Exploration of TRPM8 Binding Sites by β-Carboline-Based Antagonists and Their In Vitro Characterization and In Vivo Analgesic Activities

Transient receptor potential melastatin 8 (TRPM8) ion channel represents a valuable pharmacological option for several therapeutic areas. Here, a series of conformationally restricted derivatives of the previously described TRPM8 antagonist N,N′-dibenzyl tryptophan 4 were prepared and characterized in vitro by Ca2+-imaging and patch-clamp electrophysiology assays. Molecular modeling studies led to identification of a broad and well-defined interaction network of these derivatives inside the TRPM8 binding site, underlying their antagonist activity. The (5R,11aS)-5-(4-chlorophenyl)-2-(4-fluorobenzyl)-5,6,11,11a-tetrahydro-1H-imidazo[1′,5′:1,6]pyrido[3,4-b]indole-1,3(2H)-dione (31a) emerged as a potent (IC50 = 4.10 ± 1.2 nM), selective, and metabolically stable TRPM8 antagonist. In vivo, 31a showed significant target coverage in an icilin-induced WDS (at 11.5 mg/kg ip), an oxaliplatin-induced cold allodynia (at 10–30 μg sc), and CCI-induced thermal hyperalgesia (at 11.5 mg/kg ip) mice models. These results confirm the tryptophan moiety as a solid pharmacophore template for the design of highly potent modulators of TRPM8-mediated activities.


■ INTRODUCTION
The transient receptor potential melastatin type 8 (TRPM8) is a member of the thermo-TRP family 1 of polymodal, nonselective, and Ca 2+ permeable ion channel, identified as the physiological sensor of environmental cold. 2 TRPM8 is activated by a range of innocuous to noxious cold temperatures (10−28°C), 2c,3 natural and synthetic cooling agent, 2c,4 membrane depolarization, 5 changes in extracellular osmolarity 6 and phosphatidylinositol 4,5-biphosphate (PIP 2 ). 7 Originally expressed in a prostate cancer cell line, 8 TRPM8 was subsequently detected in a subset of primary afferent neurons in the dorsal root ganglion (DRG) and trigeminal ganglia (TG), 2c,9 which innervate cold highly sensitive tissues, such as skin, oral cavity epithelium, teeth, tongue, and cornea. 9a,10 TRPM8 is also expressed in visceral tissues innervated by pelvic or vagal nerves, 11 several tumor cells, 12 macrophages, 13 and different regions in rodents brain. 14 Regulation of the TRPM8-expression and/or -morphological changes in pathological processes involving these tissues may represent a new opportunity for the therapeutic intervention in pain, cancer, inflammation, and metabolic diseases, among others. 15 In particular, there is a large body of evidence that correlates the hypersensitivity to cold, typical of neuropathic pain models, after nerve injury or oxaliplatin-treatment with augmented expression of TRPM8 in sensory neurons, 16 suggesting that blocking the channel can be a suitable approach to treat these pain conditions. In fact, TRPM8 gene deletion 17 or pharmacological inhibition of the channel in both animal models and humans is correlated with a decreased cold hypersensitivity in neuropatic, 17,18 chronic visceral pain, 19 and also migraine. 20 Considering these findings and the potential activity of TRPM8 antagonists also in cancer and other pathologies, 21 it is easy to understand the effort of the academic groups and pharmaceutical/biotech companies to develop potent and selective TRPM8 modulators. 22 To date, two antagonists, the quinoline-3-carboxamido derivative PF-05105679 18a and the amino-2-oxoethyl nicotinic acid derivative AMG-333 20 (Chart 1), have been evaluated for the treatment of cold related pain and migraine, respectively, although they have not passed phase I studies. In 2017, two undisclosed structures, named RQ 00434739 23 and Ice 3682, 18d have reached clinical trials for the treatment of neuropathic pain in Japan and Israel, respectively.
In the past years, the information obtained through mutagenesis experiments 24 and molecular modeling studies 25 on the structure−function of TRPM8 channels has suggested the existence of several independent and overlapping pockets in the TRPM8 binding site able to interact with different antagonist chemotypes. 15c,22,26 This makes it difficult to rationalize pharmacological results, particularly in the context of neuropathic pain, where also agonists of TRPM8 are able to induce analgesia, 27 as well as to define the molecular basis for TRPM8 antagonism. Recently the group of Lee 28a,b resolved the structure of full-length TRPM8 protein from the collared flycatcher Ficedula albicollis (TRPM8 FA ) using cryoelectron microscopy. The network of interactions generated from the TRPM8 FA /menthol, icilin, or lipids lays the structural basis for the design and identification of potent and selective ligands. Importantly, in 2019 two novel structures of TRPM8 complexed with the two antagonists AMTB and TC-I 2014 (PDB codes 6O6R and 6O72) were released, thus providing further important structural details for aiding the identification of TRPM8 modulators. 28c In this context, we have also recently generated a homology model of human TRPM8 using the TRPM8 FA structure as template to rationalize the potent antagonist activity showed by tryptamine 29 and tryptophan-based 18b TRPM8 modulators (3 and 4, Chart 1). In patch-clamp recordings, these compounds were more potent (IC 50 = 367 and 0.2 nM, respectively) than the well-known TRPM8 antagonist BCTC. In vivo, compound 4 attenuated icilin-induced shaking behaviors and reversed oxaliplatin-induced cold allodynia in mice model. Docking studies disclosed the voltage sensor region (VSLD), in the transmembrane segments portion S1− S4, as a possible binding site for these derivatives, highlighting the ability of both compounds to affect the network of interactions established between TM (S1−S4) and the TRP domain at C-terminal of the channel subunits.
In order to deepen the structural requirements necessary for the TRPM8 antagonist activity of these indol-based derivatives, we designed and synthesized a new series of conformationally restricted analogues of 4 pursuing a double aim: (a) to increase the metabolic stability of our lead compound by decreasing its amino acid character; (b) to explore new TRPM8/antagonist interactions leading to the potential discovery of SAR clues. In this paper, we discuss the design and synthesis of three different series of tryptophan restricted analogues of the lead compound 4, namely, tetrahydro-β-carbolines (THBCs), THBC-based diketopiperazines, and THBC-based hydantoin derivatives, as well as the results of TRPM8 antagonist activity obtained by assays of Ca 2+ fluorescence and patch-clamp measurements. These data were rationalized by molecular modeling studies defining new structural requirements for the TRPM8 antagonist activity. Finally, the most potent compound identified was tested in three different in vivo pain models. Chemistry. Tetrahydrobetacarbolines (THBCs) 6a,b, 9, and 10−12a,b, were synthesized as depicted in Scheme 1.
On the other hand, Scheme 4 reports the synthesis of the Naryl hydantoin derivatives 36a−38a and 36b−38b. In this case, a different chemical approach is required because of the minor reactivity of anilines. Intermediates THBC 32a and 32b were coupled with 3CF 3 or 2F or 4-F-phenyl isocyanate in basic medium of TEA. In these conditions, we obtained the corresponding (5R,11aS) trans-and (5S,11aS) cis-hydantoins (36a−38a and 36b−38b, respectively), which were isolated and characterized by 2D NMR spectroscopy. In particular, the cis configuration was evidenced by the correlation between H11a and H5, corresponding to δ 4.53 ppm and δ 5.86 ppm, respectively, for compound 36b ( Figure S76). Absolute configuration was determined as described above. The formation of the cis intermediates, which was not observed with the N-benzyl or N-alkyl analogs, can be explained by the increased stability of the kinetic control species due to the higher rigidity of this structure. However, C11a epimerization was not suppressed and we noticed that the cis conformers converted to their thermodynamically more stable trans congeners (5S,11aR) 36a′−38a′, with a conversion kinetic depending on experimental conditions. High temperatures and alcoholic solvents such as methanol and ethanol favored the conversion to the trans derivative, while in aqueous media at room temperature the cis conformers were more stable ( Figure  S2). Therefore, given the spontaneous trend of cis-hydantoins toward trans-conversion, we considered inappropriate the pharmacological testing of all the cis isomers and we decided to assay only 36b for its pharmacological activity, due to its higher stability in water environment in comparison with its congeners 37b and 38b, which were almost fully converted to the trans isomers during 60 min regardless of the solvent used  Figure S1). In addition, the corresponding C-11a epimers 36a′−38a′ could be obtained directly by reaction of 32b with the corresponding isocyanates and TEA at 60°C for 30 min in 39−45% yields. Pharmacological Characterization. Screening by Ca 2+ -Imaging Assay. TRPM8 blocker activity of all synthesized compounds was tested by Ca 2+ fluorimetric assays using HEK-293 cells stably expressing the rat isoform of TRPM8 channels, using menthol and AMTB as prototypical agonist and antagonist, respectively. All the compounds showed an antagonist activity higher than the canonical TRPM8 antagonist AMTB, although lower than the lead compound 4 with IC 50 values in the 100−0.3 μM range (Table  1).
Patch-Clamp Electrophysiology Assay. Functional assay identified derivatives 6a, 9, 11a, 23, 31a, and 36b to be among the most effective and potent TRPM8 antagonist compounds with IC 50 values in the submicromolar range. To provide direct evidence for this activity, these derivatives were tested in HEK-293 cells transiently expressing the human TRPM8 isoform by whole-cell voltage clamp experiments. Moreover, we decided to perform whole-cell voltage clamp experiments also for compounds 12a and 31a′ in order to further highlight the pharmacophoric properties of the ester group in 11a and of the stereocenters of 31a. As shown in Table 2, the well-known TRPM8 antagonist BCTC (300 nM), used as reference, produced a complete inhibition of mentholgated TRPM8 currents, with an IC 50 of 501 nM. THBC-based diketopiperazine 23 and the hydantoin derivatives 31a have concentration-dependent antagonistic activity, showing IC 50 of 6.57 ± 1.21 nM and of 4.10 ± 1.52 nM, respectively. The THBC 6a and 9 showed decreased potency. The propanoic ester derivative 11a, identified as a potent inhibitor of mentholinduced increase of intracellular Ca 2+ levels (IC 50 = 0.8 μM), antagonized the effect of menthol with an EC 50 of 15.41 nM, while its acid free analogue 12a inhibited only 34% of the menthol-induced current at the maximum concentration of 300 nM. To determine the role of the relative configuration at the stereocenters in the hTRPM8-blocking activity of compound 31a, the pharmacological effect of its 5S,11aR enantiomer, namely, 31a′, was also investigated. As shown in Table 2, 31a′ weakly inhibited menthol-induced currents showing very weak efficacy (11% inhibition) when compared to the 5R,11aS enantiomer, therefore confirming the crucial role of the configurations in the pharmacological properties of this series of compounds.
The activity of compound 36b, which proved to be a powerful antagonist of TRPM8 in Ca 2+ fluorimetric assay, was confirmed by patch clamp experiment with an IC 50 of 7.67 nM, and an inhibition efficacy of the menthol evoked currents of 59.4%. In light of the reported spontaneous epimerization of the cis isomer 36b to its trans congener (36a′) we hypothesized that the cis-isomer contributed mainly to this pharmacological activity. Thus, 36b was assayed in a time course stability test, and results confirmed that the percentage of epimerization was negligible during patch-clamp electrophysiology assays ( Figure S1).
Selectivity Studies. The most potent compounds identified by patch clamp studies (6a, 9, 11a, 23, 31a, and 36b) were subjected to further in vitro characterization by assessment of their selectivity toward TRPV1, TRPA1, and Na v1.7 channels by calcium fluorimetric experiments. TRPV1 and TRPA1 channels belong to the TRP superfamily and share a high degree of homology with TRPM8. 1 On the other hand, Na v1.7 channels are reported to be involved in several neuropathic pain pathways, also modulated by TRPM8. 31 All the derivatives were unable to modulate these channels, showing no activity as agonists or antagonists. Only compounds 6a and 9 showed a negligible antagonistic activity over Na v1.7 with IC 50 > 10 μM ( Figure S2).
Molecular Modeling and Structural Rationale. The TRPM8 three-dimensional structures complexed with the two antagonists AMTB and TC-I 2014 (PDB codes 6O6R and 6O72) released by Diver et al. in 2019 revealed new important details for developing potential modulators of this protein. 28c The preliminary analysis and superposition of both of the TRPM8 structures revealed a very similar protein architecture when bound with the two different antagonists. Starting from these premises, the binding mode of the lead compound 4 was first re-evaluated by considering the TC-I 2014-bound TRPM8 protein structure (PDB code 6O72), chosen as reference system since it featured a better resolution if compared with that originally complexed with AMTB (PDB code 6O6R). In particular, the obtained docking poses of the lead compound 4 revealed a binding mode different from what was reported in the original paper, 43 in which an homology modeling structure of the protein was accounted. Indeed, in the TC-I 2014-bound protein structure (PDB code 6O72), compound 4 adopted a particular shape in which one aromatic function was in front of another one, establishing an intramolecular π−π stacking interaction. Specifically, the aromatic functions of 4 were π−π stacked with several residues stabilizing the ligand/protein complex and allowing a large set of additional interactions, such as H-bond contacts. Indeed, the indole function of 4 was involved in both π−π stacking (with Tyr736) and π−cation (with Arg998) interactions, whereas one benzyl function also established an edge-to-face π−π stacking with Phe729 ( Figure  1). Also, H-bonds were detected for compound 4 with Asn732 and Gln776 ( Figure 1).
In order to shed light about the possible mechanism of action of the reported β-carboline-based TRPM8 antagonists, molecular docking calculations (Glide software) were performed. With the aim of rationalizing the molecular basis behind the different antagonistic activity of the tested molecules, we specifically investigated the predicted protein−  , 9, 11a, 11b, 12a, 12b, 23, 31a, 31a′, 36a, 36a′, 36b. In this way, we investigated both the influence of the molecular architecture, namely, accounting the tetrahydro-β-carboline (6a, 9, 11a, 11b, 12a, 12b), tetrahydropyrazino[1′,2′:1,6]pyrido [3,4-b]indole-1,4(6H,7H)-dione (23), tetrahydro-1Himidazo[1′,5′:1,6]pyrido [3,4-b]indole-1,3(2H)-dione (31a, 31a′, 36a, 36a′, 36b) scaffolds while also considering the effects of the different substituents as well as the impact of the specific stereoarrangements for the three chemotypes on the observed biological activity. The analysis of the ligand docking poses on this specific protein structure highlighted further details for clarifying the action of the investigated compounds at a molecular level (Figures 2 and 3). First, the tetrahydro-β-carboline-based compound 6a, more conformationally restricted if compared with its parent compound 4, showed a slightly different binding mode due to the flip of the indole moiety ( Figure 2A). On the other hand, the careful analysis of the superimposed poses of 4 and 6a highlighted a similar total shape ( Figure 2B), and this was further confirmed by detecting a similar set of key interactions for both the compounds, such as the π−π stacking with Phe729 and the polar contacts with Gln776 and Asn790. Also, an additional π−π was detected with Tyr995, whereas the terminal benzyl moiety established a partial π−π contact with Tyr736 ( Figure 2A).
Compound 9 occupied the TRPM8 binding site showing π− cation interactions with Arg832 and Arg998 and further π−π interactions with Tyr736 (as in the starting compound 4; vide supra) and Phe1003 through the 4-fluorobenzyl function, whereas π−π stacking contacts were detected with Phe729 and Tyr995 through the indole moiety ( Figure 3A). The introduction of a substituent at C-1, as in compounds 11a, 11b, 12a, 12b, determined a similar accommodation in the TRPM8 binding site ( Figure 3B−E). Specifically, in the cases of compounds 11a and 12a the tetrahydro-β-carboline moiety was oriented in front of Phe729 and Tyr995 residues, while the 4-fluorobenzyl substituent interacted again with Arg998 through a π−cation and with Tyr736 through a π−π stacking ( Figure 3B and Figure 3D). Also, the acid moiety in 12a allowed a further H-bond interaction with Arg998 ( Figure  3D). On the other hand, the different stereochemical arrangements of the related analogs 11b and 12b (featuring 1S,3S configuration, instead of 1R,3S as for compounds 11a and 12a) determined a slightly different binding mode. Specifically, for compound 11b, the 4-fluorobenzyl substituent was inserted in a deep cavity in front of Phe1003, while the π−π stacking interactions with Phe729 and Tyr995 were again detected as well as further H-bonds with Asn732 and Arg998 ( Figure 3C). A quite similar binding mode was observed for compound 12b, in which the terminal carboxylate function was involved in H-bond interactions with Asn732 and Arg998, whereas the 4-fluorobenzyl substituent showed in this case a π−π interaction with Tyr736 ( Figure 3E) The introduction of a conformational restriction in compound 23, featuring four fused rings (tetrahydropyrazino-[1′,2′:1,6]pyrido [3,4-b]indole-1,4(6H,7H)-dione scaffold), determined a different placement in the binding site, namely, with the indole moiety establishing a π−cation interaction with Arg998 and Arg832, whereas the terminal benzyl moiety made further π−π contacts with Phe729 and Tyr995 ( Figure 3F). Concerning compound 31a, again featuring four fused rings (tetrahydro-1H-imidazo[1′,5′:1,6]pyrido [3,4-b]indole-1,3(2H)-dione scaffold), the presence of a substituent at C-5 determined a flip of the indole moiety, able to interact with Phe729 and Tyr995 through π−π stacking contacts, as previously observed for 11a, 11b, 12a, 12b that, interestingly, also featured an additional substituent at C-1, corresponding to C-5 in 31a/31a′. Also, the 4-Cl-phenyl substituent at C-5  Journal of Medicinal Chemistry pubs.acs.org/jmc Article determined further π−π interaction with Tyr736, whereas an H-bond contact was established with Arg998 ( Figure 3G). As expected, a similar binding mode was detected for compound 36a, featuring the same absolute configurational pattern of 31a, but the presence of a phenyl substituent at N-2 instead of a benzyl determined a slightly different accommodation of the tetrahydro-1H-imidazo[1′,5′:1,6]pyrido [3,4-b]indole-1,3(2H)dione core and the consequent lack of the π−π stacking between the aromatic substituent at C-5 and Tyr736 (as observed for 31a), replaced by an additional π−cation with Arg998 ( Figure 3I). On the other hand, the corresponding enantiomeric species of 31a and 36a, namely, compounds 31a′ and 36a′, respectively, showed a different occupation of the TRPM8 binding site due to the different stereoarrangements, especially for what concerns the position of the terminal substituted benzyl and aryl moieties, not in line with all the above-reported structure−activity observations, suggesting the poor consistency of this mode of binding that could explain the detected related decreases of antagonistic activity against TRPM8 ( Figure 3H and Figure 3J). Interestingly, compound 36b, the only one of the series featuring the 5S,11aS absolute configuration, showed a three-dimensional arrangement onto the TRPM8 compatible with the establishment of the key interactions with the receptor counterpart, namely, the π−π stacking with Phe736 through the terminal 3-(CF 3 )-aryl moiety (also able to interact with Arg998 through a π−cation) as well as the π−π interaction with Phe729 and Tyr995 with the indole moiety ( Figure 3K). In summary, the comparison of the predicted binding modes related to the reference compound 4 and of the new identified TRPM8 inhibitors disclosed a similar accommodation in the ligand binding site, with the subsequent respect of a network of specific interactions with key residues in the receptor counterpart (e.g., Phe729, Tyr736, Tyr995, Arg998). These in silico results shed light on the rationalization of the observed antagonistic activity of the new identified compounds, providing structural insights for the development of new agents able to interfere with the activity of this target. Starting from these encouraging data at a molecular level, we then moved to the investigation of specific molecular properties of the identified compounds (e.g., in vitro metabolism; vide inf ra) for selecting the most promising items and for further deepening their antagonistic pharmacological profile against TRPM8.
In Vitro Metabolism. The most potent compounds analyzed by patch-clamp electrophysiological assays were further characterized for their metabolic stability using human liver microsomes as in vitro model. Compound 4 was used as reference, considering that its main pitfall was represented by metabolic instability that the newly synthesized compounds were aimed in overcoming. As shown in Figure 4, almost all the compounds proved to be stable in the absence of metabolic cofactors (NADPH or UDP-GlcUA/NADPH) except for 11a, showing unspecific metabolic liability (black bars). In fact, after 60 min in contact with liver microsomes, in the absence of any metabolic cofactors, 11a turnover was 66.5 Figure 3. Predicted binding modes of (A) compound 9 (colored by atom type, C light violet), (B) 11a (colored by atom type, C gray), (C) 11b (colored by atom type, C purple), (D) 12a (colored by atom type, C yellow), (E) 12b (colored by atom type, C orange), (F) 23 (colored by atom type, C pale blue), (G) 31a (colored by atom type, C pale red), (H) 31a′ (colored by atom type, C red-orange), (I) 36a (colored by atom type, C violet), (J) 36a′ (colored by atom type, C light purple), (K) 36b (colored by atom type, C light green) in Journal of Medicinal Chemistry pubs.acs.org/jmc Article ± 3.8%. When the phase I metabolism conditions were mimicked (see protocol I, material and methods section), compound 4 was massively metabolized with a turnover percentage of 98.3 ± 3.1% ( Figure 4, gray bars), in accordance with our previously reported data. 18b Indeed, the newly synthesized analogues showed improved metabolic stability with a metabolic turnover in the range 1.1−72.0% under phase I metabolism conditions. In particular, compound 31a with a phase I metabolic turnover of 26.5 ± 3.9% was the most stable compound. For these reasons, stability of derivative 31a was further challenged using a different protocol that involved both phase I and phase II metabolic cofactors. As shown in Figure 4 (white bar), 31a proved to have a slow metabolic turnover (46.0 ± 2.3%) 32 in the experimental conditions used and was then selected for the in vivo pharmacological assays.
In Vivo Experiments. Effect of 4 and 31a on Icilin-Induced WDS. Initially, we have evaluated the capability of TRPM8 antagonist 31a in blocking the spontaneous wet-dog shake (WDS) induced by icilin in comparison with its precursor derivative 4 at equimolar doses. Due to the difference in metabolic stability, a prolonged pharmacological effect of 31a was expected. For this purpose, 4 and 31 were administrated 30 min before the challenge with icilin (1 mg/kg ip) and WDS was recorded for 30 min. In the vehicle-treated group, a mean of about 128 shakes were counted ( Figure 5, white column). As expected, from the metabolic stability experiments, the pretreatment with 4 (10 mg/kg ip) significantly decreased the number of icilin-induced WDS 0.5 h after the injection ( Figure 5; **p < 0.01 vs vehicle treated mice); no effect was observed at 2 h. On the contrary, 31a (11.5 mg/kg ip) showed a significant effect at both 0.5 and 2 h after the injection ( Figure 5; *p < 0.05 and **p < 0.01 vs vehicle treated mice).
Effect of 31a in Neuropathic Pain Models. TRPM8 plays a critical role in mouse models of chemotherapy-induced neuropathic pain evoked by oxaliplatin (OXP), a condition mimicking cold hypersensitivity provoked by chemotherapyinduced peripheral neuropathy (CIPN). Both acute and chronic OXP-induced cold hypersensitivity has been reproduced in rats and correlated with TRPM8 expression and function. Mizoguchi et al. 33 reported that in a rodent model, acute cold allodynia after OXP injection was alleviated by the TRPM8 blockers N-(2-aminoethyl)-N-[4-(benzyloxy)-3-methoxybenzyl]-N′-(1S)-1-(phenyl)ethyl]urea and TC-I 2014. According to these findings, we investigated the effect of our antagonist 31a in an OXP-induced cold allodynia model, using acetone for cooling stimulation. Considering that the cold pain threshold is increased from ≈12°C to ≈26°C in OXP-treated patients, acetone stimulation is considered to evoke pain in OXP-treated mice.
The activity of compound 31a was evaluated 7 days after three intraperitoneal injections of OXP (6 mg/kg) in C57/BL6 mice, when cold allodynia had developed. As shown in Figure  6, a single subcutaneous administration of 1 μg of 31a was not effective in inhibiting the (OXP)-induced cold allodynia, whereas injections of 10 and 30 μg of our compound showed a remarkable inhibitory effect, which was maximum after 15 min. This effect was still evident 30 min after administration of a 30 μg dose ( Figure 6). These data suggest that 31a may be a viable therapeutic scaffold for the treatment of CIPN.  Further we investigated the efficacy of 31a in a chronic constriction injury (CCI) model of neuropathic pain, using a thermal gradient ring assay. This assay deeply differs from the canonical reflexive measures of nociception, in which the end point is withdrawal to a noxious stimulus, a fact that has been questioned during the past years for their unsatisfactory translation. 34 In particular, this test integrates information on temperature perception distinguishing exploratory behavior from thermal preference behavior. 35a Thus, we measured the thermal preference location of sham, CCI-mice, and CCI-mice treated intraperitoneally with 31a in a thermal gradient assay equilibrated between 15 and 40°C.
The mean temperature to which the sham animal located during the observation time was 27.9°C ± 0.35°C ( Figure  7A), and no statistical differences were evidenced at the different time points ( Figure 7B). No effects on temperature preferences were observed after 31a administration in shammice (data not shown). This value slightly differs from the previously reported by Touska et al. 35a but is consistent with gender, age, and strain differences within animals used. The same temperature preference was observed in CCI-mice 7 days after ligation (mean preferred temperature 25.88°C ± 1.08°C , for CCI mice, p = 1.452 vs sham mice, Figure 7A and Figure 7B). However, 14 days after ligation, when the neuropathic pain and the related nociceptive disorders are well-known to occur, 35b,c the CCI animals displayed a marked preference for colder areas (mean temperature = 22.80°C ± 0.61°C, *p < 0.05 vs sham mice, Figure 7A), which was most prominent during the first 45 min of exposure as shown in Figure 7B (**p < 0.01 vs sham mice) and extending to 60 min. This is in accordance with the cold-seeking behaviors reported during inflammatory states. 35d Considering that thermosensation is mediated by the primary afferent Aδ and C fibers, 35e where TRPM8 is particularly represented, 2 its role in the coldseeking behaviors of CCI animal seems evident. In fact, intraperitoneal administration of the TRPM8 antagonist 31a (11.5 mg/kg) significantly reverted this behavior to 33.30°C ± 1.44°C ( Figure 7A;°p < 0.05 vs CCI 14 days). Similar enhanced thermal tolerance has been recently reported when the antihyperalgesic drug clonidine was administered in a CCI mouse model. 34 Moreover, the mice behavior is also in accordance with previous data that describe TRPM8 deficient mice (TRPM8 −/− ) as rather preferring warmer than colder areas. 35a It should be noted that mice treated with 31a immediately recognized warmer zones as preferable to colder areas compared to vehicle CCI-mice ( Figure 7B;°°°p < 0.001 and°°°°p < 0.0001 vs CCI-mice) also showing a preference for an even warmer temperature than sham animals during the first 15 min ( Figure 7B; # p < 0.05 vs sham mice). It is questionable why this transient effect was recorded, but it must be considered that TRPM8 antagonists are able to decrease the body temperature. This effect could probably account for the thermal preference expressed by animals treated with 31a at 15 min.
The efficacy and the rapid onset of action further confirm the efficacy of compound 31a as TRPM8 antagonists.

■ CONCLUSIONS
Following our interest in the TRPM8 modulation and taking into account the in vivo promising results obtained with a tryptophan-based TRPM8 antagonist (4), in this work we describe the synthesis and the pharmacological characterization of different conformationally restricted analogues of this hit compound, designed with the dual objective of exploring the structural requirements for antagonizing TRPM8 at molecular level and improving the metabolic stability of our hit compound. Some of the synthesized compounds featuring tetrahydrocarboline, tetrahydropyrazino[1′,2′:1,6]pyrido [3,4b]indole-1,4(6H,7H)-dione, and tetrahydro-1H-imidazo- Figure 6. Dose-dependent inhibition of nocifensive paw licking given by compound 31a (1, 10, and 30 μg, sc) in oxaliplatin-induced cold allodynia in C57/BL6 mice. Data are given as the mean ± SEM n = 6. Statistical analysis was two-way ANOVA followed by post hoc Bonferroni test by multiple comparison: ***p < 0.001, ****p < 0.0001.   [3,4-b]indole-1,3(2H)-dione chemical structures showed an efficient and potent TRPM8 antagonist activity in the nanomolar range. Using a new TRPM8 threedimensional protein structure, we rationalized the SAR of this series of compounds by identifying the structural and stereochemical requirements that determine their competitive antagonist activity. One of the synthesized compounds, the (5R,11aS)-5-(4-chlorophenyl)-2-(4-fluorobenzyl)-5,6,11,11atetrahydro-1H-imidazo[1′,5′:1,6]pyrido [3,4-b]indole-1,3(2H)dione, 31a, has a slow metabolic turnover and both overcomes TRPM8-mediated cold hypersensitivity over time, as measured in the WDS assay, and displays acute antinociceptive response 15 min after its application in an oxaliplatin-induced cold allodynia model. In addition, 31a also shows remarkable analgesic activity in an animal model of CCI-induced hyperalgesia. These last data are in agreement with the results obtained with 4 in other models of neuropathic pain 27d but differ with those obtained by other authors who demonstrate the efficacy of the TRPM8 agonists in animal models of injuryinduced neuropathic pain. 27a−c Our results confirm the validity of the indole nucleus in the design of potent TRPM8 modulators, adding one more piece to the puzzle that composes the TRPM8's complex biology in the transmission and modulation of pain.

■ EXPERIMENTAL SECTION
General. All reagents and solvents used were purchased from Sigma-Aldrich (Milan, Italy) unless otherwise stated. Reactions were performed under magnetic stirring in round-bottomed flasks unless otherwise noted. Moisture-sensitive reactions were conducted in oven-dried glassware under nitrogen stream, using freshly distilled solvents. TLC analysis of reaction mixtures was performed on precoated glass silica gel plates (F254, 0.25 mm, VWR International), while crude products were purified by the Isolera Spektra One automated flash chromatography system (Biotage, Uppsala, Sweden), using commercial silica gel cartridges (SNAP KP-Sil, Biotage). NMR spectra were recorded on a Bruker Avance 400 MHz apparatus, at room temperature. Chemical shifts were reported in δ values (ppm) relative to internal Me 4 Si for 1 H and 13 C NMR and to CFCl 3 for 19 F NMR. J values were reported in hertz (Hz). 1 H NMR and 19 F NMR peaks were described using the following abbreviations: s (singlet), d (doublet), t (triplet), and m (multiplet). HR-MS spectra were recorded by LTQ-Orbitrap-XL-ETD mass spectrometer (Thermo Scientific, Bremen, Germany), equipped with an ESI source. Analytical RP-HPLC analysis of final products was performed through a Nexera UHPLC system (Shimadzu, Kyoto, Japan) consisting of a CBM-20A controller, two LC-30AD pumps, a DGU-20 A5R degasser, an SPD-M20A photodiode array detector, a CTO-20AC column oven, a SIL-30AC autosampler, and a Kinetex C18 150 mm × 2.1 mm × 2.6 μm (100 Å) column (Phenomenex, Bologna, Italy). The optimal mobile phase consisted of 0.1% HCOOH/H 2 O v/ v (A) and 0.1% HCOOH/ACN v/v (B). Analysis was performed in gradient elution as follows: 0−13.00 min, 5−65% B; 13−14.00 min, 65−95% B; 14−15.00 min, isocratic to 95% B; then 3 min for column re-equilibration. Flow rate was 0.5 mL min −1 . Column oven temperature was set to 40°C. Injection volume was 5 μL of sample. The following PDA parameters were applied: sampling rate, 12.5 Hz; detector time constant, 0.160 s; cell temperature, 40°C. Data acquisition was set in the range 190−800 nm, and chromatograms were monitored at 230 nm. Analytical RP-HPLC confirmed that all final compounds had a purity of >95%. For quantitative analysis, the calibration curve was obtained in a concentration range of 2.5−40 μM with five concentration levels and triplicate injections of each level were run. Peak areas were plotted against corresponding concentrations, and the linear regression was used to generate a calibration curve with R 2 values of ≥0.999 (Table S1).
All circular dichroism spectra were recorded using a JASCO J810 spectropolarimeter at 25°C in the range λ = 260−190 nm (1 mm path length, 1 nm bandwidth, four accumulations, and a scanning speed of 10 nm min −1 ). Compounds were dissolved in methanol at a concentration of 0.100 mM. Spectra were corrected for the solvent contribution.
General Procedure A: Pictet−Spengler Reaction. 1 mmol of L-tryptophan methyl ester or (S)-2-amino-N-(4-fluorobenzyl)-3-(1Hindol-3-yl)propanamide (8) was dissolved in methanol and added with the proper aldehyde (1.5 equiv) and trifluoroacetic acid (1.5 equiv). The mixture was subjected to a microwave assisted closed vessel reaction for 45 min at 110°C. 36 The mixture was then evaporated in vacuo, and the residue was dissolved in dichloromethane and was washed three times with water. The organic phase was extracted, dried over Na 2 SO 4 , filtered, and concentrated under vacuum. The crude products were purified by flash chromatography using mixtures of n-hexane/ethyl acetate as mobile phase.
General Procedure B: Coupling Reactions. 1 mmol of the proper carboxylic acid was dissolved in dichloromethane/DMF (4:1 v:v) and added with HoBt (1.2 equiv), HBTU (1.2 equiv), DIPEA (2.4 equiv), and the corresponding amine (1.2 equiv) and stirred at room temperature overnight. Then, the solvent was evaporated in vacuum, and the residue was dissolved in dichloromethane and washed with water (3 times), a saturated solution of NaHCO 3 (3 times), and a solution of citric acid (10% w:w). The organic phase was extracted, dried over Na 2 SO 4 , filtered, and concentrated under vacuum. The crude products were purified by flash chromatography using mixtures of n-hexane/ethyl acetate as mobile phase.
General Procedure C: Boc Removal. The N-Boc protected intermediate (0.2 mmol) was dissolved in a mixture of TFA/DCM (1/3, v/v), and triisopropylsilane (TIS, 0.25 equiv) was added. Reaction was stirred at room temperature for 2 h. Then, a solution of NaOH (2 N) was added dropwise until pH 7. The mixture was diluted with water and dichloromethane, and the organic phase was extracted, dried over Na 2 SO 4 , filtered, and concentrated under vacuum. The crude products were purified by flash chromatography using mixtures of n-hexane/ethyl acetate as mobile phase.
General Procedure D: Hydantoin Synthesis. Diastereoisomerically pure tetrahydro-β-carbolines (0.2 mmol) were dissolved in THF, and 0.4 equiv of triphosgene was added. The pH was adjusted to 8 by addition of TEA, and the mixture was stirred at room temperature for 10 min. Then, the proper amine (1.2 equiv) was added and the resulting mixture was refluxed for 1 h. After cooling to room temperature, the solvent was evaporated, the residue reconstituted in dichloromethane and washed with water (3 times). The organic phase was extracted, dried over Na 2 SO 4 , filtered, and concentrated under vacuum. The crude products were purified by flash chromatography using mixtures of n-hexane/ethyl acetate as mobile phase.
General Procedure E: Hydantoin Synthesis. Tetrahydro-βcarboline 32a or 32b (0.2 mmol) was dissolved in THF, and 1.2 equiv of trimethylamine and 1.2 equiv of the proper isocyanate were added. The mixture was stirred at room temperature for 30 min. The solvent was evaporated, the residue reconstituted in dichloromethane and washed with water (3 times). The organic phase was extracted, dried over Na 2 SO 4 , filtered, and concentrated under vacuum. The crude products were purified by flash chromatography using mixtures of nhexane/ethyl acetate as mobile phase.
General Procedure F: Hydantoin Synthesis. Tetrahydro-βcarboline 32b (0.2 mmol) was dissolved in THF, and 1.2 equiv of trimethylamine and 1.2 equiv of the proper isocyanate were added. The mixture was stirred at room temperature for 30 min and then refluxed for further 30 min. After cooling to room temperature, the solvent was evaporated, the residue reconstituted in dichloromethane and washed with water (3 times). The organic phase was extracted, dried over Na 2 SO 4 , filtered, and concentrated under vacuum. The Journal of Medicinal Chemistry pubs.acs.org/jmc Article crude products were purified by flash chromatography using mixtures of n-hexane/ethyl acetate as mobile phase.
Fluorimetric Assays. The tested molecules dissolved in DMSO were added at the desired concentrations, and the plates were incubated in darkness at 37°C in a humidified atmosphere of 5% CO 2 for 60 min. The fluorescence was measured using instrument settings appropriate for excitation at 485 nm and emission at 535 nm (POLARstar Omega BMG LABtech). A baseline recording of four cycles was recorded prior to stimulation with the agonist (100 μM menthol for TRPM8). The TRPM8 antagonist, 10 μM AMTB, was added to the medium containing the corresponding agonist to induce channel blockade. The changes in fluorescence intensity were recorded during 15 cycles more. The higher concentration of DMSO used in the experiment was added to the control wells. The cells' fluorescence was measured before and after the addition of various concentrations of test compounds. The fluorescence values obtained are normalized to that prompted by the corresponding agonist (for channel activating compounds) or upon agonist and antagonist coexposure (for channel blocker compounds).
Selectivity Assays. The analysis was performed in 384-well clear bottom black walled polystyrene plates, (Thermo Scientific, Waltham, USA) for CHO-K1 cells and in 384-well clear bottom black polystyrene walled poly-D-Lys coated plates (TwinHelix, Rho, Italy) for HEK-293 cells. Compound dilution was performed in 96-well U bottom plates (Thermo Scientific), and then compounds were transferred into 384-well V bottom polypropylene barcoded plates (Thermo Scientific). To assess the activity of the selected compound over TRPA1 and TRPV1, cells were seeded in 384 MTP in complete medium (25 μL/well) at 10 000 cells/well concentration. 24 h after seeding, the culture medium was removed and cells were loaded with 20 μL/well of 0.5× calcium sensitive dye (Fluo-8 NW, AAT Bioquest, Sunnyvale, USA) in assay buffer. To assess the activity of the selected compound over Nav1.7, cells were seeded at 15 000 cells/well in 384 MTP in complete medium (25 μL/well). 24 h after seeding, the culture medium was removed and cells were loaded with 20 μL/well of 0.5× membrane potential dye (FLIPR membrane potential assay kits Blue, Molecular Devices LLC, San Jose, USA) in assay buffer. Plates were incubated for 1 h at room temperature in the dark. Then, 10 μL/well of test compounds and controls were injected at 3× concentration, and the signal of the emitted fluorescence was recorded using FLIPRTETRA apparatus (ForteBio, Fremont, USA). Then, a second injection of 15 μL/well of 3× reference activator (at ∼EC 80 ) was performed analyzing the signal of the emitted fluorescence. Allyl isothiocyanate (AITC, Sigma-Aldrich), capsaicine (Sigma-Aldrich), and veratridine (Sigma-Aldrich) were used as reference agonists, while HC-030031 (Sigma-Aldrich), capasazepine (Sigma-Aldrich), and tetrodotoxine (Tocris bioscience, Bristol, U.K.) were used as reference antagonists for TRPA1, TRPV1, and Nav1.7 assaying, respectively.
Menthol was used as reference agonist, and a stock solution (1 M, 100% DMSO) was prepared the day of the experiment from the powder; an intermediate stock of 300 mM was prepared from the 1 M stock in 100% DMSO, and the final dilution was performed in the extracellular solution to obtain a working concentration of 300 μM (1:1000, 0.1% final DMSO concentration). Stock solutions of the testing compounds (10 mM; 100% DMSO; stored at −20°C) were prepared the day of the experiment; an intermediate stock for each compound (300 μM) was prepared from the 10 mM stock in 100% DMSO, and the working dilutions were performed just before the experiments in the extracellular solution containing 300 μM menthol. The highest concentration tested was 300 nM, with serial dilutions (1:10) in the extracellular solution. DMSO was balanced to keep it constant throughout all the solutions in the same experiment (0.2% final DMSO concentration). Standard whole-cell voltage clamp experiments are performed at room temperature using the multihole technology. For the voltage clamp experiments on human TRPM8, data are sampled at 2 kHz. After establishment of the seal and the passage in the whole cell configuration, the cells are challenged by a voltage ramp (20 ms step at −60 mV; 100 ms ramp −60/+100 mV; 20 ms step at +100 mV; return to −60 mV) every 4 s. The potential antagonistic effect on human TRPM8 current of target compounds was evaluated after application of the agonist (menthol, 300 μM) alone and in the presence of the compound under investigation at increasing concentrations. Output: outward current evoked by the  40 Specifically, water molecules were deleted, cap termini were included, all hydrogen atoms were added, and bond orders were assigned. Finally, the .pdb files were converted to the .mae file.
The grids for the subsequent molecular docking calculations were generated accounting the related position of TC-I 2014 on the receptor binding sites. In this way, the cocrystallized ligands were also automatically removed from the original binding sites.
The library of investigated compounds (see Results and Discussion) was prepared using LigPrep software (Schrodinger Suite). 41 Specifically, all the possible tautomers and protonation states at pH = 7.4 ± 1.0 were generated for each compound, and finally the structures were minimized using the OPLS 2005 force field.
Molecular docking experiments were performed using Glide software (Schrodinger Suite), 42 setting the Extra Precision [XP] mode. For this step, 20 000 poses were kept in the starting phase of docking, and 1200 poses for energy minimization were selected. The scoring window for keeping the initial poses was set to 400.0, and a scaling factor of 0.8 related to van der Waals radii with a partial charge cutoff of 0.15, based on a 0.5 kcal/mol rejection cutoff for the obtained minimized poses, was considered. In the output file, 10 poses for each compound were saved.
In Vitro Metabolic Stability Using Liver Microsomes. Protocol I. Each sample (2.5 mM) was incubated with 100 mM phosphate buffer (pH 7.4) and 20 mg/mL of liver microsomes (Thermo Fisher Scientific, Bremen, Germany). After preincubation in water bath for 5 min, the mixture was incubated with 20 mM NADPH (protocol I) at 37°C for 60 min in a Thermomixer comfort (Eppendorf, Hamburg, Germany).
Protocol II. For the measurement of UGT activity the microsomes were preincubated with alamethicin, which forms pores in microsomal membranes, promoting access of substrate and cofactor to UGT enzymes. Subsequently, each sample was incubated with 100 mM phosphate buffer, 500 mM magnesium chloride, 10 mM NADPH, and 20 mM UDP-GlcUA at 37°C for 60 min.
Finally, the reactions from both protocols (protocols I and II) were stopped by the addition of 200 μL of ice-cold methanol, and then samples were centrifuged at 10 000 rpm at 25°C for 5 min (Eppendorf microcentrifuge 5424, Hamburg, Germany). The supernatants were collected and injected in UHPLC-PDA.
The control at 0 min was obtained by addition of the organic solvent immediately after incubation with microsomes. As the positive control, testosterone was used, while the negative controls were prepared by incubation up to 60 min without NADPH and UDP-GlcUA/NADPH for protocols I and II, respectively. The negative control is essential to detect problems such as nonspecific protein binding or heat instability. The extent of metabolism is expressed as a percentage of the parent compound turnover using the following equation, as previously described: 43 Ä Ç Å Å Å Å Å Å Å Å É Ö Ñ Ñ Ñ Ñ Ñ Ñ Ñ Ñ % parent compound turnover 100 concentration at 60 min concentration at 0 min 100 = − × Animals. C57-mice (males, 5 week old, ∼30 g) (Harlan, The Netherlands) were used for the oxaliplatin-induced neuropathic pain study. All experiments were approved by the Institutional Animal and Ethical Committee of the Universidad Miguel Hernandez where experiments were conducted, and they were in accordance with the guidelines of the Economic European Community and the Committee for Research and Ethical Issues of the International Association for the Study of Pain. All parts of the study concerning animal care were performed under the control of veterinarians.
The WDS was performed in Wistar male rats (300−350 g), and the thermal ring experiment was performed on male Swiss CD1 mice (30−35 g) purchased from Charles Rivers (Calco-Lecco-Italy) and then housed in the animal care facility of the Department Experimental of Pharmacology, University of Naples. The animals were acclimated to their environment for 1 week, and food and water were available ad libitum. All behavioral tests were performed between 9:00 am and 1:00 pm, and animals were used only once. Procedures involving animals and their care were conducted in conformity with international and national law and policies (EU Directive 2010/63/ EU for animal experiments, ARRIVE guidelines, and the Basel declaration including the 3R concept). All procedures reported here were approved by the Institutional Committee on the Ethics of Animal Experiments (CVS) of the University of Naples Federico II and by "Ministero della Salute" under Protocol No. 851/2016. All efforts were made to minimize animal suffering, and at the end of all experiments, the animals were euthanized by CO 2 overdose.
Drug Treatment. For the oxaliplatin-induced neuropathic pain assay, oxaliplatin (Tocris) was dissolved in water with gentle warming and was subcutaneously (sc) injected on days 1, 3, and 5 at a 6 mg/kg dose. The day 7 after administration, experiments were performed. Together with oxaliplatin injection, saline and a 5% mannitol solution were intraperitoneally injected to prevent kidney damage and dehydration. 31a stock was prepared in DMSO (Sigma-Aldrich) and diluted in saline for injections. Compound 31a at different doses (1 to 30 μg) was injected into the plantar surface (25 μL) of the right hind paw of mice.
For the other in vivo assays, compound 4 and 31a were dissolved in PEG 400 10% v/v, Tween 80 5% v/v, and sterile saline 85% v/v and injected once intraperitoneally at the equimolar doses of 10 mg/kg for 4 and 6.7 mg/kg for 31a. Control group was only treated with vehicle.
Icilin-Induced "Wet-Dog" Shaking in Rats. Icilin, a TRPM8 agonist, was used to induce shaking in mice. 44 Animals were first habituated to the testing room for 30 min. After that they were randomized into treatment groups and treated with vehicle or TRPM8 antagonists. Icilin was administered intraperitoneally (ip) at 1 mg/kg dissolved in 1% Tween 80/H 2 O 30 or 120 min after drugs. The number of intermittent but rhythmic "wet-dog-like" shakes (WDS) of neck, head, and trunk in each animal was counted for a period of 30 min following icilin administration.
Oxaliplatin-Induced Neuropatic Pain Model. Cold chemical thermal sensitivity was assessed using acetone drop method. 18b Mice were placed in a metal mesh cage and allowed to habituate for approximately 30 min in order to acclimatize them. Freshly dispensed acetone drop (10 μL) was applied gently onto the mid-plantar surface of the hind paw. Cold chemical sensitive reaction with respect to paw licking was recorded as a positive response (nociceptive pain response). The responses were measured for 20 s with a digital stopwatch. For each measurement, the paw was sampled twice and the mean was calculated. The interval between each application of acetone was approximately 5 min.
Chronic Constriction Injury (CCI) Model of Neuropathic Pain. Neuropathic pain behavior was induced by ligation of the sciatic nerve as described previously. 27d Briefly, mice were first anesthetized with xylazine (10 mg/kg ip) and ketamine (100 mg/kg ip), and the left thigh was shaved and scrubbed with betadine, and then a small incision in the middle left thigh (2 cm in length) was performed to expose the sciatic nerve. The nerve was loosely ligated at two distinct sites (spaced at a 2 mm interval) around the entire diameter of the nerve using silk sutures (7−0). The surgical area was closed and finally scrubbed with betadine. In sham-operated animals, the nerve was exposed but not ligated. Drug effects were evaluated 7 and 14 days after ligation.
Thermal Gradient Ring. We utilized the thermal gradient ring from Ugo-Basile previously using a modified protocol from Touska et al., 2016. 35 The apparatus consists of a circular running track where each side of the ring is divided into 12 zones, in which the temperature is proportionally distributed from 15 to 40°C, and each sector represents an increment of 2.27°C. Before the experiment, on day 1, all mice were habituated to the apparatus for 30 min with the Journal of Medicinal Chemistry pubs.acs.org/jmc Article aluminum floor acclimatized to room temperature (22−24°C). On day 2, mice were injected and 30 min after were placed in the apparatus and measured for 60 min using 15−40°C. Data on preference temperature in time course were collected from the videotracking software Any-Maze connected to the apparatus. Data Analysis. Data are reported as the mean ± standard error of the mean (sem) values of at least three independent experiments each in triplicate. Statistical analysis was performed by analysis of variance test, and multiple comparisons were made by Bonferroni's test by using Prism 5 (GraphPad Software, San Diego, CA, USA). p-values smaller than 0.05 were considered significant.