Calcitonin Gene-Related Peptide Receptor Antagonists: New

Jun 24, 2014 - Copyright © 2014 American Chemical Society. *Phone: 215-652-6455. E-mail: [email protected]. Biography. Ian M. Bell was educated at G...
15 downloads 7 Views 4MB Size
Perspective pubs.acs.org/jmc

Calcitonin Gene-Related Peptide Receptor Antagonists: New Therapeutic Agents for Migraine Ian M. Bell* Department of Discovery Chemistry, Merck Research Laboratories, West Point, Pennsylvania 19486, United States ABSTRACT: Calcitonin gene-related peptide (CGRP) is a potent neuromodulator and vasodilator. It has been implicated in the pathogenesis of migraine by a number of lines of evidence, although its precise role has yet to be fully defined. Compelling evidence for the importance of CGRP in migraine has been provided by clinical trials with multiple small molecule CGRP receptor antagonists. These clinical studies have shown that blockade of the CGRP receptor can produce antimigraine efficacy comparable to that of the gold standard triptan class of drugs with an incidence of adverse events that appears to be relatively low. The present review describes the discovery and development of these new antimigraine agents and highlights the challenges of identifying orally acting drugs that target a family B G-protein-coupled receptor.



INTRODUCTION Migraine is a highly disabling neurovascular disorder characterized by attacks of moderate to severe headache that are often associated with nausea, vomiting, photophobia, and phonophobia.1,2 The attacks can last from 4 to 72 h, and the average attack frequency is 1 or 2 per month.2 About 20−30% of migraine patients experience transient focal neurologic symptoms known as aura, which are usually visual and can precede or accompany the headache.1 Migraine afflicts about 11% of adults worldwide and results in a significant socioeconomic burden, in terms of both quality of life and lost productivity.3 The details of migraine pathogenesis have been debated since the 17th century, and the discourse has been dominated by two main theories: the vascular theory and the central neuronal theory.4 The vascular theory, described by Willis around 350 years ago, proposed that dilation of arteries in the meninges resulted in migraine pain.5 By the late 19th century, various neuronal theories of migraine had been advanced in opposition to the classic vascular theory.4 A notable example was Liveing’s proposal that migraine originated in the cerebral cortex as a neural disturbance he termed a “nerve storm”.5 In the mid-20th century, a series of studies by Wolff lent support to the vascular theory and specifically to the notion that vasodilation of cerebral and meningeal blood vessels could lead to activation of nociceptors and elicit a painful response.6 A more recent concept is neurogenic inflammation, in which trigeminal nerve activation is thought to lead to release of proinflammatory peptides that cause a sterile neurogenic inflammation of the meninges.7 Under these conditions, peripheral nerve fibers are hypothesized to be sensitized to normally innocuous stimuli.2 The central neuronal theory holds that migraine originates from aberrant function of neuronal circuits within the central nervous system (CNS).4 Key evidence supporting this view © XXXX American Chemical Society

includes genetic evidence associating familial hemiplegic migraine with several proteins involved in the control of CNS excitability.4 For more than 300 years, these theories have helped to advance understanding of the origins of migraine, but important questions remain unanswered and research on this neurovascular disorder continues. Acute treatment of migraine is frequently based on nonspecific pain medications, including over-the-counter analgesics or more-powerful opioids.8 Consistent with the early vascular theories of migraine pathogenesis, the first specific antimigraine drugs were potent vasoconstrictors: the ergot alkaloids ergotamine and dihydroergotamine.3 Introduced in the first half of the 20th century, the “ergots” have high affinity for multiple receptors, including those for serotonin, dopamine, and noradrenaline, and it is believed that this polypharmacology contributes to some adverse events.8 The vasoconstrictive action of the ergots has been associated with cardiovascular side effects, although it is also thought to be important in their antimigraine efficacy.3 In the 1990s, the introduction of the triptan class of drugs was a significant breakthrough for the acute treatment of migraine.9 The triptans are selective agonists of the 5-HT1B and 5-HT1D receptors, and they are thought to exert their antimigraine effects by vasoconstriction of cranial blood vessels and by inhibition of the release of neuropeptides in the perivascular nerve terminals.10 Triptans are effective for the acute treatment of migraine and have fewer side effects than the ergot alkaloids presumably because of their cleaner pharmacology.3 Unfortunately, they also cause vasoconstriction of coronary arteries and are therefore contraindicated in patients with cardiovascular disease and uncontrolled hypertension.8 There is significant Received: March 7, 2014

A

dx.doi.org/10.1021/jm500364u | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Perspective

ligand binding.23 The receptor activity-modifying proteins, such as RAMP1, are also transmembrane proteins with significant Nterminal ECDs, and both CLR and CTR can associate with different RAMPs to produce pharmacologically distinct receptors.23 As noted, the combination of CLR with RAMP1 produces the CGRP receptor, but CLR may also associate with RAMP2 or RAMP3 to produce high-affinity receptors for AM that are designated AM1 and AM2, respectively.22 Similarly, CTR may partner with RAMPs to produce receptors for AMY.22 For example, the combination of CTR and RAMP1 affords the receptor designated AMY1, which is structurally similar to the CGRP-R by virtue of the homology between CLR and CTR. A number of studies on the CGRP receptor (CLR/RAMP1), using both chimeric receptors and mutational approaches, have revealed that the ECDs of both proteins are crucial for agonist binding and selectivity.24−26 Mutagenesis studies have also demonstrated that a number of residues in the ECDs are important for binding of antagonists, most notably CLR Met42, RAMP1 Trp74, and RAMP1 Trp84 in the human CGRP receptor.22,27−30 Cross-linking studies with [125I]CGRP have indicated that the agonist apparently interacts with both CLR and RAMP1.24 Additionally, residues in the extracellular loops between the transmembrane domains in CLR, as well as in the transmembrane domains themselves, have been shown to play a role in agonist binding and receptor function.31 Although the precise details of CGRP binding to, and activation of, its receptor remain elusive, the two-domain model for family B GPCRs described by Hoare is consistent with the available data.32 The two-domain binding model for the CGRP receptor is shown in Figure 2, and it highlights the importance

interest in the development of new antimigraine agents that are safe and effective and, in particular, lack the cardiovascular liabilities seen with ergots and triptans. Although the precise details of migraine pathogenesis are not fully understood, one aspect that is generally accepted is the involvement of CGRP, a 37-amino acid neuropeptide that was first described about 30 years ago.11 CGRP is widely expressed in the peripheral and central nervous systems and is implicated in a number of biological functions.12 There is increasing evidence that CGRP plays an important role as a neuromodulator,4 but it is best known as a potent vasodilator.13 Evidence that CGRP was involved in cerebrovascular regulation led to a postulated role for the peptide in migraine pathophysiology.14 Since this original proposal, support for the involvement of CGRP in this neurovascular disorder has been provided by a number of studies. For example, a collaboration between Edvinsson and Goadsby determined that there was a significant increase in the craniovascular levels of CGRP in migraineurs during migraine headache.15 Moreover, these migraine-associated elevations of CGRP were found to be eliminated along with headache following treatment with sumatriptan.16 Another study in migraineurs found that intravenous infusion of CGRP induced a migraine-like headache.17 The growing body of evidence linking CGRP to migraine led to interest in the development of agents capable of blocking the effects of this neuropeptide. The present review will focus on the efforts to discover and develop small molecule antagonists of the CGRP receptor for acute treatment of migraine in the clinic.



CGRP RECEPTOR STRUCTURE AND FUNCTION CGRP is a member of the calcitonin (CT) family of peptides that also includes amylin (AMY) and adrenomedullin (AM).18,19 These peptides vary significantly in terms of amino acid sequence but share key structural features, including a C-terminal amide group and an N-terminal cyclic structure formed with a Cys−Cys disulfide bond.18 Two forms of CGRP are known: α-CGRP, produced by alternate splicing of the CT gene; β-CGRP, which differs from α-CGRP by three amino acids in humans and is encoded by a separate gene.19 The human amino acid sequences of the two forms of CGRP are shown in Figure 1. α-CGRP and β-CGRP have similar

Figure 2. Schematic two-domain binding model for the CGRP receptor. (A) The C-terminal region of CGRP (brown) binds to the N-terminal domains of CLR (teal) and RAMP1 (dark blue), increasing the local concentration of the N-terminal region of the peptide. (B) The N-terminal region of CGRP interacts with the juxtamembrane region of CLR, leading to a conformational change in the GPCR and consequent G-protein activation and signaling. (C) A small molecule antagonist (red) binds to the N-terminal domains of both CLR and RAMP1, blocking the association of the receptor with the CGRP Cterminal region and preventing receptor activation.

Figure 1. Amino acid sequences of human α-CGRP and β-CGRP.

biological properties but somewhat different distribution, with α-CGRP more highly expressed in primary neurons while βCGRP predominates in enteric neurons.20 Interestingly, deletion of the N-terminal cyclic structure from CGRP affords a functional antagonist of the CGRP receptor (CGRP8−37), demonstrating the importance of the first seven residues of CGRP for receptor activation.19 The CGRP receptor (CGRP-R) is a heterodimer consisting of the calcitonin receptor-like receptor (CLR) and receptor activity-modifying protein (RAMP) 1.21,22 As its name implies, CLR is homologous to the calcitonin receptor (CTR) and both are members of the secretin family (family B) of G-proteincoupled receptors (GPCRs), which are characterized by a large N-terminal extracellular domain (ECD) that is important for

of the ECD for both agonist and antagonist binding. It is thought that the C-terminal region of CGRP initially binds to the large N-terminal ECD of the receptor, likely making interactions with both CLR and RAMP1. This initial binding event greatly increases the local concentration of the Nterminal region of CGRP in the vicinity of the juxtamembrane portion of CLR, allowing their relatively weak interaction to B

dx.doi.org/10.1021/jm500364u | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Perspective

Figure 3. Crystal structures of the CLR:RAMP1 ectodomain complexes.36 CLR (teal) and RAMP1 (dark blue) are shown as (A) unliganded (PDB code 3N7P), (B) complexed with olcegepant (magenta) (PDB code 3N7S), and (C) complexed with telcagepant (green) (PDB code 3N7R).

Figure 4. Structures of olcegepant and telcagepant with the privileged structure (PS) designated.

Figure 5. Crystal structures of the CLR:RAMP1 ectodomain complexes.36 View of the antagonist binding site formed by CLR (teal) and RAMP1 (dark blue) complexed with (A) olcegepant (magenta) (PDB code 3N7S) and (B) telcagepant (green) (PDB code 3N7R). Side chains of selected residues involved in antagonist binding are shown and labeled in black (CLR residues) or white (RAMP1 residues).

receptors and the difficulty of isolating suitable quantities of purified protein have hampered these efforts, and the heterodimeric nature of the CGRP receptor represents an additional complication. On the basis of the evidence indicating that small molecule antagonists bind to the ECD of the CGRPR (vide supra), researchers at Vertex Pharmaceuticals used a reductionist approach to express and purify a stable complex of the ECDs of CLR and RAMP1.35 The purified ECD complex exhibited relatively low-affinity binding to CGRP but highaffinity binding to known small molecule CGRP receptor antagonists, consistent with the two-domain model outlined in

occur and resulting in receptor activation. Since mutagenesis experiments indicated that most small molecule antagonists interacted with the ECD of CLR/RAMP1, it was hypothesized that they bind to this region of the receptor and prevent the initial binding of CGRP to the receptor.22 A notable exception to this model of peptide binding and small molecule receptor antagonism is the hydroxypyridine class of antagonists,33 which apparently interact with transmembrane domain 7 (TM7) in CLR and not with the extracellular domain.28 Obtaining detailed structural data on GPCRs remains a significant challenge.34 The membrane-bound nature of these C

dx.doi.org/10.1021/jm500364u | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Perspective

Figure 2.35 This complex was successfully crystallized, and the structure of the CLR:RAMP1 ectodomain was solved (Figure 3).36 The N-terminal domain of CLR has a similar overall fold to other reported structures for family B GPCR ectodomains, such as those of the parathyroid hormone (PTH) or gastric inhibitory polypeptide (GIP) receptors.37,38 The RAMP1 ectodomain forms a three helix bundle, similar to an earlier report of the crystal structure of the RAMP1 ECD alone,39 and two of these helices pack against an N-terminal helix in CLR to form the interface between the two proteins. In analogy with the structures of the PTH1 and GIP receptor ectodomains,37,38 which were obtained as complexes with their respective endogenous ligands PTH and GIP, ter Haar et al. suggested that the CGRP peptide might bind in the pocket formed at the interface between CLR and RAMP1.36 This suggestion is also consistent with known mutations that affect binding of CGRP to its receptor.19 The Vertex group also solved the crystal structures of the CGRP ECD complex bound to the antagonists olcegepant (1) and telcagepant (2) (Figures 3 and 4).36 These important studies revealed structural details of how a small molecule can bind to and antagonize a family B GPCR (Figure 5). Most known small molecule CGRP receptor antagonists possess a terminal cyclic CONH moiety, a key motif in the so-called “privileged structure”40 found in these molecules (Figure 4). In the reported crystal structures of the complexes with 1 and 2, this amide functionality is engaged in hydrogen bonding interactions with the backbone elements of CLR Thr122.36 The crystal structures also shed light on mutagenesis studies that implicated key residues for antagonist binding.27−30 For example, CLR Met42, RAMP1 Trp74, and RAMP1 Trp84 come together to form a hydrophobic pocket at the interface of the two proteins. This hydrophobic pocket accommodates groups in the small molecule antagonists known to be important for receptor affinity: the dibromophenol in 1 and the difluorophenyl ring in 2. Overall, 1 and 2 were found to possess a similar binding topology. Both molecules bind in an extended conformation, bridging a distance of about 18 Å between the crucial Thr122 hydrogen bonding interactions and the hydrophobic pocket at the CLR/RAMP1 interface.36 Importantly, these crystal structures are consistent with known structure−activity relationships for analogues of olcegepant41 and telcagepant.42 Moreover, the nature of the binding site underscores the difficulty of obtaining orally acting drugs that target the CGRP receptor. In order to achieve sufficient potency and selectivity, small molecule antagonists of this receptor must apparently span the distance from CLR Thr122 to the hydrophobic pocket at the CLR:RAMP1 interface.36 These spatial considerations and the need for hydrogen bond donors and acceptors help to explain why most potent CGRP-R antagonists are relatively large, rigid molecules, typically with multiple amide moieties. Such structures tend to violate known guidelines for orally acting drugs, such as Lipinski’s rule of five, and their physicochemical properties are often not conducive to good oral pharmacokinetics. Thus, the development of oral drugs for this receptor, and by analogy for other family B GPCRs, is nontrivial. As already noted (vide supra), the indole ring of RAMP1 Trp74 makes important interactions with the dibromophenol of 1 and the difluorophenyl of 2.36 Earlier mutagenesis studies highlighted the importance of this residue, demonstrating that both of these antagonists exhibited significantly reduced affinity

for the Trp74Ala mutant receptor.30,43 Even more interestingly, this residue plays a crucial role in determining species selectivity for such antagonists. Most potent small molecule CGRP-R antagonists exhibit pronounced species selectivity, displaying high affinity for primate receptors such as those of rhesus monkey and human but significantly reduced affinity for nonprimate CGRP receptors, such as those from rats or dogs.22 One consequence of this is that it is often challenging to fully evaluate such antagonists in rodents. Indeed, most drug discovery programs have developed non-human primate models for in vivo evaluation of CGRP-R antagonism, such as those based on facial flushing in marmosets44 or capsaicininduced dermal vasodilation in rhesus monkeys.45 Studies by Mallee et al. revealed that this species selectivity is determined by the RAMP1 protein and, more specifically, by the region comprising residues 66−112 of the protein.27 Following an inspection of the amino acid sequence of this region of RAMP1 from several species, they demonstrated that mutation of residue 74 in rat RAMP1 from lysine to tryptophan produced a receptor with human-like pharmacology.27 The observation that a single residue in RAMP1 was largely responsible for the observed species selectivity paved the way to production of transgenic mice expressing human-like CGRP-R pharmacology, and such animals should help to advance our understanding of CGRP receptor function.46



CGRP RECEPTOR ANTAGONISTS FROM BOEHRINGER INGELHEIM On the basis of the substantial evidence implicating CGRP in the pathogenesis of migraine, Boehringer Ingelheim initiated an effort to discover small molecule CGRP receptor antagonists.44 A high throughput screening (HTS) campaign using a radioligand competition binding assay format identified dipeptide derivatives like 3 (Chart 1), which possessed micromolar receptor affinity (IC50 = 17 μM).41 Remarkably, Rudolf and colleagues were able to progress this micromolar HTS hit to a picomolar clinical candidate without modifying the core dipeptide. Instead, a strategy of structural rigidification and systematic modification of the terminal aryl groups led to nanomolar lead antagonists like 4 (Chart 1, IC50 = 44 nM).41 Further optimization of the benzoxazolone privileged structure in 4 led to the discovery of olcegepant (1, Chart 1), which possessed significantly enhanced potency and became the first CGRP RA to advance to clinical trials.47 The affinity of 1 for the human CGRP-R (Ki = 0.014 nM) was approximately 150-fold higher than the known peptide antagonist CGRP8−37, and it was found to be a competitive antagonist.44 This very high affinity was only observed for primate CGRP receptors, and 1 had approximately 200-fold lower affinity for the rat CGRP-R (Ki = 3.4 nM) and similarly reduced affinity for other non-primate species.44 Compound 1 was also highly selective for the CGRP receptor, with reported IC50 ≫ 1000 nM for a range of other enzymes and receptors, including the related calcitonin, amylin, and adrenomedullin receptors.44 Because of the pronouced species selectivity, in vivo antagonism of the CGRP receptor was evaluated in a marmoset model in which stimulation of the trigeminal ganglion results in a CGRP-induced increase in facial blood flow. In this model, 1 effectively blocked the effects of CGRP after iv administration of a dose of 30 μg/kg.44 Although CGRP is a potent vasodilator, no cardiovascular effects were observed after dosing D

dx.doi.org/10.1021/jm500364u | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Perspective

CGRP receptor antagonist (Ki = 0.3 nM) with moderate oral bioavailability in preclinical species.48 On the basis of its overall preclinical profile, compound 5 was taken forward for evaluation in the clinic.48

Chart 1. D-Tyrosine-Based CGRP Receptor Antagonists from Boehringer Ingelheim



AZEPANE-BASED CGRP RECEPTOR ANTAGONISTS FROM MERCK Researchers at Merck focused on the development of orally active CGRP receptor antagonists. High-throughput screening identified a benzodiazepinone-based lead (6, Chart 2), which had modest affinity for the CGRP receptor (Ki = 4.8 μM).49 Compound 6 was a suboptimal lead in several respects. The combination of micromolar potency and high molecular weight translated to a relatively poor ligand efficiency (LE = 0.18 kcal/ mol). Additionally, the high PSA of 6 (147 Å2) indicated that its passive permeability would likely be poor. Nonetheless, the structure was novel compared to previously described CGRP RAs based on a D-tyrosine motif, such as those reported by Boehringer Ingelheim, and the Merck medicinal chemistry team initiated work on this lead. One feature that compound 6 shared with other known CGRP RAs was the presence of an apparent privileged structure (vide supra)specifically, a secondary amide hydrogen bond donor−acceptor pair in a heterocycle, a spirohydantoin moiety in this case. This led to speculation that the spirohydantoin might be making similar interactions with the receptor to the dihydroquinazolinone ring of olcegepant (1), suggesting the possibility of hybridizing these structures.49 Indeed, it was discovered that the spirohydantoin privileged structure could be replaced by the piperidinyldihydroquinazolinone moiety from 1. Combining this modification with a trifluoroethyl substituent on the benzodiazepinone led to the nanomolar antagonist 7 (Chart 2).49 While 7 represented a significant improvement in receptor affinity compared with the HTS lead, the piperidinyldihydroquinazolinone privileged structure was found to have suboptimal stability. In analogy with the observations for 1 (vide supra), compound 7 was subject to benzylic oxidation under relatively mild conditions.50 An extensive survey of alternative groups identified the azabenzimidazolone represented in 8 (Ki = 23 nM), and this new privileged structure possessed suitable potency and stability.50 In general, these benzodiazepinone-based CGRP RAs possessed low oral bioavailability, which was attributed, at least in part, to poor physicochemical properties.51 In order to address these issues, the benzodiazepinone ring was reengineered to give a caprolactam core, and a systematic exploration of substitution and stereochemistry of this simple lactam was undertaken.51 These investigations established that a 6-aryl substituent with trans (3R,6S) stereochemistry was optimal, leading to compounds like 9, which combined good potency with oral bioavailability in both rat and dog.42 Optimization of the substituents on the caprolactam ring, in order to provide suitable potency, led to telcagepant (MK-0974, 2, Chart 2), in which the key 2,3-difluorophenyl substituent provided a significant increase in CGRP-R affinity.42 Compound 2 displayed high affinity for both human (Ki = 0.77 nM) and rhesus monkey (Ki = 1.2 nM) CGRP receptors but >1000-fold reduced affinity for the CGRP receptors from non-primate species.52 Compound 2 was highly selective (>10 000-fold) against a panel of over 160 receptors, transporters, and enzymes, as well as the related AM1 and AM2 receptors.52 In a cell-based functional assay of CGRP receptor antagonism,

1 up to 1 mg/kg (iv), in contrast to the known cardiovascular effects of the triptans.44 While compound 1 is a highly potent and selective CGRP-R antagonist, it is a relatively large and polar molecule with high MW (869) and calculated polar surface area (PSA = 181 Å2). Consistent with these molecular properties, 1 was found to exhibit low oral bioavailability (F < 1% in rat and dog), and it was advanced to the clinic using iv administration.41 Further work at Boehringer Ingelheim focused on reducing the molecular weight and polar surface area of 1 with the goal of identifying an orally bioavailable antagonist.48 It was found that the central urea could be replaced with a less polar carbamate group and the lysine residue could be removed, while still mantaining subnanomolar affinity for the CGRP-R. Additionally, the dihydroquinazolinone bicyclic system in the privileged structure of many potent antagonists, including olcegepant, was found to exhibit oxidative instability at its benzylic carbon.48 A simple solution was replacement of the unstable heterocycle with the stable, ring-expanded homologue. Combination of these modifications and optimization of the pyridylpiperazine end group led to the development compound BI 44370 (5, Chart 1), which was reported to be a potent E

dx.doi.org/10.1021/jm500364u | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Perspective

Chart 2. Azepane-Based CGRP Receptor Antagonists from Merck

be required to block the CIDV signal in vivo compared with that needed to block cAMP production in the cell-based assay. The preclinical pharmacokinetics of 2 have been studied in detail.56 After iv dosing, 2 exhibited low plasma clearance in rat but moderate clearance in dog and monkey, with a short plasma half-life in all three species (see Table 1).55,56 It should be

2 was found to be a potent antagonist (cAMP IC50 = 2.2 nM). In the presence of 50% human serum, the cell potency of 2 was shifted about 5-fold (cAMP + HS IC50 = 11 nM).52 This shift suggested a significant effect of plasma protein binding, consistent with the unbound fraction of 2 in human plasma (f u = 4.1%).52 Studies on the radiotracer [3H]2 revealed reversible and saturable binding to the native human CGRP-R with a KD of 1.9 nM.53 In kinetic binding studies, [3H]2 dissociated from the human CGRP-R with an off-rate of 0.51 min−1 and a half-life of 1.3 min.53 The high level of species selectivity observed with 2 and related compounds necessitated the development of a nonhuman primate pharmacodynamic model to assess their in vivo potency. In the rhesus monkey capsaicin-induced dermal vasodilation (CIDV) model, topical application of capsaicin results in release of endogenous CGRP, and this produces vasodilation and a local increase in blood flow that may be quantitated using laser Doppler imaging.45 The ability of a CGRP RA to block this CGRP-induced signal affords a direct measure of its in vivo potency. An attractive aspect of this pharmacodynamic model is that, because of its noninvasive nature, it could be readily translated to the clinical setting (vide infra).54 In this rhesus monkey pharmacodynamic assay, 2 was found to have EC50 and EC90 values of 127 and 994 nM, respectively.52 Since 2 has similar plasma protein binding and CGRP receptor affinity in rhesus monkey and human,52,55 the fact that the in vivo EC50 value (127 nM) is about 10-fold higher than the in vitro serum-shifted IC50 value (11 nM) suggests that a higher level of CGRP receptor occupancy may

Table 1. Preclinical Pharmacokinetic Properties of Telcagepant species rat dog monkey

dose (mpk) 15 (po); 2 (iv) 1 (po); 0.5 (iv) 5 (po); 10 (iv)

F (%) a

22 35b 6a

Cl c (mL min−1 kg−1)

iv t1/2 c (h)

9.4 17 14

1.6 0.67 3.2

a

Determined after dosing in 1:1 Imwitor-742:/Tween-80 vehicle. Determined after dosing in 1% Methocel vehicle. cDetermined after dosing in DMSO vehicle. b

noted that a short half-life was not considered to be a liability for acute treatment of migraine. Compound 2 had good oral bioavailability in rat (F = 22%) and dog (F = 35%), but oral bioavailability in rhesus monkey was low (F = 6% at a 5 mpk dose) and not dose-proportional, with a 6-fold increase in oral dose producing an 87-fold increase in AUC.56 Results from a number of experiments indicated that intestinal first-pass metabolism was a major contributor to the low oral bioavailability and nonlinear oral pharmacokinetics in monkeys.56 Overall, 2 possessed an attractive combination of oral bioavailability, potency, and selectivity and was advanced to F

dx.doi.org/10.1021/jm500364u | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Perspective

Chart 3. Spiroindane-Based CGRP Receptor Antagonists from Mercka

a

Position of [11C]methyl group in [11C]16 is indicated by an asterisk.

clinical trials as the first orally acting CGRP receptor antagonist to be evaluated in humans.42 The Merck team subsequently focused on identifying another orally acting CGRP receptor antagonist with a lower projected clinical dose than telcagepant.57 In order to achieve this, they targeted improvements in potency and pharmacokinetics and sought to realize a superior pharmacokinetic profile by increasing solubility and blocking metabolism.57 Replacement of the caprolactam with an imidazoazepane introduced a weakly basic heterocycle that could improve solubility, especially at acidic pH. It was also known that the primary site for metabolism of 2 was the azabenzimidazolone privileged structure, which was subject to both oxidation and complete Ndealkylation.58 To address this concern, novel spiropiperidinebased privileged structures were developed, including a spiroazabenzoxazinone.58 Combining this novel privileged structure with the imidazoazepane modification, followed by optimization of the imidazole C5 substituent, led to the discovery of MK-2918 (10, Chart 2).57 Studies on 10 revealed that it was more potent than 2 both in vitro (cAMP + HS IC50 = 0.63 nM) and in vivo (rhesus monkey CIDV EC90 = 300 nM).57 Although the oral bioavailability of 10 was relatively low in preclinical species (F = 5−16%), significant levels of an active metabolite (the alcohol derived from demethylation of the ether) were observed following oral dosing. This active metabolite (rhesus monkey CIDV EC 90 = 330 nM) had estimated oral bioavailability of 14−29% in preclinical species following dosing of 10 and was expected to contribute significantly to

clinical efficacy, leading to a lower anticipated clinical dose than 2.57 On the basis of the projected dose and its overall profile, 10 was selected as a preclinical candidate, but its current development status is unclear.



SPIROINDANE-BASED CGRP RECEPTOR ANTAGONISTS FROM MERCK

Concurrent with the work that led from the HTS lead 6 to compounds 2 and 10, a complementary approach was pursued at Merck in which the spirohydantoin portion of 6 was retained while the benzodiazepinone moiety was replaced. The goal of this effort was to identify novel CGRP-R antagonists with significant structural diversity with respect to telcagepant and a low projected clinical dose. Starting with several core spirohydantoin-based structures, rapid analogue methodology was used to screen for alternatives to the benzodiazepinone in 6.59 Initial submicromolar hits from this effort were modified to give 11 (Chart 3), a nanomolar CGRP receptor antagonist with good oral bioavailability in preclinical species.59 Further optimization of this spirohydantoin-based series followed a circuitous path to compound 12 (Chart 3), which resulted from intentional simplification of earlier constrained structures to facilitate rapid exploration of SAR.60 Interestingly, the simplified acyclic structure 12 (CGRP K i = 1.9 nM) exhibited a change in the preferred spirohydantoin chirality from (R) in progenitor molecules like 11 to (S) in 12. G

dx.doi.org/10.1021/jm500364u | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Perspective

Table 2. Data for Spiroindane-Based CGRP Receptor Antagonists compd

CGRP Ki (nM)a

cAMP IC50 (nM)b

AM2 Ki (nM)c

AM2/CGRPd

11 12 13 14 15 16

21 ± 8 (7) 1.9 ± 0.5 (13) 0.039 ± 0.012 (13) 0.021 ± 0.006 (15) 0.047 ± 0.020 (13) 0.046 ± 0.015 (7)

78 ± 23 (3) 6.8 ± 0.3 (4) 0.16 ± 0.06 (5) 0.12 ± 0.05 (6) 0.16 ± 0.06 (5) 0.097 ± 0.03 (8)

>100 000 (1) 710 ± 100 (4) 160 ± 46 (4) 160 ± 17 (7) 590 ± 200 (15) 210 ± 38 (3)

>5000 370 4100 7400 13000 4600

a

The Ki value for inhibition of [125I]hCGRP binding was determined using membranes from HEK293 cells stably expressing human CLR/RAMP1. Inhibition of CGRP-induced cAMP production in HEK293 cells stably expressing human CLR/RAMP1. cThe Ki value for inhibition of [125I]hAM binding was determined using membranes from HEK293 cells stably expressing human CLR/RAMP3. dRatio of AM2 Ki to CGRP Ki.

b

Table 3. Preclinical ADME Properties of Compounds 14 and 15 compd

species

dose (mpk)

F a (%)

Cl b (mL min−1 kg−1)

Vdss b (L/kg)

iv t1/2 b (h)

f u (%)

14

rat dog monkey rat dog monkey

10 (po); 2 (iv) 2 (po); 0.5 (iv) 2 (po); 0.5 (iv) 5 (po); 2 (iv) 2 (po); 0.5 (iv) 2 (po); 0.5 (iv)

74 67 9 28 64 7

11 8.0 15 31 7.2 22

0.3 0.6 1.7 0.81 0.67 2.4

0.6 1.0 1.5 0.6 1.3 2.3

0.8 6.1 9.6 23 17 37

15

a

Determined after dosing in 0.5% or 1% Methocel vehicle. bDetermined after dosing in DMSO vehicle.

CGRP-R slowly (t1/2 = 59 min) compared with telcagepant (t1/2 = 1.3 min), consistent with the high affinity of 14 for the CGRP receptor.63 The high level of in vitro potency translated into the rhesus monkey capsaicin-induced dermal vasodilation (CIDV) model, in which the EC90 value was 7 nM.63 Thus, 14 was determined to be approximately 50-fold more potent than telcagepant as an antagonist of the CGRP receptor in vitro and about 100-fold more potent in vivo. In analogy with other small molecule CGRP receptor antagonists, 14 had high affinity for rhesus monkey CGRP receptors (Ki = 0.024 nM) but significantly lower affinity for non-primate CGRP receptors, such as the rat CGRP receptor (Ki = 10 nM).63 Compound 14 maintained high selectivity for the CGRP receptor vs the related AM1 (>600 000-fold), AM2 (6500-fold) and AMY3 (>5000-fold) receptors but displayed only modest selectivity against the AMY1 receptor (30-fold).63 Additional testing revealed that 14 was >50 000-fold selective against a diverse panel of 169 enzymes, receptors, channels, and transporters.63 In addition to being a highly potent CGRP receptor antagonist in vitro and in vivo, compound 14 possessed an attractive preclinical pharmacokinetic profile (Table 3).62 In rats and dogs, 14 exhibited low plasma clearance and good oral bioavailability (67−74%). In rhesus monkeys, moderate clearance was observed and oral bioavailability was modest (9%) at a dose of 2 mpk. In analogy with telcagepant, the oral bioavailability of 14 in monkeys increased at higher dose (41% at 20 mpk), suggesting that saturable first-pass metabolism contributed to the low bioavailability in this species.62 The plasma half-life of 14 was found to be short in preclinical species (Table 3) but acceptable for acute treatment of migraine. On the basis of its potency, selectivity, pharmacokinetics, and overall profile, compound 14 was selected as a development candidate with the potential for a low clinically efficacious dose.64 As previously noted, most small molecule CGRP receptor antagonists display pronounced species selectivity, with significantly reduced affinity for CGRP receptors in nonprimates such as rodents. This can confound attempts to

Compound 12 was an attractive lead structure with good ligand efficiency (LE = 0.35), but its potency was not consistent with a low projected clinical dose and it lacked optimal selectivity for the CGRP receptor against the related AM2 receptor (Table 2; AM2/CGRP selectivity = 370-fold).60 Structure−activity investigations revealed that fluoro substitution of the terminal phenyl ring enhanced CGRP receptor affinity and that the acyclic N-benzylpivalamide moiety could be constrained to give a substituted piperidinone.60 Furthermore, earlier work had shown that replacement of the spirohydantoin privileged structure in such antagonists with the analogous spiroazaoxindole could provide a significant increase in potency.61 Combining all these observations led to the highly potent CGRP-R antagonist 13 (Ki = 0.039 nM; Chart 3) which also possessed enhanced selectivity against the AM2 receptor (Table 2; AM2/CGRP = 4100-fold).60 Compound 13 possessed subnanomolar activity in a cellbased functional assay in the presence of 50% human serum (cAMP + HS IC50 = 0.35 nM), but its oral bioavailability in preclinical species was suboptimal, especially in rhesus monkeys.62 On the basis of the hypothesis that poor aqueous solubility and metabolism of the δ-valerolactam both played a role in limiting oral bioavailability, polar functionality was incorporated into the lactam ring. These studies led to piperazinone-based antagonists, which demonstrated excellent potency and selectivity in addition to improved oral bioavailability in monkeys.62 Further optimization efforts demonstrated that incorporation of a spirocyclopentyl-substituted piperazinone conferred the optimal balance of potency, selectivity, and pharmacokinetics in this series, affording the clinical compound MK-3207 (14, Chart 3).62 Compound 14 had excellent CGRP-R affinity (Ki = 0.021 nM) and cell-based potency (cAMP IC50 = 0.12 nM). In the presence of 50% human serum, the cell potency of 14 was shifted slightly (cAMP + HS IC50 = 0.17 nM).62 The unbound fraction of 14 in plasma was found to be similar in human (f u = 9.4%) and rhesus monkey (f u = 9.6%) but significantly lower in rat ( f u = 0.8%, Table 3).62 Studies on the binding kinetics of [3H]14 revealed that the compound dissociated from the H

dx.doi.org/10.1021/jm500364u | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Perspective

analogous compounds based upon key pharmacophoric elements: a central aromatic amino acid with an N-terminal privileged structure and a C-terminal basic moiety.70 This approach led them to compounds like the benzothiophene 17 (Chart 4), which possessed good affinity for the CGRP-R (Ki = 0.55 nM) but was unfortunately a potent inibitor of cytochrome P450 3A4 (CYP3A4 IC50 = 0.084 μM).71

evaluate CGRP receptor antagonists in rodent models, which could provide valuable insight into the complex in vivo activity of CGRP and the effects of its blockade. Therefore, the identification of a suitable rat tool compound that could block the rat CGRP receptor at achievable plasma concentrations following oral dosing was important. For this purpose, compound 14 was an attractive starting point in terms of rat CGRP receptor affinity: although it exhibited significant species selectivity, it was a nanomolar antagonist (Ki = 10 nM) at the rat receptor and it had excellent rat oral bioavailability (vide supra).62,63 Unfortunately, the in vitro potency of 14 did not translate into good in vivo potency in the rat pharmacodynamic CIDV model of CGRP receptor antagonism (EC50 = 58 μM).65 The relatively poor in vivo activity in rats appeared to be due, at least in part, to extensive plasma protein binding (rat f u = 0.8%; Table 3). Following the observation that the plasma protein binding in rats was sensitive to the C-6 piperazinone stereocenter in compounds like 14,62 it was found that addition of a methyl substituent at this position afforded a significant increase in unbound fraction in rat plasma.65 Combining this modification with incorporation of a nitrogen atom in the indanyl moiety led to MK-8825 (15, Chart 3), which possessed nanomolar affinity for the rat CGRP receptor (Ki = 17 nM) and about a 30-fold increase in unbound fraction in rat plasma ( f u = 23%; Table 3) compared with 14.65 Compound 15 also exhibited improved aqueous solubility compared with 14 and was orally bioavailable in rats (Table 3).65 Importantly, 15 was found to have improved potency compared with 14 in the rat CIDV model (EC50 = 7.4 μM) and it has been used to study the effects of CGRP-R blockade in a number of rodent studies.65−67 In addition to attenuating rat plasma protein binding, the methyl substituent on C-6 of the piperazinone also apparently reduced the susceptibility for P-gp transport, perhaps by providing increased steric bulk close to the piperazinone amide functionality.68 Incorporation of another methyl group at N-4 of the piperazinone ring led to compound 16 (Chart 3), which exhibited significantly reduced P-gp efflux: for 14, P-gp transport ratio = 25; for 16, P-gp transport ratio = 1.7.68 The reduced level of P-gp transport suggested that compound 16 should be capable of achieving good exposure in the central compartment. Consistent with this expectation, in cisterna magna-ported rhesus monkeys it was found that the ratio of CSF concentration to unbound concentration in plasma was increased for 16 compared with 14.68 Overall, the addition of two methyl substituents to the piperazinone ring of 14 appeared to confer enhanced CNS penetrance. Moreover, the N-4 methyl substituent in 16 offered a convenient method for incorporation of a 11C-label to produce the first positron emission tomography (PET) tracer for the CGRP receptor, MK-4232 (16).68 This novel PET tracer has been used to interrogate the mechanism of action of CGRP receptor antagonists for the treatment of migraine (vide infra).69

Chart 4. Benzothiophene and Indazole-Based CGRP Receptor Antagonists from BMS

The team undertook an extensive evaluation of alternatives to the central benzothiophenyl moiety and identified indazol-5yl as an attractive replacement, affording compound 18: a subnanomolar CGRP receptor antagonist (Ki = 0.23 nM) with attenuated CYP inhibition (CYP3A4 IC50 = 4.0 μM).72 Substitution of the indazole ring with a 7-methyl substituent was found to confer an approximately 30-fold increase in CGRP-R affinity.73 Presumably, this 7-methylindazole moiety is making key interactions with the CGRP receptor that are analogous to those made by the dibromophenol of olcegepant. Importantly, this potency-enhancing modification significantly increased the window between CGRP blockade and CYP inhibition; however, the compounds did not possess the kind of aqueous solubility needed for intranasal dosing.71 A key breakthrough was the discovery that addition of an 8-fluoro substituent to the dihydroquinazolinone ring led to a dramatic increase in aqueous solubility. For example, the presence of this



INDAZOLE-BASED CGRP RECEPTOR ANTAGONISTS FROM BMS At Bristol-Myers Squibb (BMS), a program was initiated with the goal of identifying CGRP receptor antagonists suitable for the acute treatment of migraine, with rapid onset of action as a key feature.70 It was thought that this could be achieved via alternative routes of administration, such as intranasal dosing. After examining published reports of small molecule CGRP-R antagonists, such as olcegepant, the BMS group prepared I

dx.doi.org/10.1021/jm500364u | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Perspective

Chart 5. Pyridine-Based CGRP Receptor Antagonists from BMS

fluorine in compound 19 (Chart 4) conferred at least a 30-fold increase in solubility compared with the des-fluoro analogue.71 While the precise origin of this enhanced solubility is uncertain, the authors speculated that the fluorine atom may lead to enhanced H-bonding and solvation by polarizing the adjacent N−H functionality.71 For 19, the aqueous solubility was greater than 500 mg/mL between pH 1 and pH 6.8, and this compound (BMS-694153) became a development candidate.71 Compound 19 possessed exquisite affinity for the CGRP receptor (Ki = 0.013 nM) and was also a highly potent functional antagonist, inhibiting CGRP-stimulated cAMP production in SK-N-MC cells with EC50 = 0.035 nM.71 The compound was >10 000-fold selective against closely related receptors of the calcitonin receptor family (CTR, AM1, AM2, AMY1), and it exhibited pronounced species selectivity, in analogy with other small molecule CGRP RAs.71 A noninvasive marmoset model, based upon laser Doppler quantitation of changes in facial blood flow induced by iv-administered CGRP, was developed to facilitate in vivo evaluation of such small molecule antagonists. In this marmoset model, 19 effectively blocked the CGRP-induced increases in facial blood flow at plasma concentrations above 10 nM.71 Compound 19 was not orally bioavailable in rat or cynomolgus monkey (F < 1%), but it exhibited good intranasal bioavailability in rabbit (F = 59% at 1 mpk) with a short Tmax (