Solid-phase synthesis of a library of amphipatic hydantoins. Discovery of new hits for TRPV1 blockade

Abstract

Some heterocyclic systems, called privileged scaffolds, appear frequently in bioactive products and marketed drugs. The combination of a recognized privileged scaffold (hydantoin) and a functional group with high incidence in bioactive molecules (guanidine) guided the design of a library of amphipatic compounds, which allowed to discover novel TRPV1 ion channel blockers. The library was synthesized by parallel solid-phase synthesis from an orthogonally protected resin-bound Lys-Lys skeleton. Key steps of the synthetic procedure were the construction of the hydantoin ring, by reaction of the N-terminal amino group with N,N-disuccinimidyl carbonate (DSC) and subsequent base-induced cyclization, and the guanidinylation of the C-terminal Lys side-chain after removal of the Alloc protecting-group. The preliminary biological studies have allowed the identification of some of the key structural features directing the blockage of capsaicin-induced Ca²⁺ influx through TRPV1 channels, particularly, the strong preference showed for highly lipophilic acyl groups and substituted guanidine moieties. Active compounds based on this new pharmacophoric scaffold that display in vitro and in vivo inhibitory activity

Introduction

5,5-Diphenylhydantoin is being used since 1938 for the treatment of epilepsy due to its regulatory effect on the bioelectric activity of the nervous system.¹ Since then, many other hydantoin-derived compounds with a wide range of therapeutic applications have been discovered. For example, 5-arylidene derivatives have shown antituberculosis and antiproliferative activity,^2,3 whereas Pt(II) complexes with hydantoin ligands have proven to be very effective citotoxic agents.⁴ Other examples include antiviral agents,⁵ ligands for the glycine binding site of the NMDA receptor and for voltage-gated sodium channels,^6,7 as well as modulators of certain protein-protein interactions related to cell adhesion.^8,9 Thus, and due to the large variety of biological applications found for this family of compounds, the hydantoin scaffold is being considered a “privileged structure”, term first coined by Evans in 1988 to define a single molecular framework able to provide ligands for a range of different biological targets.¹⁰

The impact of hydantoins in medicinal chemistry programs has triggered an extensive development of synthetic methodologies for the generation of libraries based on this heterocyclic skeleton.^11–13 Apart from some procedures in solution, using multicomponent reactions and fluorous synthesis,^14,15 most efforts to prepare hydantoin-based libraries have been carried out on solid-phase, following two main approaches: i) cyclization of an acyclic precursor with concomitant cleavage from the resin, and ii) cyclization prior to cleavage.

The vast majority of synthetic sequences, derived from the first approach,¹⁶ comprise the attachment of an amino acid or dipeptide to the resin through the C-terminal carboxylate, further elaboration to an acyclic urea, and the final cyclative/cleavage step.^17–29 An alternative strategy used a urethane attachment of amino acids to the resin (through the α–NH₂ group), followed by base-catalyzed cyclization of linear adducts to form and detach final hydantoins.³⁰

On the other hand, many different strategies have been described to develop methodologies following the second approach, which allow further on-bead transformations. They differ mainly in the nature of the reactive intermediate or in the activating agents used to facilitate the cyclization step. Thus, for example base-promoted cyclization of dipeptide-derived phenyl carbamates, thermal ring closure from dipeptide-isocyanate intermediates, and activation of the peptide N-terminal amino group by means of diphosgene, triphosgene, disuccinimide carbonate or carbonyldiimidazole have been successfully employed.^31–39

The guanidine functional group is also a structural motif commonly found in natural products and in many therapeutically active compounds, including some marketed drugs, where it usually acts as a crucial pharmacophore entity. Thus, guanidine-containing derivatives have been shown to have antitumor,^40–42 antibiotic,^43,44 and antiviral properties.⁴⁵ Regarding the biological target, the guanidine group is frequent in both enzyme inhibitors^46,47 and ligands for different receptors and ion channels.^48–51 Therefore, and by analogy with the “privileged scaffolds”, the guanidine moiety can be considered as a “privileged functional group” in the search for new bioactive compounds.

Based on the above mentioned considerations, we hypothesized that the combination of a “privileged structure” and the guanidine “privileged functional group” could be a practical approach for the discovery of new compounds of therapeutic relevance. Hence, using the hydantoin ring as central scaffold, we have prepared a discrete library with two points of diversity, which are decorated by an arrangement of differently substituted guanidine groups and diverse acyl moieties, to confer a certain amphipatic character to library compounds. We report herein the preparation of this collection of compounds and the identification of TRPV1 channel blockers with in vitro and in vivo activity. We focused on this receptor because its central role in pain transduction. In addition, we used in parallel the NMDA receptor with the aim of identifying compounds that preferentially block the TRPV1 over the glutamatergic receptor. The rationale of using the NMDA receptor is that displays a similar Ca²⁺ permeability as the TRPV1 channel.

Results and Discussion

A common Lys-Lys-derived hydantoin scaffold was envisaged to incorporate the elements of diversity at the Lys side-chains. The new chemset of 112 compounds resulted from a combination of 14 diverse acyl moieties and 8 different guanidine groups at N- and C-terminal side-chains, respectively (1{G1-8,A1-14}, Figure 1). As diversity elements at the C-terminal Lys side-chain, three sublibraries contain unsubstituted guanidine groups, connected through spacers with different flexibility (G{1-2}) or directly linked to the carbon side-chain (G3). The other five sublibraries possess different pentasubstituted guanidine moieties (G{4-8}) linked to the tetramethylene side-chain. To further incorporate diversity at the 5-(4-aminobutyl)-substituent of the hydantoin skeleton, we decided to acylate the amino group with carboxylic acids containing: short linear and branched aliphatic chains (A{1-4}), fatty-type saturated chains (A{5-6}), carbocyclic and aromatic moieties (A{7-11}), and linear and cyclic substituents incorporating primary and secondary amino groups (A{12-14}).

Figure 1.

General structure of the hydantoin library and selected building blocks

Synthesis

The orthogonally protected Fmoc-Lys(Mtt)-Lys(Alloc)-PS (2) was prepared by conventional SPPS methods on a Fmoc-Rink amide-MBHA-Polystyrene resin (Scheme 1). Removal of the Fmoc group from this dipeptidyl resin, followed by treatment with DSC in the presence of DBU allowed the cyclization to the hydantoin scaffold,^38–39 and thus the formation of the key resin-bound intermediate 3, characterized by HPLC-MS after cleavage. The selective removal of the Alloc protecting group, using Pd(PPh₃)₄/PhSiH₃, afforded a free amino at the C-terminal Lys side-chain, used for the incorporation of the guanidine moieties. Thus, acylation with the hydrochloride salt of 4-guanidinobenzoic acid and 5-guanidinovaleric acid, in the presence of DIPCDI/HOBt as coupling agents, gave resins 4{1-2}. On the other hand, direct guanidinylation with various guanidinylating agents (G{3-8}),⁵² bearing different substituents, led to the dipeptidyl chemset resins 5{3-8}. Aliquots of every member of the chemsets 4 and 5 were subjected to cleavage and HPLC-MS analysis, revealing a >80% conversion in all cases. Next, each resin, 4{G1-2} and 5{G3-8}, was treated with the acidic cocktail TES:TFA:DCM (1:3:96) for Mtt protecting group removal, and divided into 14 portions for the incorporation of the second element of diversity (acyl groups, A{1-14}). Coupling reactions to chemsets 6 and 7 were performed with 1.5 equiv. of the reagent chemset A{1-14 } in the presence of DIPCDI/HOBt. Finally, all members of chemsets 6 and 7 were cleaved from the resin with TFA.

Scheme 1.

Reagents and conditions: (i) (a) Piperidine/DMF (1:4); (b) Fmoc-Lys(Alloc)-OH (1.5 equiv.), DIPCDI (1.5 equiv.), HOBt (1.5 equiv.), DMF; (c) Piperidine/DMF (1:4); (d) Fmoc-Lys(Mtt)-OH (1.5 equiv.), DIPCDI (1.5 equiv.), HOBt (1.5 equiv.), DMF. (ii) (a) Piperidine/DMF (1:4), DSC (5 equiv.), DMAP (0.5 equiv.); (b) 5% DBU/DMF. (iii) (a) (a) Pd[PPh₃]₄, PhSiH₃; (b) HCl.H₂N(NH)CNH-X-CO₂H (1.5 equiv.), DIPCDI (1.5 equiv.), HOBt (1.5 equiv.), DMF. (iv) (a) Pd [PPh₃]₄, PhSiH₃; (b) Guanidylation reaction. (v) (a) TES:TFA:DCM (1:3:96); (b) RCO₂H (1.5 equiv.), DIPCI (1.5 equiv.), HOBt (1.5 equiv.), DMF. (vi) TFA.

The expected final compounds 1{G1-8,A1-14} were obtained in variable yields, ranging from low to moderate for sublibraries 1{1,1-14} and 1{2,1-14}, and from moderate to good for the other subset of compounds, 1{3-8,1-14} (see supporting information, Table S1, for details). Ninety (80.4%) of the 112 compounds were isolated in ≥80% purity, and a further 14.3% in 70–80% purity. The remaining hydantoin derivatives (5.3%) showed purities that range from 48 to 68%.

It is interesting to note that main components in library 1{7,1-14} are the diguanidinyated compounds indicated in Figure 2 (>70%), while the expected monoguanidinylated derivatives were in minor proportion within the mixtures (5–15%). The presence of a second guanidine group was determined by MS, and its location at position 1 of the hydantoin ring was assigned from the ¹H NMR spectra (see the chemical shifts of contiguous H5 protons, Table S9 of supporting information). The cleavage of an aliquot of the 5{7} resin corroborated the incorporation of the second guanidine moiety during the guanidinylation reaction, a side-reaction that was also observed, but in much lower extent, during guanidinylation of resin 3 with G4 (8%). Only compounds 1{7,12} and 1{7,14} were obtained as monoguanidinylated products in high purity (>80%). This result indicates that treatment of resins 7{7,12} and 7{7,14} with piperidine for the Fmoc removal, prior to cleavage, reverted the di- to the monoguanidylated derivative.

Figure 2.

Diguanidylated compounds obtained for G7 sublibrary and cleavage product of 5{G7}

All final compounds 1{G1-8,A1-14} were obtained as an approximately 1:1 mixtures of diastereoisomers, due to epimerization during the hydantoin ring formation under the DBU basic medium. The Cα of the C-terminal Lys residue is most likely being epimerized, since it was demonstrated that epimerization takes place at the C-terminal residue on related hydantoin-derived Phe-Phe model peptides.⁵³ Although the use of the more reactive carbonylating agent CDI in combination with the weaker base DIEA could avoid epimerization,³⁹ we decided to use the epimerizing method in our combinatorial program to initially explore a greater diversity.

Screening as TRPV1 and NMDA Channel Blockers

In the search for new bioactive chemical entities, compounds of library 1 were assayed as blockers of the capsaicin-induced channel activity of TRPV1, and of the glutamate-evoked activity of NMDA, both heterogously expressed in Xenopus laevis oocytes.

TRPV1 is a neuronal receptor that integrates thermal and chemical stimuli in the peripheral nervous system.⁵⁴ From a therapeutic point of view, TRPV1 antagonists have shown efficacy in reducing nociception from inflammatory and neuropathic pain in animal models.^55–587 NMDA receptors are non-specific cation channels that directly contribute to excitatory synaptic transmission, playing a key role in a wide range of physiologic and pathologic processes, such as excitotoxicity.⁵⁹ The interest in developing safe and effective NMDA channel blockers comes from the pivotal role of glutamate and NMDA receptors in mediating multiple neurodegenerative CNS disorders.

To identify activity-dependent TRPV1 channel blockers, we used used saturating concentrations of capsaicin, and a hyperpolarized membrane potential. Similar conditions were used for the NMDA receptor by using saturating concentrations of L-glutamate and glycine. The full set of compounds blocked the capsaicin-evoked currents from TRPV1 and the L-glutamate/glycine responses from NMDA receptors to different extents (Figures S3 and S4 in Supporting Information). Noteworthy, some library members significantly inhibited the TRPV1 activity induced by capsaicin, without significantly affecting NMDA receptor function (Figures S3 and S4 in SI). Note that some of the compounds produced in increase of the response probably by binding to an allosteric site that potentiates the activity of the agonist used.

An inspection of the active components of the library provides preliminar information on the structural requirements for TRPV1 channel blockers based on this structure. Thus, highly lipophylic fatty-type substituents, like palmitoyl (A5) and litocholyl (A8), as well as m-nitrocinnamyl (A10) are preferred on the 5-(4-aminobutyl) chain. In addition, compounds with substituted guanidine groups (G{4-8}) at the C-terminal Lys side-chain showed better TRPV1 blockade than free guanidine-derived analogues (G{1-3}). In the last case, the distance between the guanidine and acyl moieties is important for the antagonist potency, with better results for derivatives with the guanidine group directly linked to the Lys side-chain, such as 1{3,5}, 1{3,6} or 1{3,8}, than for the corresponding analogues 1{1-2} with longer distances between the diversity elements.

A few of these active compounds were selected, resynthesized and evaluated as mixture of enantiomers for in vivo activity. Basically, we evaluated the effect attenuating the burning sensation evoked by intraplantar capsaicin administration,⁶⁰ which is evidenced by the duration of shaking and licking. In parallel, we evaluated the effect on the thermal nociception by measuring the latency to a response in a hot plate at 52°C. As illustrated in Figure 3, intraperitoneal administration of selected compounds did not affect the thermal nociception as evidenced by the similar latency times of animals administered with vehicle and compounds, indicating that TRPV1 blockers do not affect temperature sensitivity in physiological conditions. In contrast, compounds significantly decreased the burning sensation of intraplantar capsaicin application by reducing the duration of the flinching and licking of the inflamed paw. Administration of capsaicin induces an acute inflammatory state that results in a nocifensive response. Therefore, these findings imply that selected hits display anti-inflammatory activity in vivo by attenuating TRPV1 function in nociceptor terminals; although a synergistic effect on NMDA receptors cannot been completely discarded because of their weak interaction with this ion channel. In addition, because these compounds act as non-competitive channel blockers, they may not display the hyperthermic effects observed for competitive antagonists. Further optimization using medicinal chemistry is necessary to clearly unveil the therapeutic potential of these hits and to evolve leads for drug development.

Figure 3.

In vivo activity of 1{G1-8,A1-14} library compounds with significant TRVP1 blockade properties and . a) Black bars show latency time to the first response of animals in a hot plate at 52°C⁴². Mice were administered ip with saline (vehicle) or 30 mg/Kg of the compounds. b) Grey bars show duration time of the burning sensation evoked by intraplantar administration of capsaicin, measured as licking and shaking the paw⁵⁹. Compound was administered ip at 30 mg/Kg. Data are given as mean±sem, with n≥6.

Conclusions

A resin-bound Fmoc-Lys(Mtt)-Lys(Alloc) dipeptide derivative was used as key intermediate for the generation of a hydantoin library with two points of diversity. After the base-promoted cyclization of the corresponding N^α-DSC-activated intermediate to the hydantoin ring, the orthogonal deprotection of Lys side-chains allowed the parallel and successive incorporation of the diversity elements through simple reactions. This library, joining the privileged scaffold hydantoin and a series of differently substituted guanidine and diverse acyl groups, allowed the discovery of new hits for TRPV1 ion channel blockade, and to establish the first structural requirements for activity within this series.

Experimental Section

Preparation of resin-bound dipeptide 2

The Fmoc-AM-MBHA-PS resin (3.5 mmol) was swollen in DCM and DMF, and treated with 20% piperidine in DMF (1 × 1 min, 3 × 5 min, 1 × 10 min) for the removal of the Fmoc group, and washed with DMF (5 × 0.5 min), DCM (5 × 0.5 min) and DMF (5 × 0.5 min). Then, Fmoc-Lys(Alloc)-OH (2.375 g, 5.25 mmol), DIPCDI (0.82 mL, 5.25 mmol), and HOBt (0.71 g, 5.25 mmol) were added in DMF (15 mL) and shaken for 6 h. This coupling reaction was repeated overnight. After removing the Fmoc group in the above indicated conditions, Fmoc-Lys(Mtt)-OH (3.28 g, 5.25 mmol),), DIPCDI (0.82 mL, 5.25 mmol), and HOBt (0.71 g, 5.25 mmol) were added in DMF (15 mL) and shaken for 12 h. The second coupling was repeated twice with 1 equivalent of the amino acid derivative and coupling agents. Couplings were monitored by the Kaiser ninhydrin test.

Synthesis of Hydantoin intermediate 3

Resin 2 (3.5 mmol) was first treated with 20% piperidine in DMF (1 × 5 min, 1 × 20 min), washed with DMF (5 × 0.5 min), DCM (5 × 0.5 min) and DMF (5 × 0.5 min), and reacted with DSC (5 equiv.) and DMAP (0.5 equiv.) in DMF (minimum volume possible). The mixture was shaken for 3 h, washed with DMF (5 × 0.5 min.) and DCM (5 × 0.5 min.), and completion was monitored by the ninhydrin test. Then the resin was treated with DBU in DMF (1:4 v/v 1 × 1 min., 1 × 20 min.), and washed as previously indicated. An aliquot portion of the resin was cleaved to the expected product, (4″-Amino)butyl-3-(5′-allyloxycarbonylamino-1′-carbamoyl)pentylhydantoin, which was analyzed as follows: HPLC t_R = 12.50 min (>90%); ESI-MS: 406.2 (M+Na)⁺. At this point the resin was divided into 8 batches for further modifications.

Preparation of resins 4{1-2}

Batches 1 and 2 of resin 3 (0.43 mmol each) were individually reacted with Pd(PPh₃)₄ (0.1 equiv.) and PhSiH₃ (24 equiv.) in DCM under Ar, to remove the Alloc group. The reaction mixtures were shaken under Ar for 20 min and washed with DCM. The resulting resins were swollen in DMF and treated with G1 and G2 (1.1 mmol), repectively, DIPCDI (1.1 mmol) and HOBt (1.1 mmol). The couplings were accomplished for 12 h, and repeated twice with 1 equiv. of each reagent for 12 h, until negative ninhydrin test. HPLC and MS analysis of the cleavage products indicated a >86% conversion to the expected guanidine-derived hydantoin intermediates in both cases.

Preparation of resins 5{3-8}

After removing the Alloc group in bathes 3-8 of resin 3 (0.43 mmol each), as above indicated, DIEA (2.16 mmol in the case of G3 and 1.2 mmol in all other cases) and the corresponding guanidinylating agent (G{3-8}, 2.16 mmol for G3 and 1.2 mmol for others) were successively added. For the preparation of resin 5{3}, the guanidinylation reaction was carried out for 12 h and repeated twice in the same conditions. For the synthesis of resins 5{4-8}, the reaction was shorter (3 h), and it needed to be repeated once with half equivalents of each reagent. This reaction was monitored by the Kaiser test. In all cases, after cleavage of aliquot portions, good conversions were observed in HPLC and the (M+1)⁺ of the corresponding guanidinylated intermediates were found in MS experiments. The only exception was compound 5{7}, for which the main product corresponded to a diguanidinylated derivative with (M+1)⁺ = 601.9.

General procedure for the preparation of resin-bound intermediates 6 and 7

Syringes containing resins 4{1-2} and 5{3-8} were swollen in DCM and treated with TES:TFA:DCM (1:3:96, 3 × 10 min, and 2 × 15 min) for the removal of the Mtt group. After washing with DMF (5 × 0.5 min), DCM (5 × 0.5 min) and DMF (5 × 0.5 min), a neutralization step was carried out with 5% DIEA in DCM (3 × 0.5 min) followed by a final washing with DCM and drying under vacuum. Each resulting resin was then divided into 14 sub-batches for the incorporation of the selected carboxylic acids. For this purpose, resins of each sub-batch were treated with the corresponding carboxylic acids (1.5 equiv.), HOBt (1.5 equiv.) and DIPCDI (1.5 equiv.). When required, the coupling reaction was repeated until the Kaiser test was negative.

General procedure for the cleavage from the resin: Synthesis of final compounds 1

Cleavage from the resins was carried out by treatment with neat TFA (1 mL/100 mg of resin) for 1.5 h. After filtration, the acidic filtrate was evaporated by a N₂ or Ar stream, and the resultant residue was triturated with Et₂O and centrifuged. The resulting solid residue was separated, dissolved in H₂O or H₂O/ACN mixtures and lyophilized. All library members were characterized by HPLC, MS and ¹H NMR, while ¹³C NMR data was recorded for some representative compounds (see supporting information for details).

Recombinant rat TRPV1 and NMDAR channels expression in Xenopus oocytes and channel blockade

All the procedures have been described in detail elsewhere⁶⁰. Whole-cell currents from rat TRPV1-injected oocytes were recorded in Mg²⁺-Ringer’s solution (in mM: 10 Hepes pH 7.4, 115 NaCl, 2.8 KCl, 0.1 BaCl_2, 2.0 MgCl₂) with a two-microelectrode voltage-clamp amplifier at 20°C. TRPV1 channels were activated by application of 10 μM capsaicin in absence or presence of individual compounds at a holding potential (V_h) of −80 mV. Receptor selectivity was evaluated on heterologously expressed NMDA receptors (rat NR1:NR2A). Recombinant NMDA receptor responses were activated with 100 μM L-glutamate plus 20 μM glycine in the absence and presence of the compound in normal Ringer solution (in mM: 10 Hepes pH 7.4, 115 NaCl, 2.8 KCl, 1.8 BaCl_2,), and at a holding of −80 mV.

Behavioural nociception assays

Adult male ICR mice were habituated to the test environment for 24 h in plexiglass chambers prior the nociception assays. Thermal nociception was studied using a hot plate at 52°C. The response latency for paw shaking or licking or jumping was measured⁶⁰.

Capsaicin-induced hyperalgesia

Capsaicin (10 μL at 0.06% in 10% ethanol, 10% Tween 80 and 80% saline) was injected intradermally into the heel pad with a 0.3 mm diameter needle attached to a Hamilton syringe. The duration time of licking and shaking the paw in response to the injection was recorded⁶⁰.

Abbreviations

Ac₂O

acetic anhydride

ACN

acetonitrile

Alloc

allyloxycarbonyl

CDI

1,1′-carbonyldiimidazole

CNS

central nervous system

DBU

1,8-Diazabicyclo[5.4.0]undec-7-ene

DCM

dichloromethane

DIPCDI

N,N′-diisopropylcarbodiimide

DIEA

diisopropylethylamine

DMAP

4-Dimethylaminopyridine

DMF

dimethylformamide

DSC

N,N′-Disuccinimidyl carbonate

Fmoc

Fluorenylmethoxycarbonyl

HOBt

hydroxybenzotriazole

NMDA

N-methyl D-aspartate

Mtt

4-Methyltrityl

Fmoc-Rink amide-MBHA-PS or Fmoc-AM-MBHA-resin

4-(2′,4′dimethoxyphenyl-Fmoc-aminomethyl)-pheoxyacetamido p-methylbenhidrylamine polystyrene resin

HPLC

high performance liquid chromatography

HPLC-MS

high performance liquid chromatography-mass spectrometry

MS

mass spectrometry

ESI-MS

electrospray ionization mass spectrometry

SPPS

solid-phase peptide synthesis

TES

triethyl silane

TFA

trifluoroacetic acid

TRVP1

transient receptor potential cation channel, subfamily V, member 1. Abbreviations used for amino acids follow the IUPAC-IUB Commission of Biochemical Nomenclature in Jones, J.H. J. Pept. Sci., 2003, 9, 1–8