Backdoor opening mechanism in acetylcholinesterase based on X-ray crystallography and molecular dynamics simulations

Abstract

The transient opening of a backdoor in the active-site wall of acetylcholinesterase, one of nature's most rapid enzymes, has been suggested to contribute to the efficient traffic of substrates and products. A crystal structure of Torpedo californica acetylcholinesterase in complex with the peripheral-site inhibitor aflatoxin is now presented, in which a tyrosine at the bottom of the active-site gorge rotates to create a 3.4-Å wide exit channel. Molecular dynamics simulations show that the opening can be further enlarged by movement of Trp84. The crystallographic and molecular dynamics simulation data thus point to the interface between Tyr442 and Trp84 as the key element of a backdoor, whose opening permits rapid clearance of catalysis products from the active site. Furthermore, the crystal structure presented provides a novel template for rational design of inhibitors and reactivators, including anti-Alzheimer drugs and antidotes against organophosphate poisoning.

Acetylcholinesterase (AChE) plays a crucial role at cholinergic synapses in both the central and peripheral nervous systems, terminating impulse transmission by rapid hydrolysis of the neurotransmitter acetylcholine.¹ AChE also attracts considerable interest in pharmacological, agrochemical, and toxicological contexts, being the target of a variety of natural and man-made drugs and inhibitors, such as organophosphate nerve agents and insecticides, the snake-venom polypeptide toxin, fasciculin, and the first generation of drugs for management of Alzheimer's disease. The first crystal structure of an AChE, that of Torpedo californica (Tc) AChE, revealed an unusual and unanticipated architecture² in view of its remarkably high catalytic power (turnover number 1000–10,000 s⁻¹ depending on the species); the active site, a Ser-His-Glu catalytic triad, is buried at the bottom of a deep and narrow gorge. The crystal structure raised cogent questions as to how traffic of substrate and products to and from the active site might occur.³ Molecular dynamics (MD) simulations indicated the presence of a backdoor in the thin aspect of the gorge, adjacent to the active site, which might permit transit of substrate, products, or solvent.^4–6 It was suggested that it might open transiently, in the course of catalysis, because of movement of residues within the ω-loop (residues 67–94 in TcAChE). The residual catalytic activity observed in the complex of AChE with fasciculin, which almost completely blocks access to the active-site via the top of the gorge,⁷ lent support to the backdoor hypothesis. Yet, extensive kinetic studies, together with site-directed mutagenesis,^8,9 failed to identify the backdoor, and X-ray crystallography, so far, has provided only circumstantial evidence for its existence.^10,11

Here, we present a crystal structure of TcAChE in complex with the inhibitor aflatoxin at 2.3 Å resolution, in which a large conformational change in the side chain of Tyr442 results in the opening of a backdoor [compare Fig. 1(A,B)], without involving residues within the ω-loop that had previously been the prime candidates.⁴ The structure (PDB entry code 2XI4) was obtained after soaking orthorhombic crystals of native TcAChE with aflatoxin (see Supporting Information), which binds to the peripheral anionic site (PAS), located at the top of the gorge, about 13 Å distant from the active site (Sanson et al., unpublished). Bound aflatoxin is observed in both monomers of the dimer that constitutes the asymmetric unit of orthorhombic crystals of TcAChE (not shown). In the native structure (PDB entry code 1W75), Tyr442Oζ H-bonds to Trp84Nɛ1 (distance 3.2 Å), linking the two subdomains of TcAChE (residues 4–305 and 306–535) across the active-site gorge,¹² and the backdoor is closed [Fig. 1(C)]. In the complex, rotation of Tyr442 occurs in monomer A [Fig. 1(D) and Supporting Information Fig. S1(A)], but not in monomer B [Supporting Information Fig. S1(B)], breaking the H-bond linking Tyr442Oζ to Trp84Nɛ1 (new distance, 12.7 Å); this results in the opening of a channel between the active site and the exterior large enough to allow the passage of a water molecule [Fig. 1(B)]. The structural environments of monomers A and B in the crystal are identical, and aflatoxin binding occurs in both monomers; hence, it is unclear why Tyr442 undergoes a conformational change only in monomer A. Small variations in the conformational heterogeneity of both monomers that are not detectable by X-ray crystallography may be at the origin of the differential behavior of Tyr442. Rotation of Tyr442 in monomer A also disrupts the H-bond linking Tyr442Oζ to Trp432Nɛ1. In this new conformation, Tyr442 is stabilized mainly by hydrophobic interactions with Phe423 and Phe446 and by an H-bond to a water molecule. The conformational change in the side chain of Tyr442 displaces Ala427, leading to shifts in main- and side-chain atoms from Arg426 to Val431. The narrowest dimension of the channel is characterized by a surface-to-surface distance of 3.4 Å, calculated using the programme HOLE.¹³ Inspection of all 78 TcAChE structures deposited in the PDB revealed only one other case (accession code 1HBJ¹⁴) in which Tyr442 adopts a very similar alternate conformation [Supporting Information Fig. S2(C)]. In 1HBJ, the enzyme was crystallized in a trigonal space group (P3₁21; one monomer in the asymmetric unit), and the ligand, 4-amino-5-fluoro-2-methyl-3-(3-trifluoro-acetylbenzylthiomethyl)-quinoline (AFM-Q), was seen to occupy the “anionic” subsite of the active site (catalytic anionic site). In the AFM-Q/TcAChE complex, the backdoor channel is slightly wider [Supporting Information Fig. S2(A); smallest surface-to-surface distance 3.5 Å] than that in the AB1/TcAChE complex.

Figure 1.

Backdoor in TcAChE. A section (black) through the enzyme and its accessible surface area (blue) show (A) the catalytic serine (light blue; Ser200) at the bottom of the deep gorge and Tyr442 (green) in the native enzyme; (B) the opening of a backdoor in an aflatoxin/TcAChE complex due to rotation of Tyr442 in monomer A. Tyr442Oζ is H-bonded to Trp84Nɛ1 within the ω-loop and to Trp432Nɛ1 in the native structure (C) but not in the complex (D). The red arrows in panels A and B suggest entrance and exit pathways for substrates and products, respectively.

The alternate Tyr442 conformation seen experimentally was also observed transiently in a set of 10 20-ns MD simulations based on native TcAChE (PDB entry code 1EA5, space group P3₁21; see Supporting Information). The χ₁ and χ₂ angles reflecting the Tyr442 side-chain conformations accessed during the simulations include those observed in both 1HBJ and in 2XI4 monomer A (Fig. 2). Individual inspection revealed that the experimentally observed alternate Tyr442 conformation is accessed in only two of the 10 20-ns simulations (Supporting Information Fig. S3). The simulated Tyr442 conformations with χ₁ and χ₂ angles between 150° and 210°, and between 15° and 75°, respectively, correspond to 2.5% of the conformations accessed during the entire 200-ns simulation period (Fig. 2), which provides an idea of the frequency at which a Tyr442 conformation similar to the one observed in 2XI4 and 1HBJ is accessed. Neither the program Ringer,¹⁵ nor the qFit WEB server,¹⁶ found any evidence for an alternate conformation of Tyr442 in the crystal structure of the native enzyme (1EA5) similar to that in 2XI4. Similarly, Ringer did not detect evidence of an alternate conformation for Tyr442 in the aflatoxin complex structure (2XI4). Visual inspection suggested the presence of a native conformation in monomer A that may be populated to a very small level. Therefore, the occupancy of the primary conformation of Tyr442 was set to 80%. There was no visual evidence for an alternate conformation of Tyr442 in monomer B. We conclude that the crystallographically observed alternate Tyr442 conformation in monomer A of 2XI4 is part of the complex energy landscape of the native enzyme, even though it is rarely accessed.

Figure 2.

χ₁ and χ₂ angles of Tyr442 side-chain conformations from simulation and X-ray crystallography. Conformations from 10 20 ns-MD trajectories of native TcAChE (PDB entry code 1EA5) are represented as grey dots, derived from snapshot structures extracted at 1 ps intervals (a total of 200,000). The triangle, inverted triangle, square, circle, diamond, and pentacle represent the conformations in the corresponding crystal structures. The Tyr442 conformation in monomer A of 2XI4 and in 1HBJ is accessed in the MD simulations of the native enzyme and is thus part of its complex energy landscape.

We have recently used MD simulations to explore the exit pathways of the enzymatic hydrolysis product, thiocholine (TCh), from the active site of TcAChE.¹⁷ The crystal structure of a complex in which TCh was bound to TcAChE at both the active site and the PAS (PDB entry code 2C5G¹⁸) served as the starting model for these simulations, with a total simulation time of 2.8 μs. The pathway most frequently taken by TCh to exit the active site was through a backdoor (see Ref.17 for details). Detailed examination of the trajectory of exit of TCh via this route revealed that it involved a conformational change in Trp84 with concomitant breaking of the H-bond between Tyr442Oζ and Trp84Nɛ1 (Fig. 3).

Figure 3.

Positions of the side chains of Trp84, Trp432 and Tyr442, and of TCh taken from the crystal structure of the (TCh)₂/TcAChE complex (PDB entry code 2C5G¹⁸) (green) and from snapshots extracted from an MD simulation of the complex at 4.2 ns (cyan) and 17.1 ns (magenta), which corresponds to simulation group III in Ref.17. The distance between Tyr442Oζ and Trp84Nɛ1 in the crystal structure and the distances between Tyr442Oζ and Trp432Nɛ1 in the crystal structure and in the two MD snapshots are indicated. The arrow emphasizes the path taken by TCh to exit the enzyme through the backdoor.

The breaking of the Tyr442Oζ-Trp84Nɛ1 H-bond associated with the conformational change in Tyr442 observed in monomer A of 2XI4 may result in increased flexibility of the ω-loop [displayed in red in Fig. 1(C,D) and Supporting Information Fig. S2(C,D)], thus permitting the backdoor to open more widely. To explore this hypothesis, 10 20-ns MD simulations were performed based on the 2XI4 structure from which aflatoxin had been omitted (see Supporting Information). A snapshot from that simulation shows that Tyr442Oζ and Trp84Nɛ1 can move as far as 21.5 Å from each other [see Supporting Information Fig. S2(D) and magenta model in Fig. 4(A)], reflecting a large movement of the ω-loop that can occur when the Tyr442Oζ-Trp84Nɛ1 H-bond is broken. Analysis of the distribution of Tyr442Oζ-Trp84Nɛ1 distances in the course of the MD simulations, based on 1EA5 and 2XI4 (Fig. 4B), indicates that the two atoms are on average much farther away from each other if the corresponding H-bond is already broken when the simulation commences (as in the simulations based on 2XI4) than when it is present (as in the simulations based on 1EA5). It should be noted that the peak at 16.5 Å in the distance distribution of simulations based on 1EA5 (green graph in Fig. 4B) corresponds to Tyr442 conformations close to the one observed in 2XI4 and 1HBJ. A large Tyr442Oζ-Trp84Nɛ1 distance such as the one depicted in Supporting Information Figure S2(D) and in the magenta model in Fig. 4(A) produces a backdoor with a smallest surface-to-surface distance of 5.7 Å. Most importantly, the snapshot depicted in Supporting Information Figure S2D and Figure 4(A) leads to sealing of the active-site gorge with concomitant opening of a backdoor [Supporting Information Fig. S2(B)] much more voluminous than that created by the movement of Tyr442 observed in monomer A of 2XI4 [Fig. 1(B)]. A sealed gorge might desolvate the substrate in the active site and increase catalytic efficiency as discussed earlier.¹⁸

Figure 4.

Distance between Tyr442Oζ and Trp84Nɛ1. (A) Orientations of Tyr442 and Trp84 in the crystal structure of native TcAChE (green; PDB entry code 1EA5), and in a conformational snapshot taken 12 ns after initiation of one of the 10 20-ns MD trajectories of monomer A in the aflatoxin/TcAChE complex (magenta; PDB code 2XI4) from which aflatoxin had been omitted. The orientation of the residues is similar to that in Figure 1(C,D) and Supporting Information Figure S2(C,D). (B) Histograms of distances between Tyr442Oζ and Trp84Nɛ1 based on snapshots extracted from the MD trajectories at 1-ps intervals. Each histogram is constructed making use of 200,000 data points and a bin size of 1 Å.

The crystal structure presented, and the related MD simulations, terminate the long-standing controversy concerning the existence of a backdoor in AChE and shed light on the dynamic events accompanying catalysis by this rapid enzyme. The backdoor for whose existence we have now provided unequivocal experimental evidence provides a plausible route for the rapid clearance of the hydrolysis products from the active site. At the heart of the backdoor mechanism is a single H-bond that is linking the side chains of Tyr442 and Trp84 in the two subdomains of TcAChE. If this H-bond is broken, either due to the movement of Tyr442 seen in the 2XI4 crystal structure, or by the movement of Trp84 detected by MD simulations¹⁷ and kinetic crystallography,^11,19 a backdoor channel opens (Fig. 1B). This channel can be considerably enlarged through large-amplitude motions in the ω-loop, which can only occur after the H-bond between Trp84 and Tyr442 has snapped, as shown in the MD simulations (Supporting Information Fig. S2(B,D)]. The characterization of a new conformational substate of AChE, in which the alternate conformations of Tyr442 and of its close environment allow an alternative access route to the active-site gorge, allows better rationalization of the residual catalytic activity observed for AChE in its complex with the mamba venom neurotoxin, fasciculin,⁷ which is known to cap the active-site gorge.^20,21 It also provides a new template for the rational design of AChE inhibitors and reactivators, including anti-Alzheimer drugs and antidotes against organophosphate poisoning, respectively.