Peripheral but crucial: a hydrophobic pocket (Tyr⁷⁰⁶, Leu³³⁷, and Met³³⁶) for potent and selective inhibition of neuronal nitric oxide synthase

Abstract

Selective inhibition of the neuronal isoform of nitric oxide synthase (nNOS) over endothelial nitric oxide synthase (eNOS) and inducible nitric oxide synthase (iNOS) has become a promising strategy for the discovery of new therapeutic agents for neurodegenerative diseases. However, because of the high sequence homology of different isozymes in the substrate binding pocket, developing inhibitors with both potency and excellent isoform selectivity remains a challenging problem. Herein we report the evaluation of a recently discovered peripheral hydrophobic pocket (Tyr⁷⁰⁶, Leu³³⁷, and Met³³⁶) that opens up upon inhibitor binding and its potential in designing potent and selective nNOS inhibitors using three compounds, 2a, 2b, and 3. Crystal structure results show that inhibitors 2a and 3 adopted the same binding mode as lead compound 1. We also found that hydrophobic interactions between the 4-methyl group of the aminopyridine ring of these compounds with the side chain of Met³³⁶, as well as the π-π stacking interaction between the pyridinyl motif and the side chain of Tyr⁷⁰⁶ are important for the high potency and selectivity of these nNOS inhibitors.

Overproduction of the small molecule nitric oxide (NO) by the neuronal isoform of nitric oxide synthase (NOS) in the brain is closely associated with numerous neurodegenerative diseases, including chronic pathologies such as Parkinson’s,¹ Alzheimer’s,² Huntington’s,³ headaches,⁴ as well as neuronal damage in stroke.⁵ Inhibition of nNOS, therefore, has become a compelling strategy for the treatment of neurodegeneration.^6-8 In the past two decades, a large number of nNOS inhibitors have been reported, including derivatives of L-arginines, guanidines, isothioureas, isothioureidos, 2-iminopiperidines, amidines, aminopyridines, thioureas, thiazoles, imidazoles, and indazoles.⁹ None of these inhibitors, however, has entered clinical trials due to drawbacks with respect to their potency and isozyme selectivity.⁹ These inhibitors, when binding to nNOS, target the L-arginine binding pocket where nNOS has high sequence homology with the other two isozymes, eNOS and iNOS.⁶ This explains why most inhibitors suffer from poor isoform selectivity. At present, discovery of nNOS inhibitors with high potency and selectivity still represents a major challenge.

To overcome this limitation, a number of selective nNOS inhibitors have emerged from structure-based design.^10-12 One of the more important is pyrrolidine-based inhibitor 1, which exhibits excellent potency (K_i = 5 nM) and selectivity for rat nNOS over bovine eNOS (3,800-fold) and murine macrophage iNOS (1,200-fold).^13-15 Interestingly, crystal structures of nNOS and eNOS in complex with 1 revealed that this inhibitor adopts an unexpected “flipped” binding mode, shown in Figure 1, in which the 2-amino-4-methylpyridine motif extended beyond the edge of the heme distal site, hydrogen bonded to the heme propionate of pyrrole ring D, and fits nicely into a hydrophobic pocket (Tyr⁷⁰⁶, Leu³³⁷, and Met³³⁶) formed by an alternate rotamer of the aromatic side chain of Tyr⁷⁰⁶.¹³ A close look at this pocket indicates two potential beneficial interactions: (1) the hydrophobic interaction between the 4-methyl group of 1 and the side chains of Leu³³⁷ and Met³³⁶, and (2) the π-π stacking interaction between the aminopyridine motif of 1 and the aromatic side chain of Tyr⁷⁰⁶. More importantly, the specific residues around this pocket are significantly different in the 3 mammalian NOS isoforms. For example, instead of the three amino acid residues, Tyr⁷⁰⁶, Leu³³⁷, and Met³³⁶ for rat nNOS, for human nNOS the corresponding residues are Tyr⁷¹⁰, His³⁴¹, and Met³⁴⁰, while for bovine eNOS and murine iNOS, these are Tyr⁴⁷⁷, Leu¹⁰⁷, and Val¹⁰⁶, and Tyr⁴⁸⁷, Asn¹¹⁵, and Met¹¹⁴, respectively.¹⁶ Therefore, we speculate that this hydrophobic pocket may provide a new “hot spot” for developing isoform-selective inhibitors for NOSs.

Figure 1.

Binding of lead compound 1 to the rat nNOS active site. The major hydrogen bonds are drawn as dashed lines. The structural figures were prepared with PyMOL (www.pymol.org).

To further investigate the chemical environment of this peripheral site, and also to evaluate the possibility of utilizing it for the development of selective nNOS inhibitors, we report herein the synthesis and evaluation of three inhibitors, 2a, 2b, and 3. Compound 2a is an analog of 1, with the 4-methyl group removed from the 2-aminopyridine ring. The difference in potency between 1 and 2a for nNOS can demonstrate the importance of the methyl group to inhibitory activity. Compound 2b, the enantiomer of 2a, has opposite chirality at the pyrrolidine core. Therefore, 2b cannot adopt the “flipped” binding mode as with 2a, and thus will provide a direct comparison between the two different binding modes. Inhibitor 3 removes the aromaticity of the aminopyridine from 2a. A comparison of 3 and 2a demonstrates the role of the π-π stacking interaction between the pyridine ring of 1 and the side chain of Tyr⁷⁰⁶.

The syntheses of key intermediates 4a and 4b are shown in Scheme 1, using a recently reported method.^13-15 The amino group of 2-amino-4,6-dimethylpyridine (5) was protected using (Boc)₂O in the presence of triethylamine (TEA) to provide 6 in high yields. Compound 6 was treated with two equivalents of n-BuLi, and the resulting dianion was allowed to react with the epoxide (7)¹³ to generate the desired regioisomer (8) in modest yields. Then, the free NH group on the pyridine ring was further protected by a benzyl-protecting group using benzyl bromide to yield 9 in good yields. Next, trans alcohol 9 underwent a Mitsunobu reaction using acetic acid as a nucleophile to produce 10 in very high yields. Hydrolysis of 10 yielded cis-alcohol 11 in quantitative yields. Finally, the racemic mixture of 11 was successfully separated using a two step procedure. First, 11 was treated with (1S)-(−)-camphanic chloride in the presence of TEA and N,N-dimethylaminopyridine (DMAP) to yield two diastereomers (12a and 12b), which were successfully separated using silica chromatography. Then, each ester was subjected to aqueous Na₂CO₃ to produce 4a and 4b as single enantiomers in excellent yields.

Scheme 1.

Synthesis of 4a and 4b.

^a Reagents and conditions: (a) (Boc)₂O, TEA, t-BuOH, 50 °C, 24 h, 90%; (b) (i) n-BuLi (2 equiv.), −78 °C to r.t., 30 min, (ii) 7, −78 °C to r.t., 3 h, 55%; (c) NaH, 0 °C to r.t., 10 min, benzyl bromide, r.t., 8 h, 92%; (d) PPh₃, DIAD, acetic acid, r.t., 12 h, 95%; (e) NaOH (2N)/MeOH, 50 °C, 4 h, 90%; (f) (1S)-(−)-camphanic chloride, TEA, DMAP, r.t. 4 h, 49% for each diastereomer; (g) Na₂CO₃, H₂O/MeOH, r.t., 3 h, 99%.

Next, allylation of 4a/b gave 13a/b in excellent yields (Scheme 2).^13-15 Alkenes 13a/b were subjected to ozonolysis using Zn as the reducing reagent to provide aldehydes (14a/b), which underwent reductive aminations with 2-(3-fluorophenyl)ethanamine to provide secondary amines 15a/b. The secondary amino group was protected with a Boc-protecting group, and then the benzyl-protecting groups of 16a/b were removed by catalytic hydrogenation at 60 °C to give 17a/b. Finally, the three Boc-protecting groups were removed in a 2:1 mixture of 6 N HCl and MeOH to generate inhibitors 2a and 2b. Inhibitor 3 can be synthesized from 15a using high pressure catalytic hydrogenation conditions in very high yields.

Scheme 2.

Syntheses of 2a, 2b, and 3.

^a Reagents and conditions: (a) NaH, allyl bromide, 0 °C to r.t. 30 min, 99%; (b) (i) O₃, −78 °C, (ii) Zn, acetic acid, −78 °C to r.t., 70-75%; (c) 2-(3-fluorophenyl)ethanamine, NaHB(OAc)₃, r.t., 3 h, 52-55%; (d) Pd(OH)₂/C, H₂, 2:1 EtOH/HCl (12 N), r.t., 575 psi, 40 h, 100%; (e) (Boc)₂O, Et₃N, MeOH, r.t., 0.5 h, 99%; (f) Pd(OH)₂/C, H₂, 60 °C, 24 h, 35-50%; (g) 6 N HCl/MeOH (2:1), r.t., 4 h, 80-85%.

In the crystal structure of the active site of rat nNOS, 2a adopts the same binding mode as lead compound 1 (Figure 2A). The aminopyridine motif extends to the same peripheral hydrophobic pocket containing Tyr⁷⁰⁶, Leu³³⁷, and Met³³⁶, forming a charge-charge interaction with the heme propionate D, as well as a π-π stacking interaction with the aromatic side chain of Tyr⁷⁰⁶. However, removal of the 4-methyl group from the 2-aminopyridine motif dramatically impaired the potency (7-fold) of the inhibitor (2a vs 1). This result highlights the crucial role of the 4-methyl group for retaining the high inhibitory activity of 1 for rat nNOS. Importantly, the selectivity of 2a for rat nNOS over bovine eNOS also dropped significantly (2.3-fold). This was mainly the result of the lower sensitivity of eNOS to the presence of the 4-methyl group, only a 3-fold difference when comparing 2a to 1. This methyl group might make less favorable contacts with the smaller side chain of Val¹⁰⁶ in eNOS. Inhibitor 2b, the corresponding enantiomer of 2a, adopts the normal binding mode with its 2-aminopyridine hydrogen bonded to the side chain of Glu⁵⁹² (Figure 2B), resulting in a 4-fold lower potency for rat nNOS (2b vs 2a). However, eNOS has no preference for the two binding modes with 2b and 2a showing comparable affinities. Inhibitor 3, with the 2-aminopyridine of 2a reduced to a cyclic amidine, showed diminished potency for rat nNOS (3 vs 2a). This result indicates that the π-π stacking interaction between the pyridine ring and Tyr⁷⁰⁶ is an important factor for tight binding of 2a to nNOS. The stacking interaction provides less contribution to the binding affinity of 2a to eNOS, as its K_i values are similar for both 3 and 2a. This is probably because the Tyr⁴⁷⁷ side chain in eNOS does not interact as closely with the 2-aminopyridine ring of the inhibitors, as demonstrated previously in crystal structures for other pyrrolidine inhibitors complexed to eNOS and nNOS.¹³

Figure 2.

Active site structures of rat nNOS in complex with 2a (A, PDB code 3NNY) and 2b (B, PDB code 3NNZ). Shown also the 2Fo–Fc electron density for inhibitor at 1σ contour level. The major hydrogen bonds are drawn as dashed lines.

We also tested the inhibitory activity of 1, 2a, 2b, and 3 against the human isoform of nNOS (Table 2). Human nNOS shows very high sequence homology to rat nNOS in the active site;¹⁶ the only different residue in the peripheral site is His³⁴¹ in place of Leu³³⁷, which makes the hydrophobic pocket in human nNOS smaller than that in rat nNOS. As a result, the hydrophobic pocket of human nNOS should prefer smaller fragments. Even though inhibitors 1, 2a, 2b, and 3 are less potent with human nNOS compared to rat nNOS, the selectivity (K_i-human/K_i-rat) dropped 1.2-fold and 1.6-fold for 2a and 3, respectively, compared to 1. These results indicate that the relatively smaller hydrophobic pocket in human nNOS prefers to bind the non-substituted aminopyridine fragment.

Table 2. K_i values of 1, 2a, 2b, and 3 for rat nNOS and human nNOS.

	Kia(nM)		selectivity
inhibitor
	rat nNOS	human nNOS	rat/human
1	7	50	7.1
2a	50	290	5.8
2b	220	1,800	8.2
3	110	490	4.5

^aThe K_i values represent at least duplicate measurements with standard deviations of ±10%.

In summary, we demonstrate the feasibility and potential of using the peripheral site (Tyr⁷⁰⁶, Leu³³⁷, and Met³³⁶ for rat nNOS) as a “hot spot” for designing potent and selective nNOS inhibitors. Three new inhibitors, 2a, 2b, and 3, have been synthesized. The crystal structures show that 2a shares the same flipped binding mode as lead compound 1, while 2b binds in a normal mode as expected by their stereochemistry of the pyrrolidine ring. The inhibitory activities were tested against various NOS isoforms. Our results highlight the importance of the hydrophobic interaction between the 4-methyl group of the 2-aminopyridine motif and the side chain of Met³³⁶ for tight binding and high isoform-selectivity of the inhibitors. We also show that the π-π stacking interaction between the aminopyridine ring and the aromatic side chain of Tyr⁷⁰⁶ is important for good potency. The insensitivity of eNOS to the binding mode of pyrrolidine-based inhibitors and the weaker π-π stacking interaction from its Tyr⁴⁷⁷ to the 2-aminopyridine ring of the inhibitors are isoform specific properties valuable to future inhibitor design.

Table 1. K_i values of 1, 2a, 2b, and 3 for rat nNOS, bovine eNOS and murine iNOS.

	Kia(nM)			selectivity
inhibitor
	rat nNOS	eNOS	iNOS	n/e	n/i
1	7	19,000	5,800	2,700	830
2a	50	58,000	10,000	1,200	200
2b	220	59,000	13,200	270	60
3	110	58,000	21,800	530	200

^aThe K_i values represent at least duplicate measurements with standard deviations of ±10%