Chemical Biology of Leucine-Rich Repeat Kinase 2 (LRRK2

Apr 27, 2015 - Expert Opinion on Therapeutic Targets 2017 21 (8), 751-753 ... Synthesis of [ 11 C]HG-10-102-01 as a new potential PET agent for imagin...
2 downloads 10 Views 1MB Size
Subscriber access provided by NEW YORK UNIV

Perspective

The Chemical Biology of Leucine-Rich Repeat Kinase 2 (LRRK2) Inhibitors Anthony A Estrada, and Zachary K Sweeney J. Med. Chem., Just Accepted Manuscript • Publication Date (Web): 27 Apr 2015 Downloaded from http://pubs.acs.org on April 27, 2015

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Journal of Medicinal Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 43

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Medicinal Chemistry

The Chemical Biology of Leucine-Rich Repeat Kinase 2 (LRRK2) Inhibitors Anthony A. Estrada*,† and Zachary K. Sweeney* †

Department of Discovery Chemistry, Genentech, Inc., 1 DNA Way, South San Francisco, California 94080, United States

RECEIVED DATE (to be automatically inserted after your manuscript is accepted if required according to the journal that you are submitting your paper to)

ACS Paragon Plus Environment

Journal of Medicinal Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 43

Abstract: There is an urgent need for the development of Parkinson’s disease (PD) treatments that can slow disease progression. The Leucine-Rich Repeat Kinase 2 (LRRK2) protein has been genetically and functionally linked to PD, and modulation of LRRK2 enzymatic activity has been proposed as a novel therapeutic strategy. In this Perspective, we describe small molecules that have been used to inhibit LRRK2 activity in vitro or in vivo. These compounds are important tools for understanding the cellular biology of LRRK2 and for evaluating the potential of LRRK2 inhibitors as disease-modifying PD therapies.

Introduction Parkinson’s disease is a neurological disorder defined by a characteristic tremor, rigidity, and slowing of movement in afflicted patients.1 These cardinal symptoms are often accompanied by disabling insomnia, constipation, and depression. Postmortem examination of the brains of PD patients usually reveals pronounced degeneration of dopaminergic neurons in the substantia nigra and the presence of insoluble protein complexes.2 Approved medicines for the treatment of PD, including dopamine agonists and dopamine precursors, help to reduce the movement difficulties in the early stages of the disease.3 However, progressive neurodegeneration ultimately reduces the effectiveness of these symptomatic therapies, and there is a significant medical need for medicines that stop the progressive accumulation of disability. Advances in genomics have had a dramatic effect on our understanding of disease origin.4 Many human diseases, including neurological disorders that were historically considered to be caused by environmental effects, are now known to have a significant genetic component. Contemporary drug discovery and development efforts can focus on genetically validated targets and well-defined clinical populations, and this approach has contributed to the discovery of several innovative medicines for the treatment of cancer.5 The application of a genetically-based approach to the discovery of new medicines has also been employed by scientists studying PD.6 PD was not considered to have a significant heritable component ACS Paragon Plus Environment

Page 3 of 43

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Medicinal Chemistry

until disease risk was linked to mutations in the gene encoding the alpha-synuclein protein.7 Subsequent studies identified alpha-synuclein as a major component of the protein deposits commonly found in the brain of PD patients, and provided one of the first clear genetic links to neurodegenerative disease. Many other genetic variants, including several distinct sequences encoding the LRRK2 protein, have now been linked to an increased risk for PD.8

LRRK2 Discovery, Structure and Function The association between LRRK2 and neurodegeneration was first identified in studies of several families that presented with an autosomal dominant, late onset form of PD.9,10,11 Since this finding, seven variants

(LRRK2[G2019S],

LRRK2[I2020T],

LRRK2[S1761R],

LRRK2[Y1699C],

LRRK2[R1441C/G], and LRRK2[N1437H]) have a strong enough association with disease that they are considered to be pathogenic (Figure 1).12 In addition, the G2385R variant increases the risk of developing PD in Asian populations. PD patients with disease-linked LRRK2 mutations have an average age of disease onset, clinical presentation, and pathologies nearly indistinguishable from those of patients with idiopathic disease. Large genome-wide association studies have also identified LRRK2 as a risk factor for the development of PD, suggesting that modification of the normal function of LRRK2 may be a broadly useful approach to the treatment of this disease.6 The LRRK2 gene sequence encodes a large (2,527–amino acid) protein that is strongly expressed in many tissues, including the brain, the spleen, the lung, the kidney, and immune cells.13,14 This protein sequence contains several protein-protein interaction domains in addition to enzymatically active GTPase and kinase domains (Figure 1).15 Most of the pathogenic LRRK2 variants are located in either the enzymatically active serine-threonine kinase domain or the GTPase domain of the protein, suggesting that the genomic changes associated with increased disease risk might alter the catalytic properties of the protein. There is no published structure of LRRK2 or any of the individual protein domains. However, studies of related kinases and GTPases indicate that the LRRK2[I2020T] and LRRK2[G2019S] variants likely increase kinase activity by stabilizing the active conformation of the ACS Paragon Plus Environment

Journal of Medicinal Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 43

activation loop.16,17,18,19,20 GTP-binding in the LRRK2 GTPase domain and protein dimerization also influence LRRK2 kinase activity.21 The LRRK2[R1441C] and LRRK2[Y1699C] pathogenic variants have reduced GTPase activity.

These residue changes may impact LRRK2 kinase activity by

modulating LRRK2 dimerization kinetics or by altering interactions between the ROC and COR domains. The functional relationship between GTP binding and kinase activity is underscored by the lack of kinase activity of the LRRK2[T1348N] mutant, which is located in the ROC domain and has a greatly reduced ability to bind GTP.22 In total, while the structural biology of LRRK2 is complex, the weight of evidence suggests that LRRK2 pathogenic variants generally increase kinase activity.23 Modulation of kinase function with small molecules might be achieved by interfering either with ATPbinding in the kinase domain, GTP binding and hydrolysis in the GTPase domain, or the LRRK2 dimerization equilibrium.

pS910 pS935 pS955 pS973

LRRK2/ARM ANK

R1441C R1441G S1761R G2019S N1437H Y1699C I2020T pS1292

LRR

G2385R

pT1491

ROC

COR

pT2483

Kinase

WD40

Figure 1. Schematic of LRRK2 protein structure with reported PD pathogenic familial mutations (red), risk factors (black), and cellular phosphorylation (blue) and autophosphorylation (green) sites. The development of in vitro LRRK2 kinase assays was important for the identification of selective inhibitors, and also provided initial confirmation that disease linked mutations influenced LRRK2 kinase activity.16,24,25,26,27 This finding was consistent with the autosomal dominant inheritance pattern of LRRK2-linked PD, and supported the hypothesis that kinase inhibition could serve as a therapeutic intervention.24,28,29,30 Subsequent optimization of systems for LRRK2 expression in bacteria, and conditions for assaying kinase activity in high-throughput assays enabled the identification of LRRK2 inhibitors from large compound collections utilizing biochemical screening assays.31

ACS Paragon Plus Environment

Page 5 of 43

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Medicinal Chemistry

Early cellular studies of LRRK2 function focused on apoptosis or reduction in neurite length endpoints resulting from overexpression of LRRK2 or LRRK2 variants.32,33 Exhaustive cellular investigations have indicated that LRRK2 may be involved in cell death processes, synaptic vesicle formation, cytoskeleton assembly, and/or autophagy.17 Several selective and non-selective kinase inhibitors were found to protect neurons from the consequences of LRRK2 overexpression, but these studies were generally unable to provide a useful biomarker of kinase inhibitor activity or a convincing mechanistic link between LRRK2 cellular function and neurodegeneration.34

As PD has been

genetically and pathologically linked to chronic neuroinflammation, LRRK2 in vitro models that interrogate the LRRK2-dependence of inflammatory signalling processes have been developed.35 LRRK2 inhibitors, when incubated at high concentration, have been shown to modulate the reaction of microglial cells to stimulation.36 Many studies searching for a cellular biomarker of LRRK2 enzymatic function attempted to identify changes in cellular phosphorylation states that could be associated with differences in LRRK2 kinase activity. The most useful cellular markers of LRRK2 kinase function currently appear to be the phosphorylation status of LRRK2 at S910, S935, or S1292.23,37 While phosphorylation of S910 and S935 is generally responsive to inhibition of LRRK2 kinase activity, several mutations that increase kinase activity decrease phosphorylation at these sites.38 These findings led to the development of a model in which LRRK2-mediated phosphorylation of an unknown kinase impacts, in turn, phosphorylation of S910 and S935.

Alternatively, it has been suggested that binding of ATP-

competitive inhibitors to LRRK2 might induce a conformational change that accelerates dephosphorylation at S910 or S935.37 Recent studies have also supported the use of the LRRK2[pS1292] phosphorylation status as a biomarker of LRRK2 kinase activity.23,37 Five LRRK2 variants associated with familial PD and located either in the GTPase domain or the kinase domain (LRRK2[R1441G], LRRK2[Y1699C], LRRK2[G2019S], LRRK2[I2020T], and LRRK2[N1437H]) displayed elevated autophosphorylation activity. LRRK2[T1348N], which does not bind GTP, did not autophosphorylate at S1292 in cells. ACS Paragon Plus Environment

Journal of Medicinal Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 43

Studies using transgenic mice have also confirmed that levels of phosphorylated protein were higher in animals expressing LRRK2[G2019S] protein than in transgenic animals expressing equivalent amounts of LRRK2[wild-type] protein. Unfortunately, the LRRK2[pS1292] phosphorylation/dephosphorylation equilibrium heavily favors the dephosphorylated species, and there is only a small amount of LRRK2 phosphorylated at pS1292 present in most cellular and in vivo systems. Therefore, while LRRK2[pS1292] appears to be a direct marker of LRRK2 kinase activity, currently available antibodies are not sensitive enough to permit assessment of LRRK2 kinase activity in wild-type animal tissue using this biomarker. The more easily detected LRRK2[pS910] and LRRK2[pS935] markers are therefore more broadly applicable technologies for confirming LRRK2 inhibition in cellular models or in animal tissue. The historical association between environmental neurotoxin exposure and PD led to the development of animal models that involved the degredation of dopaminergic neurons following toxin administration.39 However, these systems do not reflect critical aspects of PD pathophysiology, and several molecules found to be neuroprotective in toxin PD models were not efficacious in clinical studies.40,41 The discovery of the genetic association between LRRK2 and PD has led to numerous attempts to develop robust, LRRK2-driven in vivo models of PD.42,43 Demonstration of the ability of an LRRK2 kinase inhibitor to modify PD-related pathology in these animals would provide strong support for the potential clinical utility of these agents.

Although initial reports indicated that LRRK2-

transgenic mice had age-dependent movement deficits or exhibited diminished dopamine levels in the striatum, the frequency of these deficits appears to be quite variable, and most phenotypes have not been confirmed.44 LRRK2[G2019S] knock-in mice and other transgenic animals present with a hyperkinetic phenotype starting at a young age, presumably due to differences in dopamine handling in midbrain neurons. This phenotype has been modified through the administration of selective kinase inhibitors.45 LRRK2 is broadly expressed, and the protein has been associated with several critical cellular pathways. LRRK2-knockout mice and rats have been used to assess the potential safety risks associated ACS Paragon Plus Environment

Page 7 of 43

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Medicinal Chemistry

with interfering with LRRK2 catalytic function. LRRK2 deficient mice display morphological abnormalities in both kidney and lung tissue. These findings may be associated with impairments in lysosomal regulation or the lysosomal-autophagy pathway.46,47,48,49,50 Changes in the kidney associated with significantly reduced levels of LRRK2 are also observed in mice expressing a LRRK2 kinase-dead protein.49 LRRK2 knockout rats also exhibit an abnormal kidney, lung, and liver phenotype.51,52 As described in the following section, these findings have stimulated researchers to determine if similar effects are observed following administration of small molecule kinase inhibitors. LRRK2 Small Molecule Inhibitors In the last several years, small molecule inhibitors have emerged as useful tools for interrogating the function of LRRK2. Biochemical assays permit the selectivity of these kinase inhibitors to be estimated, and autophosphorylation biomarkers can be used to assess the ability of these compounds to inhibit LRRK2 kinase activity in vitro and in vivo. Ideally, an LRRK2 kinase inhibitor would demonstrate good selectivity for inhibition of LRRK2 relative to other kinases and be able to inhibit LRRK2 kinase function in the brain and other organs following oral administration to animals.

The

unbound concentrations of the inhibitor required to inhibit kinase activity in vivo should be related to the biochemical potency of the compound. Similarly, the LRRK2 inhibitor should impact cellular processes at levels that can be interpreted in the context of the LRRK2 cellular inhibitory concentrations. Given their inherent promiscuity, these considerations are particularly important for ATP-competitive kinase inhibitors. Multiple literature34,53,54,55,56,57 and patent reviews31,58,59 documenting the small molecule LRRK2 inhibitor landscape have been published in the last several years. This Perspective will focus primarily on recently reported kinase inhibitors that are highly selective for LRRK2. These structurally diverse tools should enable “chemical genetic studies” that provide additional understanding of the consequences of LRRK2 kinase inhibition.

LRRK2-Selective Small Molecule Kinase Inhibitors with Low Brain Penetration ACS Paragon Plus Environment

Journal of Medicinal Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 43

A significant advancement in the LRRK2 small molecule kinase inhibitor field was achieved in 2011 when selective inhibitors CZC25146 (1) [4/184 kinases < 300 nM using Kinobeads™, 286-fold over LRRK2 IC50] and LRRK2-IN-1 (2) (12/440 kinases < 10% inhibition of control, 1667-fold over LRRK2 IC50) were disclosed in the literature (Table 1; all kinase selectivity data is represented as number of kinases identified/total number of kinases tested, and fold selectivity is defined as testing concentration/ LRRK2 Ki or IC50).60,61 Prior to these reports, LRRK2 small molecule tool compounds had limited utility for examining LRRK2 biology, and the consequences of inhibiting LRRK2 kinase activity due to the potential inhibition of other cellular kinases. Both 1 and 2 possess low nM biochemical LRRK2 potency. Using an in vitro rat model of LRRK2[G2019S]-induced neuronal toxicity (as measured by neurite retraction, overt rounding of cells, and DNA fragementation), Ramsden and Hopf et al. demonstrated the attenuation of cell injury in a concentration-dependent fashion using aryl sulfonamide 1 (rat EC50 ~ 100 nM). Using a different quantification method (computerized algorithm assessing neurite length and average branch point count), attenuation of LRRK2[G2019S]-induced neuronal toxicity in primary human cortical neurons was also reported (human EC50 ~ 4 nM). Tricyclic diaminopyrimidine 2 was employed by Alessi and Gray et al. to examine S910 and S935 phosphorylation levels, and 14-3-3 binding in a variety of cell types. Inhibitor 2 showed a dosedependent dephosphorylation of both serine biomarkers with significant dephosphorylation achieved at 1–3 µM. Additionally, 2 demonstrated in vivo inhibition of S910 and S935 phosphorylation in mouse kidney after ip dosing (100 mg kg-1). Moehle et al. also used tool compound 2 to investigate the in vitro amelioration of TLR4-induced pro-inflammatory signaling in microglia.36 While compound 2 has been utilized by a number of groups in experiments designed to assess the impact of kinase activity on cellular processes,62 it has been recognized that the promiscuity of 2 may complicate the interpretation of results generated with this inhibitor. For example, 2 reduced the induction of LRRK2 expression by IFN-γ, although other LRRK2 kinase inhibitors did not produce this effect. Ultimately it was proposed that inhibition of ERK5 by 2 mediates the expression of LRRK2

ACS Paragon Plus Environment

Page 9 of 43

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Medicinal Chemistry

induced by the cytokine.63,64 Another publication attributes the antiproliferative effects of 2 in cancer cells to inhibition of DCLK1.65 Finally, an evaluation of the changes in protein phosphorylation caused by 2 in neuronal cultures from wild-type and LRRK2 knockout mice has demonstrated that this inhibitor can impact a variety of signaling pathways in an LRRK2-independent fashion.66 As these results illustrate, it is nearly impossible to achieve complete specificity for a single kinase target using ATPcompetitive compounds. Thus, the use of multiple structurally distinct LRRK2 inhibitors at appropriate concentrations according to their selectivity in biochemical and cellular assays should be strongly encouraged.

Table 1. Physicochemical Property, Activity, Selectivity, and in Vivo PK Profiles of Reported LRRK2 Inhibitors 1–4. O

O N

F

N N H

O

N

N H

N

N

N

1 (CZC-25146)

O

N

HN

S O O N

N

N H

O

N

3 (TAE684)

N H

N

N

N N

Cl

N

N H

O

2 (LRRK2-IN-1)

N N

N

N

O N S

N

O

O S O

4 (Pfizer) LRRK2 IC50a LLE (nM)

compd

CNS MPO Score

MW

cLogP

TPSA (Å2)

160

2.4

489

3.5

117

7

261

4.4

571

2.9

97

367

1.8

614

5.8

470

5.6

368

2.7

Kinase Selectivity

mouse Cl (mL min-1 kg-1)b

Total B/Pc

4.7

4/184 50% inhibition, 345-fold over LRRK2 Ki).

Figure 2. Docking model of 5 (cyan) in LRRK2 (green with L1949 in yellow). The side chain of Y931 of JAK2 is also shown (magenta). Intermolecular hydrogen bond interactions are shown in yellow dashed lines. Following the initial disclosure, Choi et al. also published the discovery of 5 (termed HG-10-102-01 in their publication) and confirmed the high degree of kinase selectivity and biochemical and cellular activity.80 Using 2 as a benchmark, the authors demonstrated that compound 5 was able to inhibit phosphorylation of wild-type and G2019S mutant LRRK2 at S910 and S935 in HEK293 cells. Additionally, inhibitor 5 was able to show dose-dependent inhibition of pS910 and pS935 in endogenously expressed LRRK2 from lymphoblastoid cells (G2019S PD patient sample), mouse Swiss ACS Paragon Plus Environment

Journal of Medicinal Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 43

3T3 cells, and mouse embryonic fibroblast cells with similar or better potency than 2 [EC50 ~ 0.3–1.0 µM]. In vivo PK profiling of aminopyrimidine 5 by Genentech (Table 3) revealed a rat plasma clearance approaching liver blood flow, a short iv half-life (0.5 h), and a modest Bu/Pu ratio (0.17). In an effort to improve upon the pharmacological and pharmacokinetic profile of lead compound 5, a thorough lead optimization campaign using structure- and property-based drug design was executed.81 This strategy initially resulted in the identification of G1023 (6, Table 2),81 which featured an identical biochemical lipophilic ligand efficiency (LLE = 6.6),82 a high degree of kinase selectivity (0/178 kinases > 50% inhibition, 60-fold over LRRK2 Ki), a 3-fold improvement in the pS1292 cellular assay, and an improved in vivo rat PK profile (Cl = 24 mL min-1 kg-1, Bu/Pu = 0.5) compared to 5. Inhibitor 6 was subsequently used as an in vivo tool to demonstrate inhibition of in vivo kinase activity in G2019S transgenic mouse brains (pS1292), and reversal of cellular effects of LRRK2 PD mutations in cultured primary hippocampal neurons. The in vivo unbound brain concentration required to effectively reduce pS1292 autophosphorylation in vivo (IC50 = 12 nM) was nearly identical to the in vitro cellular IC50 reported for compound 6 (9 nM). Additionally, upon treatment of LRRK2[G2019S] mouse embryonic hippocampal neurons with 100 nM of inhibitor 6, a statistically significant amelioration of the neurite outgrowth defects associated with the LRRK2[G2019S] transgene was observed.23

Table 3. In Vitro and In Vivo Rat DMPK Profiles of Reported Aminopyrimidine Brain-Penetrant LRRK2 Inhibitors 5–9a

ACS Paragon Plus Environment

Page 15 of 43

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Medicinal Chemistry hep Clhep (mL min-1 kg-1)b h/rc

% rat PPB

Cl (Clu) (mL min-1 kg-1)d

iv t1/2 (h)

F (%)

MDR1e P-gp ERf (B:A/A:B)g

Total B/Ph

Bu/Pui

579-81

4.9/24

92

51 (616)

0.5

64

2.8

0.23

0.17

681

1.8/7.6

86

24 (156)

1.2

80

1.2

0.9

0.5

781

3.2/14

97

8.3 (244)

3.1

40

0.9

0.9

0.5

883

3/25

79

44 (210)

1.3

90

0.8

1.2

0.6

983

1/21

79

26 (261)

1.5

90

0.8

0.7

0.5

compd

a

Compounds 5–9 were dosed po (1 mg kg–1) as an aqueous suspension with 1% methylcellulose, iv (0.5 mg kg–1) as a 60% PEG solution or 20–60% NMP solution for systemic PK, and iv (0.5 mg kg-1) as a 60% NMP solution for brain PK. bIn vitro stability in cryopreserved hepatocytes. ch/r = human/rat. dClu = unbound clearance = total clearance/fup, where fup is the unbound plasma free fraction. eMDCK-MDR1 human P-gp transfected cell line. fEfflux ratio. gBasolateral-to-apical/apicalto-basolateral. hTotal brain/plasma AUC ratio. iUnbound brain/unbound plasma AUC ratio.

A targeted subset of kinases (e.g. kinases that shared a similar ATP binding site sequence make-up with LRRK2) that had not been previously evaluated were screened and revealed TTK (MPS1) as a kinase that was strongly inhibited by pyrimidine 6 (13-fold TTK selectivity index) and structurally related analogs.81 Structure-based drug design efforts using the JAK2-based homology model and TTK co-crystal structures suggested that substitution on the phenyl ring para to the methoxy substituent would produce a steric clash with Asp608 in TTK and increase the TTK/LRRK2 selectivity window. Thus, a few minor structural modifications, including fluorine substitution at the aforementioned phenyl ring position, resulted in the discovery of GNE-7915 (7, Table 2).81 Compound 7 possessed a nearly identical activity and DMPK profile to 6, but with an improved TTK selectivity index (53-fold) and good kinome selectivity at a concentration (0.1 µM) that satisfied a targeted 100-fold biochemical selectivity window [2/449 kinases < 30% inhibition of control (TTK and ALK), 100-fold over LRRK2 Ki, DiscoveRx KINOMEscan™]. The excellent selectivity profile of optimized inhibitor 7 at 0.1 µM was further corroborated in an Invitrogen kinase selectivity panel [1/187 kinases > 50% inhibition (TTK), 100-fold over LRRK2 Ki]. In an effort to increase the structural diversity and selectivity of the brain-penetrant diaminopyrimidine compounds, aniline bioisosteres were explored. Lead optimization efforts ACS Paragon Plus Environment

Journal of Medicinal Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 43

culminated in the identification of aminopyrazoles GNE-0877 and GNE-9605 (8 and 9, Table 2) as highly potent and specific LRRK2 inhibitors [8 = 4/188 kinases > 50% inhibition (Aurora B, RSK2, RSK3, RSK4), 0/188 kinases > 70% inhibition, 145-fold over LRRK2 Ki; 9 = 1/178 kinases > 50% inhibition (TAK1-TAB1), 0/178 kinases > 60% inhibition, 50-fold over LRRK2 Ki] with excellent DMPK profiles (Table 3).83 In vivo non-human primate (NHP) PK profiles of lead aminopyrimidines 7, 8 and 9 are summarized in Table 4. All three compounds demonstrated good in vitro–in vivo correlation, moderate to stable plasma clearance rates, and good iv half-lives. Additionally, CSF/Pu ratios extracted from low-dose PK studies suggested that all three inhibitors possessed approximately equal free brain to free plasma distribution.84,85 Desirable Bu/Pu ratios were later confirmed for compounds 7 and 8 during NHP safety assessments (vide infra).

Table 4. In Vivo NHP PK Profiles of LRRK2 Aminopyrimidines 7, 8, and 9a MDR1d P-gp ERe (B:A/A:B)f

Bu/Pug

CSF/Puh

compd

hep Clhep (mL min-1 kg-1)b NHP

781

14

95

11 (200)

7.7

17

0.9

0.6

1.1

883

19

80

20 (100)

2.2

35

0.8

0.7

1.2

983

13

82

8 (43)

4.0

74

0.8



1.1

% NHP Cl (Clu) PPB (mL min-1 kg-1)c

iv t1/2 F(%) (h) 1 mg kg-1

a

Compounds were dosed po (1 mg kg–1) with crystalline material as an aqueous suspension with 1% methylcellulose and iv (0.5 mg kg–1) as a 20–60% NMP solution. bIn vitro stability in cryopreserved hepatocytes. cClu = unbound clearance = total clearance/fup, where fup is the unbound plasma free fraction. dMDCK-MDR1 human P-gp transfected cell line. eEfflux ratio. f Basolateral-to-apical/apical-to-basolateral. gUnbound brain/unbound plasma AUC ratio. hCSF/unbound plasma AUC ratio.

In the absence of a LRRK2-dependent PD efficacy model, in vivo PD knockdown for compounds 7–9 was assessed through the use of LRRK2 BAC transgenic mice expressing human LRRK2 protein with the G2019S mutation.23 Inhibitors were evaluated for their ability to inhibit in vivo LRRK2 Ser1292 autophosphorylation. Tissue samples were harvested and examined from the hippocampus and spleen.81 Robust concentration-dependent inhibition of S1292 phosphorylation was observed for all inhibitors ACS Paragon Plus Environment

Page 17 of 43

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Medicinal Chemistry

tested in both the brain and peripheral tissues (Figure 3). Using a PD inhibition model, in vivo unbound brain IC50 values of 7 nM, 3 nM, and 20 nM were calculated for 7, 8 and 9 respectively. As was observed with tool compound 6, the unbound in vivo LRRK2[G2019S] IC50 values for 7, 8 and 9 are nearly identical to the in vitro LRRK2[G2019S/R1441G/Y1699C] cellular IC50 determinations.

Figure 3. (reproduced & modified from Estrada 2012 and Estrada 2014)81,83 In vivo G2019S LRRK2 transgenic mouse PK/PD results measuring brain pS1292 autophosphorylation. The circles represent the ACS Paragon Plus Environment

Journal of Medicinal Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 43

observed data, and the lines represent the predicted data from a direct inhibition model. Percent inhibition is normalized to pS1292 levels observed in mice dosed with vehicle alone (n = 3). Plotted data are shown for mice treated with (A) 7 [15 mg kg–1, po, at 1, 3, and 6 h (n = 4/dose); 50 mg kg–1, po, at 1, 3, 6, and 12 h (n = 4/dose) and 24 h (n = 1); 10 mg kg–1, ip, at 1 and 2.5 h (n = 3/dose); 50 mg kg–1, ip, at 1 and 6 h (n = 3/dose)], (B) 8 [10 and 50 mg kg–1, ip, at 1, 3, and 6 h (n = 3/dose)] and (C) 9 [10 and 50 mg kg–1, ip, at 1 and 6 h (n = 3/dose)]. Toxicities observed with Aminopyrimidine LRRK2 Inhibitors. In 2011, Novartis published a report indicating that lung and kidney abnormalities existed in LRRK2 genetic knockout mice.49 In order to assess these potential liabilities, aminopyrimidine compounds were advanced into safety studies. A recent Genentech publication summarizes the toxicity and toxicokinetic profiles of advanced inhibitors 7 and 8 in mice, rats and NHPs.46 Due to the aforementioned low anti-pS1292 antibody sensitivity, in vivo PD target engagement in all of these studies was assessed using pS935. Male C57BL/6 mice were dosed po with 7 (200 and 300 mg kg–1 BID) and 8 (30 and 65 mg kg–1 BID) for 15 days. Toxicokinetic analysis showed dose-dependent increases in plasma and brain levels with average free drug exposures of 5- and 36-fold above pS1292 cellular IC50 values for the higher doses. While evidence of lung, kidney, and brain PD knockdown was observed with both inhibitors, no microscopic effects were observed in the lung or kidney, and both compounds were well tolerated. Similar 7-day repeat-dosing studies were conducted in male and female Sprague-Dawley rats. Oncedaily po administration of 7 (10, 50, and 100 mg kg–1) and 8 (30, 75, and 200 mg kg–1) showed doserelated exposure increases. The highest tolerated doses for compounds 7 (100 mg kg–1) and 8 (30 mg kg– 1

) translated to maximum free drug exposures of 22- and 185-fold over the cellular IC50 of 7 and 8

respectively. No macroscopic or microscopic effects were seen in the lung or kidney with either compound. Lastly, NHPs were dosed orally with 7 (10, 25, and 65 mg kg–1 QD) for 7 days. Toxicokinetic analysis showed a dose-dependent increase in plasma exposures with average free drug levels correlating to 4-, 14-, and 35-fold above the cellular IC50. Terminal Bu/Pu and CSF/Pu AUC ratios of 0.6 and 0.8 were ACS Paragon Plus Environment

Page 19 of 43

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Medicinal Chemistry

achieved respectively showing a high degree of brain penetration. Statistically significant PD inhibition of pS935 was observed at all doses tested. Transient and reversible clinical observations included tremors (≥10 mg kg–1), and hypoactivity and decreased reactivity to stimulus (25 and 65 mg kg–1). The sole anatomic pathology observation linked to compound 7 was limited to the lung and characterized by abnormal cytoplasmic accumulation of secretory lysosome-related organelles known as lamellar bodies in type II pneumocytes of all animals administered 25 and 65 mg kg–1 in both sexes. Comparison of these findings with the published LRRK2 knockout mouse data showed identical lung phenotypes. With the goal of examining potential on-target versus off-target related effects, structurally distinct aminopyrazole 8 (6 and 20 mg kg–1 BID) and anilino-aminopyrimidine 7 (30 mg kg–1 BID) were administered to NHPs in a 29-day repeat-dose study. Significant free drug coverage above cellular IC50 values and excellent brain penetration was achieved with both test articles. Robust PK/PD knockdown of pS935 in PBMCs, brain, kidney, and lung was confirmed at all doses. Upon microscopic evaluation of lung, abnormalities identical to those observed in LRRK2 knockout mice were observed at all doses with both inhibitors. These findings are consistent with an on-target effect of reduction of LRRK2 kinase activity that leads to lamellar body accumulation in type II pneumocytes in the lung of certain species. It should be noted that the morphologic abnormality described for the LRRK2 knockout mouse kidney was absent in both NHP studies.

Table 5. Physicochemical Property, Activity, and Selectivity Profiles of Reported LRRK2 Inhibitors 10, 10b, and 11.

ACS Paragon Plus Environment

Journal of Medicinal Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

N

Page 20 of 43

N

F

N

CN

H N

H N N

NH

O

O

O

N

LLE

pLRRK2 IC50b (nM)

9

3.8

2/449 50% inhibition, 200-fold over LRRK2 IC50). Guided by the hypothesis that an amide functionality in the original HTS hit was linked to the P-gp-mediated efflux, suitable replacements were discovered such as the 3-cyano substitution in 11, which attenuated this liability. This translated to a total mouse B/P ratio of 1.3 for 3-cyanoquinoline 11, and statistically significant reduction of pS935 in the brains of G2019S LRRK2 transgenic mice at 3 h following oral doses of 30 and 100 mg/kg (Figure 4).

ACS Paragon Plus Environment

Journal of Medicinal Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 43

Figure 4. (reproduced & modified from Garofalo 2013)90 In vivo G2019S LRRK2 transgenic mouse PK/PD results with compound 11 measuring brain pSer935 levels. pSer935 levels were determined by ELISA and normalized to total LRRK2. Data are expressed as the mean ratio of pSer935 to total LRRK2 values observed in the vehicle control group. Error bars represent SEM. Plotted data are shown for mice treated with (A) 11 [30 mg kg–1, po, at 3 h (n = 5/dose); 100 mg kg–1, po, at 3 h (n = 5/dose).

Novartis Indolinones. One of the originally identified LRRK2 inhibitor tool compounds was sunitinib, however high kinase promiscuity precluded the use of this compound for reliable model studies of LRRK2 function. Recently, Troxler et al. reported the optimization of sunitinib to indolinones 12 (Table 7) and Nov-LRRK2-11 (13, Table 7) with single-digit nM LRRK2 biochemical activity and modest in vivo pharmacokinetic properties.91 Using an IRAK4-based homology model, the 5-position of the indolinone core was targeted for improving kinase selectivity. This strategy led to the introduction of 5-alkoxy substituents in 12 and 13 that demonstrated improved selectivity profiles over other

ACS Paragon Plus Environment

Page 23 of 43

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Medicinal Chemistry

indolinone-based inhibitors against a small panel of off-target kinases (ALK, KDR, LCK, PDGFRa and RET).

Table 7. Physicochemical Property, Activity, and In Vivo Mouse DMPK Profiles of Reported LRRK2 Inhibitors 12, and 13. O N N H O

O

O N

N

N N H O

O 12 (Novartis)

N H

13 (Nov-LRRK2-11)

N H

compd

CNS MPO Score

MW

cLogP

TPSA (Å2)

LRRK2 IC50a (nM)

LLE

1291

5.1

421

1.6

80

9

6.4

nrd

28

2.2

88

1391

5.5

365

2.4

74

4

6.0

380

50

0.4

57

pLRRK2 iv t1/2 Clb IC50b -1 -1 c (h) (mL min kg ) (nM)

F (%)

a

Biochemical assay; LRRK2 IC50 at 200 µM ATP. bpSer935 cellular assay (NIH3T3 cells). cCompounds 12 and 13 were dosed po (3 mg kg–1), and iv (1 mg kg–1); Clb = Blood clearance. dNot reported.

When dosed in mice, compound 13 reduced LRRK2 protein levels in the kidney, a phenotype also observed in mice expressing a LRRK2 kinase-dead mutant.49 Inhibitor 13, along with the less selective LRRK2-kinase inhibitor H-1152,92,93 was subsequently used by Longo et al. to ameliorate the observed age-dependent hyperkinetic phenotype of LRRK2[G2019S] knockin mice.45 These results suggest that the enhanced kinase activity of the LRRK2[G2019S] protein is responsible for the observed lack of agerelated decline in stepping activity and immobility time that was demonstrated by wild-type littermates. Compounds 13 and H-1152 did not cause any behavioral changes when administered to wild-type or kinase-dead mice. Brain target engagement was confirmed through PD knockdown of S935 in the striatum and the cortex of 12 month old G2019S knockin mice 30 minutes post ip injection of 10 mg/kg of compound 13.45

ACS Paragon Plus Environment

Journal of Medicinal Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 43

Pfizer Pyrrolopyrimidines. ATP-hinge-binding pyrazolopyrimidine ring systems are abundantly exemplified in the LRRK2 patent literature.94,95,96,97,98,99,100 In November of 2014, Pfizer reported the identification of a pyrrolopyrimidine scaffold that provided a highly efficient starting point with favorable CNS properties for lead optimization.101,102 Using MST3 as a crystallographic surrogate for LRRK2 (reported MST3-LRRK2 ATP-binding site residue similarity = 73%) Henderson et al. improved the off-target liabilities of early HTS leads. This led to the discovery of PF-06447475 (14, Table 8) with in vitro LRRK2[wild-type] and LRRK2[G2019S] biochemical IC50 values of 3 and 11 nM respectively, and a pS935 cellular IC50 of 25 nM. Kinase selectivity profiling at 1 µM using the DiscoveRx KINOMEscan™ platform revealed 32/449 kinases with < 30% inhibition of control (91-fold over LRRK2 IC50). Selectivity profiling of pyrrolopyrimidine 14 in a cellular context was performed using the ActivX KiNativ technology™ demonstrating good selectivity in human peripheral blood mononuclear cells at 1 µM and an ActivX LRRK2 cellular IC50 of 15 nM.103 Table 8. Physicochemical Property, Activity, and Selectivity Profile of Reported Pyrrolopyrimidine LRRK2 Inhibitor 14. O CN N 14 (PF-06447475)

N N

N H

compd

CNS MPO Score

MW

cLogP

TPSA (Å2)

LRRK2 IC50a (nM)

LLE

Kinase Selectivity

pLRRK2 IC50b (nM)

14101

5.7

305

2.2

77

11

5.8

32/449