Discovery of 1H-Pyrazole Biaryl Sulfonamides as Novel G2019S-LRRK2 Kinase Inhibitors

Abstract

G2019S (GS) is the most prevalent mutation in the leucine rich repeat protein kinase 2 gene (LRRK2), a genetic predisposition that is common for Parkinson’s disease, as well as for some forms of cancer, and is a shared risk allele for Crohn’s disease. GS-LRRK2 has a hyperactive kinase, and although numerous drug discovery programs have targeted LRRK2 kinase, few have reached clinical development. We report the discovery and preliminary development of an entirely novel structural class of potent and selective GS-LRRK2 kinase inhibitors: biaryl-1H-pyrazoles.

Autosomal dominant, missense mutations in the leucine rich repeat protein kinase 2 (LRRK2) gene are the most common genetic predisposition to Parkinson’s disease (PD).¹LRRK2 mutations account for approximately 1–5% of familial and sporadic PD.² The most common LRRK2 mutation leads to a serine substitution of Gly2019 (GS) in the kinase domain,³ which increases kinase activity 2–4-fold.⁴ Following its discovery as a common genetic cause of PD, GS-LRRK2 carriers have been shown repeatedly to also be at increased risk for hormone-related cancers, especially breast cancer in women.^5,6 Other mutations in LRRK2 have been shown to increase the risk of Crohn’s disease.⁷ As a highly validated therapeutic target for PD, multiple drug discovery programs have yielded potent LRRK2 inhibitors with some currently in clinical trials. To date, advanced LRRK2 kinase inhibitors inhibit both wild-type (WT) LRRK2 and GS-LRRK2, and many have shown significant unwanted side effects. Such side effects have not been definitively proven to arise from a lack of LRRK2 variant selectivity; however, selectively inhibiting the pathogenic G2019S-LRRK2 mutant kinase over WT would provide a more precise therapeutic, which may help alleviate such effects by sparing otherwise essential LRRK2 cellular functions.

We recently reported a series of highly potent, selective, and brain-penetrant GS-LRRK2 inhibitors utilizing a 5-amino substituted indazole pharmacophore.⁸ The lead 5-amino indazole (2, Figure 1)⁸ possessed single-digit nanomolar in vitro potency for full-length GS-LRRK2 and >2000-fold mutant selectivity in human cells; dose-dependent GS-LRRK2-selective engagement in the brain also was observed in a G2019S knock-in mouse-model following I.P. administration. Indazoles are common kinase inhibitor pharmacophores, with >70 examples in the LRRK2 field alone,⁹⁻¹² most notably MLi-2 (1, Figure 1)⁹ developed by Merck. On the basis of our work developing GS-LRRK2 selectivity in an indazole pharmacophore (2), we hypothesized that truncation to a 1H-pyrazole could provide a second structural series which maintains essential hinge-binding interactions present in the indazole core but with a lower molecular weight. A 1H-pyrazole could also enable access to novel chemical space directed at GS-LRRK2 selectivity, specifically targeting molecular interactions revealed in the development of 2. Here, we report the discovery and synthesis of substituted 1H-pyrazoles as potent and selective GS-LRRK2 kinase inhibitors.

Figure 1.

Chemical structure representations of the nonselective LRRK2 kinase inhibitor MLi-2 (1), indazole 38 (2) from our previous work, and our subsequent hypothesis for a novel LRRK2 kinase inhibitor scaffold, biaryl 1H-pyrazoles.

Our discovery of indazole 2 (Figure 1) highlighted that a hydrogen bond acceptor (HBA) motif attached to the phenyl ring of the tetrahydronapthalene provided a significant boost in potency (>10-fold) and selectivity (10-fold) toward GS-LRRK2.⁸ Our structural model of the LRRK2 kinase active site, together with inhibition data, suggest that this nitrile makes a hydrogen bonding interaction with a lysine residue (Lys132) present at the boundary of the active site. Thus, we hoped also to incorporate this interaction in our new 1H-pyrazole series.

To test our hypothesis, we designed and synthesized a series of molecules containing 3,4-substituted biaryl 1H-pyrazoles containing various HBA motifs (Figure 1). Out of our initial screen, we identified a set of pyrazoles bearing substituted phenylsulfonamides at the R₃ position that possessed some encouraging GS-LRRK2 inhibition potency (Table 1). Moreover, some of these phenylsulfonamides also possessed modest inhibition in our cell-based assay. Pyrazoles 3–5 came out of our initial search as the most potent hits. As reported with our indazole series,⁸ exchange of the R₂ 3-pyridyl moiety for a phenyl resulted in a loss of potency (Table 1, 6).

Table 1. Initial Discovery of Substituted 1H-Pyrazoles As Novel, GS-LRRK2 Kinase Inhibitors^a.

^aAll compounds could be readily synthesized through facile, microwave-assisted coupling reactions from common aryl halide intermediates, as shown in Schemes S1 and S2 in the Synthetic Chemistry section of the Supporting Information. In vitro biochemical assay (Adapta). Data presented as mean IC₅₀ (average of two independent determinations). The average coefficient of variation was less than 10% for the kinase inhibition assays conducted with compounds tested in duplicate at 10 different concentrations (ranging from 5 nM to 10 μM) using WT and GS-LRRK2 kinase assays. MLi-2 (1) was used as a nonselective reference throughout (see also Figure 3)

^bCellular assay results represent the arithmetic mean of a minimum of three determinations, and these assays generally produced results within 10% of the reported means.

^cExact EC₅₀ values were not determined for those reported as >100 μM.

^dMaximal percent activity remaining at 30 μM inhibitor is reported

We sought to improve the intrinsic potency of our newly identified phenylsulfonamide hits with further exploration around the 3-pyridyl (R₂), R₃, and pyrazole core motifs (R₁); GS-LRRK2 inhibition data for such compounds are presented in Table 2. We found that incorporation of a methyl group at the R₁ position improved potency, e.g., 8 (IC₅₀ = 15 nM), a direct analogue of 5 (IC₅₀ = 50 nM). Furthermore, this modification also improved cellular potency 7 (EC₅₀ = 2.1 μM), a direct analogue of 3 (EC₅₀ = 5.1 μM). Intriguingly, pyrazole 9 yielded a reduced biochemical potency compared to the original hit (4 vs 9), and we hypothesized that this may be due to an unfavorable conformation of the compound in the active site. Furthermore, we observed that pyrazoles bearing the 2,3-dimethyl phenylsulfonamide fragment that also possess an R₁ methyl group were not tolerant of R₂-pyridine modifications (entries 10–12). This was initially surprising, as we and others have reported that R₂-pyridine modifications can improve LRRK2 kinase binding potency significantly.^8,9,12 In fact, 12 was found to be a partial inhibitor of GS-LRRK2 in cells. At this stage, we conducted computational minimization and docking to a homology model of the GS-LRRK2 kinase-active site⁸ using Glide within Schrödinger’s Maestro platform to rationalize our data and enable further structure-based drug design.

Table 2. Further Development of Pyrazol-Biaryl Sulfonamides As Novel GS-LRRK2 Kinase Inhibitors^a.

^aSee footnotes for Table 1.

Docking studies with 3, 4, 7, and 9 suggested that incorporation of a methyl substituent at the 5-position of the pyrazole core reinforced a preferred conformation of the “central” phenyl ring, residing perpendicular to both the pyrazole and substituted phenylsulfonamide (Figure 2). Otherwise, pyrazoles presented in Figure 2 docked in an expected manner; the 1H-pyrazole core is predicted to participate in hydrogen bonding interactions with residues Leu86 and Ala87 while also making a T-shaped edge-to-face π interaction¹³ with Phe139. The R₂-pyridyl moiety extends out toward the solvent-accessible region, analogous to that observed with indazole-based LRRK2 inhibitors.^8,14 Moreover, modeling suggested that the 2,3-dimethyl moiety of 4 makes lipophilic interactions with the glycine-rich loop (Figure 2C) that may be compromised in 9 (Figure 2D) as a result of the 5-methyl group. In support of the docking outcomes shown in Figure 2, removal of the 5-methyl group of 12 to give 13 restored potency in this scaffold. In summary, our modeling and inhibition data at this stage suggested that an unfavored, induced conformation can arise from incorporation of a 5-methyl substituent on substituted pyrazole scaffolds with highly substituted phenyl fragments at the R₃ position. Modifications to the R₃ region such as moving the methyl of 8 to give 14 and moving the sulfonamide of 9 to give 15, produced no significant improvements in potency (Table 1). We previously noted that selectivity and potency toward GS-LRRK can be enhanced through HBA interaction with Lys132. Indeed, a conserved interaction was predicted between sulfonamide oxygens and Lys132, in both 3- and 4-subtituted phenylsulfonamides at R₃. The sulfonamide moiety also was predicted to interact as a hydrogen bond donor (HBD) through its sulfonamide N–H bond to either Asp148 in the case of 3-sulfonamides 3 and 7 or Asp130 for 4-sulfonamides 4 and 9.

Figure 2.

Ball and stick and chemical structure representations of pyrazoles 3 (A), 7 (B), 4 (C), and 9 (D) following Glide docking to a GS-LRRK2 active site model based on a crystal structure of CHK1 (PDB ID: 5OPB). Addition of a methyl group in 3 (A) to give 7 (B) does not affect binding conformation; however, the same modification to 4 (C) to give 9 (D) is predicted to induce an unfavored binding conformation.

In light of our discoveries thus far, we envisioned that sulfonamide 7 may be more favorable toward further investigation, and our efforts in this campaign are presented in Table 2. We initially focused on modifications in the R₂ region and observed significant boosts in potency with modifications to the pyridine group, in contrast to analogous modifications to 9 presented in Table 1. As observed with our indazole series,⁸ the addition of a methyl group adjacent to the R₂-pyridine nitrogen (16) gave a 3-fold boost in GS-LRRK2 kinase inhibition in IC₅₀ and cellular assays compared to 7. Interestingly, this boost in potency was observed with (16) or without (17) the presence of a methyl at R₁. We hoped to explore the R₁ region; however, we discontinued these efforts given that CF₃ analogue 18 significantly reduced potency. Incorporation of a morpholine group made 19 and 20 very potent GS-LRRK2 inhibitors in both IC₅₀ and cellular assays. Methoxy pyridyl derivatives 21 and 22 did not improve potency, while dimethyl substituted pyridines 23 and 24 gave comparable inhibition as 16 and 17.

We next focused our attention on the “central” ring (R₃, Table 2) that links the pyrazole core to the benzenesulfonamide (R₄) moiety. Modeling suggested structural modifications could be accommodated in this region of the active site. Indeed, incorporation of a methyl group at the R₃ 2-position to give 25 gave a 3.5-fold boost in cellular potency compared to unmodified analogue 7; however, pyrazole 26 highlighted that modifications to the R₃ phenyl ring may be regiospecific. We attempted to exchange the sulfonamide of 25 with a primary amide (27) or methylsulfonyl (28) motif; however, none were tolerated. Movement of the sulfonamide to the para position (29) did not improve potency, although the same introduction of an ethyl group in the R₃ region (30) gave a 10-fold boost in potency. Interestingly, removal of the R₁ methyl substituent of 30 to give 31 gave a 3-fold reduction in potency. We hypothesized that introduction of a pyrazole R₁ methyl group combined with “central” R₃ phenyl modifications haS an additive effect on inhibition potency, potentially via further conformational restriction. Less lipophilic modifications at the R₃ 2-position such as −CF₃ (32), −CN (33), and −OMe (34) gave no improvement in potency, although −F (35) gave appreciable inhibition. 2,6-Difluoro (36) and 2-cyclopropyl (37) analogues were comparably potent inhibitors to 30, although 2-isopropyl analogue 38 did not improve potency. Furthermore, 2,6-dimethyl (39) and 2,3-difluoro (40) analogues did not improve potency. Consistent with prior observations, we were able to improve the intrinsic potency of pyrazole 37 with incorporation of a methyl group at the R₂ pyridine to give 41.

We determined THE activity of the 1H-pyrazole series as novel LRRK2 kinase inhibitors by relying on well-established in vitro activity assays that use full length enzyme and human cell-based ELISAs (enzyme-linked immunosorbent assay) of pSer935,^15,16 as we did with our indazole series⁸ (Tables 1–3), Figure 3A and B. To support our ELISA assays, we immunoblotted lysates from HEK293 cells expressing either GFP-tagged (green fluorescent protein) wild-type (WT)-LRRK2, GS-LRRK2, or GS+A2016T-LRRK2 to evaluate the impact of pyrazoles 37 and 41 on kinase activity in these three versions of LRRK2. The HEK293 cells expressing THE GS+A2016T-LRRK2 double mutation validate target engagement because the A2016T mutation hinders compound access to the active site and thus can be used to distinguish direct target engagement from indirect effects on phosphorylation.¹⁷ We observed good correlation between pSer935 ELISA and pSer935 immunoblots comparing WT-LRRK2 and GS-LRRK2 and that GS+A2016T-LRRK2 was refractory to inhibition by both 37 and 41 up to 10 μM (Figure 3C and D). We next evaluated 37 and 41 inhibition of endogenous LRRK2 signaling to Rab10 and its phosphorylation at pSer935 in WT or homozygous GS-LRRK2 knock-in MEFs (mouse embryonic fibroblasts). We observed inhibition and selectivity for GS-LRRK2 phosphorylation of Rab10 Thr73 and LRRK2 Ser935 at levels similar to our ELISA results (Figure 3E and F). These results show entry into cells (although we did not formally determine solubility), moderate potency, and GS-LRRK2 selectivity with direct engagement of the active site for both 37 and 41.

Table 3. In Vitro ADME Profile Comparisons of Select 1H-Pyrazole-based GS-LRRK2 Inhibitors^a.

entry	GS IC50 / nMa	IC50 ratio	WT EC50 / μMa	GS EC50 / μMa	EC50 ratio	Caco-2		efflux Ratio	MLMc	HLMc	TPSAd/ Å2	kinase profilef
						A-Bb	B-Ab
13	15.5	16.3	76	1.5	51	7.14	30.4	4.3	7.4	28	114.2	n.t.
7	21.3	4.8	>100	2.1	48	10.3	38.9	3.8	28	140	101.7	n.t.
19	1.1	7.5	>100	0.29	345	4.21	40.6	9.6	1.6	57	114.2	n.t.
30	2.2	8.9	>100	0.26	385	4.27	40.5	9.5	19	119	101.7	n.t.
37	3.1	8.3	>100	0.39	256	8.78	27.0	3.1	30	159	101.7	12 (6)
41	2.4	7.1	5.4e	0.28	19.3	3.36	25.7	7.6	9.9	102	101.7	12 (6)

^aSee footnotes for Table 1. “n.t.” = not tested.

^bP_app (× 1 e^–6 cm/s).

^ct_1/2 (min).

^dCalculated at pH 7.4.

^e34% activity remaining at 30 μM.

^fNumber of kinases inhibited at >80% out of a total of 485 kinases (0.1 μM inhibitor). Kinases inhibited other than LRRK2 variants in parentheses (see Figures S1, S2).

Figure 3.

LRRK2 inhibitor screening. (A,B) ELISA screening of G2019S selective compounds. HEK293 cells expressing wildtype GFP-LRRK2 (WT) or GFP-G2019S-LRRK2 (GS) were plated in 48-well plates. Twenty-four hours after plating, cells were treated in 3-fold serial dilutions for 24 h with compounds 37 (n = 5) or 41 (n = 6) or MLi-2 (1) as a reference compound (n = 7). Cells were lysed in situ and analyzed by ELISA for LRRK2 pSer935. Percent activity (pSer935) remaining was calculated for each compound and EC₅₀ values determined by curve fitting in Prism, as reported in Tables 1–3. (C,D) Plasmids encoding GFP-WT-LRRK2, GFP-GS-LRRK2, and GFP-A2016T+G2019S-LRRK2 were transfected into HEK293 cells and treated with 37 or 41 at the indicated concentrations for 24 h. Lysates were probed by immunoblot against total LRRK2, LRRK2 pSer935, and tubulin (upper panel, N = 3). Levels of pSer935 were corrected for the amount of total LRRK2 (lower panel). (E,F) Confluent dishes of WT- or GS-LRRK2 mouse embryonic fibroblasts (MEFs) were treated with 37 or 41 at the indicated concentrations for 24 h. Cell lysates were probed by immunoblot for LRRK2 pSer935 (UDD2), total LRRK2 (N241), pRab10 pThr72, total Rab10 (4E10), and tubulin (upper panel). Levels of pSer935 and pRab10 were corrected for their respective total proteins, n = 3 (lower panel).

We conducted a preliminary in vitro ADME evaluation of key compounds from our initial work, Table 3. Generally, we found that the 1H-pyrazoles possessed excellent human liver microsomal (HLM) stability; however, mouse liver microsomes (MLMs) gave a shorter metabolic half-life (t_1/2). As expected for molecules possessing highly polar functional groups,¹⁸ the phenylsulfonamide-1H-pyrazoles had high efflux ratios (>3) in our caco-2 permeability assay, suggestive of active transport. Kinase selectivity profiling showed good selectivity toward LRRK2 variants: out of 482 kinases screened, only six other kinases were inhibited >80% (Table 3 and Figures S1, S2). Our in vitro ADME analysis combined with calculated TPSA suggested that these molecules would not penetrate the blood–brain barrier. Some of the substituted pyrazoles showed selectivity toward kinase inhibition of GS-LRRK2 over WT-LRRK2 (19, 30, 37, and to some extent 41) that ranged from about 20-fold to just under 400-fold. We previously observed a discrepancy between cell-based and IC₅₀ assays.⁸ We have found that our cell-based assays and subsequent in vivo evaluations effectively identified selective inhibition of GS-LRRK2 over WT, while our in vitro biochemical assay more effectively identified potent inhibitors but was less effective in detecting selectivity toward GS-LRRK2. We and others have reported this phenomenon previously,^8,12 and although it requires significant investigation beyond the scope of this manuscript, we hypothesize that this difference may be due to protein architectural changes or organelle distribution for GS-LRRK2 in cells versus in vitro. In combination, utilization of both assays enabled efficient identification and development of selective GS-LRRK2 inhibitors. A clear increase in selectivity toward GS-LRRK2 in our cellular assay was evident when comparing analogues based on 4 (dimethyl para-benzenesulfonamides), such as 13, to analogues based on 3 (meta-benzenesulfonamides), such as 19, 30, and 37. As described above, docking studies suggested that a significant conformational induction arises from the introduction of a methyl group at the pyrazole 5-position. Furthermore, modeling predicted significant van der Waals interactions between the kinase glycine-rich loop and lipophilic substitutions at the 2-position of the “central” ring (see Table 2), such as in 30 and 37. Similar interactions were attributed to the potency and selectivity of indazole 2.⁸

We selected 37 for preliminary pharmacokinetic (PK) profiling in CD-1 fasted mice because of its in vitro and cell-based potency against GS-LRRK2 (Table 4). Both oral and IV dosing (2 and 0.5 mg/kg, respectively) provided high plasma concentrations (C_max ∼ 250 ng/mL), acceptable half-lives, and high bioavailability. Expectedly, 37 was not detected in the mouse brain, which was also reflected in the low volume of the distribution factor (V_dss = 2), suggesting 37 is largely restricted to plasma.¹⁹ Taken together, alongside a low observed clearance, our preliminary PK analysis suggested that the 1H-pyrazoles disclosed here present as excellent starting points for further drug development.

Table 4. Preliminary Pharmacokinetic Parameters of 37 CD-1 Mouse Plasma^a.

	37
route of administration	IVb	POc
dose (mg/kg)	0.5	2
AUCinf [μM h]	0.58	2.55
Cmax [μM]	0.58	0.58
Tmax [h]	n/a	1.00
T1/2 [h]	0.8	1.25
Vdss [L/kg]d	2.0	n/a
MRT [h]	1.01	3.01
Cl [mL/min/kg]	33.5	n/a
F [%]e		111%

^aData presented as mean values. Three animals per arm, three animals per time point.

^bDosed as a solution of the free base in 10% DMSO/90% 2-hydroxypropyl-beta-cyclodextrin (HPCD; 20% w/v).

^cDosed as a solution of the free base in DMSO: methylcellulose (0.5% w/v).

^dVolume of distribution (V_d) calculated with the steady-state method.

^eCalculation for F [%] is calculated relative to IV and assumes clearance does not change from that calculated for IV. Thus, calculated values >100 can arise where compound adsorption and/or elimination is dramatically different following PO versus IV.

We previously described the design and discovery of a highly potent and GS-LRRK2-selective compound, 5-amino indazole 2.⁸ Upon the basis of the knowledge we gained developing this series, we hypothesized that selective and potent inhibitors of GS-LRRK2 kinase could be attained by reducing the indazole to a 1H-pyrazole core, linked with a 1,1-biphenyl ring to a hydrogen bond acceptor motif. These efforts led to the initial discovery of 1H-pyrazole biaryl-sulfonamide 3 that warranted further development. Taking a data-driven approach, we were able to generate several potent GS-LRRK2 inhibitors, some of which possessed impressive selectivity for GS- over WT-LRRK2. Furthermore, we observed that these compounds have excellent stability in human liver microsomes (t_1/2 = 159 min in some cases) and good kinome selectivity; 37 and 41 inhibited only 6 of 485 kinases at >80%, presenting an excellent foundation for further compound development. As a representative of our newly discovered series, 37 was selected for pharmacokinetic profiling and showed excellent bioavailability, plasma concentration, and mean retention time; these preliminary PK data suggest that compounds in this class are largely restricted to plasma.

Although initially identified as an inherited cause of PD, the missense mutation that produces GS-LRRK2 has subsequently been linked to increased risk for hormone-driven cancers, especially breast cancer in women. In addition, different missense mutations in the region of LRRK2 that encode the kinase domain also increase the risk for Crohn’s disease. Thus, there are important brain and peripheral organ targets for LRRK2 inhibitors that likely will require an ensemble of agents with different biochemical and pharmacologic properties to be matched to an individual patient’s needs. We present a substituted series of 1H-pyrazoles as potent LRRK2 kinase inhibitors with varying selectivity for GS-LRRK2 as well as acceptable in vitro and in vivo pharmacokinetic properties. We will continue to evolve this series of 1H-pyrazoles toward both brain-penetrant and nonpenetrant, highly selective GS-LRRK2 kinase inhibitors as potential precision medicines for the subset of people with PD and/or cancer who are carriers of the G2019S LRRK2 mutation.

Glossary

Abbreviations

ELISA

enzyme-linked immunosorbent assay

GFP

green fluorescent protein

MEF

mouse embryonic fibroblast

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.2c00116.

• Kinase profiles, ¹H NMR and HPLC-MS spectra, synthetic chemistry, and experimental materials (PDF)

The authors declare no competing financial interest.