆ or D₂O. Partial hydrolysis occurred with time and the mixture of monomer 2b and dimer 1b was analyzed by ¹H, ¹³C and ²⁹Si NMR. Chemical shifts of monomer 2b and dimer 1b are reported in Table 1.

Table 1 ¹H, ¹³C and ²⁹Si NMR shifts (in ppm) of siloxane dimer 1 and monomer 2

Solvent	Nucleus	Dimer 1a		Monomer 2a		Dimer 1b		Monomer 2b
		–CH2Si	CH3Si	–CH2Si	CH3Si	–CH2Si	CH3Si	–CH2Si	CH3Si
DMSO-d6	^1H	0.53	0.06	0.48	0.00	0.52	0.06	0.48	0.01
	^13C	14.6	0.2	14.5	0.0	14.5	0.1	14.4	−0.1
	^29Si	8.0		10.9		8.0		10.8
D2O	^1H	0.71	0.22	0.74	0.23	0.70	0.21	0.72	0.22
	^13C	13.6	−1.2	12.9	−2.3	13.6	−1.2	12.9	−2.3
	^29Si	11.4		18.2		11.4		18.1

---

Analyses of monomer/dimer mixtures by ²⁹Si NMR showed two signals at 8.0 and 10.8 (±0.1) ppm in DMSO-d₆ and 11.4 and 18.1 (±0.1) ppm in D₂O that corresponded to the silicon signal involved in siloxane Si–O–Si and hydroxysilane Si–OH bonds respectively (Fig. 2). Whatever the conditions of analysis, ²⁹Si NMR unambiguously allowed the identification of hybrid monomeric and dimeric species. The ¹³C NMR spectra also allowed us to readily discriminate the monomer from the dimer signals looking at the methyl and the methylene resonances connected to the Si. ²⁹Si and ¹³C NMR have the advantage to display a large spectral width and only few ²⁹Si and ¹³C signals are present in our systems. Nevertheless, due to the low natural abundance (4.7% and 1.1%, respectively) and low NMR sensitivity of ²⁹Si and ¹³C, long acquisition times are required even at 0.1 M concentration. In this context, we investigated the characterization of all derivatives using ¹H NMR spectra which are far more sensitive than ²⁹Si and ¹³C NMR despite a narrowest spectral width and the larger number of ¹H resonances. In this context, we switched our attention to the NMR signals of the hydrogen on carbon atoms in alpha position to the silicon atoms (Fig. 1). Interestingly, ¹H chemical shifts of –CH₂Si and CH₃Si of the monomer and dimer were localized in a rather clear region of the ¹H NMR spectra of the hybrid peptides (≤0.7 ppm) (Table 1). These resonances should not overlapped with those of the methyl groups of the Leu, Ile and Val expected around 0.7–0.8 ppm. In addition, we showed with the model compounds 1a–b and 2a–b that in DMSO-d₆ ¹H NMR chemical shifts of the monomer and dimer were clearly different whatever the amine counter-ion, thereby hybrid dimers and monomers could be unambiguously identified by a simple ¹H NMR. In D₂O, ¹H NMR methyl signals of monomer and dimer as hydrochloride salts were partially overlayed. However, the counter-ion switch from hydrochloride to trifluoroacetate induced a splitting of monomer and dimer signals (ESI Fig. S22†). As a consequence, TFA salts should be preferred for a better signal separation in D₂O. In contrast to the ²⁹Si and the ¹³C resonances, the ¹H chemical shifts values of the –CH₂Si (∼0.7 ppm) and CH₃Si (∼0.2 ppm) were rather close in both species and sensitive to the pH which could lead to confusion between monomer and dimer during assignment. When both monomer and dimer were present, signal assignment was easy as NMR peaks always came out in the same order: in D₂O, dimer protons were typically more shielded than monomer hydrogens. In case of any doubt about a sample containing only one species, ²⁹Si NMR analyses had to be undertaken. In order to significantly decrease the experiment time, we used a DEPT24 sequence instead of the typical ²⁹Si quantitative sequence (ESI Fig. S23†).

Fig. 2 ²⁹Si, ¹³C and ¹H NMR spectra of the dimer 1b/monomer 2b mixture recorded in DMSO-d₆. Chemical shifts of relevant signals (protons, carbons, and silicon atoms of –CH₂Si and CH₃Si) are reported in Table 1. Integrations suggested that the sample contained 83% of dimer 1b and 17% of monomer 2b.

When both monomer and dimer are present in a sample, determination of their proportion is valuable. For that, we used ¹³C and ²⁹Si quantitative sequences which require long acquisition times. Interestingly, the ²⁹Si DEPT sequence was not expected to be quantitative; however we noticed that when applied to our hybrid systems, signal integrations reflected the proportion of monomer and dimer in the sample (Fig. 2, ²⁹Si spectra). To go further, signal integrations of ¹H spectra were compared with ²⁹Si NMR spectra. Interestingly, integration of NMR signals related to dimer or monomer showed that the monomer/dimer ratio was unchanged (Fig. 2). Thus, ¹H could be used to determine the proportion of monomer and dimer in a sample. This was particularly advantageous for kinetic studies where high rate sampling was required to monitor the hydrolysis of siloxane bond. Indeed, long acquisition times provided an average ratio that did not make sense when sample evolution was fast. Furthermore, due to the high sensitivity of ¹H NMR, the spectra always presented a higher ratio signal to noise by comparison to those of ²⁹Si, which led to a lower detection limit and more accurate integrations.

We showed that simple ¹H NMR can be routinely used to analyze hybrid peptides and quantify hydroxydimethylsilane/dimethylsiloxane ratio. The presence of TFA salts seemed to be preferable in D₂O for a better signal separation. However, when possible, DMSO-d₆ should be preferred as solvent since chemical shifts in DMSO-d₆ were less sensitive to salts and pH, resulting in an easier signal assignment. The analysis of hybrid peptide ligands (4–9) used for biological assays was performed by ¹H NMR in DMSO-d₆.

Stability studies of the siloxane bond in different solutions by ¹H NMR

The applicability of the dimerization method based on the dimethylhydroxysilane peptides is subjected to the stability of the siloxane bond towards hydrolysis. It is well known that acid- or base-catalyzed polycondensation reactions are reversible and the hydrolysis rate of siloxanes is pH-dependent. This phenomenon was well studied in silica matrixes which quickly dissolve in alkali aqueous solution (pH > 10).¹⁷ However, to the best of our knowledge, no general study was reported for the type of siloxane bonds we used in the dimerization arm. Thus, we investigated the kinetic of hydrolysis of the 1,3-bis(3-aminopropyl)tetramethyl disiloxane in water at different pH by ¹H NMR (Fig. 3). First of all, dimer 1a was dissolved in DMSO-d₆ and ¹H NMR analyses were carried out regularly for one week. No monomer was detected, which confirmed the dimer was stable in the absence of water. This result was in agreement with what we already observed on siloxane polymers.¹² In the presence of water, dimer 1 was hydrolyzed. To study the influence of pH on the dimer hydrolysis rate, dimer 1a was dissolved at 0.1 M in D₂O at pH 4, 7, 7.4 and 10. ¹H signals from the –CH₂Si group of monomer and dimer were integrated and the dimer percent was plotted as a function of time (Fig. 3).

Fig. 3 ¹H NMR kinetic studies. Left panel: dimer percent reported as a function of time. ♦ DMSO; ■ pH 7; _× pH 7.4; _● pH 4; ▲ pH 10. Full sets of data for pH 7 and 7.4 are presented in the box. Right panel: ¹H NMR spectra at pH 10.

As expected, the closer the pH was to the neutrality, the slower was the dimer hydrolysis. When the media is moved to acidic and basic pH (4 and 10, respectively), the hydrolysis rate increased, particularly in basic medium. The half-life of dimer 1a was 203 min at pH 4 and 54 min at pH 10. Most of the time, bioactive peptides are administered intravenously. As a consequence, a good stability of hybrid peptide dimers at blood pH was mandatory. Importantly, at pH 7.4, the half-life of dimer 1a was 439 h, with less than 10% of hydrolysis within 2 days. This stability allows envisioning biological applications for hybrid dimers.

In order to mimic the RP-preparative HPLC experimental conditions, the behavior of dimer 1a in more acidic conditions (pH 2) was studied. As expected a few minutes after solubilization of dimer 1a in D₂O/0.1% TFA, only monomer was detected (Fig. S33†). Thus, when RP-preparative HPLC is required, hybrid compounds need to be purified as monomer species prior to dimerization.

Synthesis of GHRP-6 analogues

Having studied the stability of the dimer 2 at physiological pH, we decided to apply this strategy of dimerization to a relevant bioactive sequence. We selected the growth hormone releasing hexapeptide (GHRP-6) of sequence H-His-DTrp-Ala-Trp-DPhe-Lys-NH₂. Both binding properties and bioactivity of dimers was assayed on intact HEK293T cells expressing the GHS-R1a. Apart from controlling growth hormone secretion, this receptor is involved in many physiological processes such as energy homeostasis,¹⁸ glucose metabolism,¹⁹ gastric acid secretion²⁰ and cardiovascular activity.²¹ Besides, the study of this ligand/receptor system is particularly relevant to witness an eventual switch in the behavior of dimer analogues depending on their structure. Indeed GHSR-1a possesses a remarkable high constitutive activity. Indeed, ∼50% of its maximal activity is observed in the absence of its natural agonist, ghrelin (Ghr).²²

We evaluated the impact of GHRP-6 dimerization in terms of binding affinity and more importantly, in terms of bioactivity. Indeed, we hypothesized that, depending on the position of the bridge between monomers, the agonism behavior of GHRP-6 sequence could be impacted. To answer these questions, three hybrid derivatives of GHRP-6 were prepared (Fig. 4).

Fig. 4 Peptide 3, H-His-DTrp-Ala-Trp-DPhe-Lys-NH₂ (GHRP-6) and potential points of dimerization.

Peptide sequences of the three hybrid derivatives of GHRP-6 were prepared on solid support by conventional Fmoc/tBu strategy. For N-terminus dimerization, a β-alanine spacer was added after the last amino acid and the peptidyl resin was reacted with the 3-isocyanatopropyldimethylchlorosilane (ICPDMCS) to yield the hybrid monomer 4 after cleavage and side-chain deprotection. In contrast, the C-terminus modification was performed through introduction of an ethylene diamine. For that purpose, ethane-1,2-diamine was anchored on 2-chloro chlorotrityl PS resin and the sequence was elongated to the N-terminus via conventional Fmoc/tBu SPPS. The N-Boc, fully protected peptide was cleaved from the solid support by mild acidic treatment (HFIP/DCM 1/4 v/v, 30 min) to maintain the side chain protecting groups. The protected peptide was then reacted with ICPDMCS in solution, precipitated in diethyl ether and treated by TFA to remove the side-chain protecting groups to yield the hybrid peptide 6. To conserve the N and C-termini unchanged, dimerization within the peptide sequence was performed, which might induce a non-negligible steric hindrance in the center of the sequence. Alanine residue in position 3 of GHRP-6 was swapped with a lysine, which served as anchoring point for ICPDMS. Selective removal of Alloc protecting group on the Lys³ side-chain was performed with phenylsilane and Pd⁰(PPh₃)₄ on the sequence still anchored on solid support. The ε-amino group was reacted with ICPDMCS before cleavage of the hybrid peptide from Rink amide resin, yielding hybrid peptide 8.

In all cases, the last step of the synthesis of dimethylhydroxysilyl hybrid peptides was a TFA treatment, which removed the acid sensitive protecting groups of the side chains and cleaved the linker when the synthesis was entirely performed on solid support (compounds 4 and 8). TFA was removed under reduced pressure and the hybrid peptide was recovered as its monomeric form by precipitation in diethyl ether. Hybrid monomers 4, 6 and 8 were purified by preparative RP HPLC. We then applied the dimerization procedure by solubilization of the hybrid peptide monomers in aqueous solution at pH 7.4 at high concentration (>0.2 M) as previously described for polymers obtained from bis-silylated precursors.¹² In most cases, a precipitate appeared at this stage and condensation products were recovered by centrifugation. This procedure proved to be efficient for dimer 5, which precipitated and was recovered by centrifugation with good yield (90%). Only dimer 5 was observed by ¹³C NMR. However, a more sensitive ¹H NMR analysis indicated the presence of 3% of monomer 4. Unfortunately, no precipitation occurred for dimers 7 and 9, probably because of the presence of the free N-terminal amine, which enhanced water solubility.

Optimization of the dimerization

We hypothesized that the condensation could also happen during lyophilization. To verify it, a solution of the model monomer 2a was analyzed by ¹H NMR before and after freeze-drying (ESI Fig. S34†). The percent of dimer increased from 6% to 100% upon lyophilization. As a consequence, the lyophilization constitutes an easy procedure to obtain dimers directly from hybrid monomers species. This procedure was applied to the synthesis of three analogues of the growth hormone releasing hexapeptide (GHRP-6). In agreement with these results, monomers 6 and 8 were purified by preparative HPLC to get rid of buffer salts used during the precipitation attempts. They were freeze-dried at a concentration of 0.1 mM approximately. After lyophilisation, dimer 7 and dimer 9 were obtained. ¹H NMR analysis revealed that 10% of monomer was still present in both cases. However, they were used without further processing for biological studies.

Binding property of ligands

The affinity of ligands was determined by competition binding assay on HEK293T cells expressing the Growth Hormone Secretagogue type 1 receptor (GHS-R1a). Competition binding curves are reported in Fig. 5 and binding affinities are reported in Table 2.

Fig. 5 Binding of dimer ligands to GHS-R1a. Ligands affinities were determined by HTRF-based competition binding assay performed on intact HEK293T cells as described in Materials and methods. Results are from one representative experiment of three, each performed in triplicate.

Table 2 Binding properties of dimers compared to GHRP-6 and ghrelin

Compound	N-ter dimer 5 JMV 6187	C-ter dimer 7 JMV 6186	Internal dimer 9 JMV 6185	Compound 3 GHRP-6	Ghrelin
a Ki values (mean ± S.E.M)s were obtained from competition binding curves analysis with GraphPad Prism v5 software.
Ki^a (nM)	59.7 ± 21.4	173 ± 2	319.5 ± 49.5	18.8 ± 8	7.8 ± 1.8

---

First of all, all dimers were able to bind to ghrelin receptor in the 100 nM range. However, compared to GHRP-6, a 5 to 10-fold decrease of binding was observed for compounds 5 and 7. Internal dimerization (compound 9) induced a 15-fold loss of affinity. N-Terminus dimerization (compound 5) had the less impact on the binding affinity. These results could be related to the steric hindrance of the dimeric molecule to approach GHS-R1a binding sites, the transmembrane region 3 of the GHS-R1a in which peptide ligands and small molecules can bind.²³ However, binding behavior cannot be associated to bioactivity by a direct relationship. We thus investigated the biological activity of the ligands by measuring their ability to promote inositol phosphate (IP) production in HEK293T cells expressing the GHS-R1a. The production of IP, reported as % of ghrelin-induced maximal responses, are reported in Fig. 6 and summarized in Table 3. The peptide [D-Arg¹, D-Phe⁵, D-Trp^7,9, Leu¹¹]-substance P (noted SPA for substance P analogue)²⁴ is a non selective ligand of GHS-R1a, which behaves as an inverse agonist. It was also included in the same experiments.

Fig. 6 Signaling property of ligands. Efficacy of ligands to stimulate IP1 production was measured on HEK293T cells expressing the GHS-R1a as described in Materials and methods. Results are one representative experiment of three, each performed in triplicate.

Table 3 Efficacy of ligands to stimulate inositol phosphate production

Compound	N-ter dimer 5 JMV 6187	C-ter dimer 7 JMV 6186	Internal dimer 9 JMV 618	SPA	Compound 3 GHRP-6	Ghrelin
a EC50 and Emax were obtained from dose–response curve analysis with GraphPad Prism v5 software. For agonist compounds, Emax is expressed as the % of maximal ghrelin. For inverse agonists, Emax is expressed as the maximal basal inhibition (100% inhibition corresponding to the basal of mock-transfected HEK293T cells). Values are mean ± S.E.M of three independent experiments.
EC50^a (nM)	20.4 ± 6.8	2.57 ± 0.35	160 ± 44	27 ± 2	1.18 ± 0.82	0.85 ± 0.29
Emax (%)	86.5 ± 2.7	97.8 ± 1.7	38 ± 3	77.5 ± 7.5	99.5 ± 2.2	100
Behaviour	Partial agonist	Full agonist	Inverse agonist	Inverse agonist	Full agonist	Full agonist

---

Interestingly very different behaviors were obtained for the three dimers. N-Terminal dimer 5 which was the best binder within the dimer series, proved to be a partial agonist with an EC₅₀ of 20 nM. It induced 86.5 ± 2.7% of the maximal effect of the ghrelin. In contrast, the C-terminal dimer 7, which had a three-fold lower affinity for GHS-R1a, was a full agonist with an EC₅₀ of 2.57 nM, close to that of GHRP-6 (1.2 nM). The most striking result was obtained with dimer 9, which behaved as an inverse agonist with an efficacy of 38 ± 3% (Table 3). Interestingly, it was reported that the peptide D-Lys³-GHRP-6 displayed some inverse agonist activity in the sea bream²⁵ but behaved as an antagonist in mammals.^26,27 In the case of dimer 9, a L-lysine was introduced as an anchoring point of the dimethylhydroxysilane moiety. Therefore, it could not be excluded that the inverse agonist effect was not induced by dimerization but simply by the Lys/Ala substitution, even though the stereochemistry was not the same (D-Lys in the reported monomeric inverse agonist and L-Lys in the dimer 9). Thus, to know the significance of either the dimerization or the modification of the linear sequence itself in inducing the inverse agonist activity, two other compounds were prepared (Fig. 7). Compound 10 was obtained in the same way as monomer 8 except isocyanopropyltrimethylsilane was used instead of ICPDMCS for the formation of the urea bond. Lys³-GHRP-6 (compound 11) was also prepared. Bioactivity of 10, 11 and of Lys³ N-ε acetylated analogue 12 was also evaluated on HEK293T cells expressing the GHS-R1.

Fig. 7 Peptides 10, 11 and 12 monomeric analogues of dimer 9.

Interestingly none of them displayed any inverse agonist activity and all behaved as neutral ligands (ESI Fig. S54†). This last result showed that the dimerization in itself induced a complete switch of activity and the position of the dimerization within the active sequence is accordingly of great impact. Further pharmacological studies will be performed to investigate the structure–activity relationships of GHRP-6 analogues.

Conclusions

In this paper, we demonstrated that formation of siloxane bound could be a simple and useful tool to yield homodimers of peptide ligands. Formation of siloxane bonds starting from hydroxydimethylsilane could be easily monitored by NMR spectroscopy, especially by measuring the proton signals of the methyl groups linked to silicon atoms. This study clearly demonstrated that the bond was relatively stable at neutral pH but was hydrolyzed in basic or acidic conditions. Noteworthy, the position of dimerization and thus, the orientation of the bioactive sequence could be controlled by introduction of a hydroxydimethylsilane moiety either at the N-terminus, C-terminus or within the sequence via the side chain of a lysine. Using GHRP-6 as an example, we demonstrated that the position of the dimerization could be crucial for modulating bioactivity, dimers being agonists, inverse agonists or partial agonists.

Acknowledgements

C. Echalier's Ph.D. is partly founded by Region Languedoc Roussillon through ‘Chercheur d'avenir’ grant attributed to G. Subra. Peptide syntheses were performed using the facilities of SynBio3 IBISA platform supported by ITMO cancer. NMR analyses were performed using the facilities of the ‘Laboratoire de Mesures Physiques’ (LMP) from the University of Montpellier.

Notes and references

1. J. E. Gestwicki, C. W. Cairo, L. E. Strong, K. A. Oetjen and L. L. Kiessling, J. Am. Chem. Soc., 2002, 124, 14922–14933.
2. C. Valant, J. Robert Lane, P. M. Sexton and A. Christopoulos, Annu. Rev. Pharmacol. Toxicol., 2012, 52, 153–178.
3. C. Hiller, J. Kühhorn and P. Gmeiner, J. Med. Chem., 2013, 56, 6542–6559.
4. S. Halazy, M. Perez, C. Fourrier, I. Pallard, P. J. Pauwels, C. Palmier, G. W. John, J. P. Valentin, R. Bonnafous and J. Martinez, J. Med. Chem., 1996, 39, 4920–4927.
5. M. Perez, P. J. Pauwels, C. Fourrier, P. Chopin, J. P. Valentin, G. W. John, M. Marien and S. Halazy, Bioorg. Med. Chem. Lett., 1998, 8, 675–680.
6. Y. Shimohigashi, T. Costa, H.-C. Chen and D. Rodbard, Nature, 1982, 297, 333–335.
7. I. Daffix, M. Amblard, G. Bergé, P. Dodey, D. Pruneau, J.-L. Paquet, C. Fouchet, R.-M. Franck, E. Defrêne, J.-M. Luccarini, P. Bélichard and J. Martinez, J. Pept. Res., 1998, 52, 1–14.
8. G. Subra, M. Amblard, P. Verdie, S. Komesli, P. Dutartre, P. Durand, P. Renaut and J. Martinez, QSAR Comb. Sci., 2007, 26, 496–510.
9. D. Lecis, E. Mastrangelo, L. Belvisi, M. Bolognesi, M. Civera, F. Cossu, M. De Cesare, D. Delia, C. Drago, G. Manenti, L. Manzoni, M. Milani, E. Moroni, P. Perego, D. Potenza, V. Rizzo, C. Scavullo, C. Scolastico, F. Servida, F. Vasile and P. Seneci, Bioorg. Med. Chem., 2012, 20, 6709–6723.
10. G. Subra, M. Amblard and J. Martinez, Tetrahedron Lett., 2002, 43, 9221–9223.
11. S. Mitra, S. Duggineni, M. Koolpe, X. Zhu, Z. Huang and E. B. Pasquale, Biochemistry, 2010, 49, 6687–6695.
12. S. Jebors, J. Ciccione, S. Al-Halifa, B. Nottelet, C. Enjalbal, C. M'Kadmi, M. Amblard, A. Mehdi, J. Martinez and G. Subra, Angew. Chem., Int. Ed., 2015, 54, 3778–3782.
13. C. M'Kadmi, J.-P. Leyris, L. Onfroy, C. Galés, A. Saulière, D. Gagne, M. Damian, S. Mary, M. Maingot, S. Denoyelle, P. Verdié, J.-A. Fehrentz, J. Martinez, J.-L. Banères and J. Marie, J. Biol. Chem., 2015, 290, 27021–27039.
14. S. Coantic, G. Subra and J. Martinez, Int. J. Pept. Res. Ther., 2008, 14, 143–147.
15. C. Echalier, S. Al-Halifa, A. Kreiter, C. Enjalbal, P. Sanchez, L. Ronga, K. Puget, P. Verdié, M. Amblard, J. Martinez and G. Subra, Amino Acids, 2013, 45, 1395–1403.
16. J.-P. Leyris, T. Roux, E. Trinquet, P. Verdié, J.-A. Fehrentz, N. Oueslati, S. Douzon, E. Bourrier, L. Lamarque, D. Gagne, J.-C. Galleyrand, C. M'kadmi, J. Martinez, S. Mary, J.-L. Banères and J. Marie, Anal. Biochem., 2011, 408, 253–262.
17. R. Corriu and N. Trong Anh, Molecular Chemistry of Sol–Gel Derived Nanomaterials, Wiley, 2009.
18. M. Kojima, H. Hosoda, Y. Date, M. Nakazato, H. Matsuo and K. Kangawa, Nature, 1999, 402, 656–660.
19. W. Zhang, M. Chen, X. Chen, B. J. Segura and M. W. Mulholland, J. Physiol., 2001, 537, 231–236.
20. V. Sibilia, F. Pagani, F. Guidobono, V. Locatelli, A. Torsello, R. Deghenghi and C. Netti, Neuroendocrinology, 2002, 75, 92–97.
21. E. Bollano, C.-H. Bergh, C. Kjellström, E. Omerovic, V. Kujacic, K. Caidahl, B.-Å. Bengtsson, F. Waagstein and J. Isgaard, Eur. J. Heart Failure, 2001, 3, 651–660.
22. Y. Mear, A. Enjalbert and S. Thirion, Front. Neurosci., 2013, 7, 87.
23. S. D. Feighner, A. D. Howard, K. Prendergast, O. C. Palyha, D. L. Hreniuk, R. Nargund, D. Underwood, J. R. Tata, D. C. Dean, C. P. Tan, K. K. McKee, J. W. Woods, A. A. Patchett, R. G. Smith and L. H. Van der Ploeg, Mol. Endocrinol., 1998, 12, 137–145.
24. B. Holst, M. Lang, E. Brandt, A. Bach, A. Howard, T. M. Frimurer, A. Beck-Sickinger and T. W. Schwartz, Mol. Pharmacol., 2006, 70, 936–946.
25. C.-B. Chan, P.-K. Leung, H. Wise and C. H. K. Cheng, FEBS Lett., 2004, 577, 147–153.
26. R. G. Smith, Endocr. Rev., 2005, 26, 346–360.
27. M. Kojima and K. Kangawa, Physiol. Rev., 2005, 85, 495–522.

---

Footnote

† Electronic supplementary information (ESI) available: NMR spectra, kinetic studies, compound characterization, monomeric ligand binding assays and bioactivity. See DOI: 10.1039/c6ra06075g

---