Discovery of a Highly Selective, Brain-Penetrant Aminopyrazole LRRK2 Inhibitor

Abstract

The modulation of LRRK2 kinase activity by a selective small molecule inhibitor has been proposed as a potentially viable treatment for Parkinson's disease. By using aminopyrazoles as aniline bioisosteres, we discovered a novel series of LRRK2 inhibitors. Herein, we describe our optimization effort that resulted in the identification of a highly potent, brain-penetrant aminopyrazole LRRK2 inhibitor (18) that addressed the liabilities (e.g., poor solubility and metabolic soft spots) of our previously disclosed anilino-aminopyrimidine inhibitors. In in vivo rodent PKPD studies, 18 demonstrated good brain exposure and engendered significant reduction in brain pLRRK2 levels post-ip administration. The strategies of bioisosteric substitution of aminopyrazoles for anilines and attenuation of CYP1A2 inhibition described herein have potential applications to other drug discovery programs.

Parkinson's disease (PD) is a neurodegenerative disorder that affects approximately 1% of the world's population over the age of 65.¹ Identification of a disease-modifying or neuroprotective therapeutic for PD patients remains a significant challenge. Recently, genetic research has revealed a connection between a missense mutation (G2019S) in the leucine-rich repeat kinase 2 (LRRK2) gene and a number of familial and idiopathic PD cases.²⁻⁹ Significantly, multiple reports have shown that this mutation enhances the kinase activity of LRRK2.¹⁰⁻¹⁷ The modulation of LRRK2 kinase activity by a selective small molecule inhibitor has therefore been proposed as a potential treatment for PD.¹⁸⁻²³

We recently reported the first disclosure of a series of highly potent, selective, and brain-penetrable anilino-pyrimidine LRRK2 inhibitors as exemplified by compound 1.^22,23 Examination of the liabilities of 1 and structurally similar analogues revealed (a) moderate to poor aqueous solubility (thermodynamic solubility of 1 at pH 7.4 < 0.9 μg/mL), (b) potential for ortho-quinoneimine reactive metabolite formation, and (c) a morpholinocarboxamide motif as a major site of metabolism as indicated by metabolite identification studies (MetID). In an effort to improve upon 1 and mitigate potential safety risks, we initiated a campaign to identify a suitable bioisosteric replacement for the aniline functionality. This approach resulted in the discovery of a novel series of highly selective aminopyrazole LRRK2 inhibitors.

Docking experiments using a JAK-2-derived homology model of LRRK2 suggested that compound 1 binds in the ATP binding site of LRRK2 as shown in Figure 1.^22,23 In designing anilino-carboxamide replacements, we chose to retain the hinge binding diaminopyrimidine core as well as the C-5 trifluoromethyl group that interacts favorably with the methionine gatekeeper through van der Waal contacts. Additionally, our lead optimization efforts toward 1 established the importance of occupying the vector adjacent to the hinge-binding anilino N-H motif (methoxy group in 1) for overall kinase selectivity.^22,23 The incorporation of this “selectivity handle” was one of the major considerations in the design of possible aniline replacements. Lastly, the morpholine carboxamide moiety occupies a solvent-exposed region of the active site. This provided a flexible substitution vector to modulate physicochemical and ADME properties. Because of their favorable physicochemical properties and synthetic tractability, aminopyrazoles were explored as potential bioisosteres for the aniline motif.^24,25

Figure 1.

Docking model of 1 (gray) and 6 (pink) in LRRK2 binding site. Key hydrogen bonds are shown as yellow dashed lines. The side chain of Y931 of JAK2 is shown and colored in yellow.

The results of an initial screen of aminopyrazoles as aniline replacements are shown in Table 1. The inhibitors were also assayed for inhibition of Janus kinase 2 (JAK2), as previous efforts had confirmed that JAK2/LRRK2 biochemical selectivity correlated with biochemical selectivity against the broader kinome.^22,23 The 3- and 5-amino-linked pyrazoles (2 and 3) demonstrated good JAK2/LRRK2 selectivity but appeared to have inferior potency for LRRK2 in our biochemical screen and cell-based LRRK2 autophosphorlyation assay when compared to the 4-linked aminopyrazoles (5–7). The poor JAK2 selectivity of compound 4 (2.7×) confirmed that a nitrogen lone pair (e.g., compound 2) or 3-/5-substituent was required to achieve good selectivity. An overlap of docked structures 6 (pink) with 1 (gray) in our LRRK2 homology model is shown in Figure 1. As expected, the aminopyrazole coincided with the aniline of 1, with the C5-methyl group (pyrazole numbering) directed toward the selectivity pocket and the N1-methyl toward solvent.

Table 1. SAR of Pyrazole Isomers.

^aBiochemical assay.

^bCellular assay.²⁷

^cAll IC₅₀ values and apparent K_i values represent arithmetic means of at least two determinations. These assays generally produced results within 2-fold of the reported mean.

Compounds 5–7, along with their ethyl and isopropyl analogues (8–11, Table 2), were profiled further in in vitro ADME assays. In general, the N-alkyl pyrazoles demonstrated moderate stability in human and rat liver microsomes (LM). Permeability assays using a MDR1-transfected MDCK cell line suggested that the pyrazole inhibitors were generally not human P-glycoprotein (Pgp) substrates (efflux ratio < 3). Disappointingly, however, compounds 5 and 7–11 were found to be reversible and/or time-dependent inhibitors of cytochrome P450 (CYP) 1A2.²⁶ This phenomenon proved to be a general trend for our first-generation aminopyrazole inhibitors.

Table 2. SAR of Aminopyrazole Analogues.

^aBiochemical assay. All apparent K_i values represent arithmetic means of at least two determinations. These assays generally produced results within 2-fold of the reported mean.

^bLM predicted hepatic clearance.

^cER measured in MDCK-MDR1 permeability assay.

^dEvidence of time-dependent inhibition includes either an IC₅₀ of less than 10 μM or an increase in AUC of >30% in a IC₅₀-shift assay.

A typical strategy to attenuate inhibition of the CYP family of enzymes includes modulating the affinity of the molecules toward the CYP enzymes by reducing lipophilicity.²⁸ Introduction of polar functionalities, such as methoxy and nitrile groups (compounds 12–15), however, did not resolve the issue of CYP1A2 inhibition.

Established SAR of known substrates and inhibitors of CYP1A2 suggests that the CYP1A2 binding site is extremely compact.^29,30 We therefore opted to take advantage of the uniquely shaped active site of CYP1A2 in our optimization effort. It was hypothesized that reduced CYP1A2 inhibition could be achieved through the introduction of sterically demanding groups that would engender steric clashes with the planar and narrow CYP1A2 active site.

With the high degree of structural tolerability, bulky three-dimensional substitutions at the solvent-exposed pyrazole N-capping group were explored further. Toward this end, a gem-dimethyl group was introduced onto the cyanoethyl motif of compound 15, one of the least lipophilic analogues [cLogP = 1.9; lipophilic ligand efficiency (LLE) = 6.3] from our initial screen. We were pleased to find that dimethylated analogue 16 maintained its affinity for LRRK2, and the metabolic stability was not compromised despite the increase in lipophilicity (cLogP = 2.6; LLE = 5.9). More importantly, 16 demonstrated no reversible or time-dependent inhibition (TDI) of CYP1A2. Replacement of the nitrile with a hydroxyl group (17) further improved metabolic stability without evidence of CYP1A2 inhibition.

When dosed intravenously at 0.5 mg/kg in rat, 17 demonstrated moderate clearance and a modest unbound brain to unbound plasma area under the curve (AUC) ratio (Table 3). In vitro LM MetID studies of related compounds have shown oxidation of the C5-methyl of the pyrazole motif as one of the major routes of metabolism. To maintain favorable physicochemical properties, it was envisioned that the conversion of the methylpyrazole to chloropyrazole should obviate the metabolic liability.

Table 3. In Vivo Pharmacokinetic Parameters.

	rat PKa			dog PKb
compd	Clp (mL/min/kg)	Vd (L/kg)	Bu/Puc	Clp (mL/min/kg)	F (%)
17	35	0.88	0.26
18	21	2.5	0.37	22	31

^aCompounds were dosed in male Sprague–Dawley rats at 0.5 mg/kg iv as a 60% NMP solution (n = 3).

^bCompound 18 was dosed in male Beagle dogs at 0.5 mg/kg iv as a 30% NMP solution and at 1 mg/kg po as a 1% RC591 + 0.2% tween80 aqueous suspension (n = 3).

^cUnbound brain/unbound plasma AUC ratio determined after 0.5 mg/kg iv dose.

Gratifyingly, chloropyrazole 18 showed improved in vivo brain exposure and stability in comparison to methylpyrazole 17. Further in vivo PK studies in dogs with 18 demonstrated acceptable clearance and oral bioavailability. It is worthy of note that, despite the potential for the displacement of the chloro substituent by nucleophiles, incubation of 18 in human LMs in the presence of glutathione, KCN, or methoxyamine did not result in any observable conjugates.

Compound 18 exhibited good cellular potency (cell IC₅₀ = 28 nM). When screened at 1 μM (116-fold over LRRK2 K_i) against a 185-membered Invitrogen kinase panel, compound 18 demonstrated inhibitory activity of >75% against only one other kinase (TSSK1, 75.2%). Additionally, the thermodynamic solubility at pH 7.4 was measured to be 20 μg/mL.

The pharmacodynamic effect of compound 18 was tested in bacterial artificial chromosome (BAC) transgenic mice expressing G2019S mutant LRRK2 gene.^31,32 As shown in Figure 2, robust concentration-dependent knockdown of phospho-LRRK2 (pLRRK2) levels in the brain was observed after ip dosing at 10 or 15 mg/kg. Inhibition modeling indicates the in vivo unbound brain IC₅₀ of 18 to be 12 nM.³²

Figure 2.

In vivo G2019S LRRK2 transgenic mouse PK/PD results measuring brain pSer1292 autophosphorylation. The circles represent the observed data for mice treated with 18, and the line represents the predicted data from a direct inhibition model. The percent inhibition is normalized to pSer1292 levels observed in mice dosed with vehicle alone (n = 3).³²

The pyrazole-based LRRK2 analogues were synthesized according to Schemes 1 and 2. In Scheme 1, alkylation of 5-methyl-3-nitropyrazole (19) followed by hydrogenation afforded the corresponding aminopyrazoles (20). Acid- or palladium-catalyzed coupling of the aminopyrazole to 2-chloro-4-methylamino-5-trifluoromethylpyrimidine (21) provided the target compounds. Key pyrazole building blocks 25 and 27 were prepared as described in Scheme 2 using standard chemistry.

Scheme 1.

Scheme 2.

The incorporation of 2-alkoxyaniline motifs into the design of small molecule kinase inhibitors has become a common strategy to achieve broad kinome selectivity.^33−33d In this paper, we described the use of aminopyrazoles as effective bioisosteres for the aniline motif. Through this strategy, we discovered the novel, selective, and brain-penetrant LRRK2 inhibitor 18. In addition to eliminating the aniline toxicophore, inhibitor 18 exhibited a marked improvement in solubility as compared to 1 while maintaining excellent pharmacological and drug metabolism and pharmacokinetic profiles. Inhibitor 18 will be an effective tool to delineate the role of LRRK2 kinase activity in the etiology of PD. Moreover, during the optimization process, CYP1A2 inhibition was attenuated by incorporation of sterically demanding groups that engendered unfavorable interactions with the compact binding site of the enzyme. The general strategies of bioisosteric replacement of anilines with aminopyrazoles and attenuation of CYP1A2 inhibition described herein have potential applications to other drug discovery programs.

Glossary

Abbreviations

AUC

area under the curve

B

brain concentration

BAC

bacterial artificial chromosome

B_u

unbound brain concentration

CYP

cytochrome P450

dba

dibenzylideneacetone

DMPK

drug metabolism and pharmacokinetics

ER

efflux ratio

GSH

glutathione

H

human

ip

intraperotineal

iv

intravenous

JAK2

Janus kinase 2

LHMDS

lithium hexamethyldisilazide

LLE

lipophilic ligand efficiency

LM

liver microsome

LRRK2

leucine-rich repeat kinase 2

MDCK-MDR1

Madin–Darby canine kidney cells-multidrug resistance protein 1

MOA

methoxyamine

MW

molecular weight

P

plasma concentration

Pgp

P-glycoprotein

PK/PD

pharmacokinetics/pharmacodynamics

pLRRK2

phospho-LRRK2

po

oral administration

P_u

unbound plasma concentration

R

rat

TDI

time-dependent inhibition

TPSA

topological polar surface area

TFA

trifluoroacetic acid

V_d

volume of distribution

XPhos

2-dicyclohexylphosphino-2′,4′,6′-triisopropylbiphenyl

Supporting Information Available

Synthetic procedures, spectral data, and kinase selectivity data for key compounds and biological and ADME assay protocols. This material is available free of charge via the Internet at http://pubs.acs.org.

Author Present Address

Z.K.S.: Novartis Institutes for Biomedical Research, 4560 Horton Street, Emeryville, California 94608, United States.

The authors declare no competing financial interest.