Structural basis for isoform-selective inhibition in nitric oxide synthase

Abstract

Nitric oxide synthase (NOS) converts l-arginine into l-citrulline and releases the important signaling molecule nitric oxide (NO). In the cardiovascular system, NO produced by endothelial NOS (eNOS) relaxes smooth muscle which controls vascular tone and blood pressure. Neuronal NOS (nNOS) produces NO in the brain, where it influences a variety of neural functions such as neural transmitter release. NO can also support the immune system, serving as a cytotoxic agent during infections. Even with all of these important functions, NO is a free radical and, when overproduced, it can cause tissue damage. This mechanism can operate in many neurodegenerative diseases, and as a result the development of drugs targeting nNOS is a desirable therapeutic goal. However, the active sites of all three human isoforms are very similar, and designing inhibitors specific for nNOS is a challenging problem. It is critically important, for example, not to inhibit eNOS owing to its central role in controlling blood pressure. In this Account, we summarize our efforts in collaboration with Rick Silverman at Northwestern University to develop drug candidates that specifically target NOS using crystallography, computational chemistry, and organic synthesis. As a result, we have developed aminopyridine compounds that are 3800-fold more selective for nNOS than eNOS, some of which show excellent neuroprotective effects in animal models. Our group has solved approximately 130 NOS-inhibitor crystal structures which have provided the structural basis for our design efforts. Initial crystal structures of nNOS and eNOS bound to selective dipeptide inhibitors showed that a single amino acid difference (Asp in nNOS and Asn in eNOS) results in much tighter binding to nNOS. The NOS active site is open and rigid, which produces few large structural changes when inhibitors bind. However, we have found that relatively small changes in the active site and inhibitor chirality can account for large differences in isoform-selectivity. For example, we expected that the aminopyridine group on our inhibitors would form a hydrogen bond with a conserved Glu inside the NOS active site. Instead, in one group of inhibitors, the aminopyridine group extends outside of the active site where it interacts with a heme propionate. For this orientation to occur, a conserved Tyr side chain must swing out of the way. This unanticipated observation taught us about the importance of inhibitor chirality and active site dynamics. We also successfully used computational methods to gain insights into the contribution of the state of protonation of the inhibitors to their selectivity. Employing the lessons learned from the aminopyridine inhibitors, the Silverman lab designed and synthesized symmetric double-headed inhibitors with an aminopyridine at each end, taking advantage of their ability to make contacts both inside and outside of the active site. Crystal structures provided yet another unexpected surprise. Two of the double-headed inhibitor molecules bound to each enzyme subunit, and one molecule participated in the generation of a novel Zn(2+) site that required some side chains to adopt alternate conformations. Therefore, in addition to achieving our specific goal, the development of nNOS selective compounds, we have learned how subtle differences in dynamics and structure can control protein-ligand interactions and often in unexpected ways.

Figure 1

Rat nNOS heme domain dimer strcuture (1OM4). L-Arg is held in place by a series of H-bonds to conserved active site residues as well as one heme propionate. The cofactor, BH₄, is positioned at the dimer interface and also H-bonds with the same heme propionate.

Figure 2

A) Structure of one of the early dipeptide lead compounds, 1, that exhibits excellent isoform selectivity. B) and C) show the crystal structures of the dipeptide inhibitor 1 in the active site of eNOS (1P6L) and nNOS (1P6H). In nNOS the inhibitor “curls” which enables the inhibitor α-amino group to interact with both Glu592 and Asp597. In eNOS Asn368 is the homolog to nNOS Asp597.

Figure 3

The initial design of aminopyridine-pyrrolidine inhibitors. The aminopyridine mimics the guanidinium group of L-Arg but exhibits a much lower pKa thus increasing bioavailability. In addition the aromatic aminopyrridine should stack over the heme ring. The pyrrolidine NH group extends up toward the active site Asp597 in nNOS.

Figure 4

Crystal structures of the two enantiomers of cis-2 bound to nNOS. (3S,4S)-2 (3JWS) binds as expected with the aminopyridine positioned over the heme near Glu592. However, (3R,4R)-2 (3JWT) binds in the flipped mode which places the aminopyridine near a heme propionate. This requires Tyr706 to adopt the “out” rotamer conformation.

Figure 5

The nNOS active site showing the inhibitor in the flipped mode (3JWT). Note that the aminopyridine and Tyr706 are close to residues Met336 and Leu337. There is a break in the electron density right after Pro338 suggesting that this region may be particularly flexible in nNOS which enables Tyr706 to more readily adopt the out conformation relative to eNOS.

Figure 6

A total of 7 crystal structures of the 5 different aminopyridine inhibitors bound to either nNOS or eNOS were used for the free energy calculations. The correlation between the relative ΔG_calc and experimental ΔG_exp (extracted from measured K_i values) is excellent.

Figure 7

The initial design of double-headed symmetric inhibitors.

Figure 8

Crystal structure of 3 complexed to nNOS (3N5W). One molecule of 3 binds as predicted with one aminopyridine interacting with Glu592 and other with the heme propionate. The second molecule of 3 displaces BH₄ thus enabling an aminopyridine to interact with the second heme propionate. This places the bridging pyridine in position to complete a tetrahedral coordination sphere around a Zn²⁺ ion. In order for the Zn²⁺ to bind, Arg596, which normally interacts with BH4, must swing out of the way. In addition, the dimer interface must slightly tighten to enable His692 from molecule B of the dimer to move close enough for Zn²⁺ coordination.

Figure 9

Variations on inhibitor 2 designed for better bioavailability.