3-Amido-3-aryl-piperidines: A Novel Class of Potent, Selective, and Orally Active GlyT1 Inhibitors

Abstract

3-Amido-3-aryl-piperidines were discovered as a novel structural class of GlyT1 inhibitors. The structure–activity relationship, which was developed, led to the identification of highly potent compounds exhibiting excellent selectivity against the GlyT2 isoform, drug-like properties, and in vivo activity after oral administration.

Schizophrenia is a devastating and chronic mental illness that affects close to 1% of the population. Symptoms of schizophrenia, which typically arise at young age (adolescence or early adulthood), are categorized as positive, negative, or cognitive. Currently approved drugs are efficacious in the treatment of positive symptoms but do not address the negative and cognitive symptoms. Numerous lines of evidence suggest that hypofunction of glutamatergic transmission via N-methyl-d-aspartate (NMDA) receptors may represent the final common pathway leading to symptoms in schizophrenic patients.¹ One approach to normalize the reduced NMDA receptor activity is to elevate the concentration of the coagonist glycine at its modulatory site on the receptor through blockade of the glycine transporter type 1 (GlyT1), which is colocalized with the NMDA receptor.² In the past decade, ample efforts have been focused on the discovery and development of selective GlyT1 inhibitors.³ The first examples reported were sarcosine derivatives including 1(4) and 2(5) (Figure 1).

Figure 1.

Selection of published GlyT1 inhibitors.

More recently, nonsarcosine-based compounds⁶ have been described such as DCCCyB (3),⁷ GSK1018921 (4),⁸ and bitopertin (5),^9,10 which all progressed to clinical studies (Figure 1). In a phase II proof of concept study, Bitopertin demonstrated a beneficial effect in patients with schizophrenia characterized with predominant negative symtoms.¹¹ However, recently, Hoffmann-La Roche announced that two phase III studies of bitopertin did not meet their clinical end points.¹² Four additional phase III studies with bitopertin are ongoing.

As part of our continued effort to discover and develop novel and selective GlyT1 inhibitors, we considered the pyrrolidino-ethyl-amide 4 as a potential starting point for further structural modification. In the low energy conformation obtained for compound 4 (Figure 2), the carbon (C1) adjacent to the pyrrolidine nitrogen and the benzylic carbon (C2) bearing the phenyl and amide substituents appeared as suitable attachment points for forming, via a linker, a novel cyclic system (Figure 2). We thus decided to investigate this cyclization approach further. As connecting elements between C1 and C2, a bond, a methylene, and an ethylene unit were evaluated (in red, Figure 2). To make our designed molecules chemically tractable, the gem dimethyl group and the pyrrolidine ring present in 4 were deleted. As a result, our approach led us to the generation of the 3,3′-substituted-N-methyl-azetidine 6, pyrrolidine 7, and piperidine 8 (Figure 2). In a glycine uptake inhibition assay performed in CHO cells transfected with hGlyT1, azetine 6 and pyrrolidine 7 were found inactive or poorly active. In contrast, and to our delight, piperidine 8 displayed a promising GlyT1 potency with an IC₅₀ of 140 nM. The X-ray crystal structure of 8(13) (Figure 2) showed that the amide and phenyl substituents at the position 3 on the piperidine ring adopted, respectively, an axial and equatorial orientation. In addition, because of the presence of the two ortho groups on the benzamide side, the carbonyl function adopted a highly twisted orientation with a torsion angle of 78°. Overall, as depicted in Figure 2, the crystal structure of 8 aligned very well with the low energy conformation of 4, a result perfectly in line with the good GlyT1 potency measured for 8. A literature and patent search revealed that, although quite structurally simple, 3-amido-3-aryl-piperidines such as 8 had not been previously described. Thus, the discovery of 8 offered us the unique opportunity to develop a novel series of potent GlyT1 inhibitors having an exquisite intellectual proprietary status. In this letter, we report on our structure–activiy relationship (SAR) exploration effort in this series¹⁴ that culminated in the identification of potent, selective, and orally active GlyT1 inhibitors such as compound 29.

Figure 2.

Design principles leading to the identification of active 3-amido-3-aryl-piperidine 8 using pyrrolidino-ethyl-amide 4 as a seed structure.

N-Methyl-3-amido-3-aryl-piperidine derivatives 8 and 18–43 were synthesized under standard amide coupling conditions from N-methyl-3-amino-3-aryl-piperidines 44 (Scheme 1). Derivatives 10–17 having different R₁ groups on the piperidine nitrogen were obtained upon alkylation, reductive amination, acylation, or sulfonylation of the unsubstituted piperidine 9, itself prepared from N-Boc-protected-3-amino-3-phenyl-piperidine building block 45. N-Methyl-3-amino-3-aryl-piperidine intermediates 44a–d carrying the aromatic R₃ substituents indicated in Scheme 2 were prepared from the 3-aryl-3-piperidin-ol derivatives 46a–d under modified Ritter reaction conditions¹⁵ in the presence of sodium azide and trifluoroacetic acid, followed by reduction of the obtained azido intermediates 47a–d with lithium aluminum hydride.

Scheme 1. Synthesis of 3-Amido-3-aryl-piperidines 8–43.

Reagents and conditions: (a) R₂COCl, DIPEA, CH₂Cl₂, RT, 4–90%; (b) R₂CO₂H, HATU, DIPEA, DMF, RT, 18–68%; (c) 2- OMe,2′,4-(CF₃)₂-PhCOCl, DIPEA, CH₂Cl₂, RT, 86%; (d) HCl, dioxane, RT, 86%; (e) R₁I, DIPEA, CH₂Cl₂, RT, 58%; (f) ketone or aldehyde, NaCNBH₃, AcOH, MeOH, RT, 63–89%; (g) acyl- or sulfonyl-chloride, DIPEA, CH₂Cl₂, RT, 65–99%.

Scheme 2. Synthesis of N-Methyl-3-amino-3-aryl-piperidines 44a–d.

Reagents and conditions: (a) NaN₃, TFA, H₂O, RT, 87–100%; (b) LiAlH₄, THF, RT, 32–63%.

However, this synthetic approach turned out to be not versatile enough. In particular, the azidation reaction performed on 3-aryl-3-piperidinols carrying electron rich aryl R₃ groups such as 4-OMe-Ph failed. During the search of a more general access route, we discovered that N-methyl-3-amino-3-aryl-piperidine intermediates 44 could be efficiently prepared from easily accessible 4-aryl-4-nitro-butyric acid methyl esters 51 via a novel route involving as a key step a Mannich in situ lactamization reaction as depicted in Scheme 3. When performed in the presence of methylamine and formaldehyde, this sequence provided with good to excellent yields N-methyl-5-aryl-5-nitro-piperidin-2-one intermediates 52, which led to the target building blocks 44 after the reduction of the nitro and carbonyl functions. In addition, the synthetic access route we have established for 44 was successfully applied to the preparation of N-Boc-3-amino-3-phenyl-piperidine building block 45 using ammonium acetate and formaldehyde as reagents in the key Mannich in situ lactamization reaction (Scheme 3).

Scheme 3. Synthesis of N-Methyl-3-amino-piperidines 44f–o and N-Boc-3-amino-3-phenyl-piperidine 45.

Reagents and conditions: (a) methyl acrylate, amberlyst A-21, dioxane, RT, 41–85%; (b) cat. Pd₂dba₃, cat. di-tert-butyl-(2′-methyl-biphenyl-2-yl)-phosphane, Cs₂CO₃, DME, reflux, 10–90%; (c) MeNH₂ (41% in water), CH₂O (36% in water), dioxane, RT then 65 °C, 49–94%; (d) zinc, HCl, dioxane, RT, 13–86%; (e) LiAlH₄, THF, RT, 71–86%; (f) Lawesson’s reagent, toluene, 80 °C, 77–100%; (g) NaBH₄, MeOH, RT, 49–76%; (h) Ra–Ni, H₂ (1 atm.), THF, 0 °C, 87–100%; (i) NH₄OAc, CH₂O (36% in water), EtOH, reflux, 84%; (j) Boc₂O, Et₃N, CH₂Cl₂, RT, 70%.

Our first effort aimed at establishing the SAR at the piperidine nitrogen (Table 1). Starting from 8, deletion of the methyl group (compound 9) led to a 5-fold reduction in GlyT1 activity. A drop of activity was equally observed upon replacing the methyl group with larger alkyl (10, 11) or cycloalkyl groups (12, 13). Moreover, N-benzylation (14) as well as N-acylation and sulfonylation (15–17) practically abolished the GlyT1 potency suggesting that a certain level of basicity of the piperidine nitrogen is required for GlyT1 activity. Overall, our exploration at this position revealed a highly restricted SAR with the methyl group being the only well tolerated substituent.

Table 1. In Vitro Inhibitory Activity of 9–17 at GlyT1^a.

compd	R1	GlyT1 EC50 (μM)b
9	H	0.66
10	Et	0.23
11	i-Pr	0.23
12	c-C5H9	0.27
13	CH2-c-Pr	0.38
14	CH2–Ph	6.7
15	C(O)–Me	>10
16	C(O)–Ph	>10
17	SO2Me	>10

^aEC₅₀ values are the average of at least two independent experiments.

^b[³H]-glycine uptake inhibition assay in cells transfected with hGlyT1.¹⁰

Next, keeping the preferred N-methyl group in place, the SAR evaluation at the benzamide region was conducted (Table 2). Starting from 8, selective deletion of the ortho, ortho′, and para substituents (compounds 18 to 21) resulted in a significant drop of GlyT1 potency suggesting that substitution at all three positions on the benzamide moiety is important for reaching a good level of potency. In particular, the lack of activity observed for the p-CF₃ monosubstituted compound 21, devoid of any ortho groups suggested that for GlyT1 activity the amide carbonyl group should adopt a twisted oriention as seen in the X-ray structure of our initial compound 8. Next, starting from inactive compound 21 and keeping the p-CF₃ group in place, we explored various substituents at the ortho position (compounds 22–26). No improvement of activity was observed with polar groups in place such as nitrile (22) as well as with small substituents like fluorine (23). With larger lipophilic groups such as methyl (24) or bromine (25), the activity improved by more than 30-fold. Further exploration in that direction resulted finally in the identification of the thiomethyl group as one of the best substituents at the ortho position providing a compound displaying a low nanomolar activity (26, 67 nM). Our initial SAR studies (vide supra) suggested that a further gain in activity could be predicted upon introducing a second ortho group on 26. Pleasingly, by adding a methoxy group (27), a very potent compound was generated demonstrating an EC₅₀ of 24 nM. This compound existing as a racemic mixture was subsequently separated by chiral HPLC to provide the two enantiomers 28 ((S)-enantiomer) and 29 ((R)-enantiomer). The R-configurated isomer 29(16) showed the best GlyT1 potency, reaching an EC₅₀ as low as 10 nM at the hGlyT1 (5 nM at the mouse GlyT1).

Table 2. In Vitro Inhibitory Activity of 18–30 at GlyT1^a.

compd	R2	GlyT1 EC50 (μM)b
18	2-OMe, 6-CF3–Ph	1.5
19	2-OMe, 4-CF3–Ph	4.8
20	2-CF3, 4-CF3–Ph	0.54
21	4-CF3–Ph	>10
22	2-CN, 4-CF3–Ph	>10
23	2-F, 4-CF3–Ph	9.9
24	2-Me, 4-CF3–Ph	0.32
25	2-Br, 4-CF3–Ph	0.30
26	2-SMe, 4-CF3–Ph	0.067
27	2-SMe, 2′-OMe, 4-CF3–Ph	0.024
28	S-enantiomer of 27	0.18
29	R-enantiomer of 27	0.010
30	2,4-Cl2–Ph	0.67

^aEC₅₀ values are the average of at least two independent experiments.

^b[³H]-glycine uptake inhibition assay in cells transfected with hGlyT1.¹⁰

Finally, the SAR around the aryl group at position 3 on the piperidine ring was explored (Table 3).

Table 3. In Vitro Inhibitory Activity of 31–43 at GlyT1^a.

compd	R3	GlyT1 EC50 (μM)b
31	4-F–Ph	0.026
32	4-Cl–Ph	0.028
33	4-Me–Ph	0.23
34	4-OMe–Ph	0.17
35	3-F–Ph	0.048
36	3-Cl–Ph	0.025
37	3-OMe–Ph	0.110
38	pyridin-3-yl	0.054
39	pyridin-4-yl	0.060
40	5-F-pyridin-2-yl	0.180
41	pyrazinyl-2-yl	1.670
42	pyrimidin-2-yl	>10
43	pyrimidin-4-yl	1.7

^aEC₅₀ values are the average of at least two independent experiments.

^b[³H]-glycine uptake inhibition assay in cells transfected with hGlyT1.¹⁰

This study was performed keeping the optimized 2-thiomethyl-2-methoxy-4-trifluoromethyl-phenyl group in place at the benzamide moiety. A very similar SAR was seen at the para and meta position on the aryl moiety. Indeed, at both positions, introduction of EWGs like fluorine or chlorine (compounds 31–32 and 35–36) was well tolerated (EC₅₀ = 23–48 nM), whereas with EDGs such as methoxy group (34, 37) the potency dropped significantly (EC₅₀ = 110–170 nM). Moreover, a 6-membered-ring-heteroaryl scan revealed that the meta and para positions were the best centers to insert a nitrogen atom as exemplified with compounds 38 (54 nM) and 39 (60 nM). All the synthetized heteroaryl derivatives having a nitrogen atom in ortho (40–43) displayed much reduced potencies.

From our SAR and optimization studies, compound 29 emerged as the most potent congener from our newly discovered piperidine series. Further profiling revealed that 29 had excellent selectivity against GlyT2 (15 μM) as well as a fairly good selectivity profile in a CEREP screen performed against a panel of 80 targets including transmembrane and soluble receptors, ion channels, and monoamine transporters.¹⁷ Moreover, as indicated in Table 4, 29 displayed good physicochemical properties with a moderate lipophilicity, a good aqueous solubility and membrane permeability. The metabolic stability of 29 in mouse and human liver microsomes was found to be intermediate with a measured intrinsic clearance of 54 and 19 μL/min/mg protein, respectively. Not surprisingly, in vitro metabolism studies on 29 revealed that the corresponding N-des-methyl piperidine derivative was a major metabolite after incubation in human hepatocytes. In agreement with our SAR results (vide supra), this compound showed only weak GlyT1 activity (EC₅₀ = 0.45 μM). Gratifyingly, 29 showed low inhibitory activity at the hERG potassium channel (IC₂₀ = 5.4 μM), a result that may be attributed to the relatively low basicity of its piperidine nitrogen (pK_a = 7.72). For comparison, pyrrolidino-ethyl-amide seed structure 4, which displayed a higher basicity (pK_a = 9.12), showed a significantly higher hERG activity (IC₂₀ = 0.7 μM). Moreover, 29 demonstrated a low level of inhibition at the major drug metabolizing CYP450 enzymes (IC₅₀ > 16 μM) and was inactive in the genotoxicity assays (Ames and MNT tests). Finally, 29 was not a substrate for the human and mouse P-gp efflux systems.

Table 4. In Vitro Properties of 29.

parameter	value
log Da	3.12
solubilityb (μg/mL)	147
pampa pec (10–6 cm/s)	7.4
pKad	7.72
human, mouse CLint (μL/min/mg protein)	19, 54
hERG IC20e (μM)	5.4
CYP450 3A4, 2D6, 2C9, 2C19 IC50 (μM)	>50, 16, >50, 46
human, mouse P-gp ERf	0.9, 1.5

^aDetermined in 1-octanol/phosphate buffer, pH = 7.4.

^bMeasured in a lyophilization solubility assay (LYSA) at pH = 6.5.¹⁸

^cPe: PAMPA permeation constant through artificial membranes.¹⁹

^dDetermined by capillary electrophoresis.

^ePatch clamp assay.

^fER = efflux ratio in LLC-PK1 cells stably expressing human MDR1 and mouse MDR1A.

In vivo pharmacokinetic studies in mouse (Table 5) revealed, in addition, that 29 had an attractive profile, with, in particular, a complete oral bioavailability despite a medium systemic clearance and an excellent brain penetration (brain/plasma = 4.3), a result in agreement with its low P-gp efflux ratio (1.5) and large volume of distribution. Interestingly, 29 demonstrated a much superior brain penetration profile over seed compound 4 for which a P-gp efflux ratio of 3.9 and a brain/plasma ratio of only 0.4 was measured in the mouse.

Table 5. Pharmacokinetics Properties of 29 in Mouse.

parameter	valuea
iv dose (1.7 mg/kg)
CL (mL/min/kg)	54
Vss (L/kg)	8
po dose (7.7 mg/kg)
bioavailability F (%)	100
t1/2 (h)	4.2
brain/plasma (at 1.5 h)	4.3
PPB (% unbound)	1.7

^aValues are the average of two independent experiments.

These favorable data led us to evaluate the oral effect of 29 in the mouse L-687,414-induced hyperlocomotion assay, a novel and straightforward functional screening model,²⁰ that allows the detection of the in vivo activity of GlyT1 inhibitors (Figure 3). Pleasingly, 29 displayed a robust pharmacodynamic activity after oral administration and reached an ID₅₀ (dose producing 50% inhibition of L-687,414-induced hyperlocomotion) of 1.5 mg/kg po. Noteworthy, the in vivo efficacy of 29 was reached at a very low plasma concentration (8 ng/mL), a result that can be explained by its high GlyT1 potency and its excellent brain penetration in the mouse.

Figure 3.

Dose-dependent inhibition of L-687,414-induced hyper-locomotion by 29 in mice. The data represent mean horizontal activity counts per group recorded over a 60 min time period; error bars indicate SEM (n = 8 per group). ^##, p < 0.01 versus vehicle (Veh) alone; ^***, p < 0.001 versus L-687,414 alone.

In addition, the effect of 29 on the extracellular level of glycine in rat striatum in a microdialysis study was evaluated (Table 6). We were pleased to observe that, at an oral dose of 10 mg/kg, 29 produced a 1.7-fold glycine increase over basal levels.

Table 6. Effect of 29 on the Extracellular Glycine Levels in Rat Striatum at 10 mg/kg po^a and Its Plasma and Brain Exposures.

max. fold incr. of glycine	fold incr. of glycine at 3 h	plasmab (ng/mL)	brainb (ng/mL)	brainb GlyT1 EC50
1.7	1.5	713	60	13

^an = 6.

^bMeasured at 3 h postdosing.

After 3 h, at which time the extracellular fluid sampling was stopped, the PD effect was slightly diminished (1.5-fold). Determination of the plasma and brain exposures at 3 h postdosing revealed a much lower brain penetration (B/P: 0.08) for 29 in this rat experiment compared to its penetration measured in the mouse (B/P: 4.3) (vide supra). In spite of this, 27 achieved a total brain exposure above its GlyT1 EC₅₀ (13-fold), sufficient to produce the effect on glycine levels observed at the 3 h time point.²¹

Noteworthy, in a dedicated mechanistic study performed in the rat, we observed that the brain/plasma ratio increased by 4-fold after coadministration of 29 with the P-gp inhibitor tariquidar²² indicating that 29 is a rat P-gp substrate. This result is in sharp contrast with the low P-gp activity of 29 measured in the mouse (vide supra). The marked difference in brain penetration observed in the mouse and in the rat may thus be attributed to the differentiated P-gp activity profile of 29 in these two rodent species. Compound 29, which is not a substrate of the human P-gp, is thus predicted to display in human an excellent brain penetration, quite similar to the one measured in the mouse.

In summary, we report here on the discovery of 3-amido-3-aryl-piperidines, a novel structural class of GlyT1 inhibitors designed using the previously reported pyrrolidino-ethyl-amide 4 as a seed structure. From the initial compound 8, the exploration of the SAR at the three exit vectors has led to the identification of potent and selective compounds such as 29, which demonstrated drug-like properties, a promising in vitro safety profile and in vivo efficacy in rodents after oral administration.

Glossary

ABBREVIATIONS

CYP

cytochrome P

dba

dibenzylideneacetone

DIPEA

N,N-diisopropylethylamine

EWG

electron withdrawing group

EDG

electron donating group

HATU

1-[bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyri-dinium 3-oxid hexafluorophosphate

hERG

human ether-à-go-go-related gene

MDR

multidrug resistance protein

TFA

trifluoro acetic acid

Supporting Information Available

Experimental details for the synthesis of compounds 8–43 and intermediates. This material is available free of charge via the Internet at http://pubs.acs.org.

Author Contributions

The manuscript was written through contributions of all authors.

The authors declare no competing financial interest.