Mouse butyrylcholinesterase (mBChE) and an mBChE-based cocaine hydrolase (mCocH, i.e. the A199S/S227A/S287G/A328W/Y332G mutant) have been characterized for their catalytic activities against cocaine, i.e. naturally occurring (−)-cocaine, in comparison with the corresponding human BChE (hBChE) and an hBChE-based cocaine hydrolase (hCocH, i.e. the A199S/F227A/S287G/A328W/Y332G mutant). It has been demonstrated that mCocH and hCocH have improved the catalytic efficiency of mBChE and hBChE against (−)-cocaine by ~8- and ~2000-fold respectively, although the catalytic efficiencies of mCocH and hCocH against other substrates, including acetylcholine (ACh) and butyrylthiocholine (BTC), are close to those of the corresponding wild-type enzymes mBChE and hBChE. According to the kinetic data, the catalytic efficiency (kcat/KM) of mBChE against (−)-cocaine is comparable with that of hBChE, but the catalytic efficiency of mCocH against (−)-cocaine is remarkably lower than that of hCocH by ~250-fold. The remarkable difference in the catalytic activity between mCocH and hCocH is consistent with the difference between the enzyme–(−)-cocaine binding modes obtained from molecular modelling. Further, both mBChE and hBChE demonstrated substrate activation for all of the examined substrates [(−)-cocaine, ACh and BTC] at high concentrations, whereas both mCocH and hCocH showed substrate inhibition for all three substrates at high concentrations. The amino-acid mutations have remarkably converted substrate activation of the enzymes into substrate inhibition, implying that the rate-determining step of the reaction in mCocH and hCocH might be different from that in mBChE and hBChE.
Cocaine is a widely abused drug  with no FDA (U.S. Food and Drug Administration)-approved medication available for treatment-seeking users. A promising concept for anti-cocaine medication is to accelerate cocaine metabolism by hydrolysis at the benzoyl ester, producing biologically inactive metabolites [2–7]. In humans, butyrylcholinesterase (BChE) is the primary endogenous cocaine-metabolizing enzyme capable of catalysing this reaction in plasma. However, wild-type human BChE (hBChE) has a low catalytic activity against naturally occurring (−)-cocaine (kcat=4.1 min−1 and KM=4.5 μM) [8–12]. Our previous efforts were focused on improving the catalytic activity of hBChE against (−)-cocaine, leading to the discovery of various hBChE mutants [8,9,13–18] with a considerably improved catalytic efficiency towards that drug. The highly efficient hBChE mutants, such as A199S/S287G/A328W/Y332G (kcat=3060 min−1 and KM=3.1 μM) [14,19] or A199S/F227A/S287G/A328W/Y332G (kcat=5700 min−1 and KM=3.1 μM) , can be described as cocaine hydrolases (CocH). Initial experiments in rats and mice [16,17,20–29] showed that CocH is likely to be effective as an enzyme therapy or gene therapy for treating cocaine abuse by greatly reducing the reward value of a given drug dosage. In addition, we have neither seen any acute toxicity of hCocH (hBChE-based cocaine hydrolase) in mice or rats nor have other investigators found that wild-type hBChE elicited adverse effects in experimental animals [30–32]. Two clinical trials (NCT00333515 and NCT00333528) of hBChE have been performed by Baxter Healthcare Corporation, although the clinical data have not been made available.
Not surprisingly, some mice and rats eventually develop antibodies against hBChE and hCocH, accelerating the clearance of these enzymes and lowering their plasma levels , although no immune response is noted when mouse BChE (mBChE) is injected into mice . This outcome was expected because hBChE shares only ~80% sequence identity with its rodent counterpart . We deemed it unlikely that human beings would generate antibodies to hCocH as the mutated residues are not exposed on the surface but occupy a deep and narrow catalytic gorge. Nonetheless, the mouse response called for further experiments to test the hypothesis that mutations in the catalytic site are not antigenic. Therefore, we developed a conspecific CocH with equivalent mutations in mBChE: A199S/S227A/S287G/A328W/Y332G, designated mouse-butyrylcholinesterase-based CocH (or ‘mCocH’). The catalytic properties of mCocH were compared with those of mBChE, hBChE and hCocH and it was incorporated into a viral gene-transfer vector for in vivo studies with the aim of avoiding complications from an immune response in the animals.
In fact, initial gene transfer experiments with mCocH showed that very high levels of enzyme protein could be generated, of the order of 1000-fold above the native mBChE background level . However, the levels of cocaine hydrolysis did not increase to the extent achieved with hCocH. This outcome suggested that although mCocH and hCocH contain similar mutations, their catalytic efficiencies with (−)-cocaine are different and results from the gene transfer study were consistent with this interpretation . To explore the reason, in the present study, mBChE and mCocH proteins were compared with hBChE and hCocH with regard to catalytic properties against (−)-cocaine and various other substrates. In addition, homology modelling and molecular dynamics (MD) simulations were used to compare structural features of the two mutated enzymes. As will be shown in the present paper, the catalytic efficiency of mCocH against (−)-cocaine is indeed lower than that of hCocH and computational modelling of the detailed 3D structures provides some insight into the reasons for this conclusion, which in turn may facilitate future attempts at re-engineering enzymes for therapeutic purposes.
MATERIALS AND METHODS
The cDNA for mBChE containing N-terminal signal was kindly provided by Dr Palmer Taylor (Skaggs School of Pharmacy and Pharmaceutical Sciences, University of California, San Diego, CA, U.S.A.). Cloned pfu DNA polymerase and DpnI endonuclease were obtained from Stratagene. Restriction enzyme, alkaline phosphatase [calf intestinal alkaline phospahtase (CIP)] and T4 DNA ligase were purchased from New England Biolabs. All oligonucleotides were synthesized by Eurofins MWG Operon. Vector pCMV–MCS was obtained from Agilent Technologies. The QIAprep Spin Plasmid Miniprep Kit and QIAquick Gel Extraction Kit were obtained from QIAGEN. Chinese hamster ovary (CHO)–S cells and FreeStyle™ CHO Expression Medium were from Invitrogen. [3H](−)-Cocaine (50 Ci/mmol) and [3H]acetylcholine (ACh) were purchased from PerkinElmer Life and Analytical. Butyrylthiocholine (BTC) was obtained from Sigma-Aldrich.
Construction of eukaryotic expression plasmids
Site-directed mutagenesis for obtaining the mCocH cDNA was carried out using the QuikChange method . The cDNA for full-length mBChE or mCocH was constructed in a pCMV–MCS expression plasmid by using restriction enzyme EcoRI to digest the original vector and cDNA. Before the ligation, alkaline phosphatase (CIP) was used to dephosphorylate the 5′-end of vector. Gel-purified cDNA was ligated with pCMV–MCS vector using T4 DNA ligase. Plasmids encoding hBChE and hCocH were obtained as previously described .
Protein expression and purification
All proteins (mBChE/hBChE and mCocH/hCocH) were expressed in CHO–S cells separately. Cells were incubated at 37°C in a humidified atmosphere containing 8% CO2 and transfected with plasmids encoding various proteins using TransIT-PRO® transfection kit once cells had grown to a density of ~1.0×106 cells/ml. The culture medium (Gibco® FreeStyle™ CHO expression medium with 8 mM glutamine) was harvested 7 days after transfection. Secreted enzyme in the culture medium was purified by a two-step approach described previously , including ion exchange chromatography using a Q sepharose fast flow (QFF) anion exchanger and affinity chromatography using procainamide–sepharose. Pre-equilibrated procainamide–sepharose was added into protein sample purified by ion exchange chromatography and incubated for 3 h with occasional stirring. After washing the column with 20 mM potassium phosphate, 1 mM EDTA, pH 7.0 until A280 < 0.02, enzyme was eluted by buffer containing 0.3 M NaCl and 0.1 M procainamide–HCl. The eluate was dialysed in phosphate buffer, pH 7.4 by a Millipore centrifugal filter device. The entire purification process was carried out in a cold room at 4°C. Concentration of the active enzyme was determined through active site titration with di-isopropylfluorophosphate (DFP) as described previously . Purified enzymes were stored at 4°C before enzyme activity assays.
Enzyme activity assays
The catalytic activities of enzymes against (−)-cocaine were determined with a radiometric assay based on toluene extraction of [3H](−)-cocaine labelled on its benzene ring. A 150 μl volume of enzyme solution (100 mM phosphate buffer, pH 7.4) was added to 50 μl of [3H](−)-cocaine solution of various concentrations. The reactions were stopped by adding 200 μl of 0.1 M HCl which neutralized the liberated benzoic acid while ensuring a positive charge on the residual (−)-cocaine. [3H]Benzoic acid was extracted by 1 ml of toluene and measured by scintillation counting. The assays to determine catalytic activity with [3H]ACh differed only in that the reaction was stopped with 200 μl of 0.2 M HCl containing 2 M NaCl. To determine catalytic activities of enzymes against BTC, UV-Vis spectrophotometric assays were carried out in a GENios Pro Microplate Reader (TECAN, Research Triangle Park) with XFluor software. A 100 μl volume of enzyme solution was mixed with 50 μl of 25 mM di-thiobisnitrobenzoic acid and 50 μl of BTC at various concentrations. Reaction rates were measured by recording the time-dependent absorption at 450 nm. All measurements were performed at 25°C. Kinetic data were analysed by performing non-linear, least-squares fitting to eqn (1) (which accounts for the potential secondary binding site of the enzyme, i.e. a peripheral anionic binding site around Asp70) [36,37].
In eqn (1), S represents the concentration of the substrate, Vmax=kcat[E] in which [E] is the enzyme concentration, Kss is a binding constant for substrate at the secondary binding site and b is a factor reflecting whether or not there is a substrate activation/inhibition. When b=1, there is no substrate activation or inhibition and the enzymatic reaction follows Michaelis–Menten kinetics. There is substrate activation when b > 1 and substrate inhibition when b < 1. Kinetic data were analysed with Microsoft Excel, coding eqn (1) for non-linear fitting.
The 3D structure of mCocH was modelled based on our previously refined 3D structure [16,38] of hCocH as a template. The hCocH structure was refined through MD simulations and hybrid quantum mechanics/molecular mechanics (QM/MM) calculations [16,37] starting from the X-ray crystal structure  available for hBChE. With the refined hCocH structure as a template, a 3D structure of mCocH was constructed and refined using the Protein Modeling module of Discovery Studio (Version 2.5.5, Accelrys). The amino-acid sequence of mBChE was directly extracted from the PubMed website with the sequence changes necessary to generate the sequence of mCocH. The sequence alignment was generated by using ClustalW with the Blosum scoring function [40,41]. The best alignment was selected according to both the alignment score and the reciprocal positions of the conserved residues between human and mouse proteins, particularly the residues forming the catalytic triad (S198–H438–E325) and the oxyonion hole (G116–G117–A/S199). The sequence identity between mCocH and hCocH reached 80%. The coordinates of the conserved regions were directly transformed from the template structure, whereas the non-equivalent residues were mutated from the template to the corresponding ones of mCocH. The side chains of those non-conserved residues were relaxed during the process of homology modelling in order to remove the possible steric overlap or hindrance with the neighbouring conserved residues. The initial structure of mCocH was subject to energy minization by using the Sander module of the Amber program  with a conjugate gradient energy-minimization method and a non-bonded cut-off of 10 Å (1 Å=0.1 nm). First, the structure of mCocH was solvated in an orthorhombic box of three-site transferrable intermolecular potential (TIP3P) water molecules  with a minimum solute-wall distance of 10 Å. Standard protonation states in a physiological environment (pH ~7.4) were used for all ionizable residues of the proteins and the proton positions were set properly on the Nδ1 atom of histidine residues. Additional Cl− ions were added to the solvent as counter ions to neutralize the system. The final system size was about 94×91×87 Å3, composed of 62489 atoms, including 18555 water molecules. The first 2000 steps of the energy minimization were carried out for the backbone while the side chains were fixed and then the next 60000 steps for the side chains and water molecules. Finally, the system (mCocH) was energy-minimized for 6000 steps for all atoms and a convergence criterion of 0.001 kcal mol−1·Å−1 (1 cal ≡ 4.184 J) was achieved.
Molecular dynamics simulation
Using the homology model of mCocH, we further examined how mCocH binds with (−)-cocaine. First, (−)-cocaine was docked into the binding site, giving a binding mode similar to that for the corresponding hCocH binding with (−)-cocaine through the superposition. The atomic charges for (−)-cocaine were the restrained electrostatic potential (RESP) charges determined and used in our previous studies on hBChE and hCocH interacting with (−)-cocaine [14,16]. MD simulations were carried out on the mCocH–(−)-cocaine-binding complex by using the Sander module of the Amber program. Each system was slowly heated to 300 K by the weak-coupling method  and then equilibrated for 50 ps. During the MD simulations, a 10 Å non-bonded interaction cut-off was used and the non-bonded list was updated every 1000 steps. The particle-mesh Ewald (PME) method  was applied to treat long-range electrostatic interactions. The lengths of covalent bonds involving hydrogen atoms were fixed with the SHAKE algorithm , enabling the use of a 2-fs time step to numerically integrate the equations of motion. Finally, the production MD was kept running for 4 ns with a periodic boundary condition in the NTP (constant temperature and pressure) ensemble at T=300 K with Berendsen temperature coupling and at P=1 atm (=101.325 kPa) with anisotropic molecule-based scaling .
RESULTS AND DISCUSSION
kcat and KM
The kinetic data are depicted in Figures 1–3 and the kinetic parameters obtained are summarized in Table 1. As seen in Table 1, compared with hBChE, mBChE has a smaller kcat value (1.4 compared with 4.1 min−1) and a smaller KM value (1.6 compared with 4.5 μM) against (−)-cocaine. Overall, the catalytic efficiency of mBChE against (−)-cocaine (kcat/KM=8.8×105 min−1M−1) is comparable with that of hBChE (kcat/KM=9.1×105 min−1 M−1). Concerning the effects of the mutations, hCocH has a ~2000-fold improved catalytic efficiency (kcat/KM) against (−)-cocaine compared with hBChE. From that standpoint, one might expect that mCocH would also have considerably greater catalytic efficiency against (−)-cocaine than mBChE. In fact, as seen in Table 1, the catalytic rate constant kcat of mCocH against (−)-cocaine is ~180-fold larger than that of mBChE against (−)-cocaine, but the KM of mCocH against (−)-cocaine is also larger than that of mBChE against (−)-cocaine (~22-fold). So, the improvement in kcat is compromised by the significant increase in KM, resulting in only ~8-fold improved catalytic efficiency over mBChE (kcat/KM=7.1×106 min−1 M−1). As a result, compared with hCocH, mCocH has ~250-fold lower catalytic efficiency against (−)-cocaine.
|Enzyme .||Substrate .||kcat (min−1) .||KM (μM) .||kcat/KM (min−1·M−1) .||RCE* .||Kss (μM) .||b .|
|Enzyme .||Substrate .||kcat (min−1) .||KM (μM) .||kcat/KM (min−1·M−1) .||RCE* .||Kss (μM) .||b .|
RCE refers to the relative catalytic efficiency (kcat/KM), i.e. the ratio of the kcat/KM value of a mutant (mCocH or hCocH) to that of the corresponding wild-type enzyme (mBChE or hBChE) against the same substrate.
According to the kinetic parameters in Table 1, against substrate ACh, mBChE has a slightly smaller kcat value (38400 compared with 61200 min−1) and a slightly larger KM value (400 compared with 148 μM) compared with hBChE. Therefore, the catalytic efficiency (kcat/KM) of mBChE against ACh is ~4-fold lower than that of hBChE. Concerning the mutational effects on hydrolysis of ACh, both mCocH and hCocH exhibit catalytic efficiencies only slightly lower than those of the wild-type enzymes (mBChE and hBChE). In other words, the mutations caused no substantial effect.
Against substrate BTC, mBChE has a slightly larger kcat value than hBChE (35600 compared with 29500 min−1) but a significantly larger KM value (72 compared with 17 μM). Overall, the catalytic efficiency (kcat/KM) of mBChE against BTC is ~3-fold lower than that of hBChE. Concerning the mutational effects on enzyme activity against BTC, the catalytic efficiency of mCocH is only slightly lower than that of mBChE, whereas the catalytic efficiency of hCocH is only slightly higher than that of hBChE. Overall changes in the catalytic efficiency against BTC are probably not physiologically significant in either of the mutated enzymes.
BChE has a peripheral anionic-binding site around Asp70, similar to acetylcholinesterase (AChE) . For this reason, when substrate is abundant, an additional molecule can bind during the catalytic reaction process. The binding affinity of this ‘side reaction’ is reflected by Kss in eqn (1) and Table 1. Binding an additional substrate molecule at the peripheral anionic-binding site may either increase catalytic activity (substrate activation, reflected by b > 1) or decrease catalytic activity (substrate inhibition, b < 1). It has long been established that BChE exhibits substrate activation with ACh, whereas AChE exhibits substrate inhibition .
The data in Figures 1–3 reveal that substrate activation is a shared feature of both wild-type enzymes (mBChE and hBChE) with each of our three tested substrates (b=1.79–3.36). This means that the additional substrate molecule at the peripheral anionic-binding site can stabilize the transition state (TS) for the rate-determining step more favourably than the corresponding reactant or intermediate associated with the TS. The result is to decrease the activation free energy and facilitate the reaction. In contrast, both mutant enzymes exhibited substrate inhibition with all three substrates (b=0.19–0.88). This behaviour implies that an additional substrate molecule at the peripheral anionic-binding site stabilizes the TS for the rate-determining step less favourably than the corresponding reactant or intermediate associated with the TS. The mutations have converted substrate activation into substrate inhibition, with increased activation energy and slower reaction. This remarkable change may involve a shift in the rate-determining step of the enzymatic reaction. It seems reasonable that an additional substrate molecule binding to the peripheral anionic site may decrease the activation free energy for certain steps while increasing activation free energy for other reaction steps. For example, it has been known that the rate-determining step of hCocH-catalysed hydrolysis of (−)-cocaine occurs at the acylation stage of the chemical process , whereas the rate-determining step of hBChE-catalysed (−)-cocaine hydrolysis is formation of the pre-reactive enzyme–substrate complex [9,49]. Further experiments are required to determine the precise steps involved.
Kinetic data obtained
in vitro for (−)-cocaine hydrolysis catalysed by mCocH, mBChE, hCocH and hBChE
Kinetic data obtained
in vitro for ACh hydrolysis catalysed by mCocH, mBChE, hCocH and hBChE
Kinetic data obtained
in vitro for BTC hydrolysis catalysed by mCocH, mBChE, hCocH and hBChE
Insights from molecular modelling
To understand why mCocH has much lower catalytic efficiency against (−)-cocaine compared with hCocH, we modelled the 3D structure of mCocH binding with (−)-cocaine for comparison with the corresponding hCocH–(−)-cocaine binding. Depicted in Figure 4 are the aligned sequences of mCocH and hCocH, showing that the overall sequence identity between these two enzymes is as high as 80%. As shown in Figure 4, mCocH and hCocH share the same residues for the catalytic triad that reacts with (−)-cocaine (Ser198, His438 and Glu325) and the same oxyanion hole residues (Gly116, Gly117 and Ser199) that form hydrogen bonds with the carbonyl oxygen atom on the benzoyl group of (−)-cocaine. Another common feature of cocaine binding with hCocH and mCocH is that the cationic head of (−)-cocaine has a similar cation–π interaction with the side chain of Trp82.
Sequence alignment between mCocH and hCocH
The main difference between hCocH and mCocH is that the cationic head of (−)-cocaine interacts more favourably with the protein environment including side chains of Phe73 and Trp328 in hCocH, compared with the corresponding interactions in mCocH. This appears to be due to a difference in the detailed shape of the binding pockets in the two enzymes. For example, residue #72 is alanine in mCocH and serine in hCocH. In hCocH, the hydroxy group of Ser72 side chain forms a strong hydrogen bond with an oxygen atom in the carboxylate moiety of the Asp70 side chain (Figures 5C and 5D). This hydrogen bond apparently influences the orientation of the aromatic ring in Phe73 such that the cationic head of (−)-cocaine aligns nearly parallel to the vector normal to the plane of the aromatic ring of Phe73 side chain. As a result, the MD-simulated average distance between the positively charged N atom of (−)-cocaine and the centre of the hCocH Phe73 side chain aromatic ring was 7.06 Å and the MD-simulated average distance between the positively charged N atom of (−)-cocaine and the centre of aromatic ring of Trp328 side chain was 5.99 Å.
In mCocH, with no hydrogen bond between the side chains of Ala72 and Asp70, the side chain of altered residue #72 is farther away from Asp70 than it is in hCocH. This causes the orientation of the aromatic ring of the Phe73 side chain in mCocH to differ substantially from that in hCocH. Due to this alteration, the hydrogen atoms on the aromatic ring of Phe73 side chain in the initial model of mCocH–(−)-cocaine-binding structure seemed too close to the (−)-cocaine atoms. After further simulation, the Phe73 side chain pushed away from the (−)-cocaine atoms and the MD-simulated average distance between the positively charged N atom of (−)-cocaine and the centre of aromatic ring of Phe73 side chain grew to 8.58 Å (1.52 Å longer than that in hCocH). The difference in residue #72 also indirectly affected the interaction of (−)-cocaine with Trp328. The MD-simulated average distance between the positively charged N atom of (−)-cocaine and the centre of the Trp328 side-chain aromatic ring was 6.90 Å in mCocH (0.91 Å longer than that in hCocH). Due to the less favourable interactions of the cationic head of (−)-cocaine with Phe73 and Trp328 side chains in mCocH, the overall binding of (−)-cocaine with mCocH can be expected to be weaker, which is consistent with the experimental observation that, compared with hCocH, mCocH has a significantly larger KM value and a significantly smaller kcat value against (−)-cocaine.
Kinetic analysis reveals that the catalytic efficiencies (kcat/KM) of mBChE against ACh, BTC and (−)-cocaine resemble those of hBChE. After comparable substitutions at five homologous sites in the catalytic gorge, the corresponding mutant forms, mCocH and hCocH, both retain similar activities against ACh and BTC and both show enhanced hydrolysis of (−)-cocaine. However, the magnitude of enhancement differs radically between the two enzymes: ~8-fold with mCocH and ~2000-fold with hCocH, leaving the mouse protein ~250-fold less efficient with (−)-cocaine than its human counterpart. A second surprise was that ACh, BTC and (−)-cocaine all showed substrate activation in wild-type mouse and human BChE, but uniformly caused substrate inhibition in both of the mutated enzymes. That result implies that the rate-determining step of the reactions in mCocH and hCocH may differ from that in mBChE and hBChE. These unexpected outcomes posed an interesting challenge to rational-, structure- and mechanism-based enzyme mutation. However, homology modelling and MD simulations shed light on the underlying causes. In other words, the observed behaviour was consistent with the enzyme–(−)-cocaine binding structures obtained from molecular modelling.
Xiabin Chen, Liyi Geng, Liu Xue, Shurong Hou and Xirong Zheng carried out the in vitro experimental tests. Xiaoqin Huang performed the computational modelling. Stephen Brimijoin, Fang Zheng and Chang-Guo Zhan re-analysed the data. Xiabin Chen, Stephen Brimijoin and Fang Zheng contributed to the manuscript preparation. Chang-Guo Zhan finalized the paper.
Chinese hamster ovary
calf intestinal alkaline phospahtase
human-butyrylcholinesterase-based cocaine hydrolase
mouse butyrylcholinesterase-based cocaine hydrolase
We are grateful to Dr Palmer Taylor (Skaggs School of Pharmacy and Pharmaceutical Sciences, University of California, San Diego, California, U.S.A.) for providing the cDNA encoding mBChE. The authors also acknowledge the Computer Center at the University of Kentucky for supercomputing time on a Dell X-series Cluster with 384 nodes or 4768 processors.
This work was supported by the National Institute of Health [grant numbers R01 DA035552, R01 DA032910, R01 DA013930, R01 DA025100 and DP1 DA031340].