Discovery of Brain-Penetrant Glucosylceramide Synthase Inhibitors with a Novel Pharmacophore.

Inhibition of glucosylceramide synthase (GCS) is a major therapeutic strategy for Gaucher's disease and has been suggested as a potential target for treating Parkinson's disease. Herein, we report the discovery of novel brain-penetrant GCS inhibitors. Assessment of the structure-activity relationship revealed a unique pharmacophore in this series. The lipophilic ortho-substituent of aromatic ring A and the appropriate directionality of aromatic ring B were key for potency. Optimization of the absorption, distribution, metabolism, elimination, toxicity (ADMETox) profile resulted in the discovery of T-036, a potent GCS inhibitor in vivo. Pharmacophore-based scaffold hopping was performed to mitigate safety concerns associated with T-036. The ring opening of T-036 resulted in another potent GCS inhibitor with a lower toxicological risk, T-690, which reduced glucosylceramide in a dose-dependent manner in the plasma and cortex of mice. Finally, we discuss the structural aspects of the compounds that impart a unique inhibition mode and lower the cardiovascular risk.


■ INTRODUCTION
Glucosylceramide synthase (GCS), known to catalyze glucosylceramide (GlcCer) synthesis from ceramide (Cer) and uridine diphosphate-glucose (UDP-glucose), is a key enzyme in glycolipid synthesis, as GlcCer is the starting point for the synthesis of other glycolipids. 1 GCS inhibition halts GlcCer production and potentially reduces downstream glycosphingolipids, such as glucosylsphingosine (GlcSph), which is directly generated from GlcCer via de-acylation. 2 Accumulation of specific glycosphingolipids is the key feature of most lysosomal storage disorders. 3 Gaucher's disease (GD) is one of the most prevalent lysosomal storage disorders attributed to mutations in β-glucocerebrosidase (GCase) gene. Clinical features of GD include anemia, thrombocytopenia, enlargement of the liver and spleen, and bone dysplasia, and there are cases exhibiting neurological symptoms such as seizures, cognitive impairment, ataxia, and lack of coordination, categorized as GD type 2 or 3. 4 Dysfunction of GCase leads to pathological accumulation of its substrate, GlcCer and GlcSph, in various tissues, including the brain. GlcSph is considered a reliable clinical biomarker for the progression of GD. 5,6 Reduction of these glycosphingolipids is the core therapeutic strategy for GD, and systemic administration of recombinant glucocerebrosidase (enzyme replacement therapy) or GCS inhibition (substrate reduction therapy) can effectively manage peripheral symptoms of GD. 7,8 However, current treatments lack efficacy in combating neuropathic symptoms of GD. 9 Development of a brain-penetrant GCS inhibitor would greatly benefit GD patients, potentially offering therapeutic options for neuropathic symptoms of GD.
Mutations in the GCase gene are known to be a major risk factor for developing α-synucleinopathies such as Parkinson's disease and dementia with Lewy bodies; 10−12 however, the underlying molecular mechanism is not well understood. In vitro and in vivo studies have shown that impairment of GCase causes α-synuclein accumulation, possibly due to an abnormal glycosphingolipid environment. 2,13−15 GlcCer and GlcSph were shown to promote α-synuclein aggregation in vitro. 2 Therefore, reducing GlcCer and GlcSph by employing GCS inhibitors could be a disease-modifying therapeutic strategy for α-synucleinopathies. Figure 1 presents representative GCS inhibitors. Both miglustat and eliglustat are used to treat GD and are effective in treating peripheral symptoms. 7,8 Exelixis reported another potent GCS inhibitor (EXEL-0346). 16,17 Recently, the brainpenetrant GCS inhibitor venglustat, developed by Sanofi Genzyme, underwent clinical assessment for several indications, including neuropathic GD and Parkinson's disease. 18 Using a GD mouse model, venglustat reportedly reduced GlcCer, GlcSph, and proteinase K-resistant α-synuclein in the brain. 19,20 Regarding Parkinson's disease indication, Sanofi Genzyme recently announced that a clinical trial of venglustat for the treatment of Parkinson's disease carrying a GBA mutation had missed its primary endpoint. However, it is desirable to validate the mechanism of action by multiple chemotypes. Very recently, another brain-penetrant GCS inhibitor (BZ1) was reported by Merck. 21 Eliglustat, EXEL-0346, venglustat, BZ1, potent GCS inhibitors, possess a common structural feature: an aliphatic amine and lipophilic moiety (cationic amphiphilic drug [CAD]). The structure−activity relationship (SAR) of EXEL-0346 analogs indicates the importance of the basic amine moiety for potency in these series. 16 In general, CAD demonstrates safety liabilities such as modulation of ion channel activity, potentially increasing the risk of cardiovascular (CV) adverse effects. 22−24 Given the interest in GCS inhibitors as a therapeutic target for GD and Parkinson's disease, we explored a brain-penetrant GCS inhibitor without a basic amine that generates a cationic structure to mitigate CV risks. In addition to these compounds, we have also recently reported the detailed biological evaluation of our GCS inhibitor, T-036. 25 Herein, we examined the SAR of our novel GCS inhibitors, subsequently identifying T-036. We observed that our library compound 1 exhibited GCS inhibitory activity ( Figure 2). Notably, the structure of compound 1 was distinct from previously reported GCS inhibitors, as it lacked a basic amine moiety. SAR study revealed that coplanarity of aromatic ring B with the central core, and the presence of lipophilic substituents at the ortho-position of aromatic ring A and para-position of ring B, are key for potency. In addition, we performed pharmacophore-based scaffold hopping of T-036 to reduce safety-associated risks, which led to the discovery of T-690. Finally, we demonstrate that the novel pharmacophore of  these two compounds is reflected in its unique biological profile in comparison with other GCS inhibitors.

■ RESULTS AND DISCUSSION
Identification of Key Pharmacophore and Discovery of Potent Inhibitor T-036. To identify novel compounds that decrease levels of GlcCer, we conducted a highthroughput screening campaign using a cellular assay, which measured the decrease of GlcCer in fibroblasts obtained from patients with GD. Accordingly, we identified hit compound 1 and confirmed through target identification study that compound 1 inhibits GCS enzyme. Next, to identify the key pharmacophore of our novel GCS inhibitor 1, a ligand-based SAR study was conducted. We began by varying the central core to various bicyclic cores (Table 1). First, we removed the nitrogen atom at the 7-position on the 7-aza-isoindolinone core to determine its effect on potency. Isoindolinone core (2) showed improved potency, although compound 2 demon-strated significantly reduced solubility owing to enhanced lipophilicity. Then, we introduced a nitrogen atom at another position on the core to reduce lipophilicity. The introduction of a nitrogen atom at the 5-position (3) and 6-position (4) was well tolerated in terms of GCS inhibitory potency, and compound 3 showed significantly improved solubility consistent with reduced lipophilicity. Next, compounds with different substitution patterns and ring sizes were examined. ortho-Methoxybenzene (ring A) attached at the 5-position (5) resulted in a significant loss of potency. Replacement of a 6-5membered core with a 5-5-membered ring (6) slightly reduced potency, indicating that a minor change in the orientation of phenyl substituents (ring A and ring B) is acceptable. The carbonyl moiety on the core is not essential for potency, as the 2,3-dihydro-1H-pyrrolo [3,4-c]pyridine core afforded comparable potency ( Figure S1), although the structure was chemically unstable. As compound 3 demonstrated strong potency, reduced lipophilicity, and improved solubility, we selected this compound for further optimization. We then conducted SAR studies around the left (ring A) and the bottom (ring B) aryl groups of compound 3 (Table 2). Considering the left side (ring A), ortho-substitution on the benzene ring was necessary for potency, since no substitution (7) or meta-substitution (8) resulted in a loss of potency. Regarding the bottom substituent (ring B), an aromatic ring was necessary as the replacement of the phenyl ring with cyclohexyl group (9) resulted in a loss of potency. We then evaluated the effect of the substitution pattern on ring B to the potency. ortho-Methyl substitution resulted in a loss of potency (10), indicating that the planar dihedral angle of the core and ring B is another key factor responsible for potency. Along with aromaticity, aromatic ring B on the same plane as the central core can be suggested as a key pharmacophore in this series. Regarding the substitution effect at the meta and para positions of ring B, para-substitution enhanced potency (12), while meta-substitution (11) resulted in potency comparable with no substitution (3). We next focused on improving metabolic stability by blocking the potential metabolic sites of compound 12. The metabolically labile benzylic methyl on ring B was replaced with a metabolically stable fluorine atom (13). Compound 13 exhibited improved metabolic stability, although its GCS inhibitory activity was reduced. Blocking the para-position of both ring A and ring B by fluorine (14) further improved the metabolic stability. Finally, replacing the methoxy substituent on ring A with a trifluoroethoxy group resulted in a metabolically stable compound 15, which showed significantly improved potency as well. Numerous orthosubstituents on ring A were screened, however, the fluorinated alkoxy group conferred robust potency compared with other substituents (data not shown). 26,27 The potent and metabolically stable GCS inhibitor 15 exhibited moderate brain penetration (K puu,brain = 0.20; see Table S1 for detailed results); however, 15 showed human ether-a-go-go-related gene (hERG) inhibition (51.4% inhibition at 10 μM) and poor solubility (<0.15 μg/mL). To overcome these issues, various polar substituents were introduced. SAR studies revealed that methyl substitution at the para-position of ring B (12) substantially enhanced potency. Accordingly, we investigated whether this region could be utilized to optimize the absorption, distribution, metabolism, elimination, toxicity (ADMETox) profile (Table   3). First, we introduced a polar moiety directly connected to ring B (16,17), resulting in a significant reduction of potency. As lipophilic methyl substitution (12) enhanced potency, we introduced substituents with polar and lipophilic moieties. As expected, a nitrile substituent with a dimethyl moiety (18) enhanced potency, although hERG inhibition was not improved. Conversely, introducing a tertiary alcohol moiety (T-036) significantly improved the hERG inhibition profile while maintaining strong potency. In addition, T-036 also showed significantly improved solubility (5.2 μg/mL).
Although T-036 has a hydrogen bonding donor, efflux by multidrug resistance-1 (MDR1) was moderate and acceptable brain penetration was expected, whereas compounds with HBA moiety (16 and 17) showed significant MDR1 efflux in accordance with increased polarity. Given the good profile of T-036, other substituents containing a tertiary alcohol moiety were investigated. We observed that a tertiary alcohol group with a methyleneoxy linker (19) also improved hERG inhibition.
In summary, the SAR study revealed key pharmacophores for GCS inhibitory activity of this series: an aromatic ring A attached at the 4-position of the core, lipophilic orthosubstituent present on ring A, aromatic ring B on the same plane as the core, and lipophilic para-substituent on ring B ( Figure 2). Concerning ortho-substitution of ring A, the fluorinated alkoxy moiety specifically enhanced potency. We identified the tertiary alcohol group as the key functional group Table 3. Improvement of ADMETox Profile to Discover T-036 a for improving the ADMETox profile, leading to the discovery of T-036.
The pharmacological, physicochemical, and pharmacokinetic parameters of T-036 are summarized in Tables 5 and 6. T-036 potently inhibited both human GCS (IC 50 = 0.031 μM) and mouse GCS (IC 50 = 0.051 μM). In addition, T-036 potently reduced the GCS product, GlcCer, in the fibroblasts of patients with GD (EC 50 = 0.0076 μM). Regarding the in vitro ADMETox profile, T-036 showed good metabolic stability and lacked strong CYP inhibitory activity. We have previously  reported the detailed in vivo profile of T-036. 25 In brief, T-036 has good oral exposure (BA = 67%) and moderate brain penetration (K puu,brain = 0.11; see Table S1 for detailed results). Administration of a single dose of T-036 reduced GlcCer in the plasma and cerebral cortex of wild-type mice, and administration of T-036 for 2 months significantly reduced GlcSph in the cerebral cortex of the GD mouse model.
Given the significant in vivo activity of T-036, we conducted a preliminary short-term toxicity study to identify the potential risks associated with the generated compound. Oral administration of T-036 at 30 or 100 mg/(kg day) for 3 days reduced body weight (see Table S2 for detailed results), whereas no loss in body weight was observed in mice during the pharmacodynamic (PD) study.
Scaffold Expansion to Discover T-690 with Lower Toxicological Risk. Although the rationale for body weight reduction was unclear, we speculated whether this finding could be attributed to off-target effects. Off-target activity of T-036 was measured for 47 targets and found that T-036 generally has a clean off-target profile where IC 50 values for 44 out of 47 targets were >10 μM. A notable off-target activity of T-036 was measured against serotonin transporter (SERT), exhibiting an IC 50 value of 0.31 μM (Table S3 for detailed results). SERT inhibition reportedly induces hypophagia and reduces body weight in rats. 28 Although scant evidence indicates that body weight loss observed in short-term toxicity studies can be attributed to SERT inhibition, we used SERT inhibitory activity as an indicator to determine whether the new compound exhibits a different off-target profile.
As a risk mitigation strategy, we focused our attention to vary the core structure of T-036 in aim to alter its off-target profile ( Figure 3). Based on the pharmacophores identified above, we postulated that the core ring placed the substituents in the correct direction that was essential for GCS inhibition. Therefore, we hypothesized that the core structure could be replaced without forfeiting the GCS inhibitory activity if the directionality of the key pharmacophores is maintained. We considered the replacement of a bicyclic core with a monocyclic core. As planarity between the core structure and ring B is needed for potency, we envisioned utilizing internal hydrogen bonding between the core ring and amide linker to afford a planar scaffold (23). To predict the stable conformation, in silico conformational analysis of T-036 and monocyclic core compounds (20 and 23) were conducted ( Figure S2). For the compound with internal hydrogen bonding (23), the core ring and ring B were on the same plane in the most stable conformation, which is markedly similar to T-036. On the other hand, a twisted conformation between the core ring and ring B was predicted for the compound without internal hydrogen bonding (20). Based on this analysis, compounds having a monocyclic core scaffold with intramolecular hydrogen bonding to implement the crucial factors required for potency were examined.
Results of scaffold hopping are presented in Table 4. We began by replacing the bicyclic core of T-036 with a phenyl ring and amide moiety (20) as a reference compound without internal hydrogen bonding which resulted in a significant decrease in potency. Then, we introduced HBA on the core ring to induce internal hydrogen bonding. As expected, replacing the benzene core with a pyridine core (21) significantly improved potency. Introducing a fluorine atom on the benzene core (22) or replacing the benzene core with a pyridone core (23) further improved potency, indicating that internal hydrogen bonding to generate the pseudo-sixmembered ring was superior to the pseudo-five-membered ring for potency. These results demonstrate the success of our pharmacophore-based scaffold hopping strategy. Regarding the brain penetrability, despite its additional HBD and increased polar surface area, pyridone 23 did not show a large increase in the MDR1 efflux ratio compared with T-036, presumably due to masking of HBD by internal hydrogen bonding. As pyridone 23 was attractive owing to its strong potency and low lipophilicity (Log D = 2.12), we further investigated the core structure containing a carbonyl moiety to improve brain penetration. Accordingly, we observed that pyridazin-3-one core (T-690) exhibited strong potency as that of a pyridone (23). Furthermore, MDR 1 efflux of T-690 was significantly improved compared with that of 23, consistent with the reduced polarity (Log D = 2.42). For this series of compounds, para-fluorine substitution on ring A was not essential for good metabolic stability (MLM CL int = 1 μL/(min mg protein) for T-690), and para-fluorinated compound (24) showed reduced potency. Therefore, we selected T-690 to further investigate its profile.
First, we conducted off-target screening to determine whether the off-target profile of T-690 differed from that of T-036. We confirmed that T-690 has no SERT inhibitory activity (IC 50 > 10 μM), which is the representative off-target activity of T-036 (Table S3 for detailed results). This finding indicates that we successfully obtained a novel chemical series with different off-target profiles.
The in vitro pharmacological, physicochemical, and pharmacokinetic parameters of T-690 are summarized in Table 5. Regarding GCS inhibitory activity, T-690 demon- strated species difference, eliciting an IC 50 value of 0.19 μM for the mouse GCS enzyme, which is 12.7 times weaker than human enzyme. Potent compounds in monocyclic core series (22−24 and T-690) generally showed species difference, while most of the derivatives with bicyclic core exhibited less than 10 times difference of potency toward human and mouse GCS (Table S4 for detailed results). Based on the SAR information ( Figure S4), we assume that the protein structure around ring B might be different between the two species. Scaffold hopping from bicyclic core to monocyclic core slightly changed the relative position of the key pharmacophore, which affected inhibitory activity toward mouse GCS while having a negligible effect on human GCS inhibitory activity. Next, we determined the potency of T-690 in a cellular assay by measuring the reduction of GlcCer levels and found that the compound exhibited strong potency (EC 50 = 0.0044 μM). In addition, we have confirmed that T-690 did not affect GCase activity (EC 50 > 300 μM, Table S4 for detailed results) as GlcCer is degraded by GCase. Considering the ADMETox profile, T-690 showed no strong CYP inhibition and good in vitro metabolic stability. Lastly, in vivo PK profile of T-690 in mice was obtained ( Table 6). T-690 showed good oral exposure (BA = 31%) when administered with a dosage of 5 mg/kg. Notably, brain penetration of T-690 was significantly improved (K puu,brain = 0.26) compared with T-036 (K puu,brain = 0.11). T-690 revealed good brain exposure (C u,brain = 0.21 μM at 30 mg/kg dosing, 1 h), comparable with that of T-036 (C u,brain = 0.24 μM at 30 mg/kg dosing, 1 h), while peripheral exposure to T-690 was lower than that of T-036.
As good peripheral and brain exposure of T-690 was confirmed, a short-term toxicity study with rats was conducted by oral administration of T-690 with the dose of 30, 100, or 300 mg/(kg day) for 3 days (Table S2 for detailed results). The AUC 24 of T-690 at 300 mg/(kg day) was comparable with that of T-036 at 100 mg/(kg day). To our satisfaction, T-690 showed no reduction in body weight at any dose examined, indicating that T-690 has a safer off-target toxicology profile than T-036 in terms of body weight loss. However, the causal relationship between SERT inhibition and body weight loss remains elusive.
Biological Profiles Reflecting Unique Pharmacophore of T-036 and T-690. Following the discovery of T-036 and T-690 as GCS inhibitors with novel pharmacophores, we investigated how the structural uniqueness of these two compounds impacted their biological profiles. We previously reported the inhibition mode of T-036 to be noncompetitive against UDP-glucose. Other potent GCS inhibitors, such as eliglustat and venglustat, demonstrate uncompetitive type inhibition against UDP-glucose. 25 Herein, we investigated the inhibition mode of T-690 to determine whether this UDPglucose noncompetitive inhibition mode is typical of our GCS inhibitors. We measured the GCS inhibitory activity of compounds at different concentrations of GCS substrates, C8-ceramide (1 or 70 μM) and UDP-glucose (0.1 or 200 μM) ( Figure 4). The GCS inhibitory activity of T-690 was not significantly altered regardless of C8-ceramide or UDP-glucose concentration. This result indicates that T-690 exhibits noncompetitive type inhibition with C8-ceramide and UDPglucose. In contrast, the GCS inhibitory activity of eliglustat at low concentration of UDP-glucose (0.1 μM) was significantly weaker than that at a high UDP-glucose concentration (200 μM), indicating uncompetitive type inhibition. Overall, these results revealed that noncompetitive type GCS inhibition is a general feature of the developed GCS inhibitor series. Furthermore, the unique inhibition mode indicated that the binding site of our GCS inhibitors is not the same as previously reported GCS inhibitors.
The noncationic structure is a characteristic feature of the developed GCS inhibitors and is deemed beneficial to avoid potential CV risks. Accordingly, the CV risk of our GCS inhibitors was evaluated to confirm whether the noncationic ICR mouse, n = 3; iv: 0.5 mg/kg, po: 5 mg/kg. Major parameters, including plasma clearance (CL), volume of distribution at steady state (Vdss), mean residence time (MRT), maximal concentration (C max ), area under the curve (AUC), and oral bioavailability (BA), have been reported. b Fraction of unbound drug determined from equilibrium dialysis. c Concentration of unbound drug in brain determined from brain exposures in ICR mice 60 min after 30 mg/ kg oral dosing; n = 3. d Ratio of C u,brain to C u,plasma determined from plasma and brain exposures in ICR mice 60 min after 30 mg/kg oral dosing; n = 3. structure affords a good safety profile (Table 7). First, the inhibitory activity of ion channels was evaluated in relation to CV adverse effects. At 30 μM, both T-036 and T-690 did not potently inhibit hERG, Ca V 1.2, and Na V 1.5 channels. Next, we conducted a cardiac proarrhythmia assay using human-induced pluripotent stem cell-derived cardiomyocytes. In this assay, the calcium transient duration was measured as a predictor for QT prolongation. 29,30 Exposure of cardiomyocytes to T-036 at 30 μM or T-690 at 10 μM did not affect calcium transient duration. These results indicated the low CV risks associated with our GCS inhibitors.
In Vivo Pharmacodynamic Assessment of T-690. To confirm the in vivo activity of T-690, we measured the level of the GCS product, GlcCer, in C57BL/6J mice 6 h after administering vehicle (Veh) or T-690 (30, 100, and 300 mg/ kg) ( Figure 5). We observed that T-690 reduced GlcCer concentrations in the plasma and cerebral cortex in a dosedependent manner. The effective dosage for mice was high due to moderate inhibitory activity for mouse GCS and fast clearance of T-690 in mice. However, considering that T-690 shows 12.7 times stronger potency against human GCS than mouse GCS, this result indicates the clinical potential of T-690.
The synthesis of 5 is shown in Scheme 2. Bromination of 32 and following cyclization using ammonia gave isoindolinone intermediate (34). Suzuki−Miyaura coupling with 29a and subsequent Buchwald−Hartwig cross-coupling with 36 furnished 5.
To collect basic SAR around the bottom (ring B) side, we developed a synthetic route that enables the introduction of a variety of ring B moieties at the last step (Scheme 4). The synthesis began with the amidation of 42 with aniline (27a or 43) to give 44. Lithiation of the pyridine ring followed by nucleophilic addition to DMF and spontaneous cyclization resulted in bicyclic intermediate (45). Suzuki−Miyaura coupling with 29a and subsequent hydrolysis provided 47. A variety of amine (48 or 43) were introduced as a ring B moiety by reductive amination to afford 9−13. Compounds 14 and 15 were synthesized by reduction of 45b and following Suzuki− Miyaura coupling with 50 and 30, respectively.
The exploration of the substitution on ring B was facilitated by utilizing a synthetic route similar to the synthesis of 5. Ring B with a variety of substitutions was introduced by Buchwald− Hartwig cross-coupling of 52 with corresponding aryl bromides (53) at the last step to yield 16−18 and T-036 (Scheme 5).
Synthetic routes for monocyclic core compounds are outlined in Schemes 6−8. Compounds 20−22 were synthesized by Suzuki−Miyaura coupling of 56 with 55 followed by hydrolysis and amidation with 59 (Scheme 6).   (30,100, and 300 mg/kg), n = 5. The plasma and the cerebral cortex of the mice were harvested 6 h after administration. Relative amount of GlcCer in the plasma and cortex compared with vehicle. The results for the most abundant GlcCer (C24:1 GlcCer for plasma and C18:0 GlcCer for cortex) are presented. Data are presented as scatter plots and the mean ± standard error of the mean (SEM). *p < 0.025 and **p < 0.01 in two-tailed Williams' test.
The pyridone core was constructed by the condensation of 60 with aniline (61) (Scheme 7). Subsequent hydrolysis and amidation afforded 23.
Scheme 2. Synthesis of 5 a a Reagents and conditions: (a) NBS, AIBN, CF 3 Ph, 80°C, 98%; (b) ammonia, MeOH, 40°C, 62%; (c) 29a, Pd(dppf)Cl 2 ·CH 2 Cl 2 , Cs 2 CO 3 , DME, water, 80°C, 59%; (d) 36, Pd 2 (dba) 3 , XANTPHOS, Cs 2 CO 3 , DMF, 100°C, 38%. Scheme 3. Synthesis of 6 a a safety concerns found for T-036. We utilized internal hydrogen bonding to maintain key pharmacophores in an appropriate orientation, which led to the discovery of T-690, another potent GCS inhibitor with distinct off-target profiles and enhanced brain penetration. In vivo studies of T-690 demonstrated dose-dependent GlcCer reduction activity in the cortex of mouse. The structural uniqueness of our GCS inhibitor series is well reflected in the characteristic UDPglucose noncompetitive inhibition mode, indicating that the binding site of our compounds is not the same with previously reported GCS inhibitors. The noncationic structure of this series resulted in desirable safety profiles for CV risk measurement. These features of our developed novel GCS inhibitors could potentially offer therapeutic options for substrate reduction therapy in neuropathic GD, as well as a disease-modifying therapy for Parkinson's disease, with suppressed risk of CV adverse effects.
RapidFire Mass Spectrometry Data Processing. For Rapid-Fire mass analyses, enzyme reaction solution was aspirated from the quenched assay plates and loaded onto a C18 solid-phase extraction (SPE) cartridge (catalog # G9203-80105, Agilent Technologies) with a 70:30 (v/v) methanol−water solution containing 5 mM ammonium formate and 0.2% HCOOH for 2000 ms. The analytes were then eluted into the mass spectrometer using a 95:5 (v/v) methanol−water solution containing 5 mM ammonium formate and 0.2% HCOOH for 5000 ms. C8-glycosylceramide and C18-d35 ceramide were detected using a multiple reaction monitoring (MRM) method with Q1/Q3 transitions at m/z 588.5 to 246.5 and m/z 762.9 to 265.5, respectively, on Sciex API4000 triple quadrupole mass spectrometer (Applied Biosystems) in the positive electrospray ionization mode. C18-d35 ceramide was used as the internal standard for data analysis. The extracted ion chromatograms for each transition were integrated and processed using the RapidFire Integrator software. The data for each well was normalized by monitoring product conversion with (product)/(internal standard) ratio. The compound activities were determined by plotting the percent inhibition data through normalization from 0% (DMSO only) and 100% inhibition (no-enzyme control) and modeling a four-parameter logistic fit to obtain IC 50 values. As for cell-based glucosylceramide lowering assay, the cell lysate sample was loaded onto a C4 cartridge (catalog # G9203A, Agilent Technologies). C16-glucosylceramid, C16-ceramide, and C18-d35 ceramide were detected using MRM with Q1/Q3 transitions at m/z 700.8 to 264.2, m/z 538.7 to 264.3, and m/z 762.9 to 265.5, respectively. GlcCer and Cer represent values of C16-glucosylceramide and C16-ceramide, respectively, divided by the internal standard. The control group was exposed to DMSO instead of test compounds. The reduction of glucosylceramide against the control group was evaluated as EC 50 values using percent of control for each data point by curve-fitting to a four-parameter logistic model. All curve fittings were performed with GraphPad Prism 6.07 for Windows.
Cell-Based Glucosylceramide Lowering Assay. Using GBA mutant human fibroblasts containing mutated glucosylceramidase, the effect of test compounds on glucosylceramide reduction was evaluated. GBA mutant human fibroblasts GM008760 (Coriel Institute) were seeded in a 96-well multiwell plate at a density of 5000 cells/100 μL culture solution/well. GlutaMAX (Thermo Fisher Scientific), penicillin−streptomycin, and minimum essential media (Thermo Fisher Scientific) supplemented with 15% fetal bovine serum were used as a culture solution. After culturing for 2 days, the compounds were tested in triplicate using a 10-fold, six-point serial dilution. After culturing the cells for 4 days in the presence of the test compound, the culture solution was removed, and the cells adhered to the 96-well plate were washed with a phosphate buffer (Thermo Fisher Scientific, product number 14190-144). A solvent composed of 50% ethanol and 50% isopropanol was added to the 96-well plate by 100 μL, and the intracellular lipids were dissolved by pipetting. A solution prepared by dissolving N-octadecanoyl-D 35 -psychosine (50 ng/mL, Matreya LLC, Inc.) as an internal standard in 50% ethanol and 50% isopropanol was added to the sample in an equal volume, and 2.5 μL of the centrifuged supernatant was used as the test compound group. In the same manner as above, a control group was prepared using DMSO instead of the DMSO solution of the test compound. For the test compound group and the control group, molar concentration of glucosylceramide [GlcCer (C16:0)] was quantified using Rapidfire-MS/MS (API-5000, Turbo-ESI, SRM), and the glucosylceramide lowering effect of the test compound group relative to the control group was evaluated.
In Vivo PD Experiments. Animal: Male C57BL/6J mice (6 weeks old) were purchased from CLEA Japan, Inc. (Tokyo, Japan). The animals were reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) of Takeda Pharmaceutical Company Limited, accredited by Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC). T-690 was reconstituted with a 0.5% (w/v) methylcellulose solution (Fujifilm Wako, Osaka, Japan) and was orally administered to 8-week-old male C57BL/6J mice at the indicated doses (n = 5). The mice were sacrificed at 6 h after the administration to collect their cerebral cortex and plasma. The samples were frozen and stored at −80°C until analysis.
Enzymatic assay, pharmacokinetics, toxicity study, offtarget profiling, conformation analysis, in vitro and in vivo study protocols, and HPLC traces (PDF) Molecular formula strings (CSV)

■ AUTHOR INFORMATION
Corresponding Authors