Structure-Activity Relationship Studies of Venglustat on NTMT1 Inhibition

Abstract

The protein N-terminal methyltransferase 1 (NTMT1) is implicated in neurogenesis, retinoblastoma, and cervical cancer. However, its pharmacological potentials have not been elucidated due to the lack of drug-like inhibitors. Here we report the discovery of the first NTMT1 in vivo chemical probe GD433 by structure-guided optimization of our previously reported lead compound venglustat. GD433 (IC₅₀ = 27 ± 1.1 nM) displays improved potency and selectivity than venglustat across biochemical, biophysical, and cellular assays. GD433 also displays good oral bioavailability and can serve as an in vivo chemical probe to dissect the pharmacological roles of Nα methylation. In addition, we also identified a close analogue (YD2160) that is inactive against NTMT1. The active inhibitor and negative control will serve as valuable tools to examine the physiological and pharmacological functions of NTMT1 catalytic activity.

Graphical Abstract

Introduction

Protein methyltransferases are a major class of enzymes that regulate proper protein methylation levels, essential for numerous biological processes including gene transcription, DNA repair, signal transduction, and RNA metabolism.^1–4 Perturbing the activities of protein methyltransferases has been associated with cancers, metabolic, inflammatory, and neurodegenerative disorders. Thus, protein methyltransferases are regarded as a class of novel therapeutic targets with multiple inhibitors in clinical trials.⁵ For instance, Tazemetostat is an FDA-approved EZH2 inhibitor for lymphoma.⁶

Protein N-terminal methyltransferase 1 (NTMT1) is ubiquitously expressed and primarily methylates the Nα-terminal amino group of proteins with a canonical motif X-P-K/R (X represents any amino acid other than D/E).^7–9 Despite of a similar recognition motif, its close homolog NTMT2 is mainly expressed in the heart and muscle tissues.¹⁰ Genetic methods suggest that NTMT1 regulates mitosis, DNA damage repair, neurogenesis, and stem cell maintenance.^11–15 Furthermore, the knockdown of NTMT1 suppresses the proliferation and migration of cervical cancer in both cellular and animal models.¹⁶ In addition, NTMT1 serves as an oncogene for neuroblastoma, as its knockdown suppresses the tumor growth and increases the sensitivity to cisplatin treatment.¹⁷ More recently, NTMT1 has demonstrated an important role in stem cell regulation. Knockout of NTMT1 in mice results in premature aging phenotypes, including neurodegeneration by depleting neuronal stem cell pools.¹⁸ Loss of NTMT1 in C2C12 mouse myoblasts leads to expression of osteogenic markers.¹⁹ However, the detailed mechanisms and pathways regulated by NTMT1 are not yet completely understood. In this regard, highly cell-active and specific inhibitors would be valuable to probe the physiological and pharmacological functions of NTMT1/2.

Different approaches have been applied to discover cell-potent NTMT1/2 inhibitors as chemical probes to dissect their biological function (Figure 1). Among them, bisubstrate inhibitors selectively inhibit recombinant NTMT1 at 140 pM but are impermeable to cell membranes.^20–23 Compared to bisubstrate inhibitors, peptidomimetic inhibitors inhibit both NTMT1 and 2 (NTMT1/2), displaying modest cellular potency in the μM range.^24–26 Following up on our recent discovery of venglustat as the first cell-potent and small molecule inhibitors of NTMT1 (Figure 1),²⁷ we describe here our efforts to systematically examine structure-activity relationship (SAR) of venglustat to boost its potency and selectivity, as well as demonstrating its applicability for inhibiting Nα methylation that catalyzed by NTMT1/2.

Figure 1.

The structure of representative NTMT1 inhibitors.

Results and Discussion

1. Design

Venglustat competitively binds at the NTMT1 substrate-binding site, exhibiting a comparable IC₅₀ value of 0.5 μM in both biochemical and cellular inhibition assays.²⁷ As venglustat is also an allosteric inhibitor for ceramide glycosyltransferase (GCS), a comprehensive SAR study is needed to provide a foundation to tune the selectivity for NTMT1 over GCS. As shown before, the N atom of quinuclidine moiety in the left region is important to the NTMT1 inhibition of venglustat through two hydrogen bonds with Asp 177 and 180 (Figure 2).²⁷ The middle region does not display any direct interaction with NTMT1 but orients venglustat in a U shape to fit the peptide binding pocket well (Figure 2B). The thiazole moiety connecting the middle and the right regions forms a hydrogen bond with Asp 168 and a pi-pi interaction with Trp136. The fluorophenyl ring in the right region is inserted into the pocket. To improve the binding affinity and selectivity, we thoroughly investigated the contribution of the left, middle, and right regions of venglustat to NTMT1 inhibition (Figure 2C).

Figure 2.

The structure of NTMT1-venglustat complex. (A) Crystal structure of NTMT1-SAH-venglustat complex (PDB ID, 7U1M). SAH (green) and venglustat (yellow). (B) Molecular interactions of venglustat with NTMT1. (C) Three regions of venglustat for SAR examinations.

2. Synthesis

Compounds 1 – 11 were synthesized following the reported methods (Scheme 1).²⁷ Starting with 4-fluorobenzothioamide, 1 was efficiently synthesized via a substitution-cyclization reaction. The dimethylation reaction followed by ester hydrolysis produced the intermediate 2, which was then subjected to the Curtius rearrangement to prepare 3 – 9. Condensation of 2 with cyclohexylamine and 3-aminoquinuclidine generated 10. Meanwhile, hydrolysis of 1 released its corresponding acid, which was then reacted with isopropanol and (3S)-quinuclidin-3-ol to generate 11 through Curtius rearrangement.

Scheme 1.

Synthesis of Venglustat Analogues 4 – 11^a

^a Reagents and conditions: (a) ethyl 4-chloroacetoacetate (1.1 eq), refluxed in EtOH for 72 h, 80%. (b) t-BuOK (2.5 eq), MeI (3.0 eq), anhydrous DMF, 0 °C for 1 h. (c) hydrolyzed with 2 M NaOH in MeOH, r.t., overnight, 65% in two steps. (d) RH (1.5 eq), DPPA (1.1 eq), Et₃N (1.1 eq), toluene, under reflux for 24 - 48 h, 31 - 36%. (e) (S)-quinuclidin-3-amine (1.2 eq), PyBOP (1.2 eq), Et₃N (3.0 eq), DCM, r.t., overnight, 62%.

Compound 12 was efficiently prepared from thiourea (Scheme 2). Starting from 12, the synthesis of intermediate 13 was challenging. We obtained 13 in 31% yield after optimizing the solvent, the concentration of the substrates, and reaction time. Dimethylation of 13 followed by hydrolysis produced 14, which was undergone the Curtius rearrangement to produce 15. Protection of the amino group of 12 with a Boc group generated 16. Subsequent methylation and hydrolysis led to 17. Subsequent Curtius rearrangement and Boc deprotection yielded 18.

Scheme 2.

Synthesis of Venglustat Analogues 15 and 18^a

^a Reagents and conditions: (a) ethyl 4-chloroacetoacetate (0.93 eq), EtOH, under reflux for 12 h, 91%. (b) t-BuONO (1.4 eq), CuI (1.5 eq), MeCN, 65 °C, 20 min, 31%. (c) t-BuOK (2.5 eq), MeI (3.0 eq), anhydrous DMF, 0 °C for 1 h, 72 - 81%. (d) 2 M LiOH in MeOH, r.t., overnight, 84 - 90%. (e) (S)-3-oxylquinuclidine (1.5 eq), DPPA (1.1 eq), Et₃N (1.1 eq), toluene, under reflux for 24 h, 30 - 36%. (f) Boc₂O (1.2 eq), Na₂CO₃ (1.5 eq), anhydrous THF, r.t., 24 h, 71%. (g) TFA, DCM, r.t., 5 h, 92%.

Starting from thiourea, 19 was efficiently prepared following the Hantzsch thiazole synthesis (Scheme 3).²⁸ Dimethylation of 19 followed by hydrolysis led to 20, which was subjected to Curtius rearrangement to yield 21. Starting from 15, 22 – 44 were synthesized through the Suzuki reaction.²⁹

Scheme 3.

Synthesis of Venglustat Analogues 21 – 44^a

^a Reagents and conditions: (a) ethyl 4-chloroacetoacetate (0.93 eq), EtOH, under reflux for 12 h, 61%. (b) t-BuOK (2.5 eq), MeI (3.0 eq), anhydrous DMF, 0 °C for 1 h. (c) 2M LiOH in MeOH, r.t., overnight, 62% in 2 steps. (d) (S)-3-oxylquinuclidine (1.5 eq), DPPA (1.1 eq), Et₃N (1.1 eq), toluene, under reflux for 24 h, 33%. (e) boronic substrates (1.2 eq), Pd(PPh₃)₄Cl₂ (0.15 eq), Na₂CO₃ (3.0 eq), dioxane/H₂O, under reflux for 2 – 8 h, 26 - 84%.

3. SAR

The quinuclidine moiety of the venglustat (3, YD206) at the left region is a rigid cage structure and interacts with NTMT1 through electrostatic interactions. Thus, removal of the amine abolished the inhibition.²⁷ As the quinuclidine portion is a key component for venglusat to inhibit GCS, we attempted to replace it with other amine-containing groups.³⁰ With a 3-piperidine group replacement, 4 failed to inhibit 50% of NTMT1 activity at 100 μM (Table 1). Considering 4 containing a secondary amine, we speculated that conversion of the secondary amine to a tertiary would rescue the inhibition. Thus, 4 was alkylated with methyl or propyl to produce 5 or 6. Both 5 and 6 showed improved inhibition than 4, indicating that tertiary amine is favored for NTMT1 inhibition. Compound 6 exhibited a lower IC₅₀ value, offering a marginal preference for the propyl group over the methyl group. Nevertheless, both compounds lost over 40-fold potency compared to venglustat. Meanwhile, cleavage of the single bond between C3 – C4 of the quinuclidine ring provided 7, causing over a 5-fold reduction than 6. Replacing the alkyl amine with a 3-pyridine moiety led to 8, resulting in a null activity. These results indicated that the proper orientation of a tertiary amine by the quinuclidine moiety at the left region is critical for NTMT1 inhibition.

Table 1.

The IC₅₀ Values of the Analogues with Modifications in the Left and Middle Regions.

The carbamate group of venglustat is exposed to the surface of NTMT1 with no direct interaction with NTMT1. To understand the importance of the carbamate group of venglustat for its inhibition on NTMT1, we replaced the carbamate group with a urea group to produce 9, causing over a 100-fold inhibition loss (Table 1). Compound 10 with an amide group did not show any inhibition to NTMT1. Given that the middle region also contains a dimethyl group, we removed the dimethyl group to produce 11, losing the inhibitory activity even at 100 μM. Considering the decrease or loss of activity of 9 – 11, we speculated that both dimethyl and carbamate moieties facilitate a right conformation of venglustat to interact with NTMT1 (Table 1).

The right part of venglustat is inserted into the binding pocket, interacting with NTMT1 through van der waals interaction. Substitution of the fluorophenyl with the iodide or methylamine resulted in about 100-fold reduced activity in 15 and 18, supporting the contribution of the phenyl ring to the NTMT1 interaction (Table 2). Meanwhile, replacement of the fluorophenyl by an isopropyl group afforded 21, showing over 150-fold reduced activity. Thus, we focused on the possibility of replacing the phenyl ring with other aromatic groups. The introduction of 1-naphthyl group generated 22 but caused a 20-fold decrease compared to the phenyl ring analogue,²⁷ suggesting the aversion to a bulky ring. For the pyridine ring, the substitution position impacted the inhibition with ~50-fold difference. Specifically, 23 and 24 with a pyridine showed similar potency as the phenyl ring alone, but 25 with the para substitution did not inhibit 50% of NTMT1 activity even at 100 μM. However, introducing a pyrimidine ring with one additional N atom at the meta position of pyridine produced inactive 26. The preference for the substitution position was also the case for both furan and thiofuran. For instance, 27 and 29 with the 2-substitute furan and thiofuran are more potent than the 28 and 30 with 3-substitution ones. Interestingly, thiofuran (28 and 30) was tenfold better than the respective furan analogues (27 and 28). In summary, 2-pyridine and 1-thiofuran replacement displayed comparable inhibition as the phenyl ring. Further introduction of a fluoride at the ortho position of 2-pyridine produced 31, exhibiting a twofold improvement than 23 with 2-pyridine alone but still 3-fold less potent than venglustat.

Table 2.

The IC₅₀ Values of Analogues with Modifications in the Right Region.

Thus, we focus on examining the effects of various substitutions at the phenyl ring on NTMT1 inhibition (Table 2). Substitution the fluoride with chloride, bromide, methyl, or trifluoromethyl at the para position resulted in a comparable activity loss by 2- to 6-fold, suggesting that a smaller size is preferred at the para position.²⁷ This can be explained by the limited space in the NTMT1 binding pocket where the fluorophenyl ring bound. Considering the position effects for the pyridine, furan, and thiofuran substitution, we then examined the effects of hydroxyl and amino groups at different positions on the phenyl ring as they are commonly used hydrogen bond donors and acceptors. Amino substitution at different positions displayed marginal differences, but there was a ~50-fold difference for the hydroxyl group at different substitutions on the phenyl ring. Among them, meta- is preferred followed by ortho- and para-substitution, as 36 showed a comparable IC₅₀ of 0.56 μM as that of venglustat. Given the preference of the hydroxyl group over the amino group at the meta position, we hypothesized that electron donating group at the meta position is preferred for NTMT1 inhibition. We then synthesized 37 and 39 with fluoride and nitro groups to test this hypothesis. The ~50-fold reduction in inhibition supported our idea. Merging the 3-hydroxyl substitution with 4-chloride on the phenyl ring yielded 42 (YD2200, IC₅₀ = 0.11 μM), with 5-fold improved inhibitory activity than venglustat. Likewise, introducing a 3-hydroxyl group on venglustat produced 43 (GD433) with an IC₅₀ value of 27 nM. Notably, 43 is about 20-fold more potent than venglustat. However, moving the hydroxyl group to the 2-position generated 44, 10-fold less potent than 43, further suggesting the preference of the hydroxyl group at the meta position.

4. Inhibition Mechanism

Given the improved potency of GD433, we then determined its inhibition mechanism by the SAHH-coupled fluorescence-based assay.^{21, 23} The IC₅₀ values of GD433 increased linearly with the ratio of peptide/K_m (Figure 3A–B), indicating GD433 is competitive with the NTMT1 peptide substrate. Like venglustat, the IC₅₀ of GD433 was slightly decreased when the ratio of SAM/K_m increased from 0.25K_m to 1K_m, but remained constant when the ratio of SAM/K_m was higher than 1K_m (Figure 3C–D). These results demonstrated that GD433 retained the exact inhibition mechanism as its parent compound venglustat, exhibiting an uncompetitive pattern with SAM.

Figure 3.

GD433 is a peptide substrate competitive inhibitor of NTMT1. (A) Linear regression plot of IC₅₀ values for GD433 with an increased ratio of peptide/K_m; (B) IC₅₀ curves of GD433 at varying concentrations of peptide substrate GPKRIA with a fixed concentration of SAM; (C) Plot of IC₅₀ values for GD433 with an increased ratio of SAM/K_m value; (D) IC₅₀ curves of GD433 at varying concentrations of SAM with a fixed concentration of GPKRIA. All the experiments were performed in triplicate (n=3) and presented as mean ± SD. (E, F) Representative ITC data of GD433 bound to NTMT1 in the absence of SAM (E) and the presence of SAM (F). All the experiments were performed in duplicate (n=2).

Next, the binding affinity was determined by titrating NTMT1 with increasing volumes of GD433 in the presence or absence of SAM. In the absence of SAM, GD433 interacted with NTMT1 with a dissociation constant (K_d) of 36.9 ± 8.18 nM (Figure 3E). But the K_d value (8.28 ± 2.58 nM) increased more than fourfold in the presence of SAM (Figure 3F), indicating the positive effect of SAM on the interaction of GD433 to NTMT1 as observed in the inhibition mechanism experiments (Figure 3C–D). Meanwhile, YD2200 showed a similar pattern, but its K_d values were about threefold less than GD433 in both cases, consistent with the difference in the biochemical assay (Figure S1).

5. Selectivity Study of GD433

Selectivity is critical for chemical probes in biological applications. First, we assessed the inhibitory activity of GD433 against five different in-house methyltransferases (G9a, PRMT1, PRMT7, NNMT, and SETD7) and the coupling enzyme SAHH (Figure S2). At 10 and 30 μM, GD433 displayed no significant inhibition against these six enzymes. Even at 100 μM, GD433 did not exhibit inhibition against the aforementioned enzymes, except 50% inhibition for PRMT1. To gain a broad understanding of its selectivity, GD433 was then evaluated for its inhibition against a panel of 40 SAM-dependent methyltransferases at a single dose of 50 μM (Figure 4). GD433 potently inhibited NTMT1 and its close homolog NTMT2, which shares the same X-P-K/R recognition motif as NTMT1.¹⁰ GD433 did not show significant inhibition against other methyltransferases including PKMTs, PRMTs, and DNA methyltransferases, except ~25% inhibition for the MLL1 complex.

Figure 4.

Selectivity study of GD433 on 41 SAM-dependent protein methyltransferases. GD433 was tested at 50 μM in duplicate. Both SAM and substrates are at their physiological concentrations.

6. Crystal Structure of NTMT1-SAH-GD433 Complex.

To further confirm the binding mode of GD433 and elucidate the molecular interactions between the NTMT1 and GD433, we determined the co-crystal structure of NTMT1 with GD433 and SAH (PDB ID:7SS1). Like venglustat, GD433 occupied the peptide substrate binding site of NTMT1 (Figure 5A). Superimposition of the ternary complex of NTMT1-GD433-SAH with the previously solved ternary complex NTMT1-Venglustat (PDB: 7U1M) gave an RMSD value of 0.099 Å (across all residues of chain A) (Figure 5B). GD433 retains all the important interactions with NTMT1 as venglustat in the NTMT1-venglustat-SAH complex (Figure 5C).³² For instance, the tertiary amine of the quinuclidine forms an electrostatic interaction with Asp177 and 180. The nitrogen atom of the thiazole ring interacts with the Asn168 through hydrogen bonding. Although pi-pi interaction between thiazole and Typ136 was not observed, GD433 gained two more hydrogen bond interactions with NTMT1 than venglustat. A direct hydrogen bond exists between the Tyr215 and the carbonyl group of carbamate of GD433. Besides, the hydroxy group of the phenyl ring forms a hydrogen bond with Gly139.

Figure 5.

X-ray co-crystal structure of NTMT1 (gray cartoon)–GD433 (green stick)–SAH (PDB ID: 7SS1). (A) Interactions of GD433 with NTMT1. H-bond interactions are shown as dotted lines. (B) Structural alignment of NTMT1 (gray cartoon)–GD433 (green stick)–SAH (green stick) and NTMT1 (gray cartoon)–venglustat (organge stick)–SAH (organge stick). (C) Structural of NTMT1 (gray cartoon)–venglustat (orange stick)–SAH (orange stick).

6. Docking study

Docking studies of all synthesized compounds were taken by using venglustat-NTMT1 co-crystal structure as a template. The predicted binding poses were then used to calculate ΔG_bind by using MM-GBSA rescoring methods. In figure 6A, the predicted ΔG_bind vs pIC₅₀ values shows a good linear regression, indicating the docking and MM-GBSA model can be used to interpret our experimental data. In figure 6B, the predicted binding pose of GD433 aligned well with its the co-crystal structure, supporting the predictability of our docking and MM-GMSA model. Next, the predicted binding poses of two potent compounds YD2200 and YD2192 were extracted and illustrated in figure 6C and 6D. Both compounds have a meta hydroxyl group at the right part phenyl ring and exhibit similar binding modes. YD2200 has the exact same interaction as GD433. Although YD2192 lost the interaction with Gly139, it forms a new interaction with Asp180 at the fluorophenyl ring moiety. Besides that, both YD2200 and YD2192 obtained additional interactions with NTMT1 than venglustat, providing the molecular basis for their improved potency. Compared to GD433, YD2160 with a pyrimidine substitution lost interaction with Gly139 of NTMT1 (figure 6E). Compound YD2158 bearing a para-carbonic acid showed a different binding pose from GD433 and resulted in a much higher predicted ΔG_bind of YD2158 than GD433 (figure 6A), indicating acidic substitutions at phenyl ring not favored.

Figure 6.

Docking studies and MM-GBSA rescoring. (A) predicted ΔG_bind vs pIC₅₀ values of all synthesized compounds. (B) structural alignment of experimental (green) and predicted (purple) binding modes of GD433. (C, D, E, F) predicted binding mode of YD2200, YD2192, YD2160 and YD2158.

7. Cellular Inhibition

Venglustat significantly suppressed me3-RCC1 and me3-SET at 1 μM in HEK293 cells, displaying an approximate IC₅₀ value of 0.3 μM.²⁷ We then assessed the cellular inhibition activities of two potent inhibitors (YD2200 and GD433) and one inactive analogue (YD2160) as described before.²⁷ As shown in figure 7B, YD2200 significantly reduces me3-RCC1 and me3-SET at 1 μM with an approximate IC₅₀ value of 0.3 μM, exhibiting a comparable inhibition as venglustat (Figure 7A). Notably, GD433 significantly reduced me3-RCC1 and me3-SET with an approximate IC₅₀ value of <30 nM (Figure 7C), exhibiting over 10-fold higher cellular inhibition than venglustat. The inactive analogue YD2160 did not induce any change in either me3-RCC1 or me3-SET up to 30 μM (Figure 7D). These results demonstrated that YD2200 and GD433 are cell-potent NTMT1 inhibitors, while YD2160 is a negative control compound.

Figure 7.

Inhibition on cellular Nα methylation. (A, B, C, D) Representative western blot results of effects of YD206, YD2200, GD433, and YD2160 on the cellular methylation level in HEK293 cells (n = 3).

Since the initial hit venglustat is a GCS inhibitor, we then evaluated the cellular effects of both venglustat and GD433 on cellular GCS through an HPLC assay based on the published method with minor modification (Figure S3).³¹ Our results confirmed that venglustat is a potent inhibitor for GCS as it significantly inhibits GCS even at 30 nM. On the other hand, GD433 did not inhibit cellular GCS activity at 30 nM though it was able to inhibit GCS activity at 300 nM. Compared to venglustat, GD433 exhibits about 100-fold improvement regarding the cellular inhibition ratio of NTMT1/GCS.

8. Maximum Tolerated Dose and Pharmacokinetic Profile of GD433

The maximum tolerated dose (MTD) study of GD433 was initially tested at doses of 100, 200, and 400 mg/kg in a single oral administration to Balb/c mice. Mice dosed with 100 and 200 mg/kg of GD433 remained responsive and body weight throughout two weeks, but mouse dosed at 400 mg/kg died at 15 min post-administration. Subsequent MTD study investigated three doses at 250, 300, and 350 mg/kg. Mice of all dosed groups remained bright, alert, and responsive without significant weight loss throughout 14 days, indicating an MTD of GD433 at 350 mg/kg (Figure 8A).

Figure 8.

Maximum tolerated dose (MTD) and pharmacokinetic (PK) studies of GD433. (A) MTD study of GD433 at a single dose of 250, 300, and 350 mg/kg in saline with 5% DMSO given to Balb/c mice through oral gavage (n=3). The mice were observed a period of 14 days. (B) Pharmacokinetic profile of GD433 dosed orally at 25 mg/kg. Plasma was collected at 0.083, 0.25, 0.5, 1, 2, 4, 8, and 24 h (n=3).

Next, a pharmacokinetic (PK) study was conducted with an oral dose of 25 mg/kg GD433 in CD1 mice at Pharmaron. The plasma concentration of GD433 reached the highest plasma concentration (C_max) of 81.9 ng/mL at 0.25 h (Figure 8B), 7-fold higher than the IC₅₀ value. Although the half-life is 2.7 h, the plasma concentration of GD433 remained higher than 18.8 ng/mL within the first 4 h after administration, about 2-fold higher than the IC₅₀ value of GD433 (27 nM, 10.9 ng/mL).

Conclusion

In summary, we designed, synthesized, and characterized a series of novel compounds aimed at increasing the potency and selectivity of our previously reported NTMT1 inhibitor venglustat. Our SAR studies of venglustat discovered GD433 enhanced inhibition for NTMT1 in both biochemical and cellular assays (IC₅₀ values, 27 ± 1.1 nM), displaying over 10-fold improvement. GD433 exerts the inhibition of NTMT1 through a competitive mechanism with the peptide substrate and uncompetitive with the cofactor SAM. GD433 is selective for NTMT1/2 over 40 other methyltransferases as it exhibited less than 50% inhibition at 100 μM. Although GD433 retained inhibition on cellular GCS because of the quinuclidine ring, GD433 demonstrated 10-fold increased potency to NTMT1 and 10-fold decreased potency to GCS than venglustat, resulting in 100-fold increased selectivity for NTMT1 over GCS. Future investigations would explore the quinuclidine surrogates to abolish its inhibitory activity on GCS but retain potent inhibition on NTMT1. Nevertheless, inactive analogue YD2160 will serve as a negative control for GD433 to dissect the biological roles of NTMT1 from GCS to shed light on the functions of Nα-methylation catalyzed by NTMT1/2. Notably, compound 37 (IC₅₀=32 μM) is also a good candidate as a negative control given its close structural similarity with GD433 but caution must be taken when used at higher concentrations (10 μM). Furthermore, GD433 is orally bioavailable and ready to serve as an in vivo probe to elucidate the functions of Nα-methylation catalyzed by NTMT1/2 in animal models.

In the meantime, our SAR investigation may also guide the future optimization of venglustat selectivity for GCS over NTMT1. For instance, it would be interesting to test the activity on GCS for those venglustat analogues that contain the quinuclidine moiety but inactive to NTMT1 to identify more selective GCS inhibitors.

Experimental Section

Chemistry.

Starting materials, reagents, and solvents were obtained from commercial sources. Analytic and preparative high-pressure liquid chromatography (RP-HPLC) was performed on Agilent 1260 Series system. Systems were run with a 5-95% acetonitrile/water gradient with a 0.1% TFA. The peptide GPKRIA was synthesized using a Liberty Automated Microwave Peptide Synthesizer (CEM) with the manufacturer’s standard coupling cycles at 0.1 mmol scales and cleaved from the resin in a cocktail of TFA/2,2’-(ethylenedioxy)diethanethiol/H₂O/triisopropylsilane (94:2.5:2.5:1) and confirmed by MS. All ¹H and ¹³C NMR spectra were recorded on a Brucker 500 MHz spectrometer. HRMS spectra were recorded on an Agilent high resolution 6550 quadrupole time-of-flight (Q-TOF) LC-MS instrument. All compounds used for their IC₅₀ determination and animal experiments possessed a purity of at least 95%, based on HPLC analysis.

Ethyl 2-(2-(4-Fluorophenyl)thiazol-4-yl)acetate (1).

Compound 1 was synthesized by following our previously reported method.²⁶

2-(2-(4-Fluorophenyl)thiazol-4-yl)-2-methylpropanoic Acid (2).

t-BuOK (280 mg, 2.5 mmol) was added to a stirred anhydrous DMF solution (10 mL) of 1 (265 mg, 1.0 mmol) in portion. After stirring for 15 min at room temperature, the mixture was cooled down in an ice bath. Next, iodomethane (156.0 μl, 2.5 mmol) was added. And the reaction mixture was stirred 1 h in ice bath. After the reaction was completed, the mixture was acidified with 0.1 N HCl, diluted with water, and extracted with EtOAc (3 × 30 mL). The combined extracts were washed with water, brine, and dried over anhydrous sodium sulfate. The volatile was removed under vacuo to yield a crude oil, which was re-dissolved in 15 mL MeOH and 7 mL 2 M NaOH to undergo hydrolysis. The reaction mixture was then stirred overnight at room temperature. After the reaction was finished, the mixture was quenched with 2 N HCl and extracted with EtOAc (3 × 30 mL). The combined extracts were washed with water, brine, and dried with anhydrous sodium sulfate. The volatile was removed under vacuo to yield the crude product, which was purified by flash column (EtOAc / Hexane, 4 / 6) to afford 2 (172 mg, 65% yield).

(S)-quinuclidin-3-yl (2-(2-(4-Fluorophenyl)thiazol-4-yl)propan-2-yl)carbamate (3, YD206).

Compound 3 was synthesized by following our previously reported method.²⁶

(S)-piperidin-3-yl (2-(2-(4-Fluorophenyl)thiazol-4-yl)propan-2-yl)carbamate (4).

To a stirred solution of compound 2 (265 mg, 1.0 mmol) in anhydrous toluene (20 mL) was added Et₃N (111.0 mg, 1.1 mmol) under nitrogen. The mixture was pre-heated to 100 °C and then DPPA (303 mg, 1.1 mmol) was added. 1-Boc-(S)-3-hydroxypiperidine (302 mg, 1.5 mmol) was added after 3 h of reflux. The reaction was refluxed for another 24 h. After the reaction was completed, the mixture was diluted with water and extracted with EtOAc (3 × 50 mL). The combined extracts were washed with water, brine, and dried with anhydrous sodium sulfate. The solvent was removed under vacuo to generate the crude product, which was dissolved in 20 mL DCM. Then, 6 mL TFA was slowly added and stirred at room temperature for 1 h. After the reaction was completed, the solvent was removed under vacuo to yield a crude oil which was purified with prep-HPLC (MeCN / H₂O) to afford 4 (127 mg, 35% yield). ¹H NMR (500 MHz, CDCl₃) δ 8.03 – 7.80 (m, 2H), 7.09 (t, J = 8.2 Hz, 2H), 7.05 (s, 1H), 5.63 (s, 1H), 3.74 (s, 1H), 3.54 (d, J = 15.4 Hz, 1H), 3.39 – 3.29 (m, 1H), 3.28 – 3.16 (m, 2H), 1.82 (dt, J = 8.3, 3.9 Hz, 1H), 1.75 (d, J = 15.6 Hz, 6H), 1.55 (dt, J = 13.6, 4.8 Hz, 1H), 1.47 – 1.41 (m, 1H). ¹³C NMR (126 MHz, CDCl₃) δ 166.26, 163.94, 162.76, 157.61, 130.12, 128.42, 128.35, 116.03, 115.85, 112.24, 66.00, 54.55, 51.04, 44.90, 32.35, 28.67, 28.27, 22.13. HRMS (ESI) m/z calcd for C₁₈H₂₂FN₃O₂S [M+H]⁺: 364.1490; Found: 364.1490.

(S)-1-methylpiperidin-3-yl (2-(2-(4-Fluorophenyl)thiazol-4-yl)propan-2-yl)carbamate (5).

Compound 5 (23 mg, 31% yield) was prepared with the same synthetic method as 3. ¹H NMR (500 MHz, CDCl₃) δ 7.92 (td, J = 5.8, 2.9 Hz, 2H), 7.19 – 7.02 (m, 3H), 6.30 (s, 1H), 5.02 (s, 1H), 3.69 (d, J = 12.0 Hz, 1H), 3.54 (d, J = 12.7 Hz, 1H), 2.82 (s, 3H), 2.78 (d, J = 13.1 Hz, 1H), 2.65 (t, J = 12.5 Hz, 1H), 2.30 (q, J = 14.2 Hz, 1H), 2.04 (d, J = 14.6 Hz, 1H), 1.78 (s, 1H), 1.73 (d, J = 4.8 Hz, 6H), 1.54 (t, J = 14.3 Hz, 1H). ¹³C NMR (126 MHz, CDCl₃) δ 166.36, 164.77, 162.44, 153.09, 130.12, 128.53, 128.46, 115.96, 115.78, 112.70, 64.45, 57.37, 54.88, 54.36, 44.32, 27.91, 27.81, 25.68, 18.25. HRMS (ESI) m/z calcd for C₁₉H₂₄FN₃O₂S [M+H]⁺: 378.1646; Found: 378.1647.

(S)-1-propylpiperidin-3-yl (2-(2-(4-Fluorophenyl)thiazol-4-yl)propan-2-yl)carbamate (6).

Compound 6 (27 mg, 33% yield) was prepared with the same synthetic method as 3. ¹H NMR (500 MHz, CDCl₃) δ 7.92 (ddd, J = 8.9, 5.3, 1.9 Hz, 2H), 7.14 (d, J = 1.7 Hz, 1H), 7.09 (td, J = 7.7, 2.0 Hz, 2H), 6.53 (s, 1H), 5.01 (s, 1H), 3.72 (d, J = 11.9 Hz, 1H), 3.60 – 3.47 (m, 1H), 3.09 – 2.97 (m, 1H), 2.96 – 2.85 (m, 1H), 2.71 (d, J = 12.8 Hz, 1H), 2.67 – 2.57 (m, 1H), 2.34 (d, J = 14.1 Hz, 1H), 2.04 (d, J = 14.6 Hz, 1H), 1.74 (d, J = 1.7 Hz, 9H), 1.60 – 1.51 (m, 1H), 1.04 – 0.86 (m, 3H). ¹³C NMR (126 MHz, CDCl₃) δ 166.17, 164.72, 162.63, 153.25, 130.26, 128.49, 128.42, 115.92, 115.74, 112.64, 64.31, 59.22, 54.86, 54.35, 53.08, 28.08, 27.78, 26.27, 18.18, 16.92, 10.98. HRMS (ESI) m/z calcd for C₂₁H₂₈FN₃O₂S [M+H]⁺: 406.1959; Found: 406.1961.

2-(Piperidin-1-yl)ethyl (2-(2-(4-Fluorophenyl)thiazol-4-yl)propan-2-yl)carbamate (7).

Compound 7 (27 mg, 35% yield) was prepared with the same synthetic method as 3. ¹H NMR (500 MHz, CDCl₃) δ 7.91 (dd, J = 8.7, 5.3 Hz, 2H), 7.20 – 7.00 (m, 3H), 6.13 (s, 1H), 4.32 (s, 2H), 3.61 (s, 2H), 3.24 (s, 2H), 2.67 (s, 2H), 1.93 (d, J = 11.8 Hz, 2H), 1.81 (d, J = 11.8 Hz, 3H), 1.72 (s, 6H), 1.36 (s, 1H). ¹³C NMR (126 MHz, CDCl₃) δ 166.86, 164.95, 161.97, 153.66, 129.64, 128.64, 128.58, 116.10, 115.93, 112.99, 57.27, 56.02, 54.24, 53.68, 27.71, 22.56, 21.69. HRMS (ESI) m/z calcd for C₂₀H₂₆FN₃O₂S [M+H]⁺: 392.1803; Found: 392.1803.

Pyridin-3-yl (2-(2-(4-Fluorophenyl)thiazol-4-yl)propan-2-yl)carbamate (8).

Compound 8 (22 mg, 31% yield) was prepared with the same synthetic method as 3. ¹H NMR (500 MHz, CDCl₃) δ 8.66 (s, 1H), 8.53 (s, 1H), 7.95 (dd, J = 8.9, 5.2 Hz, 2H), 7.63 (s, 1H), 7.23 (d, J = 5.8 Hz, 1H), 7.16 – 7.11 (m, 3H), 6.40 (s, 1H), 1.84 (s, 6H). ¹³C NMR (126 MHz, CDCl₃) δ 167.12, 163.02, 161.41, 150.60, 140.64, 138.78, 134.70, 129.45, 128.55, 128.48, 125.78, 124.09, 116.20, 116.02, 113.01, 54.96, 27.44. HRMS (ESI) m/z calcd for C₁₈H₁₆FN₃O₂S, [M+H]⁺: 358.1020; Found: 358.1020.

1-(2-(2-(4-Fluorophenyl)thiazol-4-yl)propan-2-yl)-3-(quinuclidin-3-yl)urea (9).

Compound 9 (28 mg, 36% yield) was prepared following the same synthetic method as 3. ¹H NMR (500 MHz, CDCl₃) δ 7.93 – 7.88 (m, 2H), 7.22 – 7.05 (m, 3H), 4.18 – 4.08 (m, 1H), 3.51 (t, J = 11.3 Hz, 1H), 3.37 (s, 1H), 3.14 (d, J = 10.9 Hz, 4H), 2.20 (s, 2H), 1.91 (s, 2H), 1.80 – 1.67 (m, 7H). ¹³C NMR (126 MHz, CDCl₃) δ 167.36, 163.30, 161.96, 157.92, 128.90, 128.83, 128.60, 116.43, 116.25, 113.21, 53.82, 53.36, 46.46, 45.85, 44.57, 28.45, 28.28, 24.90, 21.95, 17.38. HRMS (ESI) m/z calcd for C₂₀H₂₅FN₄OS [M+H]⁺: 389.1806; Found: 389.1805.

2-(2-(4-Fluorophenyl)thiazol-4-yl)-2-methyl-N-(quinuclidin-3-yl)propenamide (10).

To a stirred solution of 2 (53 mg, 0.2 mmol) in 10 mL DCM was added Et₃N (61 mg, 0.6 mmol) and PyBOP (123 mg, 0.24 mmol) in sequence. The reaction mixture was stirred at room temperature for 30 min. After that, 3-aminoquinuclidine (30 mg, 0.24 mmol) was added. The mixture was stirred overnight at room temperature. After the reaction was completed, the mixture was diluted with 30 mL DCM, washed with water, brine, dried with anhydrous sodium sulfate. The solvent was removed under vacuo. The generated residue was re-dissolved in 10 mL MeOH, which was purified with pre-HPLC (MeCN / H₂O) to afford the desired product 10 (46 mg, 62% yield). ¹H NMR (500 MHz, CDCl₃) δ 8.02 – 7.94 (m, 1H), 7.89 (dt, J = 8.4, 5.6 Hz, 2H), 7.19 – 7.08 (m, 3H), 5.13 (s, 1H), 4.38 – 4.15 (m, 1H), 3.90 – 3.69 (m, 1H), 3.63 – 3.38 (m, 2H), 3.27 – 2.97 (m, 2H), 2.42 – 2.17 (m, 1H), 2.08 (s, 2H), 1.84 (s, 1H), 1.63 (d, J = 12.5 Hz, 7H). ¹³C NMR (126 MHz, CDCl₃) δ 176.31, 167.17, 165.02, 160.50, 129.55, 128.21, 116.30, 116.12, 114.14, 68.40, 59.17, 53.48, 45.63, 44.31, 25.99, 25.73, 24.42, 21.98, 17.55. HRMS (ESI) m/z calcd for C₂₀H₂₄FN₃OS [M+H]⁺: 374.1697; Found: 374.1702.

(S)-quinuclidin-3-yl ((2-(4-Fluorophenyl)thiazol-4-yl)methyl)carbamate (11).

Compound 11 (22 mg, 31% yield) was prepared with the same synthetic method as 3. ¹H NMR (500 MHz, CDCl₃) δ 7.88 (dd, J = 8.7, 5.2 Hz, 2H), 7.18 (s, 1H), 7.12 (t, J = 8.6 Hz, 2H), 6.05 (s, 1H), 5.06 – 4.97 (m, 1H), 4.47 (d, J = 5.1 Hz, 2H), 3.60 (dd, J = 14.1, 8.6 Hz, 1H), 3.37 – 3.14 (m, 5H), 2.41 (s, 1H), 2.25 – 2.11 (m, 1H), 2.00 (dq, J = 10.9, 6.8 Hz, 1H), 1.92 – 1.83 (m, 1H), 1.81 – 1.71 (m, 1H). ¹³C NMR (126 MHz, CDCl₃) δ 168.22, 163.12, 155.17, 153.22, 129.12, 128.70, 128.63, 116.32, 116.14, 115.72, 67.10, 53.60, 46.48, 45.55, 40.87, 24.13, 20.41, 16.90. HRMS (ESI) m/z calcd for C₁₈H₂₀FN₃O₂S [M+H]⁺: 362.1333; Found: 362.1330.

Ethyl 2-(2-Aminothiazol-4-yl)acetate (12).

Ethyl 4-Chloro-3-oxobutanoate (4.0 g, 24.0 mmol) and thiourea (2.0 g, 26.0 mmol) were dissolved in 50 mL EtOH and the mixture was refluxed for 12 h. The crude mixture was concentrated to dryness and the obtained residue was subsequently dissolved in 200 mL water. After neutralization of the water layer (pH 7.0) with saturated NaHCO₃, the aqueous layer was extracted with EtOAc (3 × 50 mL). The combined organic layers were dried over anhydrous sodium sulfate, evaporated under vacuo, and purified by flash column (EtOAc / hexane, 1 / 1) to provide 12 (4.0 g, 91% yield). ¹H NMR (500 MHz, CDCl₃) δ 6.32 (s, 1H), 4.23 – 4.10 (m, 2H), 3.54 (s, 2H), 1.26 (t, J = 7.1 Hz, 3H). ¹³C NMR (126 MHz, CDCl₃) δ 170.46, 167.99, 144.27, 105.29, 61.04, 37.20, 14.20.

Ethyl 2-(2-Iodothiazol-4-yl)acetate (13).

To a solution of compound 12 (3.72 g, 20 mmol) in 50 mL MeCN was added CuI (5.71 g, 30 mmol) and n-butyl nitrite (2.88 g, 28 mmol) in sequence. The reaction mixture was stirred at 65 °C in an oil bath for 20 minutes. The reaction mixture was then evaporated to dryness under vacuo. The residue was dissolved in 200 mL EtOAc, washed with 0.1 M ammonia solution (2 × 100 mL), water, brine, and dried with anhydrous sodium sulfate. The volatile was removed under vacuo to generate a crude oil, which was purified with flash column (EtOAc / hexane, 1 / 9) to afford 13 (1.84 g, 31% yield). ¹H NMR (500 MHz, CDCl₃) δ 7.19 (s, 1H), 4.17 (q, J = 7.1 Hz, 2H), 3.84 (s, 2H), 1.26 (t, J = 7.1 Hz, 3H). ¹³C NMR (126 MHz, CDCl₃) δ 169.87, 151.37, 122.72, 99.91, 61.22, 36.57, 14.18.

2-(2-Iodothiazol-4-yl)-2-methylpropanoic Acid (14).

t-BuOK (280 mg, 2.5 mmol) was added to a stirred anhydrous DMF solution (10 mL) of 13 (297 mg, 1.0 mmol) in portion. After stirring for 15 min at room temperature, the mixture was then cooled in an ice bath. Next, iodomethane (156 μl, 2.5 mmol) was added to the mixture and stirred 1 h in an ice bath. When the reaction was completed, the mixture was acidified with 0.1 N HCl, diluted with water, and extracted with EtOAc (3 × 30 mL). The combined organic layer was washed with water, brine, and dried with anhydrous sodium sulfate. The volatile was removed under vacuo to yield a crude oil, which was purified with flash column (EtOAc / hexane, 1 / 9) to afford ethyl 2-(2-iodothiazol-4-yl)-2-methylpropanoate (233 mg, 72% yield). ¹H NMR (500 MHz, CDCl₃) δ 7.02 (s, 1H), 4.13 (q, J = 7.1 Hz, 2H), 1.58 (s, 6H), 1.19 (t, J = 7.1 Hz, 3H). ¹³C NMR (126 MHz, CDCl₃) δ 175.05, 162.49, 119.47, 99.18, 61.08, 45.61, 25.65, 14.10.

Compound ethyl 2-(2-iodothiazol-4-yl)-2-methylpropanoate (163 mg, 0.5 mmol) was dissolved in 10 mL MeOH and 3 mL 2M LiOH to undergo hydrolysis. The mixture was stirred overnight at room temperature. When the reaction was completed, the mixture was quenched with 2 N HCl and extracted with EtOAc (3 × 30 mL). The combined extracts were washed with water, brine, and dried with anhydrous sodium sulfate. The solvent was removed under vacuo to generate the crude product, which was separated with flash column (EtOAc / hexane, 2 / 8) to afford compound 14 (125 mg, 84% yield). ¹H NMR (500 MHz, CD₃OD) δ 7.32 (s, 1H), 1.57 (s, 6H). ¹³C NMR (126 MHz, CD₃OD) δ 177.38, 162.46, 119.91, 100.79, 45.13, 24.64.

(S)-quinuclidin-3-yl (2-(2-Iodothiazol-4-yl)propan-2-yl)carbamate (15).

Compound 15 (444 mg, 30% yield) was prepared with the same synthetic method as 3. ¹H NMR (500 MHz, CDCl₃) δ 7.10 (s, 1H), 5.62 (s, 1H), 4.91 (s, 1H), 3.55 (dd, J = 13.9, 8.7 Hz, 1H), 3.33 (s, 2H), 3.23 (d, J = 6.9 Hz, 3H), 2.37 (s, 1H), 2.22 (s, 1H), 2.00 (s, 1H), 1.88 – 1.76 (m, 2H), 1.68 (s, 6H). ¹³C NMR (126 MHz, CDCl₃) δ 163.38, 153.31, 119.74, 100.07, 66.55, 53.90, 53.55, 46.32, 45.39, 27.88, 27.43, 24.19, 20.58, 17.04. HRMS (ESI) m/z calcd for C₁₄H₂₀IN₃O₂S [M+H]⁺: 422.0394; Found: 422.0396.

Ethyl 2-(2-((Tert-butoxycarbonyl)amino)thiazol-4-yl)acetate (16).

Compound 12 (1.2 g, 6.5 mmol) and Boc₂O (1.7 g, 7.8 mmol) were dissolved in 20 mL EtOH. Sodium bicarbonate (1.1 g, 13.0 mmol) was added to the mixture and stirred at room temperature for 24 h. After the reaction was completed, the solvent was removed under vacuo. The generated crude oil was purified with flash column (EtOAc / hexane, 1/ 9) to afford the desired product 16 (1.3 g, 71% yield). ¹H NMR (500 MHz, CDCl₃) δ 6.75 (s, 1H), 4.16 (q, J = 7.1 Hz, 2H), 3.73 (s, 2H), 1.54 (s, 9H), 1.23 (t, J = 7.6 Hz, 3H). ¹³C NMR (126 MHz, CDCl₃) δ 170.26, 160.35, 143.33, 109.58, 61.00, 36.90, 28.22, 14.14.

2-(2-((Tert-butoxycarbonyl)(methyl)amino)thiazol-4-yl)-2-methylpropanoic Acid (17).

t-BuOK (448 mg, 4.0 mmol) was added to a stirred anhydrous DMF solution (15 mL) of 16 (286 mg, 1.0 mmol) in portion. After stirring for 15 min at room temperature, the mixture was cooled in an ice bath. Next, iodomethane (218 μl, 3.5 mmol) was added to the mixture and stirred 1 h in ice bath. After the reaction was completed, the mixture was acidified with 0.1 N HCl, diluted with water, and extracted with EtOAc (3 × 30 mL). The combined extracts were washed with water, brine, and dried with anhydrous sodium sulfate. The solvent was removed under vacuo. The generated crude oil was re-dissolved in 15 mL MeOH and 7 mL 2M NaOH to take a hydrolyzation. The mixture was then stirred overnight at room temperature. After the reaction was completed, the mixture was quenched with 0.1 N HCl and extracted with EtOAc (3 × 30 mL). The combined extracts were washed with water, brine, and dried with anhydrous sodium sulfate. The solvent was removed under vacuo to generate the crude product, which was separated with flash column (EtOAc / hexane, 4 / 6) to afford compound 17 (219 mg, 73% yield). ¹H NMR (500 MHz, CDCl₃) δ 6.70 (s, 1H), 3.53 (s, 3H), 1.57 (s, 17H). ¹³C NMR (126 MHz, CDCl₃) δ 177.19, 162.26, 152.87, 152.51, 108.42, 45.12, 34.43, 28.17, 26.03.

(S)-quinuclidin-3-yl (2-(2-(Methylamino)thiazol-4-yl)propan-2-yl)carbamate (18).

To a stirred solution of 17 (300 mg, 1.0 mmol) in anhydrous toluene (20 mL) was added Et₃N (303 mg, 3.0 mmol) under nitrogen. The mixture was pre-heated to 100 °C and then DPPA (330 mg, 1.2 mmol) was added. The (3S)-quinucildin-3-ol (192.0 mg, 1.5 mmol) was added after 3 h of reflux. The reaction was refluxed for another 24 h. After the reaction was completed, the mixture was diluted with water and extracted with EtOAc (3 × 50 mL). The combined extracts were washed with water, brine, and dried with anhydrous sodium sulfate. The solvent was removed under vacuo to generate the crude product, which was re-dissolved in 16 mL DCM. Then, 6 mL TFA was slowly added and stirred at room temperature for 5 h. After the reaction was completed, the solvent was removed under vacuo to generate the crude product, which was separated with prep-HPLC (MeCN / H₂O) to afford 18 (107 mg, 33% yield). ¹H NMR (500 MHz, CDCl₃) δ 6.34 (s, 1H), 6.25 (s, 1H), 4.86 (s, 1H), 3.97 (s, 1H), 3.52 (dd, J = 14.1, 8.4 Hz, 1H), 3.30 (t, J = 7.9 Hz, 2H), 3.20 (q, J = 10.5 Hz, 3H), 3.02 (s, 3H), 2.33 (s, 1H), 2.23 (s, 1H), 1.99 (dt, J = 7.5, 3.5 Hz, 1H), 1.87 – 1.70 (m, 2H), 1.62 (s, 6H). ¹³C NMR (126 MHz, CDCl₃) δ 171.79, 153.57, 147.89, 98.81, 66.79, 53.73, 52.00, 46.40, 45.45, 33.10, 27.34, 26.73, 24.02, 20.44, 16.76. HRMS (ESI) m/z calcd for C₁₅H₂₄N₄O₂S [M+H]⁺: 325.1693; Found: 325.1692.

Ethyl 2-(2-Isopropylthiazol-4-yl)acetate (19).

Compound 19 (650 mg, yield 61%) was prepared with the same synthetic method as 1. ¹H NMR (500 MHz, CDCl₃) δ 7.36 (s, 1H), 4.20 (q, J = 7.1 Hz, 2H), 3.95 (s, 2H), 3.64 (p, J = 6.9 Hz, 1H), 1.46 (d, J = 6.9 Hz, 6H), 1.27 (t, J = 7.1 Hz, 3H). ¹³C NMR (126 MHz, CDCl₃) δ 182.34, 168.73, 144.48, 116.70, 61.81, 34.22, 31.88, 22.94, 13.96.

2-(2-Isopropylthiazol-4-yl)acetic Acid (20).

Compound 20 (66 mg, yield 62%) was prepared with the same synthetic method as 2. ¹H NMR (500 MHz, CDCl₃) δ 7.21 (s, 1H), 3.60 (d, J = 34.4 Hz, 1H), 1.64 (s, 6H), 1.43 (d, J = 6.9 Hz, 6H). ¹³C NMR (126 MHz, CDCl₃) δ 182.48, 177.28, 155.04, 113.93, 44.89, 32.11, 25.68, 22.93.

(S)-quinuclidin-3-yl (2-(2-Isopropylthiazol-4-yl)propan-2-yl)carbamate (21).

Compound 21 (55 mg, 33% yield) was prepared with the same synthetic method as 3. ¹H NMR (500 MHz, CDCl₃) δ 7.18 (s, 1H), 6.88 (s, 1H), 4.82 (s, 1H), 3.76 – 3.59 (m, 1H), 3.53 (dd, J = 14.1, 8.4 Hz, 1H), 3.31 (d, J = 8.2 Hz, 2H), 3.22 (t, J = 8.0 Hz, 2H), 3.15 (d, J = 13.6 Hz, 1H), 2.32 (s, 1H), 2.23 (s, 1H), 1.99 (t, J = 12.9 Hz, 1H), 1.88 – 1.76 (m, 2H), 1.73 (d, J = 6.4 Hz, 6H), 1.44 (d, J = 6.8 Hz, 6H). ¹³C NMR (126 MHz, CDCl₃) δ 183.67, 157.09, 153.81, 113.27, 66.73, 53.84, 52.79, 46.51, 45.55, 31.83, 27.64, 27.41, 23.97, 23.01, 20.46, 16.81. HRMS (ESI) m/z calcd for C₁₇H₂₇N₃O₂S, [M+H]⁺: 338.1897; Found: 338.1897.

General Procedures for the Preparation of Compounds 22 – 44.

To a solution of 15 (21 mg, 0.05 mmol) in 4 mL dioxane and 1 mL H₂O was added corresponding boronic acid (9 mg, 0.075 mmol), Pd(PPh₃)₄ (9 mg, 7.5 μmol) and Na₂CO₃ (16 mg, 0.15 mmol). The mixture was refluxed under nitrogen for 2 – 8 h and monitored with TLC. After the reaction was completed, the mixture was diluted with water and extracted with EtOAc (3 × 15 mL). The combined extracts were washed with water, brine, and dried with anhydrous sodium sulfate. The organic solvent was removed under vacuo to yield a crude oil, which was purified with preparative HPLC (MeCN / H₂O) to afford the desired product 22 – 44.

Cell Culture and Antibodies.

HCT116 and HEK293 cells were obtained from ATCC and were cultured in McCoy’s 5A (modified) medium (Gibco, #16600082) and DMEM medium (Gibco, #16600082), supplemented with 10% fetal bovine serum (FBS) (Gibco, #16000044) and 1% antibiotic–antimycotic (Gibco, #15240062). Antibodies against the following proteins were used: me3-SPK (Schaner Tooley lab), RCC1 (Proteintech, #22142-1-AP), SET (Proteintech, #55201-1-AP), and horseradish peroxidase (HRP)-linked anti-rabbit IgG (cell signaling, #7074S).

Protein Expression and Purification.

Expression and purification of NTMT1 was performed as previously described.^10,27 Briefly, Plasmid pET28-LLC with full-length open reading frame of NTMT1 (amino acids 1-222) was obtained from Addgene ( #25502). Protein was expressed in E.coli BL21-CodonPlus(DE3)-RIL and induced by 1 mM isopropyl-D-1-thiogalactopyranoside at 16 °C overnight. Harvested cells were resuspended in 10 volumesof 1× PBS (pH = 7.4) with 10 mM imidazole, 5 mM β-mercaptoethanol, 200 μM PMSF, then lysed by three passing through an Emulsiflex C3 (Avestin) at 15,000-20,000 psi. The cell lysate was centrifuged at 25,000 × g for 20 minutes at 4 °C and the supernatant was then applied to a 5 mL His-Pure Ni-NTA column on a GE Pure purification system using 1× PBS with 5 mM BME, washed and eluted using a step gradient of imidazole (0.02, 0.05, 0.1, 0.25 and 0.5 M). The peak fractions were verified by SDS-PAGE analysis and the purest fractions were combined. The His tag was cleaved with TEV protease and then passed through Ni-NTA column. For enzymatic assays, combined NTMT1 were dialyzed in the dialysis buffer (25 mM Tris, pH = 7.5, 150 mM NaCl, 50 mM KCl) and concentrated to 1.5 mg/mL for biochemical assays. For crystallography study, the sample was then applied to an S200 Sephacryl HR size exclusion column (GE) using 1× TBS, pH= 7.6, with 0.5 mM TCEP. Fractions containing the purest NTMT1 were combined, concentrated to 30 mg/mL and supplemented with additional TCEP to a final concentration of 1 mM. G9a, SETD7, PRMT1, TbPRMT7 and NNMT were expressed and purified as described before.^33–37

Kinetic Analysis of NTMT1 inhibitors.

A fluorescence-based SAHH-coupled assay was applied to study the IC₅₀ values of all synthesized compounds, and inhibition mechanism of GD433 and YD2200. The methylation assay was performed under the following conditions in a final well volume of 40 μL: 25 mM Tris (pH = 7.5), 50 mM KCl, 0.01% Triton X-100, 5 μM SAHH, 0.2 μM NTMT1, 6 μM AdoMet, and 10 μM ThioGlo4. The inhibitors were added at concentrations ranging from 0.15 nM to 33 μM. After 10 min incubation, reactions were initiated by the addition of 1 μM GPKRIA peptide (Km value). To study the inhibition mechanism, varying concentrations of SAM (from 3 to 48 μM) with 1 μM fixed concentration of GPKRIA or varying concentration of GPKRIA (from 0.5 to 10 μM) with 6 μM fixed concentration of SAM were included in reactions at concentration of inhibitors ranging from 0.15 nM to 10 μM. All the IC₅₀ values were determined in triplicate. Fluorescence was monitored on a BMG CLARIOstar microplate reader with excitation 400–415 nm and emission 460–485 nm. Data were processed by using GraphPad Prism software 8.0.

Selectivity Study.

A similar fluorescence-based SAHH-coupled assay described above was applied to study the effects of GD433 on methyltransferase activities of G9a, SETD7, PRMT1 NNMT, and TbPRMT7. GD433 was tested at three final concentrations: 10 μM, 30 μM, and 100 μM. For G9a, the assay was performed in a final well volume of 100 μL: 25 mM potassium phosphate buffer (pH = 7.6), 1 mM EDTA, 2 mM MgCl2, 0.01% Triton X-100, 5 μM SAHH, 0.1 μM His-G9a, 13 μM SAM, and 10 μM ThioGlo4. After 10 min incubation with GD433, reactions were initiated by the addition of 2.5 μM H3-21 peptide (final concentration). For SETD7, the assay was performed in a final well volume of 100 μL: 25 mM potassium phosphate buffer (pH = 7.6), 0.01% Triton X-100, 5 μM SAHH, 1 μM His-SETD7, 2 μM SAM, and 10 μM ThioGlo4. After 10 min incubation with GD433, reactions were initiated by the addition of 90 μM H3-21 peptide (final concentration). For PRMT1, the assay was also performed in a final well volume of 100 μL: 2.5 mM HEPES (pH = 7.0), 25 mM NaCl, 25 μM EDTA, 50 μM TCEP, 0.01% Triton X-100, 5 μM SAHH, 0.2 μM PRMT1, 10 μM SAM, and 10 μM ThioGlo4. After 10 min incubation with GD433, reactions were initiated by the addition of 5 μM H4-21 peptide (final concentration). For NNMT, the assay was performed in a final well volume of 100 μL: 25 mM Tris (pH = 7.5), 50 mM KCl, 0.01% Triton X-100, 5 μM SAHH, 0.2 μM NNMT, 2.5 μM SAM, and 10 μM ThioGlo4. After 10 min incubation with GD433, reactions were initiated by the addition of 2.5 μM nicotinamide (final concentration). For TbPRMT7, the assay was performed in a final well volume of 100 μL: 25 mM Tris (pH = 7.5), 50 mM KCl, 0.01% Triton X-100, 5 μM SAHH, 0.2 μM TbPRMT7, 3 μM SAM, and 10 μM ThioGlo4. After 10 min incubation with GD433, reactions were initiated by the addition of 60 μM H4-21 peptide (final concentration). All experiments were determined in duplicate. Fluorescence was monitored on a BMG Clariostar microplate reader with excitation 400 nm and emission 465 nm. The rates were fit to the log[inhibitor] vs response model using least squares non-linear regression through GraphPad Prism 8 software. All experiments are performed in duplicate.

GD433 was evaluated against a panel of 37 SAM dependent methyltransferases at Reaction Biology Corp., where all protein substrates were at their physiological concentrations.

GCS selectivity was performed based on the published literature with some minor modification.³¹ HCT116 cells were seeded as 250000 cells/well in 10% FBS McCoy 5A’s growth medium on 12-well plates in the presence of DMSO or the small molecule inhibitor at different concentrations for 48 h. The growth medium was replaced by 1% FBS McCoy 5A’s growth medium containing 5 μM NBD-C6-Ceramide and incubated with either DMSO or small molecule inhibitors at different concentrations for 4 h. Cells suspended in ice-cold acidic methanol (acetic acid: methanol, 1:50, v/v) were transferred into glass vials (1 mL), and mixed with dichloromethane (1 mL) and water (1 mL) to extract cellular lipids. After separation, the organic lower phase was collected and evaporated to dryness, and the residue was redissolved in 100 μL MeOH, filtered, and then analyzed by LC-MS.

Isothermal Titration Calorimetry.

ITC measurements were performed at 25 °C using a MicroCal PEAQ-ITC calorimeter. For titrations without SAM, purified NTMT1 was diluted with 1x buffer (20 mM Tris-HCl, 50 mM KCl, pH 7.4) to the final concentration of 20 μM, and the compound was dissolved to 100 μM in the same buffer. For titrations in the presence of SAM, 20 μM NTMT1, and 200 μM compound solutions were added with 100 μM (final concentration) SAM in 1x buffer. The compound was titrated into the protein solution with 18 injections of 2 μL each. Injections were spaced 150 s with a reference power of 10 μcal/s. The titration curves were fitted to obtain the association constant (1/K_d), enthalpy of binding (ΔH), and stoichiometry of binding (N) using MicroCal PEAQ-ITC analysis software. All data were fitted using the one-site binding model.

Crystallization and Structure Determination.

The co-crystallization and structure determination of GD433 with NTMT1 was performed as previously described.²⁴ To 5 mL of purified NTMT1 (0.9 mg/mL in 1x TBS buffer, 0.036 mM) was added 270 μL (1.5 eq) of SAM (1 mM in 1x TBS buffer), the resulting mixture was incubated for 30 min under 4 °C. Then, 20 μL (11.3 eq) of GD433 (100 mM in ddH₂O) was diluted 10 folds with 1x TBS buffer, which was added to the previous mixture and incubated at 4 °C for 1 h. The mixture was concentrated to 25 – 30 mg/mL of NTMT1 with Amicon Ultra centrifugal filters (10.0 KDa cutoff MW) at 1800 rpm under 4 °C. The sample was then filtered using spin filters (0.22 mM) to remove any precipitates. Broad matrix crystallization screening was performed using a mosquito-LCP high-throughput crystallization robot (SPT LabTech) using the hanging drop vapor diffusion method at 20 °C. Crystals containing GD433 were grown in 0.1 M CHES: NaOH (pH 9.5), 10% (w/v) PEG 3000. The crystals were then harvested and stored in liquid nitrogen. Data were collected on single crystals at 12.0 keV on the GM/CA-CAT ID-D beamline at the Advanced Photon Source, Argonne National Laboratory. The data was processed and the structure was solved as previously described.²⁴

Molecular Docking and MM-GBSA Rescoring.

Glide of Schrödinger Maestro, NY, USA (Version 10.1) was used to predict the binding of all synthesized compounds against the active site of NTMT1. The ligand chemical structures were drawn with Ligprep module and ionization state was generated at pH 7.0 ± 2.0 using Epik module. The complex structure (PDB ID, 7U1M) of NTMT1 bound with venglustat was obtained from the Protein Data Bank (www.rcsb.org). Structures were prepared and refined with the protein preparation module and the energy was minimized using OPLS_2005 force field. Grids were generated with Glide by adopting the default parameters. A cubic box of specific dimensions centered around the active site residues was generated for the receptors. The bounding box was set to 15 Å × 15 Å × 15 Å. Flexible ligand docking was performed. Glide extra precision docking was performed by keeping all docking parameters as default. Ligand poses were generated by using Monte Carlo random search algorithm, and its binding affinities to NTMT1 were predicted with Glide docking score. Post-docking minimizations were taken under OPLS_2005 force field, and 3 poses per ligand were saved. The predicted binding poses of each compound were then rescored by using Prime MM-GBSA method. Finally, the pIC₅₀ value vs MM-GBSA dG binding energy of all compounds was plotted in GraphPad prism software 8.0.

Cellular N-Methylation Level Study.

HEK293 cells were seeded as 3,000 cells/well on 24-well plates in the presence of DMSO or inhibitors at different concentrations for 3 days. Medium with inhibitors or DMSO was removed, and cells were washed with PBS for two times. Cells were pelleted to 1.5 mL centrifuge tubes and then lysed with cell lysis buffer (25 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 5% glycerol, 1% NP-40, 0.1% SDS, and protease inhibitors) and incubated for 30 min on ice. The cell lysates were centrifuged at 15,000 g for 25 min, and the supernatant was collected. The concentration of total protein was quantified using a BCA protein assay kit (Thermo Fisher, #23228). Cell lysates were mixed with 5× loading dye and equal amounts of total protein was loaded onto a 10% SDS-PAGE gel and separated. The gel was transferred to a polyvinylidene difluoride membrane using a Bio-Rad Trans-Blot Turbo system. The membrane was then blocked for 1 h in 5% milk TBST solution and washed with 1× TBST solution 3 times. The membrane was incubated with the anti-me3-RCC1 antibody (1:2000), SET (1:1000, Proteintech #55201-1-AP), and RCC1 antibody (1:1000, Proteintech # 22142-1-AP) at 4 °C overnight, washed with 1× TBST solution 3 times, and then incubated with rabbit IgG-HRP antibody (1:1000, cell signaling #7074) for 1 h at 4 °C. The membrane was again washed with 1× TBST solution 3 times and detected using a Protein Simple FluorChem imaging system.

Pharmacokinetic and Maximum Tolerated Dose Study.

Pharmacokinetic study of GD433 was tested at 25 mg/kg with oral dose in CD1 mice. Plasma was collected at 0083, 0.25, 0.5, 1, 2, 4, 8, and 24 h. And the services were provided by Pharmaron Corp. MTD studies were performed at the Purdue University Center for Cancer Research Biological Evaluation Shared Resource. Briefly, balb/c mice were randomly divided into 3 groups, 3 in each group. A saline solution with 5% DMSO of GD433 (40 mg/mL) was administrated through oral gavage. The doses of each group were 250, 300 and 350 mg/kg. The body weight of the mice was weighed daily and observed for animal death. MTD is defined as the dose of drug that does not cause animal death or a 20% reduction in animal weight during the course of the experiment.

ABBREVIATIONS

NTMT

protein N-terminal methyltransferase

SAM

S-5’-adenosyl-L-methionine

SAH

S-5’-adenosyl-L-homocysteines

SAHH

SAH hydrolase

RCC1

regulator of chromosome condensation 1

PKMT

protein lysine methyltransferase

PRMT

protein arginine methyltransferase

PRMT1

protein arginine methyltransferase 1

MRG15

MORF-related gene 15

GCS

glucosylceramide synthase

rt

room temperature

TFA

trifluoroacetic acid

EtOAc

Ethyl acetate

DME

Glycol dimethyl ether

LiOH

Lithium hydrate

THF

Tetrahydrofuran

HCl

Hydrochloric acid

DMF

Dimethylformamide

Et₃N

Triethylamine