Portal de Periódicos da CAPES

Crystal Structure of Human Butyrylcholinesterase and of Its Complexes with Substrate and Products

2003; Elsevier BV; Volume: 278; Issue: 42 Linguagem: Inglês

10.1074/jbc.m210241200

1083-351X

Yvain Nicolet, Oksana Lockridge, Patrick Masson, Juan C. Fontecilla‐Camps, Florian Nachon,

Pesticide Exposure and Toxicity

Cholinesterases are among the most efficient enzymes known. They are divided into two groups: acetylcholinesterase, involved in the hydrolysis of the neurotransmitter acetylcholine, and butyrylcholinesterase of unknown function. Several crystal structures of the former have shown that the active site is located at the bottom of a deep and narrow gorge, raising the question of how substrate and products enter and leave. Human butyrylcholinesterase (BChE) has attracted attention because it can hydrolyze toxic esters such as cocaine or scavenge organophosphorus pesticides and nerve agents. Here we report the crystal structures of several recombinant truncated human BChE complexes and conjugates and provide a description for mechanistically relevant non-productive substrate and product binding. As expected, the structure of BChE is similar to a previously published theoretical model of this enzyme and to the structure of Torpedo acetylcholinesterase. The main difference between the experimentally determined BChE structure and its model is found at the acyl binding pocket that is significantly bigger than expected. An electron density peak close to the catalytic Ser198 has been modeled as bound butyrate. Cholinesterases are among the most efficient enzymes known. They are divided into two groups: acetylcholinesterase, involved in the hydrolysis of the neurotransmitter acetylcholine, and butyrylcholinesterase of unknown function. Several crystal structures of the former have shown that the active site is located at the bottom of a deep and narrow gorge, raising the question of how substrate and products enter and leave. Human butyrylcholinesterase (BChE) has attracted attention because it can hydrolyze toxic esters such as cocaine or scavenge organophosphorus pesticides and nerve agents. Here we report the crystal structures of several recombinant truncated human BChE complexes and conjugates and provide a description for mechanistically relevant non-productive substrate and product binding. As expected, the structure of BChE is similar to a previously published theoretical model of this enzyme and to the structure of Torpedo acetylcholinesterase. The main difference between the experimentally determined BChE structure and its model is found at the acyl binding pocket that is significantly bigger than expected. An electron density peak close to the catalytic Ser198 has been modeled as bound butyrate. Cholinesterases are divided into two subfamilies according to their substrate and inhibitor specificities: acetylcholinesterase (AChE 1The abbreviations used are: AchE, acetylcholinesterase; BchE, butyrylcholinesterase; TcAChE, Torpedo californica acetylcholinesterase; DmAChE, Drosophila melanogaster acetylcholinesterase; BTC, butyrylthiocholine; Bicine, N,N-bis(2-hydroxyethyl)glycine; MES, 4-morpholineethanesulfonic acid.; EC 3.1.1.7) and butyrylcholinesterase (BChE; EC 3.1.1.8). Acetylcholinesterase is responsible for the hydrolysis of acetylcholine released at the synaptic cleft and the neuromuscular junction in response to nerve action potential (1Massoulie J. Sussman J. Bon S. Silman I. Prog. Brain Res. 1993; 98: 139-146Crossref PubMed Scopus (125) Google Scholar). In addition, both AChE and BChE seem to be involved in roles that are independent of their catalytic activities, such as cell differentiation and development (2Meshorer E. Erb C. Gazit R. Pavlovsky L. Kaufer D. Friedman A. Glick D. Ben-Arie N. Soreq H. Science. 2002; 295: 508-512Crossref PubMed Scopus (205) Google Scholar, 3Behra M. Cousin X. Bertrand C. Vonesch J.L. Biellmann D. Chatonnet A. Strahle U. Nat. Neurosci. 2002; 5: 111-118Crossref PubMed Scopus (307) Google Scholar). The catalytic mechanism of AChE is extremely efficient approaching diffusion-controlled rates (4Quinn D. Chem. Rev. 1987; 87: 955-979Crossref Scopus (946) Google Scholar). Unexpectedly, the crystal structure of the Torpedo californica enzyme (TcAChE) showed that the active site catalytic Ser-His-Glu triad is found at the bottom of a 20-Å deep gorge lined mostly with aromatic residues (5Sussman J.L. Harel M. Frolow F. Oefner C. Goldman A. Toker L. Silman I. Science. 1991; 253: 872-879Crossref PubMed Scopus (2426) Google Scholar). The structure also revealed the nature and the location of the previously described peripheral and "anionic" sites; the former, located at the outer rim of the gorge, has been postulated to be the initial substrate binding site (6Szegletes T. Mallender W.D. Thomas P.J. Rosenberry T.L. Biochemistry. 1999; 38: 122-133Crossref PubMed Scopus (152) Google Scholar). The binding of ligand to this site has been proposed to slow down the traffic of substrate and product at the acylation site (6Szegletes T. Mallender W.D. Thomas P.J. Rosenberry T.L. Biochemistry. 1999; 38: 122-133Crossref PubMed Scopus (152) Google Scholar, 7De Ferrari G.V. Mallender W.D. Inestrosa N.C. Rosenberry T.L. J. Biol. Chem. 2001; 276: 23282-23287Abstract Full Text Full Text PDF PubMed Scopus (127) Google Scholar). Although a similar peripheral site has been described for human BChE, site-directed mutagenesis and photo-affinity labeling studies showed that its location and the response upon ligand binding differ significantly from those of AChE (8Nachon F. Ehret-Sabatier L. Loew D. Colas C. van Dorsselaer A. Goeldner M. Biochemistry. 1998; 37: 10507-10513Crossref PubMed Scopus (52) Google Scholar, 9Masson P. Legrand P. Bartels C.F. Froment M.T. Schopfer L.M. Lockridge O. Biochemistry. 1997; 36: 2266-2277Crossref PubMed Scopus (133) Google Scholar). The site to which the positively charged quaternary ammonium of choline moiety productively binds is found half-way down the gorge, in between the peripheral and acylation sites. Originally, there was a great deal of controversy concerning the nature of the residues involved in this site. Both the crystal structure and labeling experiments showed that positively charged ligands form π-cation interactions with Phe 330 and Trp 84 (numbering in italics corresponds to that of torpedo AChE) (10Harel M. Schalk I. Ehret-Sabatier L. Bouet F. Goeldner M. Hirth C. Axelsen P.H. Silman I. Sussman J.L. Proc. Natl. Acad. Sci. U. S. A. 1993; 90: 9031-9035Crossref PubMed Scopus (845) Google Scholar). The physiological role of BChE remains unclair (11Chatonnet A. Lockridge O. Biochem. J. 1989; 260: 625-634Crossref PubMed Scopus (472) Google Scholar, 12Mack A. Robitzki A. Prog. Neurobiol. 2000; 60: 607-628Crossref PubMed Scopus (122) Google Scholar). Although it is capable of hydrolyzing ACh and other acylcholines, so far no endogenous natural substrate has been described for this enzyme. Because BChE is relatively abundant in plasma (about 3 mg/liter), and can degrade a large number of ester-containing compounds, it plays important pharmacological and toxicological roles (13Lockridge O. Masson P. Neurotoxicology. 2000; 21: 113-126PubMed Google Scholar). For instance, BChE is a potential detoxifying enzyme to be used as a prophylactic scavenger against neurotoxic organophosphates such as the nerve gas soman (14Allon N. Raveh L. Gilat E. Cohen E. Grunwald J. Ashani Y. Toxicol. Sci. 1998; 43: 121-128PubMed Google Scholar, 15Broomfield C.A. Maxwell D.M. Solana R.P. Castro C.A. Finger A.V. Lenz D.E. J. Pharmacol. Exp. Ther. 1991; 259: 633-638PubMed Google Scholar, 16Raveh L. Grunwald J. Marcus D. Papier Y. Cohen E. Ashani Y. Biochem. Pharmacol. 1993; 45: 2465-2474Crossref PubMed Scopus (173) Google Scholar). We have recently published the engineering and crystallization of a monomeric and partially glycosylated recombinant human BchE (17Nachon F. Nicolet Y. Viguie N. Masson P. Fontecilla-Camps J.C. Lockridge O. Eur. J. Biochem. 2002; 269: 630-637Crossref PubMed Scopus (121) Google Scholar). Here we report several crystal structures of BChE complexed with a substrate, products, and conjugated to soman after aging. From these structures we propose alternative substrate and product binding that may be related to the high catalytic efficiency of the choline esterases. Crystallization of Recombinant BChE and Its Complexes—Recombinant human BChE suitable for crystallization was obtained, purified, and crystallized as described previously (17Nachon F. Nicolet Y. Viguie N. Masson P. Fontecilla-Camps J.C. Lockridge O. Eur. J. Biochem. 2002; 269: 630-637Crossref PubMed Scopus (121) Google Scholar). No butyrate was present in the culture medium and none was added during any step of the purification procedure. The 3-bromopropionate-BChE complex was obtained by soaking crystals for a few minutes in the mother liquor containing 100 mm bromopropionate (Sigma). The BChE-choline complex was obtained by soaking BChE crystals grown from a 2.1 m (NH4)2SO4 100 mm Bicine (Fluka), pH 9.0, crystallization solution, in the mother liquor containing 100 mm choline chloride. Crystallization of soman-aged BChE: racemic soman (pinacolyl methylphosphonofluoridate) was obtained from CEB Le Bouchet (Vert-le-Petit, France). The purified enzyme (6.6 mg/ml) was inhibited in the presence of 0.5 mm soman (∼5.5-fold molar excess) in 10 mm MES buffer, pH 6.5. The reaction mixture was further incubated for 3 days at 4 °C, allowing enough time for completion of the aging reaction and the disappearance of the remaining unreacted soman. The inhibited enzyme was crystallized under the same conditions as the uninhibited BChE except that the mother liquor was buffered at pH 8.0 using a 0.1 m Tris/HCl buffer solution. Soman-aged BChE and butyrylthiocholine (BTC) were cocrystallized at pH 6.5 (0.1 m MES buffer, 2.1 m (NH4)2SO4) with 10 mm BTC (Sigma). X-ray data were collected from a 4-day-old crystal to limit the loss of substrate by spontaneous hydrolysis. X-ray Data Collection and Structure Solution—All crystals were flash-cooled at 100 K in a nitrogen stream using 15–20% glycerol in the mother liquor as cryoprotectant. Data sets were collected at the following beamlines of the European Synchrotron Radiation Facility (Grenoble, France): ID14-eh1 for the soman-aged BChE, ID14-eh2 for the native BChE, BM30 for the choline-BChE complex and the somanaged BTC complex, and ID14-eh4 for the 3-bromopropionate-BChE complex (see Table I). Data sets for the native, the soman-aged BchE, and soman-aged BChE-BTC complex crystals were integrated, scaled, and reduced using MOSFLM, SCALA, and TRUNCATE from the CCP4 suite (18Collaborative Computational Project No. 4Acta Crystallogr. Sect. D Biol. Crystallogr. 1994; 50: 760-763Crossref PubMed Scopus (19770) Google Scholar). Data sets from the choline-BChE complex and the 3-bromopropionate-BChE species were processed using the program XDS (19Kabsch W. J. Appl. Crystallogr. 1993; 26: 795-800Crossref Scopus (3233) Google Scholar). Data collection from the 3-bromopropionate-BChE crystal was performed at a wavelength of 0.915 Å to measure the Br anomalous scattering signal. Subsequent data processing was performed without merging the Friedel mates to better evaluate the anomalous signal of the bromine atom. Molecular replacement was carried out with the native data set between 15- and 3.5-Å resolution using the program AMoRe (20Navaza J. Acta Crystallogr. Sect. A. 1994; 50: 157-163Crossref Scopus (5029) Google Scholar) and the TcAChE structure (Protein Data Bank ID code: 2ACE) as a search model. A well constrasted solution with R-factor = 42.4% and correlation coefficient = 46.7% was obtained for one monomer per asymmetric unit. This solution was used as a starting model for manual rebuilding and refinement of BChE against all data to 2.0-Å resolution with the programs TURBO (21Roussel A. Cambillaud C. TURBO computer program. Silicon Graphics, Mountain View, CA1989Google Scholar) and CNS (22Brünger A.T. Adams P.D. Clore G.M. DeLano W.L. Gross P. Grosse-Kunstleve R.W. Jiang J.S. Kuszewski J. Nilges M. Pannu N.S. Read R.J. Rice L.M. Simonson T. Warren G.L. Acta Crystallogr. 1998; D54: 905-921Crossref Scopus (16967) Google Scholar), respectively. Observed structure factors were scaled anisotropically and a bulk solvent correction was applied. Several cycles of refinement, manual rebuilding, and solvent addition led to a model with good statistics (Table I). Residues 1–3, 378–379, and 455 did not have matching electron density and, consequently, were not included in the model. Carbohydrate chains corresponding to five of the six expected glycosylation sites (17Nachon F. Nicolet Y. Viguie N. Masson P. Fontecilla-Camps J.C. Lockridge O. Eur. J. Biochem. 2002; 269: 630-637Crossref PubMed Scopus (121) Google Scholar) have been included in the crystallographic model. They correspond to those connected to Asn57, Asn106, Asn241, Asn341, and Asn485. Although Asn256 was expected to be glycosylated, no electron density was observed for carbohydrate. Further examination of the electron density maps led to the inclusion of three molecules of glycerol, used as cryoprotectant, two sulfate ions, the precipitanting agent, one molecule of MES buffer, and two chloride ions. During refinement, an unexpected residual positive electron density was observed around the Oγ atom of the catalytic serine. At the final stages of refinement and after several attempts at modeling this density as corresponding to chemicals added either during purification or crystallization, a very good fit was only obtained when butyrate was used as a model (see Fig. 2).Table IData and refinement statisticsData setNative BChE3-BromopropionateCholineaThe choline-BChE complex was refined to 2.38-Å resolution.Soman-agedSoman-aged/BSChSpace groupI422I422I422I422I422Unit cell axes, a = b, c (Å)154.659 127.890154.920 126.939154.657 128.570154.616 127.307154.924 127.909X-ray sourceID14-eh2ID14-eh4 (λ = 0.915Å)BM30ID14-eh1ID14-eh2Resolution range (Å)55.0-2.02.72.3240.6-2.4354.7-2.3No. of unique reflections51,32637,108b3-Bromopropionate-BChE data processing was carried out without merging the Friedel mates to evaluate the anomalous signal from the bromine atom.29,37825,86828,730Completeness (%)98.6 (99.6)94.1 (69.9)87.4 (77.7)90.8 (88.6)86.5 (49.6)R sym0.074 (0.448)0.103 (0.506)0.061 (0.398)0.071 (0.217)0.081 (0.264)I σ(I)6.8 (1.7)11.75 (2.72)23.7 (4.3)7.6 (3.3)8.1 (2.8)Multiplicity7.3 (6.4)4.97 (4.08)9.9 (6.7)10.2 (6.2)6.4 (5.5)R-factor (free R-factor)0.195 (0.225)0.227 (0.229)0.215 (0.244)0.199 (0.235)0.199 (0.242)No. atomsProtein4,1994,2074,1574,175Others155154126147Solvent496264286337B (Å2)36.549.238.335.1Root mean square from idealityBond length (Å)0.0140.0080.0060.006Angles (°)1.61.41.31.4Dihedrals (°)23.723.423.423.6Impropers (°)1.080.880.830.81a The choline-BChE complex was refined to 2.38-Å resolution.b 3-Bromopropionate-BChE data processing was carried out without merging the Friedel mates to evaluate the anomalous signal from the bromine atom. Open table in a new tab The soman-aged BChE structure was solved using the rigid body refinement from the program CNS. The refined model statistics are shown in Table I. Residues 1–3, 378–379, and 455 and carbohydrates bonded to Asn256 and Asn485 had no significant matching electron density and consequently were not included in the final model. The soman-aged BChE structure also contains the three glycerol molecules, the two chloride ions, and one of the two sulfate ions previously observed in the native structure. The same refinement scheme was applied to this and the choline-BChE structure (Table I). The structure of the 3-bromopropionate-BChE complex was solved using the native BChE structure as a starting coordinate set in which the modeled butyrate had been removed. Anomalous difference and F obs – F calc difference Fourier maps were calculated with the program CNS. The starting Protein Data Bank coordinates, parameters, and topology files for 3-bromopropionate were obtained using PRODRG (23van Aalten D.M.F. Bywater R. Findlay J.B.C. Hendlich M. Hooft R.W. Vriend G.W. J. Comput. Aided Mol. Des. 1996; 10: 255-262Crossref PubMed Scopus (569) Google Scholar). Subsequently, it was manually positioned to fit the electron density maps. Comparable B-factors for bromine and the other 3-bromopropionate atoms were obtained when fixing the occupancy to 0.5, suggesting that the putative butyryl moiety was only partially substituted during the soaking process. Crystallographic Analysis of Uninhibited Butyrylcholinesterase—The BChE crystal structure was originally solved by the molecular replacement method and subsequently refined to 2.0-Å resolution using CNS (see "Experimental Procedures" and Table I). As expected (24Harel M. Sussman J.L. Krejci E. Bon S. Chanal P. Massoulie J. Silman I. Proc. Natl. Acad. Sci. U. S. A. 1992; 89: 10827-10831Crossref PubMed Scopus (292) Google Scholar, 25Kaplan D. Ordentlich A. Barak D. Ariel N. Kronman C. Velan B. Shafferman A. Biochemistry. 2001; 40: 7433-7445Crossref PubMed Scopus (59) Google Scholar, 26Radic Z. Pickering N.A. Vellom D.C. Camp S. Taylor P. Biochemistry. 1993; 32: 12074-12084Crossref PubMed Scopus (420) Google Scholar, 27Saxena A. Redman A.M.G. Jiang X.L. Lockridge O. Doctor B.P. Biochemistry. 1997; 36: 14642-14651Crossref PubMed Scopus (226) Google Scholar, 28Vellom D.C. Radic Z. Li Y. Pickering N.A. Camp S. Taylor P. Biochemistry. 1993; 32: 12-17Crossref PubMed Scopus (258) Google Scholar), the overall structure of BChE is very similar to that of TcAChE. However, BChE does not form the dimer observed in previous structures of TcAChE (5Sussman J.L. Harel M. Frolow F. Oefner C. Goldman A. Toker L. Silman I. Science. 1991; 253: 872-879Crossref PubMed Scopus (2426) Google Scholar, 29Harel M. Kleywegt G.J. Ravelli R.B. Silman I. Sussman J.L. Structure (Lond.). 1995; 3: 1355-1366Abstract Full Text Full Text PDF PubMed Scopus (221) Google Scholar), mouse AChE (30Bourne Y. Taylor P. Marchot P. Cell. 1995; 83: 503-512Abstract Full Text PDF PubMed Scopus (319) Google Scholar, 31Bourne Y. Taylor P. Bougis P.E. Marchot P. J. Biol. Chem. 1999; 274: 2963-2970Abstract Full Text Full Text PDF PubMed Scopus (137) Google Scholar), and human AChE (32Kryger G. Harel M. Giles K. Toker L. Velan B. Lazar A. Kronman C. Barak D. Ariel N. Shafferman A. Silman I. Sussman J.L. Acta Crystallogr. Sect. D Biol. Crystallogr. 2000; 56: 1385-1394Crossref PubMed Scopus (303) Google Scholar). In AChEs, the four helices involved in the subunit-subunit interaction are antiparallel and the active site openings are located on opposite sides of the dimer. Although homologous helices are found at the BChE dimer interface (residues 362–375 and 514–529), they are not antiparallel but form a ≈45° angle, and the active site openings are located on the same side of the dimer (not shown). Constraints resulting from the crystal packing could be responsible for these differences. Most differences between BChE and TcAChE are confined to 1) the residues lining the gorge, where the former enzyme has replaced several of the aromatic groups of the latter by hydrophobic ones; 2) the acyl-binding pocket, with the replacements of Phe 288 and Phe 290 of TcAChE by Leu286 and Val288, respectively; these changes make it possible for the binding of the bulkier butyrate substrate moiety in BChE (as discussed below); and 3) the conformation of the acyl loop (Fig. 1). In addition, the catalytic serine is connected to a large electron density peak whose identity is discussed in the next section. We note, in passing, that a previously modeled structure of BChE (24Harel M. Sussman J.L. Krejci E. Bon S. Chanal P. Massoulie J. Silman I. Proc. Natl. Acad. Sci. U. S. A. 1992; 89: 10827-10831Crossref PubMed Scopus (292) Google Scholar) differs from the actual structure precisely at these regions, indicating a strong bias toward the AChE starting atomic coordinate set. As an example, the acyl pocket of the model is significantly smaller than the one observed in the crystal structure. A Putative Butyryl Moiety Bound to the Catalytic Serine— During the crystallographic refinement procedure, it became evident that the catalytic serine 198 was bound to an unidentified moiety that was eventually modeled as butyrate with well matching electron density (Fig. 2a and "Experimental Procedures"). To confirm this result, we soaked native crystals with the butyrate isosteric analog 3-bromopropionate in which the γ carbon is replaced by a bromine atom. Subsequently, we collected x-ray data at 0.915-Å wavelength where bromine gives rise to a strong anomalous signal. Fig. 2b clearly indicates extensive substitution of the putative butyryl moiety by its brominated analog with essentially no changes in the shape of the corresponding electron density. In the case of the butyrated enzyme the length of the bond connecting the acyl group to the catalytic serine refined to values >2.0 Å, with no residual peaks in the Fourier difference electron density maps. On the other hand, maps calculated after crystallographic refinement of either restrained non-bonded butyrate or the tetrahedral intermediate display significant negative peaks, suggesting that neither one is a good model for the Ser198-butyrate interaction (Fig. 3). We are currently investigating the plausibility of the long bond using density function theory calculations. Although the source of the butyryl moiety is unknown, it seems to be essential for crystallization because "butyrate-depleted" samples were extremely difficult to crystallize. In fact, the few crystals grown from these samples had the putative butyrate still bound to Ser198. Furthermore, crystals were readily obtained when butyrate, or BTC, was added to the crystallization solutions. 2J. Colletíer, personal communication. This phenomenon may be related to the stabilization of the extensive acyl pocket in BChE by butyrate. A Mobile Acyl Loop—As we already noticed when comparing the native hBChE, TcAChE, and DmAChE structures, although the omega loop has the same conformation (see Fig. 1), residues belonging to the acyl loop have appreciably different orientations in the three cholinesterases. Furthermore, in BChE the acyl loop changes conformation depending on the substitution at the catalytic serine by butyrate or soman (Fig. 4). The reaction of soman and other related organophosphate esters with the catalytic serine of cholinesterases can be reversed initially by nucleophilic agents such as oximes (reviewed in (33Dawson R.M. J. Appl. Toxicol. 1994; 14: 317-331Crossref PubMed Scopus (318) Google Scholar)). However, with time, serine phosphonylation becomes irreversible through dealkylation of an alkoxy chain on the phosphorus atom, a process called "aging" (34Harris L.W. Fleisher J.H. Clark J. Cliff W.J. Science. 1966; 154: 404-407Crossref PubMed Scopus (46) Google Scholar). Except for its size, the resulting "aged" phosphonyl adduct is a structural analog of both the butyryl-serine association we have detected in the original BChE crystals and of the postulated deacylation tetrahedral intermediate. The protonated His438 forms a salt bridge with one oxygen atom of the methylphosphonyl moiety. Such a bridge was also observed in the crystal structures of soman-, sarin-, and phosphonylfluoridate-aged TcAChE (35Millard C.B. Kryger G. Ordentlich A. Greenblatt H.M. Harel M. Raves M.L. Segall Y. Barak D. Shafferman A. Silman I. Sussman J.L. Biochemistry. 1999; 38: 7032-7039Crossref PubMed Scopus (258) Google Scholar). A conformational change of the acyl loop has also been reported for the crystal structure of phosphonylfluoridate-aged TcAChE. Taken together, these changes in cholinesterases show that the acyl loop is flexible and are consistent with molecular dynamic studies, indicating that acetate could leave the AChE active site either directly through the gorge, "above" this loop or through an alternative exit located "below" it (36Enyedy I.J. Kovach I.M. Brooks B.R. J. Am. Chem. Soc. 1998; 120: 8043-8050Crossref Scopus (28) Google Scholar). As mentioned above, stabilization of the acyl loop by butyrate seems to be required for crystallization. Alternative Choline Binding Site at the Active Site—To obtain an image of the unsubstituted Ser198 in BChE, we soaked our crystals in a 100 mm choline chloride solution, pH 9.0. Indeed, maps calculated with x-ray data collected from a choline-soaked crystal displayed no electron density corresponding to the putative butyryl moiety. In this structure the Oγ from Ser198 displays two approximately equally occupied conformations: a classical one where it forms a hydrogen bond with the catalytic His438, and a much less expected one, where it has turned away from the histidine and establishes a hydrogen bond with the main chain N-H from Ala199 at the oxyanion hole (Fig. 5, a and b). It is unlikely that this is solely due to the deprotonation of His438 at pH 9.0, because identical crystals grown at pH 6.5 display similar Ser198 conformations when soaked with cocaine, tacrine, or decamethonium, and a native structure solved at pH 9.0 still has the putative butyrate bound to the active site (not shown). Another surprising feature in this structure is the orientation of the bound choline. Although, as expected, its quaternary ammonium group binds close to Trp82 through a π-cation interaction, the alcohol group points roughly toward the gorge entrance and forms a hydrogen bond with a water molecule that, in turn, interacts with the main chain carbonyl O from His438. Although choline binding may not be physiologically relevant (given the high choline concentration in the soaking solution and the basic pH), it shows that this reaction product may adopt at least two conformations at the active site: the productive one, as part of butyrylcholine bound to the active site region, and an alternative one, as observed here (Fig. 5, a and b). The existence of an alternative choline binding mode prompted us to perform an additional experiment where crystals obtained from protein previously inhibited with soman were soaked in a solution containing the substrate BTC. In the complex between soman-aged BChE and BTC (pH 6.5) (Fig. 5c), the latter binds very much like choline, with its ammonium group close to Trp82 and its carbonyl group hydrogen-bonded to a water molecule topologically equivalent to the one described above for the BChE-choline complex (Fig. 5). This confirms the above-stated notion that substrate binding through π-cation interactions is bimodal and that a tetrahedral adduct bound to the catalytic serine is sterically compatible with intermediate site substrate binding as defined here by the soman-aged BChE-BTC complex. Crystallographic data for all the structures are presented in Table I. Taken together, the results presented here, in conjuction with data from the literature, shed new light concerning the catalytic mechanism of cholinesterases. These enzymes appear to function through an assembly line approach where every step in the reaction is tightly controlled by the protein environment and extremely efficient use is made of the available substrate binding sites. Peripheral sites serve as attraction centers for substrate at the outer rim of the gorge (an observation confirmed by a BChE-decamethonium complex; not shown). From one or several of these positions, substrate can move into an intermediate site, equivalent to the one observed in the soman-aged BChE-BTC complex. Computer graphics simulations show that, as predicted by Masson et al. (9Masson P. Legrand P. Bartels C.F. Froment M.T. Schopfer L.M. Lockridge O. Biochemistry. 1997; 36: 2266-2277Crossref PubMed Scopus (133) Google Scholar), a simple rotation of the molecule around its quaternary ammonium group, bound at the π-cation site, can bring the substrate to productive binding. Hydrogen bonding of the carbonyl group of BTC to a water molecule, rather than to protein, may facilitate this transition. It has been proposed that the position of the catalytic triad in cholinesterases, at the bottom of a 20-Å-deep gorge allows the exothermic formation of the transition state in a "dry" environment (37Harel M. Quinn D.M. Haridasan K.N. Silman I. Sussman J.L. J. Am. Chem. Soc. 1996; 118: 2340-2346Crossref Scopus (342) Google Scholar). The results presented here provide an additional explanation for the location of the buried catalytic triad as the presence of peripheral and intermediate sites at the gorge allow for a drastic reduction of substrate mobility, leading it to the active site in a most efficient way. The presence of a putative butyrate bound to Ser198 could be related to the reactivity and/or stability of the catalytic serine. This is supported by the recent observation by Bourne et al. (38Bourne Y. Taylor P. Radic Z. Marchot P. EMBO J. 2003; 22: 1-12Crossref PubMed Scopus (328) Google Scholar) who described a carbonate or acetate molecule found close to the catalytic serine of mouse AChE in complex with a peripheral site inhibitor. Interestingly, other structures of AChE also display electron densitiy peaks very close to the catalytic serine (39Harel M. Kryger G. Rosenberry T.L. Mallender W.D. Lewis T. Fletcher R.J. Guss J.M. Silman I. Sussman J.L. Protein Sci. 2000; 9: 1063-1072Crossref PubMed Scopus (263) Google Scholar, 40Weik M. Ravelli R.B. Kryger G. McSweeney S. Raves M.L. Harel M. Gros P. Silman I. Kroon J. Sussman J.L. Proc. Natl. Acad. Sci. U. S. A. 2000; 97: 623-628Crossref PubMed Scopus (461) Google Scholar). Although the distance between Oγ of Ser198 and Nϵ2 of His438 of 2.8 Å is compatible with hydrogen bonding, the angle between Cδ2, Nϵ2, and Oγ is about 170°, a value that is unfavorable for that type of interaction. In fact, if this hydrogen bond geometry were better, proton transfer to the alcoholate of the choline moiety would be less favorable (similar observations have been made on serine proteases. For example, the Streptomyces griseus protease A structure refined at 1.8-Å resolution (41Matthews D.A. Alden R.A. Birktoft J.J. Freer T. Kraut J. J. Biol. Chem. 1977; 252: 8875-8883Abstract Full Text PDF PubMed Google Scholar, 42Sielecki A.R. Hendrickson W.A. Broughton C.G. Delbaere L.T. Brayer G.D. James M.N. J. Mol. Biol. 1979; 134: 781-804Crossref PubMed Scopus (91) Google Scholar) shows a Ser-His hydrogen bond distance of 3.1 Å and the orientation is far from optimal, suggesting that the bond is very weak or even absent). One way to overcome the problem of poor geometry of the His-Ser hydrogen bond is to propose that the side chain of His438 rotates to optimize hydrogen bonding to Ser198 when this residue is free. However, our observations do not support this hypothesis: His438 seems to be well ordered and be similarly oriented in all the structures discussed here. On the other hand, we have observed that in several butyrate-free BChE complexes the Oγ atom of Ser198 is highly disordered and shows no tendency to bind to His438 (see, for example, Fig. 5). Furthermore, a very recent report by Masson et al. (43Masson P. Nachon F. Bartels C.F. Froment M.T. Ribes F. Matthews C. Lockridge O. Eur. J. Biochem. 2003; 270: 315-324Crossref PubMed Scopus (27) Google Scholar) indicates that in the presence of excess substrate and low pH, BChE is still active, although His438 should remain protonated under these conditions. Consequently, neither our results, nor previously reported data (41Matthews D.A. Alden R.A. Birktoft J.J. Freer T. Kraut J. J. Biol. Chem. 1977; 252: 8875-8883Abstract Full Text PDF PubMed Google Scholar, 42Sielecki A.R. Hendrickson W.A. Broughton C.G. Delbaere L.T. Brayer G.D. James M.N. J. Mol. Biol. 1979; 134: 781-804Crossref PubMed Scopus (91) Google Scholar, 43Masson P. Nachon F. Bartels C.F. Froment M.T. Ribes F. Matthews C. Lockridge O. Eur. J. Biochem. 2003; 270: 315-324Crossref PubMed Scopus (27) Google Scholar), favor the deprotonation of the catalytic serine by His438 during turnover. One alternative to deprotonation of the catalytic serine by His438 during turnover would be a concerted product release/substrate binding at Ser198, whose Oγ would remain deprotonated throughout the catalytic cycle. The presence of a putative serine-bound butyrate moiety in the purified BChE hints at this possibility. As discussed above, upon productive substrate binding at the active site the C=O bond becomes polarized through its interaction with the oxyanion hole: a lone pair orbital develops on the oxygen atom, and an empty orbital forms on the carbon atom, facilitating the nucleophilic attack of the Oγ atom of Ser198 on the partially positive carbon atom. This makes the first step of the reaction easier by lowering the energy barrier toward the formation of the first tetrahedral intermediate. A similar mechanism would operate when a water molecule hydrolyzes the acyl-enzyme intermediate. X-ray structures of both "native" and soman-aged BChE have a glycerol molecule bound to the π-cation site. As glycerol was used only as cryoprotectant (see the "Experimental Procedures"), its presence in the active site gorge confirms the broad specificity of the π-cation site and that the substrate specificity is mainly due to residues Val288 and Leu286 lining the acyl pocket. The presence of a moiety next to the catalytic serine is intriguing. Several lines of evidence indicate that in our case it is most likely butyrate as it can be replaced by 3-bromopropionate, it is displaced by choline, and the presence of butyrate is essential for crystrallization. However, additional experiments will be required to definetively establish its identity. Atomic coordinates and structure factors have been deposited with the Protein Data Bank codes 1P0I (BChE with butyrate), 1P0Q (BChE with aged soman), 1P0M (BChE plus choline), and 1P0P (BChE with aged soman and BTC) and will be released upon publication. We are grateful to the staffs of ID14 and FIP beamlines from the European Synchrotron Radiation facility. Julien Lescar is especially thanked for his help with data collection from the 3-bromopropionate-BChE crystal.