1521-0103/355/3/397–409$25.00 http://dx.doi.org/10.1124/jpet.115.227587THE JOURNAL OF PHARMACOLOGY AND EXPERIMENTAL THERAPEUTICS J Pharmacol Exp Ther 355:397–409, December 2015Copyright ª 2015 by The American Society for Pharmacology and Experimental Therapeutics

MLi-2, a Potent, Selective, and Centrally Active Compound forExploring the Therapeutic Potential and Safety of LRRK2Kinase Inhibition s

Matthew J. Fell,1 Christian Mirescu,1 Kallol Basu, Boonlert Cheewatrakoolpong,Duane E. DeMong, J. Michael Ellis, Lynn A. Hyde, Yinghui Lin, Carrie G. Markgraf,Hong Mei, Michael Miller, Frederique M. Poulet, Jack D. Scott, Michelle D. Smith,Zhizhang Yin, Xiaoping Zhou, Eric M. Parker, Matthew E. Kennedy, and John A. MorrowNeuroscience Discovery (M.J.F., C.M., M.E.K) and Discovery Chemistry (J.M.E), Merck Research Laboratories, Boston,Massachusetts; Discovery Chemistry (K.B., D.E.D., M.M., J.D.S.), Pharmacology (B.C., L.A.H., Y.L., E.M.P., M.D.S., Z.Y., X.Z.),Pathology and Cellular Toxicology (C.G.M., F.M.P), and Pharmacokinetics (H.M.), Merck Research Laboratories, Kenilworth, NewJersey; Neuroscience Discovery, Merck Research Laboratories, West Point, Pennsylvania (J.A.M.)

Received August 6, 2015; accepted September 24, 2015

ABSTRACTMutations in the leucine-rich repeat kinase 2 (LRRK2) gene are themost common genetic cause of familial and sporadic Parkinson’sdisease (PD). That the most prevalent mutation, G2019S, leads toincreased kinase activity has led to a concerted effort to identifyLRRK2 kinase inhibitors as a potential disease-modifying therapyfor PD. An internal medicinal chemistry effort identified severalpotent and highly selective compounds with favorable drug-likeproperties. Here, we characterize the pharmacological propertiesof cis-2,6-dimethyl-4-(6-(5-(1-methylcyclopropoxy)-1H-indazol-3-yl)pyrimidin-4-yl)morpholine (MLi-2), a structurally novel, highlypotent, and selective LRRK2 kinase inhibitor with central nervoussystem activity. MLi-2 exhibits exceptional potency in a purifiedLRRK2 kinase assay in vitro (IC50 5 0.76 nM), a cellular assaymonitoring dephosphorylation of LRRK2 pSer935 LRRK2 (IC50 5

1.4 nM), and a radioligand competition binding assay (IC505 3.4 nM).MLi-2 has greater than 295-fold selectivity for over 300 kinases inaddition to a diverse panel of receptors and ion channels. Acuteoral and subchronic dosing in MLi-2 mice resulted in dose-dependent central and peripheral target inhibition over a 24-hourperiod as measured by dephosphorylation of pSer935 LRRK2.Treatment of MitoPark mice with MLi-2 was well tolerated over a15-week period at brain and plasma exposures .100� the invivo plasma IC50 for LRRK2 kinase inhibition as measured bypSer935 dephosphorylation. Morphologic changes in the lung,consistent with enlarged type II pneumocytes, were observed inMLi-2-treated MitoPark mice. These data demonstrate thesuitability of MLi-2 as a compound to explore LRRK2 biologyin cellular and animal models.

IntroductionParkinson’s disease (PD) is a progressive neurodegenera-

tive disorder characterized by resting tremor, slowness ofmovement, muscular rigidity, and postural instability. Post-mortem PD brains reveal a distinctive loss of dopaminergicneurons in the substantia nigra along with insoluble a-synuclein-rich intracellular protein aggregates, termed Lewy bodies.Although motor symptoms of PD can be managed early in thedisease with dopaminergic replacement therapies (e.g., levo-dopa, MAO inhibitors, and dopamine agonists), there are notherapies that can slow or stop disease progression. Thefailure to identify disease-modifying therapies stems largely

from a lack of understanding of the pathologic mechanismsthat culminate in clinical PD. In recent years, genomicadvances have begun to shed light on the underlying patho-logic mechanisms of PD, and although the vast majority of PDremains sporadic in origin, approximately 5–10% of cases cannow be linked to a genetic origin. Mutations in the leucine-richrepeat kinase 2 (LRRK2) gene are the most common cause ofmonogenic PD and account for 4% of familial PD and 1% ofsporadic PD (Healy et al., 2008). In contrast to other Mende-lian forms of PD, LRRK2-PD appears clinically indistinguish-able from idiopathic PDwith regard to age of onset, presence ofLewy bodies, and responsivity to dopaminergic replacement(Healy et al., 2008).LRRK2 is a large multidomain soluble protein encompass-

ing two core enzymatic functions comprised of a GTPasedomain and a kinase domain. The enzymatic core is flankedby multiple protein-protein interacting domains, including

1M.J.F. and C.M. contributed equally to the writing of the manuscript.dx.doi.org/10.1124/jpet.115.227587.s This article has supplemental material available at jpet.aspetjournals.org.

ABBREVIATIONS: ANOVA, Analysis of variance; DOPAC, 3,4-dihydroxy-phenyl-acetic acid; GAPDH, glyceraldehyde-3-phosphate dehydroge-nase; HVA, homovanillic acid, 2-(4-hydroxy-3-methoxy-phenyl)acetic acid; KO, knockout; LRRK2, leucine-rich repeat kinase 2; LMA, locomotoractivity; MLi-2, Merck LRRK2 inhbitor-2, cis-2,6-dimethyl-4-(6-(5-(1-methylcyclopropoxy)-1H-indazol-3-yl)pyrimidin-4-yl)morpholine; PD,Parkinson’s disease.

397

http://jpet.aspetjournals.org/content/suppl/2015/09/25/jpet.115.227587.DC1Supplemental material to this article can be found at:

at ASPE

T Journals on M

ay 13, 2020jpet.aspetjournals.org

Dow

nloaded from

armadillo, ankyrin, and leucine-rich repeat (LRR) structuralmotifs at the N-terminal, and the WD-40 domain at theC-terminal (Cookson, 2010). At least seven LRRK2 pointmutations clearly segregate with familial PD (Paisan-Ruizet al., 2004; Zimprich et al., 2004; Kachergus et al., 2005;Bonifati, 2007). The most common variant, G2019S, occurswithin the LRRK2 kinase activation loop and results in in-creased kinase activity (West et al., 2005; Greggio et al., 2006).Preclinical observations of increased susceptibility to dopami-nergic neurotoxins, mitochondrial dysfunction, and elevateda-synuclein in human LRRK2 mutant carrier–derived inducedpluripotent stem cell (iPSC) models and G2019S knockin micefurther confirm a toxic gain-of-function for kinase-activatedmutations (Nguyen et al., 2011; Reinhardt et al., 2013; Sanderset al., 2014; Yue et al., 2015). Interestingly, GTPase domainvariants R1441G, Y1699C, and N1437H also increase LRRK2kinase activity (Sheng et al., 2012), suggesting the kinasedomain is commonly and consistently dysregulated by manypathogenic LRRK2 mutations. The strength of the associationof the LRRK2-G2019S mutation for causing PD, coupled withthe phenotypic finding that the LRRK2-G2019S mutationresults in increased kinase activity, collectively point to apharmacologically tractable LRRK2 biology. Thus, the develop-ment of LRRK2 selective kinase inhibitors has been a dominanttherapeutic focus for the treatment of PD.In the past five years, significant medicinal chemistry

advances have been made in the development of potent andselective inhibitors of LRRK2 that are beginning to enablemechanism-of-action studies related to PD pathogenesisand assessments of LRRK2 pleiotropy beyond neurodegen-eration. Initial studies reported a number of structurallydiverse inhibitors of LRRK2 (Dzamko et al., 2010; Lee et al.,2010; Ramsden et al., 2011; Zhang et al., 2012) that wereultimately found to possess activities on a number of off-target kinases, thus limiting their utility for definingLRRK2 specificity of results. Subsequently, a number ofpotent inhibitors of LRRK2 kinase activity, with improvedselectivity profiles, were disclosed, including LRRK2-IN-1(Deng et al., 2011), GSK2578215A (Reith et al., 2012), andHG-10-102-01 (Choi et al., 2012). The utility of thesemolecules is, however, largely confined to in vitro assaysowing to poor pharmacokinetic qualities and a lack of brainpenetration. More recently, two novel LRRK2 kinase inhib-itors (GNE-7915 and GNE-0877) were disclosed and used toprobe the safety and tolerability profile of LRRK2 kinaseinhibition in rodents and nonhuman primates (Estradaet al., 2012; Fuji et al., 2015). However, despite possessingimproved biodistribution properties, both compounds in-hibit several off-target kinases, and GNE-0877 has beenassociated with dose limiting toxicity in rats (Fuji et al.,2015).Despite these significant advances in developing LRRK2

kinase inhibitors over recent years, the field could greatlybenefit from the generation of additional structurally diverse,potent, and selective LRRK2 kinase inhibitors that are moresuitable for detailed exploration of LRRK2 biology in vivo aswell as in vitro. In the present study, we report the discoveryand pharmacology of cis-2,6-dimethyl-4-(6-(5-(1-methylcyclo-propoxy)-1H-indazol-3-yl)pyrimidin-4-yl)morpholine (MLi-2;Fig. 1A), a potent, selective, and centrally active inhibitor ofLRRK2 kinase activity that is suitable for exploring thetherapeutic potential and safety of LRRK2 kinase inhibition.

Materials and MethodsIn Vitro Experiments

LRRK2 Biochemical Assay. LRRK2 kinase activity was de-termined with LanthaScreen technology from Life Technologies(Grand Island, NY) using GST-tagged truncated human mutantG2019S LRRK2 (cat. no. PV4881) in the presence of the fluorescein-labeled peptide substrate LRRKtide (cat. no. PV4901). Assays wereperformed in the presence of 134 mMATP (KmATP). Upon completion,the assay was stopped and phosphorylated substrate detected with aterbium (Tb)-labeled anti-phospho-ERM antibody (cat. no. PR8975A).The compound dose response was prepared by diluting a 10 mM stockof compound to a top concentration of 98mM in 100% dimethylsulfoxidefollowed by serial dilution by 1:2.15 in dimethylsulfoxide nine times.Four hundred nanoliters of each dilution was spotted via a LabcyteEcho onto a 384-well black-sided plate followed by 10 ml of a 2.5 nMenzyme solution in 1� assay buffer (50 mMTris pH 8.5, 10 mMMgCl2,0.01% Brij-35, 1 mM EGTA, 2 mM dithiothreitol, 0.05 mM sodiumorthovanadate). Following a 30-minute incubation at room tempera-ture, the kinase reaction was started with the addition of 10 ml of800 nM fluorescein-labeled LRRKtide peptide substrate and 186 mMATP solution in 1� assay buffer. The reaction was allowed to progressat 25°C for 1 hour. The reaction was then stopped by the addition of20 ml of TR-FRET Dilution Buffer (Invitrogen cat. no. PV3756B)containing 4 nM Tb-labeled anti-phospho LRRKtide antibody and20 mMEDTA. After an incubation of 1 hour at room temperature, theplate was read on an EnVisionmultimode plate reader (Perkin Elmer,Waltham,MA) with an excitation wavelength of 340 nm and a readingemission at both 520 and 495 nm. Compound IC50s were interpolatedfrom nonlinear regression best fits of the log of the final compoundconcentration, plotted as a function of the 520/495-nm emission ratiousing GraphPad Prism.

Stable Cell Line and Compound Treatments. A humanneuroblastoma cell line, SH-SY5Y, was used for the LRRK2 stablecell line construction. SH-SY5Y cells were cultured in Dulbecco’smodified Eagle’s medium/F-12 supplemented with GlutaMax (LifeTechnologies), 10% tetracycline (Tet)-free fetal bovine serum (Clontech),nonessential amino acids (HyClone), pen-strep (100 mg/ml; HyClone) at37°C and 5% carbon dioxide. Parental cells were transfected withplasmid constructs that overexpress full-length human LRRK2 wild-type ormutant (G2019S) under the control of a Tet-inducible promoter.Transfected cells were selected and maintained with hygromycin(2 mg/ml) and zeocin (100mg/ml). Cells were seeded into six-well platesand induced with Tet (2 mg/ml) for 72 hours prior to treatment. After90 minutes of compound incubation, cells were mechanically lifted,pelleted, and lysed with lysis buffer (MSD Lysis Buffer, cat. no.R60TX-3) supplemented with protease (Roche cOmplete Mini, cat. no.11836170001), and phosphatase inhibitors (Halt Phosphatase In-hibitor, cat. no. 78420; Life Technologies). Pellets were further beadhomogenized (1.5 minutes, frequency 5 30, TissueLyser; Qiagen,Valencia, CA) and then spun at 13,200 rpm for 20 minutes at 4°C.Supernatants were then removed for subsequent Western blotanalyses.

Whole Cell Binding. SHSY-5Y cells heterologously stablyexpressing Tet-inducible human wild-type or G2019S LRRK2 wereinduced with tetracycline at 1 mg/ml for 48 hours and then plated ina suspension at 200K cells per well in 1� kinase assay buffer in a96-well plate. Test compounds (10 serial 3-fold dilutions; topconcentration 5 mM) were added to the cells in a total volume of250 ml. Following a 30-minute incubation at room temperature,250 ml of 0.7 nM [35S]MLi-A was added. Following a further30-minute incubation at room temperature, reactions were termi-nated by rapid filtration on GF/B filterplates presoaked in 0.3%polyethylenimine using a Brandel Cell Harvester (Gaithersburg,MD). Filters were washed five times in ice-cold wash buffer (5.0 mMTris, pH 7.5) then dried at 55°C for 45 minutes. Fifty microliters ofMicroscint 20 was added to each well, and plates were counted on aPackard TopCount. Ki values were determined from data points

398 Fell et al.

at ASPE

T Journals on M

ay 13, 2020jpet.aspetjournals.org

Dow

nloaded from

(the average of replicate determinations) using GraphPad Prism,by nonlinear regression–one site fit.

LRRK2 Immunoblotting. Protein content per sample was de-termined by a bicinchoninic acid colorimetric assay, using bovineserum albumin as a standard (cat. no. 23225; Life Technologies). One-hundredmicrograms of proteinwas reduced and denatured at 70°C for10 minutes, then resolved on 3–8% Tris-acetate gels (cat. no. EA0378;Life Technologies) and transferred to polyvinylidene fluoride mem-branes (cat. no. LC2005; Life Technologies). Membranes were thenblocked in 5% dry milk in Tris-buffered saline plus Tween-20 (cat. no.T9039; Sigma-Aldrich, St. Louis, MO) for 1 hour at 4°C and probedwith rabbit anti-LRRK2-pSer935 LRRK2 (1 mg/ml, cat. no. AB133450;Abcam, Cambridge, MA) overnight at 4°C. Membranes were thenincubated with donkey anti-rabbit conjugated to horseradish peroxi-dase (HRP) (1 mg/ml, cat. no. A16035; Life Technologies), combinedwith the IRDye 680CW (cat. no. 926-68076; LI-COR Biosciences,Lincoln, NE) for 30 minutes at 4°C. LRRK2-pSer935-HRP signalswere subsequently developed by luminol-enhanced chemiluminescence(SuperSignal Substrate, cat. no. 34075; ThermoFisher Scientific,Sunnyvale, CA) and then visualized and analyzed on a LI-COROdyssey system. For total LRRK2 detection, membranes were sub-sequently stripped (Restore PLUS, cat. no. 46430; Life Technologies),reblocked as above, and probed with rabbit anti-LRRK2 antibody(1:500 v/v, MJFF2 clone 41-2, cat. no. AB133474; Abcam) combinedwith mouse anti-GAPDH (cat. no. MAB374; Millipore) overnight at4°C. Membranes were then incubated with donkey anti-rabbitconjugated to HRP (1 mg/ml, cat. no. A16035; Life Technologies),combined with the IRDye 800CW goat anti-mouse antibody (cat. no.926-32210; LI-COR) for 30 minutes at 4°C. Horseradish peroxidasewas then developed, visualized, and analyzed as above. For quantifi-cation, LRRK2-pSer935 signals were normalized to total LRRK2 andexpressed as percentage of within-gel vehicle controls. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) levels were used for furthernormalization of protein loading for all samples to enable total LRRK2quantification.

Off-Target Selectivity Panel. Merck LRRK2 inhbitor-2 (MLi-2)was profiled for in vitro activity against 308 protein kinases using theInvitrogen SelectScreen protein kinase profiling service (http://www.thermofisher.com/us/en/home/products-and-services/services/custom-services/screening-and-profiling-services/selectscreen-profiling-service/selectscreen-kinase-profiling-service.html). All assays were performedat ATP concentrations near the Km (ATP). Compounds were tested at10 mM, 1 mM, and 0.1 mM and the percent inhibition observed at thethree concentrations were fitted to the four-parameter logistic equa-tion in which maximum percent inhibition, minimum percent in-hibition, and slope were held constant at 100, 0, and 1.0 respectively.Selectivity is reported as a percentage of kinases within 100� theLRRK2 IC50 in the LanthaScreen TR-FRET assay (ThermoFisherScientific). The off-target profile of MLi-2 was further assessed acrossadditional receptors, ion channels, and other proteins in a selectivityscreen at a standard concentration of 10 mM (Eurofins Panlabs,Taipei, Taiwan). The full methods and references can be found at:www.eurofinspanlabs.com/Panlabs. The percentage of inhibition isgiven as the average of three determinations. When significantdisplacement of radioligand was observed (.50% inhibition at 10 mM),complete concentration-dependent displacement curves (in triplicate)were constructed to generate IC50 values. IC50 values were determinedby a nonlinear, least-squares regression analysis using MathIQTM (IDBusiness Solutions Ltd., Guildford, Surrey, UK).

In Vivo Experiments

Animals. All experiments were performed according to the policiesof the Animal Care and Use Committee of Merck Research Labora-tories, Kenilworth, NJ, in conjunction with the American Associationfor the Accreditation of Laboratory Animal Care approved guidelinesand the Guide for Care and Use of Laboratory Animals (Institute ofLaboratory Animal Resources, 1996). Unless otherwise stated, the

studies involved male C57Bl/6 mice weighing 20–25g and werepurchased from The Jackson Laboratory (Bar Harbor, ME). Uponarrival to the facility, mice were housed five animals per cage andmaintained on a 12-hour light-dark cycle (lights on at 7:00 AM, lightsoff at 7:00 PM) under constant temperature (22 6 2°C) and humidity(.45%) conditions. Standard laboratory diet (LabDiet 5001; LabDiet,St. Louis, MO) and water were available ad libitum. Mice wereacclimated for at least 5 days before experimentation.

Pharmacokinetic Properties of MLi-2 in Mice. The pharma-cokinetic profile of MLi-2 was evaluated in male C57Bl/6 micefollowing a single intravenous (2 mg/kg; IV) or oral (10 mg/kg; PO)administration. Blood samples were collected at 5, 15, 30, 60, 120, 240,360, and 480 minutes after dosing and plasma was prepared bycentrifugation. Plasma levels of MLi-2 were determined by liquidchromatography and tandem mass spectrometry analyses (LC-MS/MS) on an API-5000 instrument (Applied Biosystems, Waltham,MA).The plasma and brain binding of MLi-2 was determined by equilib-rium dialysis at 1 mM.

Acute Effect of MLi-2 on Phosphorylation of LRRK2-Ser935in Mouse Brain Cortex. Merck LRRK2 inhbitor-2 (MLi-2) wassynthesized atMerck Research Laboratories. MLi-2 was suspended in30%Captisol (Ligand Technologies, La Jolla, CA) and administered ina volume of 10 ml/kg. Dose calculations were on the basis of activemoiety. Mice received MLi-2 [1-100 mg/kg; by mouth (PO)], or vehicle1 hour prior to euthanasia by excess CO2. Immediately followingeuthanasia, mouse brain cortex was dissected and frozen on a steelplate over dry ice for analysis of pSer935 LRRK2 via Western Blot.Plasma and brain samples were collected and frozen as describedabove for determination of MLi-2 levels by LC-MS/MS. pSer935LRRK2 data are expressed as the ratio of pSer935 LRRK2/totalLRRK2 and expressed as mean 6 S.E.M. The effect of MLi-2 onpSer935 LRRK2 levels in the brain and peripheral tissues wereanalyzed by one-way ANOVA and then post hoc comparisons weremade by Dunnett’s test using Prism software (GraphPad Software,Inc.), and the level of significance was P , 0.05.

Effect of 11-Days of In-Diet Treatment with MLi-2 (3–120mg/kg per day) on Phosphorylation of Ser935 in Mice. MaleC57Bl/6 mice (20–25g) from The Jackson Laboratory were singlyhoused in a room on a reversed 12-hour light-dark cycle (lights off at9:00 AM and lights on at 9:00 PM) with rodent diet (Research DietsD01060501; Research Diets, New Brunswick, NJ) and water availablead libitum. Mice were allowed to acclimate to the reversed light cyclefor one week before being placed on study. On day 1 of the study, micewere assigned to one of six treatment groups. Five groups of micereceived rodent diet (Research Diets D01060501) containing MLi-2and formulated by Research Diets targeted to provide concentrationsof 3, 10, 30, 60, or 120 mg/kg per day on the basis of an average foodintake of 3 g/day. The final group of mice received untreated diet(Research Diets D01060501) and served as the control group. Bodyweight and food intake were assessed daily (with the exception of theweekend). On day 10, 1 hour before the onset of the dark cycle, micewere transported to an adjacent room to assess the impact of MLi-2 onspontaneous locomotor activity (LMA). Spontaneous LMA was mea-sured using a TruScan Photo Beam Activity System (CoulbournInstruments, Whitehall, PA) that uses a three-plane tracking systemwith infrared sensors to record the animal’s movement. Mice wereplaced individually into the arena (26.67 � 26.67 � 39.37 cm) anddistance traveled (cm) and rearing behavior was recorded for a periodof 60 minutes. Mice were returned to their home cage immediatelyafter LMA assessment. On day 11, mice were euthanized by excessCO2 at 4, 8, or 24 hours after the start of the dark cycle (n5 5 mice pertreatment group). Immediately following euthanasia, mouse braincortex and kidneys were dissected and frozen on a steel plate over dryice for analysis of pSer935 LRRK2. Plasma and brain samples werecollected and frozen for determination of MLi-2 levels. The effect ofMLi-2 on pSer935 LRRK2 levels were analyzed by a two-way analysisof variance (ANOVA) and then posthoc comparisonsmade byDunnett’stest. Locomotor activity data were analyzed by one-way ANOVA and

MLi-2, a Potent and Selective Inhibitor of LRRK2 399

at ASPE

T Journals on M

ay 13, 2020jpet.aspetjournals.org

Dow

nloaded from

then posthoc comparisons were made by Dunnett’s test. The level ofsignificance was P , 0.05.

Effect of 15 Weeks In-Diet Treatment with MLi-2 (30 mg/kgper day) in the MitoPark Mouse Model of PD. Male MitoPark(Ekstrand et al., 2007) and littermate wild-type mice arrived at theMerck testing facility at 4 weeks of age and were singly housed on a12-hour light/dark cycle (lights on at 7:00 AM, lights off at 7:00 PM)with rodent diet (Research Diets D01060501) and water available adlibitum. Mice were acclimated to the facility for 4 days beforeundergoing a baseline LMA assessment for 60 minutes using thepreviously described method. At 5 weeks of age, which is prior to theonset of dopaminergic neuronal decline and the behavioral phenotype,the MitoPark mice were allocated to one of two treatment groupscounterbalanced by initial baseline LMA (distance traveled). The firstgroup received untreated rodent diet (ResearchDiets D01060501) andserved as the control group (n 5 30 per group). The second groupreceived rodent diet containing MLi-2 targeted to provide 30 mg/kgper day (n5 30 per group). The dose of 30 mg/kg per day was selectedon the basis of the previous in-diet dosing studies in C57Bl/6 mice andMitoPark mice, which demonstrated robust reductions (.90%) incortical pSer935 LRRK2 at steady state. A third group of wild-typelittermate mice received vehicle rodent diet (n 5 30 per group). Bodyweight and food intake were recorded weekly and rodent dietcontaining MLi-2 was reformulated by Research Diets every 3 weeksto take into account changes in body weight and food consumption tocontinue to target 30 mg/kg per day. All three groups of miceunderwent a 60 minutes LMA assessment at 8, 11, 14, and 17 weeksof age. Immediately after assessment of LMA at 8 and 14weeks of age,10 mice from each of the three treatment groups were taken to anadjacent room and euthanized by excess CO2. Mouse brain cortex wasdissected and frozen on a steel plate over dry ice for analysis ofpSer935 LRRK2. Additional brain tissue samples (cortex, striatum,and cerebellum) were dissected and frozen for neurochemical analysisand analysis of tyrosine hydroxylase levels (striatum only). Plasma andbrain sampleswere collected for determination ofMLi-2 levels. The finalcohort of mice (n 5 10 mice per treatment group) was euthanized byexcess CO2 at 20 weeks of age. As in the previous necropsies, the brainwas removed from all mice (n5 10mice per group) and cortex dissectedand frozen for analysis of pSer935LRRK2 (n5 10 per treatment group).An additional set of peripheral tissues, including lung, kidney, andspleen was collected from five mice per treatment group for analysis ofperipheral pSer935 LRRK2. Plasma and brain samples were collectedfrom the same five mice for MLi-2 levels. The remaining five mice pergroup underwent a safety assessment. Kidney, lung, and brain werefixed in 10%neutral buffered formalin. Tissueswere paraffin-embeddedand sectioned at 5 mM followed by staining with hematoxylin and eosinand microscopic examination. The effect of MLi-2 on pSer935 LRRK2levels in the brain and peripheral tissues were analyzed by one-wayANOVA, and then posthoc comparisons weremade by Tukey’s multiplecomparison test. Bodyweight, food intake, andLMAdatawere analyzedby a two-way ANOVA, and then posthoc comparisons made by Tukey’smultiple comparison test. The level of significance was P , 0.05.

Determination of Striatal Dopamine, DOPAC, and HVALevels. Tissue samples were weighed individually and stored at–80°C in plastic tubes until analyzed for dopamine and metabolitelevels. Immediately before analysis, samples were thawed on ice, andperchloric acid (0.1 M) was added to the sample tube to achieve0.1 g/ml according to tissue weight. Samples were homogenized on iceusing ultra sonication. Samples were then vortexed and stored at 4°Cfor 60 minutes. The samples were next centrifuged for 5 minutes at12,000 rpm, and the dopamine and metabolite content of the super-natant was analyzed by high-performance liquid chromatographywith electrochemical detection. An MD-150-150 column (3 � 150 mm,3 mm; ESA Bioscience, Chelmsford, MA) was used, and the mobilephase was MD-TM purchased from ESA Bioscience. The flow rate forthe analytical column was 0.25 ml/min, which was maintained at40°C with a column heater. A Coulochem III with Guard Cell 5020(1350mV) and Analytical Cell 5014 (E1 –150mV, E2 1220mV) was

used to detect dopamine and metabolites. Tissue concentrations ofdopamine and metabolites were determined by comparing the ratioof peak areas for unknowns versus an internal standard (3,4dihydroxybenzoic acid) to standard curves generated from knownamounts of analytes of interest (using ratio of standard peak areasverse internal standard). Linear regression curves were plotted fordopamine (DA) and major metabolites (DOPAC, HVA). Absolutetissue concentration of dopamine and metabolites are calculated onthe basis of the linear regression curve. DA, HVA, and DOPAClevels were analyzed by one-way ANOVA, and then posthoccomparisons were made by Tukey’s multiple comparison test. Thelevel of significance was P , 0.05.

Striatal Tyrosine Hydroxylase. Striatal tissue was bead ho-mogenized and lysed in 300 ml MSD Lysis Buffer (R60TX-3) withprotease (cOmplete Mini, cat. no. 11836170001; Roche) and phospha-tase inhibitors (Halt Phosphatase Inhibitor, cat. no. 78420; LifeTechnologies). Protein content per sample was determined by abicinchoninic acid colorimetric assay, as described above. Twentymicrograms of striatal protein was reduced and denatured at 70°C for10 minutes, then run on NuPAGE 3–8% Tris-acetate gels (LifeTechnologies, EA0378) and transferred to polyvinylidene fluoridemembranes (cat. no. LC2005, Life Technologies). Membranes wereblocked in 5% dry milk in Tris-buffered saline plus Tween-20 for1 hour and probed with mouse anti-Tyrosine Hydroxylase (1mg/ml,cat. no. MAB5280; Millipore) and mouse anti-GAPDH (cat. no.MAB374;Millipore) overnight at 4°C. Membranes were then incubatedwith goat anti-mouse-HRP (cat. no. sc2005; Santa Cruz Biotechnology,Dallas, TX) for 1 hour at room temperature. HRP signals weresubsequently developed by chemiluminescence (SuperSignal sub-strate, cat. no. 34075; Thermo Fisher Scientific) and were thenvisualized and analyzed on a LI-COR Odyssey system. Tyrosinehydroxylase levels were analyzed by one-way ANOVA and then posthoc comparisons were made by Tukey’s multiple comparison test. Thelevel of significance was P , 0.05.

ResultsIn Vitro Potency of MLi-2 and Selectivity Profile of

MLi-2. MLi-2 (Fig. 1A) was discovered as part of a small-molecule discovery program aimed at the identification ofpotent and selective LRRK2 inhibitors. The discovery andevolution of the indazole series of LRRK2 inhibitors fromwhichMLi-2 emerged will be described in detail elsewhere (Milleret al., in preparation). MLi-2 exhibits potent inhibition ofpurified G2019S LRRK2 kinase with an IC50 of 0.76 nM (Fig.1B). MLi-2 had an IC50 , 1 mM for five kinases tested otherthan LRRK2: CLK4, 225 nM; MAP3K14, 244 nM; MAP3K5,428 nM;CLK2, 605 nM; andTTK, 935 nM.One-hundred percentof kinases tested for MLi-2 had an IC50 greater than 100 timesthe LRRK2 IC50 (0.76 nM). Further off-target profiling revealedthat MLi-2 inhibited five nonkinases with IC50s , 10 mM:serotonin 5-HT2B, 1.2 mM; norepinephrine transporter, 3.8 mM;muscarinic M2, 6.4 mM; peroxisome proliferator-activated recep-tors g, 6.5 mM; and adenosine transporter, 9.7 mM.

Cellular Potency of MLi-2 by Ser935 LRRK2 De-phosphorylation. The phosphorylation-status of LRRK2amino acid Ser935 was assessed as an indicator of LRRK2kinase inhibition. Although this site is not directly phosphor-ylated by LRRK2, agreement in rank-order potencies wasmaintained between biochemical and cellular Ser935 IC50sacross various MLi-2 analogs (see Supplemental Table 2),demonstrating the value of evaluating Ser935 dephosphory-lation to confirm cellular LRRK2 kinase inhibition. In Tet-inducible LRRK2 SH-SY5Y cells, MLi-2 exhibited a pSer935

400 Fell et al.

at ASPE

T Journals on M

ay 13, 2020jpet.aspetjournals.org

Dow

nloaded from

LRRK2 IC50 of 1.4 nM (Fig. 1, C and D), in line with its highpotency in the biochemical assay (LanthaScreen Assay, IC50

0.76 nM) and indicative of MLi-2 possessing high cellularpermeability.

Radioligand Binding. A key issue with the use ofpSer935 LRRK2 dephosphorylation as a measure of cellularpotency and a surrogate for in vivo target engagement is thatthis is not a direct LRRK2 autophosphorylation site. Thisraises the possibility that this readout may not directlycorrelate with enzyme occupancy by LRRK2 inhibitors.To investigate this we developed a whole-cell binding assayusing a novel LRRK2 inhibitor radioligand, [35S]MLi-A, inTet-inducible LRRK2 SH-SY5Y cells (Fig. 2A). [35S]MLi-A

Fig. 1. Structure and enzymatic activity of MLi-2. (A) Chemical Structureof cis-2,6-dimethyl-4-(6-(5-(1-methylcyclopropoxy)-1H-indazol-3-yl)pyrimidin-4-yl)morpholine (MLi-2). (B) GST-LRRK2 [G2019S] (1326–2527) was assayedusing 400 nM LRRKtide in the presence of 134 mM ATP with the indicatedconcentrations of MLi-2. The results are presented as percentage inhibition ofkinase activity. Results are averages of quadruplicate reactions. (C) Westernblot and (D) quantification of the MLi-2 dose response for inhibition ofpSer935 LRRK2 in Tet-inducible LRRK2 SHSY5Y cells. The results arepresented as mean percent inhibition by MLi-2, calculated as the changefrom dimethylsulfoxide (DMSO) control samples.

Fig. 2. Whole-cell LRRK2 radioligand binding. (A) Chemical structure ofLRRK2 radioligand ([35S]MLi-A). (B) Whole-cell saturation binding withincreasing concentrations of [35S]MLi-A in Tet-inducible wild-type LRRK2expressing SHSY5Y cells. Binding was performed in the absence ofinduction (noninduced) or where expression was induced for 48 hours inthe presence of tetracycline (Tet-induced). The results are presented asspecific binding and are averages of triplicate reactions. The KD value, innM, is derived from the graph. (C) Displacement of [35S]MLi-A bindingwith MLi-2. Experiments were performed with 0.7 nM [35S]MLi-A andincreasing concentrations of MLi-2. The results are presented as specificbinding and are averages of triplicate reactions. The Ki value, in nM, isderived from the graph. CPM, counts per minute.

MLi-2, a Potent and Selective Inhibitor of LRRK2 401

at ASPE

T Journals on M

ay 13, 2020jpet.aspetjournals.org

Dow

nloaded from

exhibited high levels of displaceable and saturable high-affinity binding (KD 5 0.2 nM) with negligible nondisplaceablebinding in the absence of Tet-induction (Fig. 2B). Comparablebinding was observed using cells expressing G2019S LRRK2(KD 5 0.52 nM; Supplemental Fig. 1). Whole-cell competitionbinding studies were subsequently performed with multipleLRRK2 inhibitors. In this assay, MLi-2 exhibited a Ki of3.4 nM, very close to its cellular IC50 (1.4 nM) in the pSer935LRRK2 dephosphorylation assay (Fig. 2C). Indeed, IC50

values generated in the whole-cell LRRK2 binding assaywere found to be in line with values obtained from thecellular pSer935 LRRK2 dephosphorylation assay with avariety of inhibitors (see Supplemental Table 1). These datatherefore led us to conclude that pSer935 LRRK2 dephos-phorylation assay closely approximates target occupancy,as defined in cell lines expressing LRRK2 and, therefore,presumably in vivo as well.Pharmacokinetic Profile of MLi-2 in Mice. Pharma-

cokinetic and oral bioavailability data for MLi-2 in mice aresummarized in Table 1. In mice, the mean maximum plasmaconcentrations following a single 10-mg/kg oral dose occurredat 0.75 hours postdose. The mean residence time (MRT) was2.7 hours following intravenous dosing, and ∼11 hoursfollowing oral dosing, indicating a relatively long meanabsorption time. Oral bioavailability for MLi-2 in mice was∼45%. Unbound fraction in plasma and brain was determinedto be 0.008 and 0.009, respectively.Effect of MLi-2 on Phosphorylation of Ser935 in

Mouse Brain Cortex. On the basis of the pharmacokineticprofile of MLi-2, we evaluated the ability of MLi-2 to inhibitphosphorylation of LRRK2 Ser935 in the brain of wild-typemice after acute oral dosing. As shown in Fig. 3, A–C, 1 hourafter oral administration unbound brain concentrations ofMLi-2 increased in a dose-dependent fashion (up to 273 nMat 100 mg/kg) whereas phosphorylation of Ser935 [F(5,29) 513.44; P, 0.001] decreased in a dose-dependent manner (P,0.05, P , 0.001) with maximal suppression (.90% reduction)of pSer935 LRRK2 achieved at 10 mg/kg and higher.Effect of 11 Days In-Diet Dosing with MLi-2 (3–120

mg/kg per day) on Phosphorylation of Ser935 in MouseCortex and Kidney. To determine if MLi-2 might be asuitable compound to explore the chronic effects of LRRK2kinase inhibition (e.g., excellent in vivo oral potency and welltolerated), we evaluated the effect MLi-2 on the phosphoryla-tion status of Ser935 in mouse cortex and kidney after in-dietdelivery for 11 days. In-diet administration of MLi-2 was welltolerated and there were no differences in body weight in theMLi-2-treated groups compared with the mice receivingvehicle over the 11-day treatment period (data not shown). Food intake in all MLi-2-treated groups tended to be lower

than in the vehicle-treated group during the first 1–3 days ofthe study but returned to normal feeding levels thereafter. Asa result, the targeted doses of MLi-2 on a mg/kg per day basiswas achieved in all animals. Figure 4, A and B, shows theeffect of 11 days of in-diet administration ofMLi-2 (3-120mg/kgper day) on pSer935 LRRK2 levels in the cortex of mice. A two-way ANOVA revealed a significant effect of drug [F(5,24) 512.0; P , 0.001], time [F(2,48) 5 7.34; P , 0.001] and drugx time interaction [F(10,48)5 3.4;P, 0.001]. Post-hoc analysisrevealed that MLi-2 administration resulted in a dose-dependent inhibition of Ser935 in the brain that was sustainedacross the 24-hour sampling period. The reduction in pSer935LRRK2 was statistically significant compared with the

TABLE 1Pharmacokinetic and oral bioavailability data for MLi-2 in micea

Parameter 2 mg/kg IV 10 mg/kg PO

AUC (mM×h) 0.664 (7%) 1.50 (52%)Cmax (mM) 0.716 (11%) 0.298 (72%)Tmax (h) 0.75 (58%)MRT (h) 2.7 (17%) 11 (63%)%F 45 (52%)

AUC, area under the curve; Cmax, maximum observed drug concentration; Tmax,time to maximum observed drug concentration; MRT, mean residence time; %F, oralbioavailability.

aValues are presented as mean (CV%), N = 3.

Fig. 3. Effect of oral administration of various doses of MLi-2 on pSer935LRRK2 in the brains of mice after acute, single administration. (A)Western blot and (B) quantification of LRRK2 pSer935 LRRK2 levels. (C)Unbound brain concentration of MLi-2. Data are mean 6 S.E.M.; n = 5mice per treatment group.*P, 0.05–P, 0.001 compared with the vehicle-treated animals.

402 Fell et al.

at ASPE

T Journals on M

ay 13, 2020jpet.aspetjournals.org

Dow

nloaded from

vehicle-treated group at all doses at the 4-hour time point (P,0.001) and at 10, 30, 60, and 120 mg/kg per day at 8- and24-hour time points (P, 0.001). Across all time points amaximalreduction in pSer935 LRRK2 was observed at 30 mg/kg per day.Similar reductions in pSer935 LRRK2 were observed in thekidney after 11 days in diet dosing with MLi-2 (Supplemen-tal Fig. 2, A and B); drug [F(5,24) 5 262.9; P , 0.001], time[F(2,48) 5 11.1; P , 0.001] and drug x time interaction[F(10,48) 5 3.67; P , 0.001]. Posthoc analysis revealed

statistically significant reduction in pSer935 LRRK2 com-pared with the group receiving vehicle diet at all doses andtime points (P , 0.001). As shown in Fig. 4C, there was noeffect of MLi-2 on total LRRK2 protein levels in the brain after11 days of in-diet dosing [drug: F(5,24)5 0.8; n.s.]. However, asignificant reduction in LRRK2 protein levels, detected byWestern blot using MJFF-2 antibody, was observed in thekidneys of mice treated with MLi-2 for 11 days (see Supple-mental Fig. 2, A and B); drug [F(5,24)5 41.1; P, 0.001], time

Fig. 4. Effect of MLi-2 (3–120 mg/kg per day) on pSer935 LRRK2 (ratio of pSer935/total LRRK2) and total LRRK2 (total LRRK2/GAPDH) in the brainsof mice after 11 days of in-diet dosing. (A) Western blot showing pSer935 LRRK2 and total LRRK2. (B) and (C) shows the quantification of pSer935LRRK2 and total LRRK2, respectively. (D) Unbound brain concentrations of MLi-2 on day 11 in mice administered MLi-2 in diet for 11 days. Opencircles, 3 mg/kg per day; 10 mg/kg per day, open squares; 30 mg/kg per day, closed triangles; 60 mg/kg per day, open triangles; and 100 mg/kg, closedcircles. (E) pSer935 LRRK2 levels (ratio of pSer935/total LRRK2 expressed as a percentage of the vehicle control group) in the brains of mice versusunbound brain concentrations of MLi-2 (3–120 mg/kg per day) on day 11 of in-diet dosing. Mice were euthanized a 4, 8, and 24 hours after the start of thedark cycle on day 11 of dosing. Data are mean 6 S.E.M.; n = 5 per treatment group; ***P , 0.001 compared with the vehicle diet–treated animals.

MLi-2, a Potent and Selective Inhibitor of LRRK2 403

at ASPE

T Journals on M

ay 13, 2020jpet.aspetjournals.org

Dow

nloaded from

[F(2,48) 5 141.2; P , 0.001], and drug � time interaction[F(10,48) 5 3.5; P , 0.001]. The effect of total LRRK2 proteinlevels in the kidney was dependent on dose and was statisti-cally significant compared with the vehicle-treated animals atall doses at the 4-hour time point (P, 0.001 in all cases); at 10,30, 60, and 120mg/kg per day at the 8-hour point (P, 0.001 inall cases); and at 30, 60, and 120 mg/kg per day doses at the24-hour point (P , 0.001 in all cases). Analysis of the unboundpharmacokinetic data shown in Fig. 4D revealed that in-dietdosing with MLi-2 resulted in dose-dependent increases inMLi-2 exposures in the brain. SimilarMLi-2 unbound exposureswere observed in the plasma (data not shown). Moreover, the in-diet dosing approach led to consistent MLi-2 exposure in brainand plasma over the 24-hour sampling period with minimalpeak-to-trough variation, which is in excellent agreement withthe sustained pharmacodynamic effect of MLi-2 on pSer935LRRK2 in the brain and periphery. Using a pharmacodynamicEmax model, we determined the brain pSer935 LRRK2 IC50 forMLi-2 to be 0.8 nM using unbound brain exposure (see Fig. 4E)and 1.1 nM (data not shown) using unbound plasma exposure.As Parkinson’s disease is primarily a motor disorder and theevaluation of pharmacological treatments in animal models ofPD relies heavily on locomotor activity endpoints, we alsoevaluated the effect of MLi-2 on spontaneous LMA to controlfor possible interference in subsequent locomotor assessments.Eleven days of in-diet dosing with MLi-2 did not impactspontaneous locomotor activity asmeasuredbydistance traveled[F (5, 24)5 1.26; P5 0.31] or rearing behavior [F (5, 24)5 0.35;P 5 0.87] in mice at any of the doses examined (SupplementalFig. 3, A and B).Effect of 15 Weeks In-Diet Treatment with MLi-2

(30 mg/kg per day) on Phosphorylation of Ser935,Behavior, and Neurochemistry in the MitoPark MouseModel of PD. The connection between LRRK2 andmitochon-drial function (Biskup et al., 2006; Saha et al., 2009) suggestedthat LRRK2 kinase inhibition may confer neuroprotectiveeffects in PD models of mitochondrial dysfunction. To evaluatethe disease-modifying potential of LRRK2 kinase inhibition wetherefore evaluated the effect of MLi-2 in the MitoPark mousemodel of PD; a model in which a PD-like phenotype develops asa result of DAT-Cre-dependent knockout of the mitochondrialtranscription factor (TFAM).Asymptomatic, 5-week-oldMitoParkmice were treated for up to 15 weeks with MLi-2 in diet at adose of 30mg/kg per day At this concentration,MLi-2 has beenshown to inhibit pSer935 LRRK2 by greater than 90% over a24-hour period in both wild-type (see Fig. 4, A and B) andMitoPark mice (data not shown). Consistent with short-termin-diet administration, chronic treatment with MLi-2 (30 mg/kgper day) was well tolerated. All groups gained weight through-out the study and no differences in body weight were observedbetweenMLi-2-treatedmice and the vehicle-treatedMitoParkmice (Fig. 5A). There were also no differences in terms of foodintake between the MLi-2-treated mice and the vehicle-treated MitoPark mice (Fig. 5B), and the targeted 30 mg/kgper day dose of MLi-2 was achieved (Fig. 5C). Analysis ofMLi-2 levels in the plasma and brain revealed that theunbound brain exposure of MLi-2 was in excess of the in vivobrain pSer935 LRRK2 IC50 (0.8 nM unbound brain; 1.1 nMunbound plasma) or unbound brain throughout the study(unbound plasma week 8, 5.516 1.38 nM; week 14, 8.486 2.29nM;week 20, 7.646 2.74 nMandunbound brainweek 8, 5.0461.09 nM;week 14, 6.486 1.49 nM, andweek 20, 7.366 2.3 nM).

Fig. 5. Chronic in-diet dosing with MLi-2 (30 mg/kg per day) is welltolerated and leads to sustained inhibition of CNS LRRK2 kinase activity.(A) Bodyweight, (B) food intake, and (C) dose achieved per week of thestudy. Effect of MLi-2 at 30 mg/kg per day on pSer935 LRRK2 levels inthe cortex of mice after (D) 3 weeks on treatment, (F) 9 weeks ontreatment, and (H) 15 weeks on treatment. Panels (E), (G), and (I) showthe Western blot data for pSer935 LRRK2, total LRRK2, and GAPDHfrom each time point. MLi-2 treatment started at 5 weeks of age. N = 10mice/group per time point. Data are mean 6 S.E.M.; n = 10 pertreatment group; *P , 0.001 compared with the vehicle diet–treatedanimals.

404 Fell et al.

at ASPE

T Journals on M

ay 13, 2020jpet.aspetjournals.org

Dow

nloaded from

In agreement with the pharmacokinetic data, analysis of thepharmacodynamic response (Fig. 5, D–I) revealed that MLi-2markedly inhibited pSer935 LRRK2 in the brain at all of thetime points measured compared with the vehicle-treatedMitoPark mice. Total LRRK2 protein levels, measured byWestern blot using MJFF-2 antibody in the brains of MLi-2-treated MitoPark mice were not different from vehicle-treatedMitoPark mice at any time point (data not shown). Despitedemonstrating significant pharmacodynamic effect, as mea-sured by inhibition of pSer935 LRRK2, MLi-2 did not stop theprogressive motor phenotype that is observed in MitoPark

mice. Locomotor activity and rearing behaviorwere comparableacross the three groups at baseline and at 8weeks of age (3 weekson treatment). It should be noted, however, that at baselinethe MitoPark mice that were subsequently placed on MLi-2displayed higher rearing behavior than the wild-type cohortat baseline; P , 0.05). Activity levels in both the vehicle (P ,0.001 in all cases) and MLi-2-treated MitoPark mice (P ,0.001 in all cases) declined significantly by week 11 (6 weekson treatment) compared with the wild-type littermate controlgroups (Fig. 6, A and B). This decline in locomotor and rearingbehavior continued throughout the study and at no time pointdid the MitoPark mice-treated with MLi-2 statistically differ-entiate from the vehicle-treated MitoPark group on eitherbehavioral measure. In agreement with the lack of efficacy ofMLi-2 on the behavioral endpoints, neurochemical analysisalso revealed that MLi-2 treatment failed to prevent thedecline in striatal dopamine, DOPAC, or tyrosine hydroxylaselevels that were evident in MitoPark mice at 8 weeks of age(3 weeks on treatment;P, 0.05 in all cases compared with thewild-type-vehicle-treated group) and prior to the onset of thebehavioral deficit at week 11 (see Table 2).In addition to evaluating the disease-modifying potential

of LRRK2 kinase inhibitors in the MitoPark mouse model,we took advantage of the extended duration of this study toalso explore the long-term tolerability of LRRK2 kinaseinhibition. In particular, histologic and morphologic assess-ments were made of kidney and lungs that were demon-strated to have been altered in LRRK2 knockout (KO) mice.As shown in Fig. 7, A, B, and D, MLi-2 markedly inhibitedlung and kidney pSer935 LRRK2 levels at this time point.Significant reductions in LRRK2 total protein levels alsowere observed at 20 weeks of age in both the lung and kidneyof MLi-2-treated mice (Fig. 7, C and E). Unlike observationsof kidney darkening in LRRK2 KOmice (Herzig et al., 2011;Baptista et al., 2013; Tong et al., 2012), MitoPark micereceiving rodent diet containing MLi-2 at 30 mg/kg per dayfor 15 weeks did not exhibit kidney darkening upon grossexamination nor accumulation of pigment in renal corticaltubules on tissue section. MLi-2 treatment at 30 mg/kg perday for 15 weeks was associated with very slight enlarge-ment of randomly scattered alveolar epithelial cells thathad features consistent with type II pneumocytes. Theenlargement consisted of increased prominence of the typeII pneumocytes owing to variable degrees of distension ofthe cytoplasm by numerous clear, round vacuoles, whichwere larger and more numerous than those observed in typeII pneumocytes of untreated control mice. The number ofaffected cells varied greatly per 400�-field, from very few in

TABLE 2Effect of chronic in-diet dosing with MLi-2 (30 mg/kg per day) on dopamine, DOPAC, HVA, and TH levels in striatum ofMitoPark mice

Week Genotype Dose DA DOPAC HVA Metabolite/DA Ratio TH

mg/kgper day

ng/g ng/g ng/g % control

8 Wild-type 0 1715 6 50.71 14502 6 652.1 270.06 6.4 9.083 6 0.8 100.0 6 24.7MitoPark 0 1110 6 53.4*** 9625 6 332.8*** 263.5 6 8.5 9.044 6 0.4 54.56 6 10.2MitoPark 30 1114 6 124.6*** 8966 6 375.4*** 267.6 6 8.3 8.332 6 0.2 49.89 6 10.4

14 Wild-type 0 2175 6 232.6 20076 6 1371 391.8 6 53.3 9.864 6 0.6 100.0 6 19.8MitoPark 0 1037 6 93.1*** 6193 6 533.3*** 310.1 6 21.5 6.306 6 0.2*** 36.70 6 4.7***MitoPark 30 1105 6 176.2*** 7521 6 2145*** 323.7 6 36.7 6.556 6 0.5*** 42.60 6 7.1***

Data are expressed as the mean and 6 S.E.M. ***P , 0.001 compared with the vehicle-treated wild-type group.

Fig. 6. Chronic in-diet dosing with MLi-2 (30 mg/kg per day) does notattenuate the behavioral phenotype in in MitoPark mice. Effect of MLi-2on (A) distance traveled (B) rearing measured for 60 minutes. MLi-2treatment started at 5 weeks of age. Data are mean 6 S.E.M.; n = 10 pertreatment group. *P , 0.05; ***P , 0.001 compared with the vehiclediet–treated animals.

MLi-2, a Potent and Selective Inhibitor of LRRK2 405

at ASPE

T Journals on M

ay 13, 2020jpet.aspetjournals.org

Dow

nloaded from

Fig. 7. Effect of chronic in-diet dosing with MLi-2 (30 mg/kg per day) on LRRK2 kinase activity and histology in lung and kidney of MitoPark mice.(A) Representative Western blot data for the effect of MLi-2 on pSer935 LRRK2 and total LRRK2 in the lung and kidney of MitoPark mice.Quantification is shown in (B) for lung pSer935, in (C) for lung total LRRK2, in (D) for kidney pSer935, and in (E) for kidney total LRRK2. Mice wereeuthanized at 20 weeks of age, 15 weeks after treatment started. Lung histology images from vehicle- (F) and MLi-2- (G) treated MitoPark mice. MLi-2

406 Fell et al.

at ASPE

T Journals on M

ay 13, 2020jpet.aspetjournals.org

Dow

nloaded from

some fields to several affected type II pneumocytes in otherfields. This finding was observed in five out of five of theMLi-2-treated animals examined. No findings of enlargedtype II pneumocytes were observed in wild-type orMitoParkmice receiving vehicle diet for the same duration.

DiscussionThe current article details the pharmacological properties of

a structurally novel inhibitor of LRRK2 kinase activity,MLi-2.This potent ATP-competitive active-site kinase inhibitorpossesses excellent selectivity for LRRK2 over a broad rangeof kinases and a diverse panel of receptors and ion channels. Invivo, acute or subchronic dosingwithMLi-2 inmice resulted indose-dependent, central and peripheral target inhibition asmeasured by dephosphorylation of LRRK2 pSer935 LRRK2.To date the only phospho-antibodies that have been usedsuccessfully to monitor cellular LRRK2 kinase activity havebeen directed toward phosphorylation at Ser910 and Ser935residues. Although not genuine autophosphorylation sites,pharmacological inhibition of LRRK2 kinase activity has beendemonstrated to reduce the phosphorylation of Ser910 andSer935 in an indirect, yet dose-dependent fashion (Dzamkoet al., 2010). This indirect relationship raises the issue of theextent to which LRRK2 pSer935 LRRK2 dephosphorylationcorrelates with enzyme occupancy of a LRRK2 inhibitor. Totry and address this, we developed a whole-cell binding assaywith a novel LRRK2 radioligand [35S]MLi-A. Specific bindingin Tet-inducible LRRK2-overexpressing SHSY5Y cells wasdemonstrated to be entirely LRRK2-specific and was dis-placed by MLi-2 and, indeed, by other LRRK2 inhibitors indose-dependent fashion. Furthermore, IC50 values derivedfrom competition binding correlated verywell with IC50 valuesfor pSer935 LRRK2 dephosphorylation in the same cell line.These observations led us to conclude that for this structuralclass of LRRK2 inhibitors, pSer935 LRRK2 dephosphoryla-tionmeasures give a good approximation of enzyme occupancyin cells, and therefore supports its utility as a suitable in vivopharmacodynamic proxy of target engagement.Although multiple potent LRRK2 kinase inhibitors, from

structurally distinct chemotypes, have been reported in theliterature, their utility to probe the tolerability and therapeu-tic potential of LRRK2 kinase inhibition in vivo has beenhindered by a combination of poor selectivity, suboptimalpharmacokinetic qualities and/or a lack of brain penetration.Given the in vitro data demonstrating that MLi-2 is a highlypotent and selective inhibitor of LRRK2 kinase activity,coupled with the pharmacokinetic data demonstrating goodoral bioavailability and mean residence time, we sought todetermine if MLi-2 was suitable for exploring LRRK2 biologyin vivo. Acute and subchronic dosing with MLi-2 in miceresulted in dose-dependent reductions in both peripheraland brain LRRK2 kinase activity as measured by the de-phosphorylation of LRRK2 pSer935 LRRK2 with IC50 vales of0.8 nM and 1.1 nM using unbound brain and plasma exposure

respectively. The lack of potency shift between unbound brainversus plasma underscores the lack of Pgp sensitivity and highbrain penetration of MLi-2.Treatment with MLi-2 was foundto be well tolerated, and no adverse effects of MLi-2 on bodyweight, food intake, or behavioral activity were observed atbrain and plasma exposures .100� the in vivo IC50 for CNSLRRK2 kinase inhibition. The excellent tolerability of MLi-2after subchronic dosing is particularly encouraging as arecent publication by Fuji et al., (2015) raised significantconcerns over the tolerability of repeated administration ofLRRK2 kinase inhibitors. That LRRK2 inhibition by MLi-2is well-tolerated by rodents agrees well with a recent reportdemonstrating good tolerability with another LRRK2 ki-nase inhibitor (PFE-06447475) after chronic administration(Daher et al., 2015). Both MLi-2 and PFE-06447475 havehigher potency and selectivity for LRRK2 (Henderson et al.,2015) compared with GNE-0877 and GNE-7915 (Fuji et al.,2015), thus it seems probable that adverse effects observedwith GNE-0877 and GNE-7915 are not LRRK2-mediated andare most probably the result of an as yet to be determined offtarget activity.The ability to potently inhibit in vivo LRRK2 kinase

activity, coupled with its excellent tolerability profile, under-scores the utility of MLi-2 for confirming presumed LRRK2kinase activity–mediated phenotypes. Consistent with obser-vations made in LRRK2 KO rodents and nonhuman primatestreated with less selective LRRK2 kinase inhibitors (Fujiet al., 2015), enlargement of type II pneumocytes (defined asvery slight) was found in the lungs of all mice that receivedMLi-2 for 15 weeks. In addition to Ser935 dephosphorylation,we observed a significant reduction in total LRRK2 proteinlevels in the lung, although total LRRK2 levels were notcompletely ablated. To the best of our knowledge these are thefirst data showing that pharmacological inhibition of LRRK2kinase activity in rodents can recapitulate the lung phenotypeobserved in LRRK2 KO animals. Previous reports have failedto observe the induction of the lung phenotype or the reductionin LRRK2 protein levels in mice and rats treated with LRRK2for up to 28 days (Henderson et al., 2015; Daher et al., 2015;Fuji et al., 2015). However, the induction of lung phenotypes inrodents may require sustained LRRK2 inhibition and/or,longer treatment durations than previously achieved. In-dietdosing paradigms similar to that employed in the presentstudy may be critical to precipitating this phenotype in rodentsto support a detailed characterization of type II pneumocytebiology.In addition to exploring the role of LRRK2 kinase activity in

lung, we also evaluated the disease-modifying potential ofMLi-2 in the MitoPark mouse model of PD (Ekstrand et al.,2007). MitoPark mice display a gradual loss of DA neuronsand progressive motor impairment resulting from DAT-Cre-dependent knockout of the mitochondrial T-FAM tran-scription factorMitoParkmicemore closely approximates thatseen in Parkinson’s disease than pharmacological or neuro-toxin induced models of PD. In addition to its regional

treatment at 30 mg/kg per day for 15 weeks was associated with very slight enlargement of randomly scattered alveolar epithelial cells that had featuresconsistent with type II pneumocytes. Kidney histology images from vehicle- (H) and MLi-2- (I) treated MitoPark mice. No accumulation of pigment inrenal cortical tubules was observed in the vehicle- (H) orMLi-2- (I) treatedMitoPark mice. Histology of the lung and kidney was performed at 20 weeks ofage, 15 weeks after treatment started, and on n = 5 mice per treatment group. Representative images for the vehicle-treated wild-type littermate groupare not shown; **P , 0.01, ***P , 0.001 compared with the vehicle-treated wild-type group.

MLi-2, a Potent and Selective Inhibitor of LRRK2 407

at ASPE

T Journals on M

ay 13, 2020jpet.aspetjournals.org

Dow

nloaded from

localization to the substantia nigra and reported subcellularcompartmentalization to mitochondria (Biskup et al., 2006),LRRK2 is thought to modulate vulnerability to inhibitionof mitochondrial respiration function (Saha et al., 2009).Furthermore, pathogenic LRRK2 G2019S has been associ-ated with impaired mitochondrial structure, function, andmtDNA damage (Mortiboys et al., 2010; Cooper et al., 2012;Sanders et al., 2014), suggesting that LRRK2 kinase inhibitionmay confer neuroprotective effects in PD models of mitochon-drial dysfunction, such as the MitoPark mouse. However,despite demonstrating sustained inhibition of LRRK2 kinaseactivity in the brain over the 15-week treatment period, MLi-2did not stop or slow the progression of the behavioral andneurochemical phenotype observed in the MitoPark mousemodel. That MLi-2 did not ameliorate the progression of theMitoPark phenotype may be viewed more as a reflection of theinability to faithfully model PD disease progression preclini-cally rather than a failure of the LRRK2 hypothesis, in theabsence of data confirming neuroprotective effects of LRRK2deletion in this model.In summary, the current work describes a structurally

novel, potent, and highly-selective LRRK2 inhibitor discov-ered from internal medicinal chemistry efforts with ourindazole series of LRRK2 inhibitors. Here, we detailed thepharmacological and pharmacokinetic properties of MLi-2in mice, which demonstrate its suitability for exploringLRRK2 kinase biology in cellular and animal models.Through the use of the in-diet dosing approach, clearguidelines were defined to achieve chronic LRRK2 inhibi-tion, along with extensive reference pharmacodynamicdata, which will enable future lung de-risking studies andfurther evaluation in LRRK2 neuroprotection paradigms.As progress on these fronts has been hampered by thevarious limitations of currently available pharmacologicaltools, we hope access to MLi-2 will help advance preclinicalefforts at understanding LRRK2 biology and, throughcollaborative efforts, shed light on novel pathways down-stream of LRRK2 that may enrich the target space for PDtherapeutics.

Authorship Contributions

Participated in research design: Cheewatrakoolpong, Fell, Hyde,Kennedy, Markgraf, Mei, Mirescu, Morrow, Lin, Smith, Parker, Yin,Zhou.

Contributed new reagents or analytic tools: Basu, DeMong, Miller,Scott.

Performed data analysis: Cheewatrakoolpong, Fell, Mirescu, Mei,Lin, Poulet, Smith, Yin, Zhou.

Wrote or contributed to the writing of the manuscript: Ellis, Fell,Kennedy, Markgraf, Mirescu, Morrow, Poulet.

References

Baptista MA, Dave KD, Frasier MA, Sherer TB, Greeley M, Beck MJ, Varsho JS,Parker GA, Moore C, and Churchill MJ, et al. (2013) Loss of leucine-rich repeatkinase 2 (LRRK2) in rats leads to progressive abnormal phenotypes in peripheralorgans. PLoS One 8:e80705.

Biskup S, Moore DJ, Celsi F, Higashi S, West AB, Andrabi SA, Kurkinen K, Yu SW,Savitt JM, and Waldvogel HJ, et al. (2006) Localization of LRRK2 to membranousand vesicular structures in mammalian brain. Ann Neurol 60:557–569.

Bonifati V (2007) LRRK2 low-penetrance mutations (Gly2019Ser) and risk alleles(Gly2385Arg)-linking familial and sporadic Parkinson’s disease. Neurochem Res32:1700–1708.

Choi HG, Zhang J, Deng X, Hatcher JM, Patricelli MP, Zhao Z, Alessi DR, and GrayNS (2012) Brain Penetrant LRRK2 Inhibitor. ACS Med Chem Lett 3:658–662.

Cookson MR (2010) The role of leucine-rich repeat kinase 2 (LRRK2) in Parkinson’sdisease. Nat Rev Neurosci 11:791–797.

Cooper O, Seo H, Andrabi S, Guardia-Laguarta C, Graziotto J, Sundberg M, McLeanJR, Carrillo-Reid L, Xie Z, and Osborn T, et al. (2012) Pharmacological rescue of

mitochondrial deficits in iPSC-derived neural cells from patients with familialParkinson’s disease. Sci Transl Med 4:141ra90.

Daher JP, Abdelmotilib HA, Hu X, Volpicelli-Daley LA, Moehle MS, Fraser KB,Needle E, Chen Y, Steyn SJ, Galatsis P, Hirst WD and West AB (2015) LRRK2Pharmacological Inhibition Abates alpha-Synuclein Induced Neurodegeneration.J Biol Chem.

Deng X, Dzamko N, Prescott A, Davies P, Liu Q, Yang Q, Lee JD, Patricelli MP,Nomanbhoy TK, and Alessi DR, et al. (2011) Characterization of a selectiveinhibitor of the Parkinson’s disease kinase LRRK2. Nat Chem Biol 7:203–205.

Dzamko N, Deak M, Hentati F, Reith AD, Prescott AR, Alessi DR, and Nichols RJ(2010) Inhibition of LRRK2 kinase activity leads to dephosphorylation of Ser(910)/Ser(935), disruption of 14-3-3 binding and altered cytoplasmic localization. Bio-chem J 430:405–413.

Ekstrand MI, Terzioglu M, Galter D, Zhu S, Hofstetter C, Lindqvist E, Thams S,Bergstrand A, Hansson FS, and Trifunovic A, et al. (2007) Progressive parkin-sonism in mice with respiratory-chain-deficient dopamine neurons. Proc Natl AcadSci USA 104:1325–1330.

Estrada AA, Liu X, Baker-Glenn C, Beresford A, Burdick DJ, Chambers M, Chan BK,Chen H, Ding X, and DiPasquale AG, et al. (2012) Discovery of highly potent,selective, and brain-penetrable leucine-rich repeat kinase 2 (LRRK2) small mole-cule inhibitors. J Med Chem 55:9416–9433.

Fuji RN, Flagella M, Baca M, Baptista MA, Brodbeck J, Chan BK, Fiske BK, Hon-igberg L, Jubb AM, and Katavolos P, et al. (2015) Effect of selective LRRK2 kinaseinhibition on nonhuman primate lung. Sci Transl Med 7:273ra15.

Greggio E, Jain S, Kingsbury A, Bandopadhyay R, Lewis P, Kaganovich A, van derBrug MP, Beilina A, Blackinton J, and Thomas KJ, et al. (2006) Kinase activity isrequired for the toxic effects of mutant LRRK2/dardarin. Neurobiol Dis 23:329–341.

Healy DG, Falchi M, O’Sullivan SS, Bonifati V, Durr A, Bressman S, Brice A, Aasly J,Zabetian CP, and Goldwurm S, et al.; International LRRK2 Consortium (2008)Phenotype, genotype, and worldwide genetic penetrance of LRRK2-associatedParkinson’s disease: a case-control study. Lancet Neurol 7:583–590.

Henderson JL, Kormos BL, Hayward MM, Coffman KJ, Jasti J, Kurumbail RG,Wager TT, Verhoest PR, Noell GS, and Chen Y, et al. (2015) Discovery andpreclinical profiling of 3-[4-(morpholin-4-yl)-7H-pyrrolo[2,3-d]pyrimidin-5-yl]benzonitrile (PF-06447475), a highly potent, selective, brain penetrant, and invivo active LRRK2 kinase inhibitor. J Med Chem 58:419–432.

Herzig MC, Kolly C, Persohn E, Theil D, Schweizer T, Hafner T, Stemmelen C,Troxler TJ, Schmid P, and Danner S, et al. (2011) LRRK2 protein levels are de-termined by kinase function and are crucial for kidney and lung homeostasis inmice. Hum Mol Genet 20:4209–4223.

Kachergus J, Mata IF, Hulihan M, Taylor JP, Lincoln S, Aasly J, Gibson JM, RossOA, Lynch T, and Wiley J, et al. (2005) Identification of a novel LRRK2 mutationlinked to autosomal dominant parkinsonism: evidence of a common founder acrossEuropean populations. Am J Hum Genet 76:672–680.

Lee BD, Shin JH, VanKampen J, Petrucelli L, West AB, Ko HS, Lee YI, Maguire-Zeiss KA, Bowers WJ, and Federoff HJ, et al. (2010) Inhibitors of leucine-richrepeat kinase-2 protect against models of Parkinson’s disease. Nat Med 16:998–1000.

Mortiboys H, Johansen KK, Aasly JO, and Bandmann O (2010) Mitochondrial im-pairment in patients with Parkinson disease with the G2019S mutation in LRRK2.Neurology 75:2017–2020.

Nguyen HN, Byers B, Cord B, Shcheglovitov A, Byrne J, Gujar P, Kee K, Schüle B,Dolmetsch RE, and Langston W, et al. (2011) LRRK2 mutant iPSC-derived DAneurons demonstrate increased susceptibility to oxidative stress. Cell Stem Cell 8:267–280.

Paisán-Ruíz C, Jain S, Evans EW, Gilks WP, Simón J, van der Brug M, López deMunain A, Aparicio S, Gil AM, and Khan N, et al. (2004) Cloning of the genecontaining mutations that cause PARK8-linked Parkinson’s disease. Neuron 44:595–600.

Ramsden N, Perrin J, Ren Z, Lee BD, Zinn N, Dawson VL, Tam D, Bova M, Lang M,and Drewes G, et al. (2011) Chemoproteomics-based design of potent LRRK2-selective lead compounds that attenuate Parkinson’s disease-related toxicity inhuman neurons. ACS Chem Biol 6:1021–1028.

Reinhardt P, Schmid B, Burbulla LF, Schöndorf DC, Wagner L, Glatza M, Höing S,Hargus G, Heck SA, and Dhingra A, et al. (2013) Genetic correction of a LRRK2mutation in human iPSCs links parkinsonian neurodegeneration to ERK-dependent changes in gene expression. Cell Stem Cell 12:354–367.

Reith AD, Bamborough P, Jandu K, Andreotti D, Mensah L, Dossang P, Choi HG,Deng X, Zhang J, and Alessi DR, et al. (2012) GSK2578215A; a potent and highlyselective 2-arylmethyloxy-5-substitutent-N-arylbenzamide LRRK2 kinase in-hibitor. Bioorg Med Chem Lett 22:5625–5629.

Saha S, Guillily MD, Ferree A, Lanceta J, Chan D, Ghosh J, Hsu CH, Segal L,Raghavan K, and Matsumoto K, et al. (2009) LRRK2 modulates vulnerabilityto mitochondrial dysfunction in Caenorhabditis elegans. J Neurosci 29:9210–9218.

Sanders LH, Laganière J, Cooper O, Mak SK, Vu BJ, Huang YA, Paschon DE,Vangipuram M, Sundararajan R, and Urnov FD, et al. (2014) LRRK2 muta-tions cause mitochondrial DNA damage in iPSC-derived neural cells fromParkinson’s disease patients: reversal by gene correction. Neurobiol Dis 62:381–386.

Sheng Z, Zhang S, Bustos D, Kleinheinz T, Le Pichon CE, Dominguez SL, SolanoyHO, Drummond J, Zhang X, and Ding X, et al. (2012) Ser1292 autophosphorylationis an indicator of LRRK2 kinase activity and contributes to the cellular effects ofPD mutations. Sci Transl Med 4:164ra161.

Tong Y, Giaime E, Yamaguchi H, Ichimura T, Liu Y, Si H, Cai H, Bonventre JV,and Shen J (2012) Loss of leucine-rich repeat kinase 2 causes age-dependent bi-phasic alterations of the autophagy pathway. Mol Neurodegener 7:2.

408 Fell et al.

at ASPE

T Journals on M

ay 13, 2020jpet.aspetjournals.org

Dow

nloaded from

West AB, Moore DJ, Biskup S, Bugayenko A, Smith WW, Ross CA, Dawson VL,and Dawson TM (2005) Parkinson’s disease-associated mutations in leucine-richrepeat kinase 2 augment kinase activity. Proc Natl Acad Sci USA 102:16842–16847.

Yue M, Hinkle KM, Davies P, Trushina E, Fiesel FC, Christenson TA, Schroeder AS,Zhang L, Bowles E, and Behrouz B, et al. (2015) Progressive dopaminergic alter-ations and mitochondrial abnormalities in LRRK2 G2019S knock-in mice. Neuro-biol Dis 78:172–195.

Zhang J, Deng X, Choi HG, Alessi DR, and Gray NS (2012) Characterization ofTAE684 as a potent LRRK2 kinase inhibitor. Bioorg Med Chem Lett 22:1864–1869.

Zimprich A, Biskup S, Leitner P, Lichtner P, Farrer M, Lincoln S, Kachergus J,Hulihan M, Uitti RJ, and Calne DB, et al. (2004) Mutations in LRRK2 causeautosomal-dominant parkinsonism with pleomorphic pathology. Neuron 44:601–607.

Address correspondence to: Dr. Matthew J. Fell, Neuroscience EarlyDiscovery, Merck Research Laboratories, 33 Avenue Louis Pasteur, Boston,MA, 02115. matthew.fell@merck.com

MLi-2, a Potent and Selective Inhibitor of LRRK2 409

at ASPE

T Journals on M

ay 13, 2020jpet.aspetjournals.org

Dow

nloaded from