Fatty acid amide hydrolase (FAAH)is an integral, membrane-bound enzyme responsible for the breakdownof fatty acid ethanolamide (FAE) signaling molecules, such as theendocanabinoid arachidonyl ethanolamide (anandamide, AEA),N-palmitoyl ethanolamide (PEA), andN-oleoylethanolamide (OEA). FAAH is a member of the serine hydrolase amidasesignature family, which utilizes an unusual serine–serine–lysinecatalytic triad.^1,2 Inhibition of FAAH leads to elevatedlevels of these endogenous FAEs, which act on cannabinoid receptorsimplicated in the suppression of pain transmission.³ Levels of these FAEs were shown to be significantly elevatedin FAAH knockout (KO) mice as compared to wild-type controls.⁴ Both genetic knockout of FAAH and pharmacologicalmodulation of FAAH activity demonstrated reduced sensitivity to pain.⁵ Thus, FAAH inhibitors are expected to providetherapeutic benefits in the management of inflammatory and neuropathicpain.⁶⁻⁹

Several classes of covalent and noncovalent FAAH inhibitorshavebeen reported to date (Figure1). Several covalentFAAH inhibitors that irreversibly inhibit FAAH by carbamylation ofSer241 have been reported and are exemplified by URB-597,¹⁰⁻¹² PF-3845, and PF-750.⁸ A second subclassof FAAH inhibitors are the keto-oxazole class of FAAH inhibitors asexemplified by OL-135,¹³ which reversiblyforms an enzyme-stabilized hemiketal through a particularly reactiveelectrophilic carbonyl. More recently, however, several scaffoldshave been disclosed as reversible noncovalent modifying inhibitorsof FAAH. Aminopyrimidine5(14) and sulfonamide6(15) arechief among these novel classes of FAAH inhibitors. This approach,in our view, would decrease potential safety concerns over the creationof a long-lived covalent adduct between a compound and the FAAH enzyme.

In this letter, we report the identification ofa fully reversible,noncovalently modifying FAAH inhibitor derived from a high throughputscreening campaign culminating in pyrazole7 (Figure2). Pyrazole7 suffers from moderatepotency as well as poor preclinical PK.¹⁶ However, when dosed iv, the compound did show moderate efficacy(at 30 mg/kg) in the complete Freund’s adjuvant (CFA) inflammatorypain model, which generated considerable interest within the team.¹⁷ In addition, because of a lack of the presenceof any obvious covalent modifying site, we directed our studies towardoptimization of this pyrazole lead series to identify a selective,potent analogue with desirable pharmacokinetic properties in threepreclinical species (rat, dog, and rhesus). Coupled with excellentin vitro potency (<10 nM), we sought to identify a compound thatwas fully CNS penetrant, maintained enzyme inhibition over a 24 htime frame, displayed selectivity over related enzymes, and possesseddesirable efficacy in the CFA pain model at 10 mg/kg or less.

The synthesis of our key analogues was achieved through severaldifferent routes. To synthesize pyrazole analogues11 and12, 4-chlorothiophenol was condensed with a bromoacetophenoneof interest (Scheme1). The resulting adductwas then treated with DMF-DMA and hydrazine to afford the desiredpyrazoles19a–c. Arylation at theN-1 position was accomplished through a copper mediatedcoupling to yield analogues11 and12. Alternativelyfor analogue13, copper mediated C–N bond formationwas achieved on the corresponding 4-CN phenyl derivative18c followed by 1,2,4-oxadiazole formation under standard conditionsto afford13.¹⁸

For the oxazole derivatives16 and17, the corresponding oxazole triflatesas previously described byPanek and co-workers¹⁹ were used as thekey starting material for each analogue (Scheme2). One-pot boronate formation and subsequent Suzuki coupling wasaccomplished for compounds23a and23b.²⁰ Halogenation and cross coupling with the desiredaryl thiol gave rise to16 and17.²¹

Our initial structure–activityrelationship (SAR) effortsfocused primarily upon improvements in potency and pharmaceuticalproperties in the pyrazole-containing system7. Giventhe low molecular weight and moderate in vitro potency, we believedthat numerous modifications could be made to compound7 to optimize for drug-like properties. We initially sought to eliminatethe sulfur-linked aryl ring. Heteroatom modification of the S atomled to a deleterious effect on in vitro potency for FAAH inhibition(S = 119 nM and O = 1007 nM) for compound8 (Figure2). Oxidation of the S atom to its various higheroxidation states also led to a reduction in potency (S(O) = 2192 nM,SO₂ = >3000 nM) as seen in compounds9 and10. This data led us to abandon this strategy and to explorecapping of the free pyrazole with various groups. A simple Ph ringdirectly attached to the N-1 of the pyrazole led to a greater than4-fold improvement to the intrinsic FAAH potency as exemplified bycompound11.

We next turned our attention to scissionof the dioxane ring, whichwas deemed imperative in order to improve upon the PK properties ofthis series of compounds. Initial data pointed out clearly that thepara position has the greatest potential and flexibility toward structuraldiversification. Methyl sulfone and 1,2,4-oxadiazole replacementsfor the dioxane ring were potency-enhancing. The optimized N-subtituentis shown in compound12 as apara-Fphenyl or as a 3-pyridyl group as exemplified in compound13. Subsequent SAR studies showed importantly that a 4-fold improvementwas realized upon introduction of various heterocyclic tertiary alcoholmoieties at the 4-position such as that shown in compound14 while maintaining thepara-F phenyl ring, whichcapped the pyrazole. Compound7,11,12, and13 were further evaluated for off-targetactivities as well as pharmacokinetic (PK) studies (Table1). An off-target activity detected in a Pan Laboratoriesscreen for compounds12–14 was humanvesicular monoamine transporter (VMAT) inhibition. This is of particularconcern since this particular transporter is centrally expressed.²²

Additionally,the pyrazole lead series was also plagued by hERGactivity. Potent compounds were screened against the hERG K⁺ channel using a binding assay that measures the displacement of[³⁵S]MK-499, a well characterized hERG K⁺ channelblocker.²³ During this process, we hadnoticed that upon introduction of an oxazole central ring as a replacementfor the pyrazole, a noticeable improvement in MK-499 binding potencywas observed. The closely related analogues12 and15, which both maintained a phenyl methyl sulfone moiety asthe active pharmacophore, had MK-499 binding values of 637 nM and28 μM, respectively, implying that the oxazole core had quitean advantageous effect on MK-499 over the pyrazole (Table1). In addition, with the data collected for compounds16 and17, it was readily apparent that VMATinhibition had serendipitously improved upon modification of the centralcore to the oxazole. Lastly, compound17 took advantageof the fact that throughout our earlier SAR, we found that optimizationof the core thioether ring system to a 2-pyridyl substitutent notonly helped with PK, but also improved in vivo efficacy in our currentpain models. Several of these compounds were further studied in vivofor rat PK and efficacy based on these improvements in the in vitroprofiles.

Compounds12 and13 weredosed for ratPK studies at 1 mg/kg iv and 2 mg/kg po (Table2), and both compounds showed similar half-life, clearance, and volumeof distribution. A 3-fold lower bioavailability for compound13 to compound12 may be attributed to poor oralabsorption despite the improved solubility of compound13 as compared to compound12 (FaSSIF solubility for12 was measured to be 0.003 mg/mL versus 0.027 mg/mL for compound13). Furthermore, changes of the pharmacophore from thep-methanesulfonyl benzene ring to a pyrazine ring substitutedwith a 4-tertiary alcohol (14) provided a rather substantialboost in terminal half-life while maintaining the oral exposure observedin compound12. Lastly, compound17, whichcombined optimized substituents by replacing the 4-pyrazinyl tertiaryalcohol moiety with a 4-pyridyl tertiary alcohol, along with the thiopyridinesubstitutent drastically improved the rat PK to levels, which couldtranslate to QD dosing for in vivo efficacy models. To further characterizethese compounds, they were dosed in our in vivo efficacy pain models.

In order to ascertain in vivo efficacy,inflammatory pain was inducedby intraplantar injection of complete Freund’s adjuvant (CFA)into the hind paws of rats.¹⁷ Mechanicalallodynia was assessed at 1 and 3 h post a single oral dose. Thisstudy was performed on compounds13,14,16, and17 (Table3).A major objective was to try to achieve reasonable efficacy in theCFA at a moderate dose (10 mg/kg) as a criterion for a potential developmentcompound. Compound13 when dosed at 30 mg/kg showed efficacyat the 1 and 3 h time points (49 and 62% inhibition, respectively).However, the VMAT and MK-499 binding activity led us to focus on othermore structurally diverse analogues. The dose responses of compound14 at both 10 and 30 mg/kg indicated comparable efficacy tocompound13 at both doses, which suggested to us thatthe heterocyclic tertiary alcohol modification not only improved PKbut also provided comparable in vivo efficacy to compound13 but at a lower dose. More importantly, with both an improved off-targetprofile and PK profile, compound17 also showed reductionof inflammatory pain on par with the standard of care, naproxen, atboth 10 and 30 mg/kg in a model for inflammatory pain. Taking compound17 one step further, we decided to evaluate its efficacy inthe rat spinal nerve ligation model (SNL) for neuropathic pain.²⁴

Using the SNL model as a measureof efficacy for neuropathic pain,compound17 was evaluated in vivo at doses of 3, 30,and 100 mg/kg with excellent response observed at all three doses(Figure3). The efficacy of compound17 at a level of roughly 50% analgesia when compared to standardsof care Pregabalin, Gabapentin, and Duloxetine was shown to be quitefavorable. A secondary effect of dosing these standards of care intorodents is loss in cognition due to drowsiness and motor skill impairment.In order to ensure that a FAAH inhibitor would not cause the sametype of motor skill impairment seen with other analgesics, we alsoimplemented the rota-rod assay as a secondary assay along with theSNL as a standard counterscreen for these undesirable cognitive effects.Compound17 not only provided 50% reversal of allodyniaat a low dose (3 mg/kg) but also did not show any effect in the concomitantroto-rod assay at doses as high as 100 mg/kg (Figure4). This was in line with URB-597 (1), a previouslydescribed covalent modifying FAAH inhibitor that required a 21 mg/kgdose to achieve comparable efficacy to17 in our hands.

On the basis of the promising in vivo and in vitrodata generatedfor compound17, a full PK profile was obtained acrosspreclinical species.²⁵ The resulting PKparameters supported the development of a QD compound in human. Additionally,free fraction was measured across species (rat, 2.9%; dog, 1.2%; rhesus,0.8%; human, 1.6%) indicating a high level of plasma protein bindingfor compound17. An additional concern we had for thismechanism was tachyphylaxis. To alleviate this concern, we initiateda chronic paradigm in the CFA pain model versus a CB₁ andCB₂ agonist, WIN 55,212-2 known to produce CB1 desensitizationand loss of antinociception under chronic dosing.²⁶

The pan CB agonist WIN 55,212-2 loses its analgesicpropertiesin the CFA assay at roughly day 5. Compound17, however,continues to show excellent efficacy out to 10 days (Figure5). Plasma exposures were also obtained at both the3 and 24 h time points following dosing at 30 mg/kg. One interestingobservation is that compound17 reduced edema observedin the animals over several days, which may contribute to its anti-inflammatoryproperties.

In conclusion, we discovered compound17 (MK-4409)as a potent and selective reversible noncovalent modifying FAAH inhibitor.MK-4409 showed excellent efficacy in numerous preclinical models includingthe CFA and SNL pain models. In addition, no cognitive effects wereobserved for this brain penetrant FAAH inhibitor (measured rat brain/plasmaratio was 2). A chronic dosing paradigm in the CFA assay showed overallanti-inflammatory activity and no tachyphylaxis apparent after 10days of dosing. MK-4409 (17) was accepted as a developmentcandidate based on the promising preclinical profile. Additional humanclinical data will be reported in subsequent publications.