Improved Synthesis of Chiral Pyrrolidine Inhibitors and Their Binding Properties to Neuronal Nitric Oxide Synthase

Abstract

We report an efficient synthetic route to chiral pyrrolidine inhibitors of neuronal nitric oxide synthase (nNOS) and crystal structures of the inhibitors bound to nNOS and to endothelial NOS. The new route enables versatile structure activity relationship studies on the pyrrolidine-based scaffold, which can be beneficial for further development of nNOS inhibitors. The X-ray crystal structures of three new fluorine-containing inhibitors bound to nNOS provide insights into the effect of the fluorine atoms on binding.

INTRODUCTION

Selective inhibition of neuronal nitric oxide synthase (nNOS) over its closely related isozymes, inducible nitric oxide synthase (iNOS) and endothelial nitric oxide synthase (eNOS), has attracted tremendous drug discovery efforts for neurotoxicity and certain neurodegenerative conditions such as Parkinson’s, Alzheimer’s, Huntington’s diseases, and cerebral palsy.^1–5 As part of our research program directed toward developing novel nNOS inhibitors, we recently disclosed a series of gem-difluorinated monocationic inhibitors including compounds 1a and 1b, which are potent and highly selective inhibitors of nNOS.^6,7 Moreover, according to the results of detailed pharmacokinetic studies,^7,8 compound 1b has demonstrated intriguing cell membrane permeability and oral bioavailability, which makes it a promising drug candidate for neurodegenerative diseases.

Although they possess desirable potency, selectivity, and pharmacokinetic properties, it is difficult to prepare gram-scale quantities of 1b for a comprehensive preclinical study.^7,8 The previously reported route takes 16 steps starting from 4,6-dimethyl-2-aminopyridine (Scheme 1). Several of the steps suffered from unsatisfactory yields, resulting in an overall yield of less than 1%.^7,8 In particular, the late stage benzyl deprotection step, employing a high temperature catalytic hydrogenation of the N-Boc-N-benzyl-protected intermediate 2, proceeded poorly to give 3. As a result, the utility of this route was dramatically compromised by the present lack of scalability. More importantly, the strong reducing conditions used to remove the benzyl protecting group prohibited the possibility of incorporating reduction sensitive functional groups (e.g., nitriles, ketones, alkenes, and halophenyls) into the inhibitors, which significantly impaired structure-activity optimization based on this scaffold. For instance, attempts to remove the benzyl protecting group of 4 led to dechlorination (5, Scheme 2). Removal of the benzyl protecting group of 6 by catalytic hydrogenation led to a partial reduction of the cyclopropyl ring in 7.⁹

Scheme 1.

Previous Synthetic Route to 1a–b.

Scheme 2.

We report here a synthesis with improved overall yield and functional group compatibility. In the current method, the problematic benzyl protecting group in the previous synthesis is replaced by a Boc-protecting group. The new route not only produces 1b in an enhanced overall yield, but also provides a way to synthesize inhibitors with reduction-sensitive chlorophenyl groups.

CHEMISTRY

As shown in Scheme 3, the synthesis of key precursor 9 began with Boc-protected 4,6-dimethylaminopyridine 10. Compound 10 was treated with two equivalents of n-butyllithium (n-BuLi), and the resulting dianion was allowed to react with tert-butyl 6-oxa-3-azabicyclo[3.1.0]hexane-3- carboxylate⁶ to generate the trans-alcohol (11) in modest yields. Attempts to convert 11 directly to 14 failed because the free hydroxyl group of 11 is more reactive toward (Boc)₂O than the carbamate NH group, and the hydrolytic stability of the formed carbonate is very similar to that of the di-Boc-protected aminopyridine. Therefore, we had to use a three-step procedure to convert 11 to 14. Di-Boc protection of 10 also was fruitless because one of the Boc groups is transferred to the oxyanion of 11 upon condensation of di-Boc-9 with tert-butyl 6-oxa-3-azabicyclo[3.1.0]hexane-3-carboxylate, as described earlier.^10a The hydroxyl group of 11 was first protected with t-butyldimethylsilyl chloride (TBSCl) in the presence of imidazole to give the silyl ether (12) in excellent yields. Next, the NH of the carbamate group on the pyridine ring was further protected with another Boc protecting group using (Boc)₂O in the presence of 4-dimethylaminopyridine (DMAP) to yield 13 in high yields, then the silyl ether was cleaved using tetrabutylammonium fluoride (TBAF) to provide the tri-Boc protected alcohol (14) in very high yields. Finally, the two enantiomers of 14 were resolved through their camphanic ester derivatives (15a and 15b) employing a Mitsunobu reaction using (S)-(−)-camphanic acid as the nucleophile. The ester linkage of 15b was hydrolyzed carefully using Na₂CO₃ in MeOH/H₂O to provide chiral precursor 9 in excellent yields.

Scheme 3. Synthesis of 9.^a.

^a Reagents and conditions: (a) (i) n-BuLi (2eq.), −78 °C to rt, (ii) epoxide, −78 °C to rt, 41%; (b) TBSCl, imidazole, DMF, rt, 30 h, 93%; (c) (Boc)₂O, DMAP, rt, 24 h, 95%; (d) TBAF, rt, 15 min, 99%; (e) (S)-(−)-camphanic acid, DIAD, rt, 16 h, 94%; (f) Na₂CO₃, H₂O/MeOH, 35 °C, 30 min, 97%.

Allylation of chiral cis-alcohol 9 in the presence of allyl methyl carbonate using Pd(PPh₃)₄ as a catalyst provided 16 in good yields (Scheme 4).^10–11 Alkene 16 was treated with ozone, followed by dimethyl sulfide to generate 17 in excellent yields. Aldehyde 17 underwent a reductive amination reaction with the corresponding amines in the presence of NaHB(OAc)₃ to give 18, 19a–e in high yields. Finally, the three Boc-protecting groups were removed simultaneously in HCl to provide the final inhibitors in excellent yields.

Scheme 4. Synthesis of 1b and 8a–e.^a.

^a Reagents and conditions: (a) allyl methyl carbonate, Pd(PPh₃)₄, 45 °C, 5 h, 66%; (b) O₃, −78 °C, 30 min, (ii) Me₂S, −78 °C to rt, 2 h, 87%; (c) amine hydrochloride, triethylamine, NaHB(OAc)₃, rt, 3 h, 86–91%; (d) 6 N HCl in MeOH (2:1), rt, 12 h, 90–99%.

RESULTS AND DISCUSSION

We have determined the crystal structures of rat nNOS with inhibitors 8a–e bound and that of bovine eNOS in complex with 8c. The pdb codes are given in Supporting Information Table 1. As expected from the chirality of their pyrrolidine moiety, inhibitors 8a–e adopt the same flipped binding mode as lead compound 1a–b (Figure 1).⁷ The aminopyridine motif extends to the surface hydrophobic pocket (Tyr706, Leu337, and Met336) in nNOS, forming a charge-charge interaction with the heme propionate D, and a π–π stacking interaction with the aromatic side chain of Tyr706 in its newly adapted conformation. Substitution of the m-fluoro atom in the phenyl tail of 1b with a larger chlorine atom slightly increases the potency (1.3-fold) of the inhibitor (8a vs 1b, Table 1). A comparison between the structures of nNOS-8a (Fig. 1A) and nNOS-1b provides some clues as to why nNOS prefers the m-chloro over m-fluoro atom on the phenyl group. Similar to what was observed for nNOS-1b⁷ the 2,2-difluoro-2-(m-chlorophenyl)ethyl moiety of 8a has adopted two conformations, with the CF₂ group pointing both away from (major) and toward (minor) the heme. Residual Fo-Fc difference density is present especially around the fluorine atoms if only one conformation is modeled clearly, indicating two conformations. The two different conformations also result in two slightly different orientations of the chlorophenyl group, which is consistent with two conformations. However, because of the bulkier chlorine atom in 8a there is increased Van der Waals contact and the phenyl ring has been pushed closer to the Glu592 side chain, but is still well tolerated. The Glu592 side chain shows two rotamers in the nNOS-8a structure, which were seen in nNOS-1b as well. The new Glu592 rotamer forms a new hydrogen bond to the amino nitrogen of the inhibitor (Fig. 1A). The active site of eNOS might alsoaccommodate the m-chlorophenyl ring better; therefore, the selectivity of 8a for nNOS over eNOS has dropped 20-fold, which makes this inhibitor comparable to inhibitor 1a.

Figure 1.

The active site structures in enzyme-inhibitor complexes: (A) nNOS-8a, (B) nNOS-8b, (C) nNOS-8c, (D) eNOS-8c, (E) nNOS-8d, and (F) nNOS-8e. The sigmaA weighted 2Fo – Fc electron density map for each bound inhibitor at 1σ contour level is shown for panels A to D and the omit Fo – Fc density map at 2.5 σ contour level shown for panels E and F. Major hydrogen bonds are drawn with dashed lines. Alternate side chain rotamers are observed for Glu592 in nNOS or Glu363 in eNOS in panels A to D. The inhibitor 8a also shows two conformations in its lipophilic tail (colored as gray and magenta in panel A). The density for part of inhibitor 8e is weak as shown, but the model presented in panel F is supported by the 2Fo – Fc map at lowered contour level (0.5 σ). Figures were prepared with PyMol (www.pymol.org). The PDB codes are as follows: nNOS-8a: 3PNE; nNOS-8b: 3PNF; nNOS- 8c: 3PNG; eNOS-8c: 3PNH; nNOS-8d: 3SVP; nNOS-8e: 3SVQ.

Table 1.

K_i Values of Inhibitors for Rat nNOS, Bovine eNOS and Murine iNOS^a

Compound	nNOS (μM)	eNOS (μM)	iNOS (μM)	selectivityb
				n/e	n/i
1a	0.16	31	190	194	1188
1bc	0.11	130	25	1182	227
8a	0.086	16	78	186	907
8b	0.17	27	91	159	535
8c	0.026	19	26	731	1000
8d	0.077	17	17	221	221
8e	0.18	49	130	272	722

^aThe K_i values were calculated based on the directly measured IC₅₀ values, which represent at least duplicate measurements with standard deviations of ±10%. There is high homology among these isozymes from different species.⁵

^bThe ratio of K_i (eNOS or iNOS) to K_i (nNOS).

^cThe K_i (nNOS) for inhibitor 1b is different from that previously reported because of the use of a new high throughput assay improving both speed and reproducibility.

Inhibitor 8b, the corresponding o-chloro isomer of compound 8a, has 2-fold lower potency for rat nNOS (8b vs 8a, Table 1). However, eNOS does not show a preference for inhibitors 8b and 8a. The structure of nNOS-8b (Fig. 1B) reveals only one conformation for the inhibitor where the CF₂ group points toward the heme. In this conformation one of the fluorine atoms makes an unfavorable intramolecular contact with the chlorine atom on the phenyl ring. However, there is no electron density indicating an altered orientation for the CF₂ group that avoids this unfavorable contact. This steric hindrance may partially account for the 2-fold drop in potency for 8b compared to 8a. When both meta positions of the phenyl ring are substituted (8d), the conformation of the phenyl ring is determined by the bulkier chlorine atom and the conformation is more like 8a than 1b. This ring is still well tolerated in the active site, and, of the CF₂ analogs reported here (1a–b, 8a–b, 8d–e), 8d is the most potent of the series. Interestingly, the side chain of 8d shows only one conformation because the meta disubstituted ring is too large to be accommodated in the alternate conformation. Inhibitor 8e is the least potent analog of the series. Similar to 8b, ortho substitution in 8e leads to an intramolecular clash between fluorine of the ring and the CF₂ group, and again, only one side chain conformation is observed with CF₂ pointing downward to the heme (Figures 1B and 1F).

Inhibitor 8c, a mixture of diastereomers, is the monofluoro methylene derivative of 1b with a chiral pyrrolidine core. The mixture shows increased potency for rat nNOS (8c vs 1b, Table 1). This may result from the higher basicity of the amino group in the lipophilic tail of 8c compared to that of 1b.⁸ However, the nNOS selectivity of 8c over eNOS is somewhat diminished compared with that for 1b because of its relatively higher potency toward eNOS. The structures of 8c bound to both nNOS and eNOS were obtained (Fig. 1C and 1D). In both structures, the same binding mode for 8c is observed, with its single fluorine atom pointing down toward the heme. Apparently, despite being a mixture of diastereomers, the (R,R,R)-diastereomer has greater binding affinity to the isozymes than the (R,R,S)-diastereomer because the observed electron density for the fluorine atom only supports one position (R). This may be because there is no unfavorable contact between this fluorine atom and heme propionate A in contrast to the clashes observed for the CF₂ group in its “downward” conformation in the nNOS-8a structure (Fig. 1A). The amino group in the lipophilic tail of the inhibitor lures Glu592 of nNOS (Glu363 of eNOS) into an alternate rotamer by hydrogen bonding.

Structure activity relationship (SAR) studies demonstrate the key role of the meta substituent in 1b, 8c, and 8d for retaining high inhibitory activity for rat nNOS. A bulkier m-chloro group in 8a increases potency relative to the m-fluoro substituent, but leads to a loss in selectivity over the other two isoforms. Although m-chloro and m-fluoro disubstitution (8d) achieves another small boost in potency, it also results in an additional drop in selectivity. The o-chloro group in 8b or o-fluoro group in 8e leads to tight intramolecular contacts with its CF₂ group, which is accompanied by an unfavorable side chain conformation and a drop in potency. The p-fluoro in 1a forces its CF₂ group into a downward conformation, weakening the hydrogen bond from the amino group to the Glu592 side chain. It seems then, that only meta substituents place the phenyl ring in the right position to optimize van der Waals contact with the hydrophobic pocket surrounded by Val567 and Phe584 in nNOS,⁷ and m-fluoro inhibitor 1b achieves the optimum balance between potency and selectivity.

CONCLUSION

In summary, an improved synthesis of chiral pyrrolidine inhibitors of nNOS has been developed. Compared with the reported synthesis, it is three steps shorter with an overall yield of approximately 10% (>10-fold larger). It also enables expanded SAR studies on the pyrrolidine-based scaffold, which can be beneficial for further development of nNOS inhibitors.

EXPERIMENTAL SECTION

The purity of the final compounds was determined by HPLC analysis to be ≥95%. For experimental details, see the Supporting Information.

6-(((3R,4R)-4-(2-((2,2-Difluoro-2-(3-fluorophenyl)ethyl)amino)ethoxy)pyrrolidin-3-yl)methyl)-4-methylpyridin-2-amine (1b)

To a solution of 18a (60 mg, 85 umol) in MeOH (2 mL) was added 6 N HCl (4 mL) at room temperature. The mixture was stirred for 12 h and then concentrated. The crude product was purified by recrystallization (EtOH/H₂O) to give inhibitor 1b (40 mg, 99%) as a tri-HCl salt; [α] ²⁰ +6.25 (c 4, MeOH).

6-(((3R,4R)-4-(2-((2,2-Difluoro-2-(3-chlorophenyl)ethyl)amino)ethoxy)pyrrolidin-3-yl)methyl)-4-methylpyridin-2-amine (8a)

To a solution of 18b (70 mg, 0.10 mmol) in MeOH (2 mL) was added 6 N HCl (4 mL) at room temperature. The mixture was stirred for 12 h and then concentrated. The crude product was purified by recrystallization (EtOH/H₂O) to give inhibitor 8a (50 mg, 95%) as a tri-HCl salt: ¹H NMR (500 MHz, D₂O) δ 2.19 (s, 3H), 2.60–2.75 (m, 1H), 2.85–2.95 (m, 1H), 3.00–3.10 (m, 2H), 3.20–3.30 (m, 1H), 3.30–3.45 (m, 3H), 3.55–3.60 (d, J = 13.0 Hz, 1H), 3.65–3.70 (m, 1H), 3.75–3.90 (m, 3H), 4.15 (d, J = 3.0 Hz, 1H), 6.41 (s, 1H), 6.55 (s, 1H), 7.30–7.45 (m, 3H), 7.52 (s, 1H); ¹³C NMR (125 MHz, D₂O) δ 20.0, 29.2, 41.3, 41.4, 47.0, 47.4, 49.1, 51.7, 63.6, 78.3, 110.4, 114.0, 118.2, 123.39, 123.42, 123.47, 125.07, 125.12, 130.7, 131.5, 133.9, 134.4, 145.4, 153.9, 158.1; LC-TOF (M+H⁺) calcd for C₂₁H₂₈ClF₂N₄O 425.1920, found 425.1919.

6-(((3R,4R)-4-(2-((2,2-Difluoro-2-(2-chlorophenyl)ethyl)amino)ethoxy)pyrrolidin-3-yl)methyl)-4-methylpyridin-2-amine (8b)

Inhibitor 8b was synthesized using a similar procedure to that for 8a (50 mg, 95%) as a tri-HCl salt: ¹H NMR (500 MHz, D₂O) δ 2.19 (s, 3H), 2.60–2.75 (m, 1H), 2.85–2.95 (m, 1H), 3.03–3.08 (t, J = 11.5 Hz, 1H), 3.19 (s, 1H), 3.21–3.25 (dd, J = 3.0, 13.0 Hz, 1H), 3.35–3.42 (m, 3H), 3.52–3.58 (d, J = 13.0 Hz, 1H), 3.63–3.66 (m, 1H), 3.82–3.88 (m, 1H), 3.90–4.00 (m, 2H), 4.14–4.16 (t, J = 3.5 Hz, 1H), 6.42 (s, 1H), 6.54 (s, 1H), 7.30–7.35 (m, 1H), 7.40–7.45 (m, 2H), 7.55–7.60 (m, 1H); ¹³C NMR (125 MHz, D₂O) δ 21.0, 29.0, 41.2, 47.0, 47.4, 48.8, 49.2, 50.3, 50.5, 63.4, 78.2, 110.4, 113.9, 118.1, 127.57, 127.63, 127.70, 129.1, 130.8, 131.4, 131.5, 133.1, 145.5, 153.9, 158.1; LC-TOF (M+H⁺) calcd for C₂₁H₂₈ClF₂N₄O 425.1920, found 425.1919.

6-(((3R,4R)-4-(2-((2-fluoro-2-(3-fluorophenyl)ethyl)amino)ethoxy)pyrrolidin-3-yl)methyl)-4-methylpyridin-2-amine (8c)

Inhibitor 8c was synthesized as a mixture of two diastereomers using a similar procedure to that for 8a (32 mg, 90%) as a tri-HCl salt: ¹H NMR (500 MHz, D₂O) δ 2.19 (s, 3H), 2.65–2.75 (m, 1H), 2.85–2.95 (m, 1H), 3.03–3.11 (m, 1H), 3.20 (s, 1H), 3.21–3.25 (dd, J = 3.0, 13.0 Hz, 1H), 3.30–3.45 (m, 4H), 3.50–3.58 (m, 2H), 3.60–3.66 (m, 1H), 3.80–3.85 (m, 1H), 4.14–4.16 (m, 1H), 5.80–6.00 (m, 1H), 6.46 (s, 1H), 6.50–6.55 (m, 1H), 7.00–7.15 (m, 3H), 7.30–7.41 (m, 1H); ¹³C NMR (125 MHz, D₂O) δ 21.0, 29.0, 29.1, 41.3, 41.4, 43.7, 43.9, 47.0, 47.1, 47.3, 48.7, 49.2, 49.3, 51.3, 51.5, 51.7, 51.8, 63.6, 63.9, 78.2, 88.4, 89.8, 110.4, 112.48, 112.54, 112.56, 112.61, 112.69, 112.72, 112.74, 114.0, 114.1, 116.3, 116.5, 116.6, 121.35, 121.40, 121.46, 121.49, 121.51, 130.85, 130.92, 131.0, 136.8, 136.99, 137.02, 145.60, 145.64, 153.85, 153.86, 158.11, 158.12, 161.6, 163.5; LC-TOF (M+H⁺) calcd for C₂₁H₂₉F₂N₄O 391.2309, found 391.2288.

6-(((3R,4R)-4-(2-((2,2-Difluoro-2-(3-chloro-5-fluorophenyl)ethyl)amino)ethoxy)pyrrolidin-3-yl)methyl)-4-methylpyridin-2-amine (8d)

Inhibitor 8d was synthesized as a mixture of two diastereomers using a similar procedure to that for 8a (7 mg, 80%) as a tri-HCl salt: ¹H NMR (500 MHz, MeOD) δ 2.35 (s, 3H), 2.80–2.87 (m, 1H), 2.93–2.98 (m, 1H), 3.07–3.13 (m, 1H), 3.23–3.27 (m, 1H), 3.38–3.45 (m, 1H), 3.51–3.56 (m, 2H), 3.50–3.58 (m, 2H), 3.65–3.68 (d, J = 13.0 Hz, 1H), 3.73–3.79 (m, 2H), 3.94–3.99 (m, 1H), 4.00–4.07 (m, 1H), 4.19–4.21 (m, 1H), 6.66 (s, 1H), 6.67 (s, 1H), 7.42–7.46 (m, 2H), 7.56 (s, 1H); ¹³C NMR (125 MHz, MeOD) δ 22.0, 30.4, 43.5, 52.5, 52.7, 64.9, 80.0, 111.5, 112.8, 112.9, 115.1, 120.2, 120.4, 123.1, 137.47, 137.48, 147.95, 156.0, 159.2, 163.2, 165.2; LC-TOF (M+H⁺) calcd for C₂₁H₂₆ClF₃N₄O 443.1747, found 443.3306.

6-(((3R,4R)-4-(2-((2,2-Difluoro-2-(2,3-difluorophenyl)ethyl)amino)ethoxy)pyrrolidin-3-yl)methyl)-4-methylpyridin-2-amine (8e)

Inhibitor 8e was synthesized as a mixture of two diastereomers using a similar procedure to that for 8a (12 mg, 75%) as a tri-HCl salt: ¹H NMR (500 MHz, MeOD) δ 2.35 (s, 3H), 2.82–2.88 (m, 1H), 2.95–3.00 (m, 1H), 3.08–3.14 (m, 1H), 3.23–3.30 (m, 1H), 3.40–3.44 (m, 1H), 3.53–3.55 (m, 2H), 3.60 (d, J = 13 Hz, 1H), 3.77–3.79 (m, 1H), 3.95–3.97 (m, 1H), 4.04–4.10 (m, 2H), 4.19–4.20 (m, 1H), 6.66 (s, 1H), 6.67 (s, 1H), 7.32–7.55 (m, 3H); ¹³C NMR (125 MHz, MeOD) δ 22.0, 30.35, 30.40, 43.5, 49.9, 50.4, 52.5, 65.0, 79.9, 111.5, 115.1, 122.2, 122.3, 123.5, 126.7, 137.47, 137.48, 147.9, 148.0, 156.0, 159.2; LC-TOF (M+H⁺) calcd for C₂₁H₂₆F₄N₄O 427.2043, found 427.4021.

Abbreviations

nNOS

neuronal nitric oxide synthase

NO

nitric oxide

eNOS

endothelial nitric oxide synthase

iNOS

inducible nitric oxide synthase

n-BuLi

n-butyllithium

TBSCl

t-butyldimethylsilyl chloride

DMAP

4-dimethylaminopyridine

TBAF

tetrabutylammonium fluoride

SAR

structure activity relationship