Long-acting cocaine hydrolase for addiction therapy Xiabin Chena,b, Liu Xueb, Shurong Houb, Zhenyu Jina,b, Ting Zhanga,b, Fang Zhenga,b,1, and Chang-Guo Zhana,b,1 a Molecular Modeling and Biopharmaceutical Center, College of Pharmacy, University of Kentucky, Lexington, KY 40536; and bDepartment of Pharmaceutical Sciences, College of Pharmacy, University of Kentucky, Lexington, KY 40536

Edited by Joanna S. Fowler, Brookhaven National Laboratory, Upton, NY, and approved November 30, 2015 (received for review September 4, 2015)

Cocaine abuse is a world-wide public health and social problem without a US Food and Drug Administration-approved medication. An ideal anticocaine medication would accelerate cocaine metabolism, producing biologically inactive metabolites by administration of an efficient cocaine-specific exogenous enzyme. Our recent studies have led to the discovery of the desirable, highly efficient cocaine hydrolases (CocHs) that can efficiently detoxify and inactivate cocaine without affecting normal functions of the CNS. Preclinical and clinical data have demonstrated that these CocHs are safe for use in humans and are effective for accelerating cocaine metabolism. However, the actual therapeutic use of a CocH in cocaine addiction treatment is limited by its short biological half-life (e.g., 8 h or shorter in rats). Here we demonstrate a novel CocH form, a catalytic antibody analog, which is a fragment crystallizable (Fc)-fused CocH dimer (CocH-Fc) constructed by using CocH to replace the Fab region of human IgG1. The CocH-Fc not only has a high catalytic efficiency against cocaine but also, like an antibody, has a considerably longer biological halflife (e.g., ∼107 h in rats). A single dose of CocH-Fc was able to accelerate cocaine metabolism in rats even after 20 d and thus block cocaine-induced hyperactivity and toxicity for a long period. Given the general observation that the biological half-life of a protein drug is significantly longer in humans than in rodents, the CocH-Fc reported in this study could allow dosing once every 2–4 wk, or longer, for treatment of cocaine addiction in humans. enzyme therapy

| cocaine addiction | drug abuse | protein engineering

A

s is well known, cocaine is one of the most reinforcing abused drugs, stimulating the reward pathway of the brain and teaching the user to take it again (1–3). Despite decades of effort, the classical pharmacodynamic approach of using small molecules to block or counteract the drug’s neuropharmacological actions has not proven successful for cocaine, because it would be extremely difficult to antagonize cocaine’s physiological effects without affecting normal functions of the CNS (4). In principle, pharmacological treatment for a drug of abuse can be pharmacodynamic or pharmacokinetic (5). Most current medications for other drugs of abuse use the classical pharmacodynamics approach, using small molecules to block or counteract the drug’s neuropharmacological actions at one or more neuronal binding sites. The inherent difficulties of antagonizing cocaine in the CNS led to the development of protein-based pharmacokinetic approaches with biologics such as monoclonal antibodies, vaccines that produce antibodies in the body, and enzymes (4, 6). The pharmacokinetic approach with an efficient enzyme is recognized as the most promising treatment strategy for cocaine overdose and addiction (4, 7–9). Unlike the stoichiometric binding of an antibody with drug, one enzyme molecule can degrade multiple drug molecules, depending on the turnover number (catalytic rate constant, kcat) and Michaelis–Menten constant (Km). In humans, the principal metabolic enzyme of cocaine is a plasma enzyme known as butyrylcholinesterase (BChE), producing biologically inactive metabolites. Unfortunately, the catalytic efficiency (kcat/Km) of wild-type BChE against (-)-cocaine (which is the naturally occurring enantiomer of cocaine) is too low (kcat = 4.1/min and Km = 4.5 μM) (10) to be effective for cocaine metabolism. A mutant of human BChE with a considerably improved catalytic efficiency against (–)-cocaine is greatly desired. Through structure- and mechanism-based computational modeling and simulation, we have successfully designed and identified human 422–427 | PNAS | January 12, 2016 | vol. 113 | no. 2

BChE mutants, recognized as true cocaine hydrolases (CocHs) in the literature (9) when they have at least 1,000-fold improved catalytic efficiency against (-)-cocaine compared with wild-type human BChE (11–14). The first of our designed CocHs, known as “CocH1” (the A199S/S287G/A328W/Y332G mutant of human BChE) (11, 15), truncated after amino acid 529, was fused with human serum albumin (HSA) to prolong the biological t1/2 (9). This HSAfused BChE mutant is also known as “Albu-CocH,” “Albu-CocH1,” “AlbuBChE,” or “TV-1380” (Fig. 1B) in the literature (7–9, 16). TV-1380 has been proven safe and effective for use in animals and humans (7, 8), but its actual therapeutic value for cocaine addiction treatment is still limited by the moderate biological t1/2, which is ∼8 h in rats (9) and 43–77 h in humans (7). [In general, the biological t1/2 of a therapeutic protein is significantly longer in humans than in rats (9).] A biological t1/2 of 43–77 h in humans might be adequate for a twice-weekly therapy, depending on the dose of the enzyme used. In addition, our more recently designed and identified CocHs (12–14) are significantly more active than TV-1380 against (-)-cocaine. Further engineering a more active CocH with a biological t1/2 longer than that of TV-1380 is highly desired. Our current design strategy for enzyme engineering is based on the observation that an antibody such as human IgG has a very long biological t1/2, because the fragment crystallizable (Fc) region of IgG can bind with the neonatal Fc receptor (FcRn) in the acidic environment of the endosome and later is transported to the cell surface where, upon exposure to a neutral pH, IgG is released back into the main bloodstream (17). In comparison, a recombinant enzyme such as BChE usually has a very short biological t1/2, and an antibody usually has no catalytic activity. Using a stable analog of the transition state for cocaine hydrolysis, Landry et al. (1) successfully generated the first anticocaine catalytic antibody (CAb). Further effort generated the most active anticocaine CAb (Fig. 1C), known as “15A10” (kcat = 2.3/min and Km = 220 μM) (18, 19), whose catalytic activity Significance It is essential for a truly effective addiction medication to block the drug’s physiological effects effectively without affecting normal functions of the brain and other critical organs such as the heart and while still preventing relapse during abstinence. Most popularly used pharmacological approaches to addiction treatment, including all currently available addiction therapies, either affect normal functions of brain receptors/transporters or are unable to prevent relapse. The long-acting enzyme approach may provide a novel, truly promising therapy capable of effectively blocking the physiological and toxic effects of cocaine without affecting normal functions of the brain and other critical organs and prevent relapse during abstinence. New insights obtained in this study also may be valuable in guiding development of other therapeutic proteins. Author contributions: F.Z. and C.-G.Z. designed research; X.C., L.X., S.H., Z.J., and T.Z. performed research; X.C., T.Z., F.Z., and C.-G.Z. analyzed data; and X.C., F.Z., and C.-G.Z. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. 1

To whom correspondence may be addressed. Email: [email protected] or fzhen2@email. uky.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1517713113/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1517713113

against (-)-cocaine is still significantly lower than that of wild-type BChE. It has been difficult to improve the CAb activity further because a CAb, unlike an enzyme, can help stabilize the transition state only for nonenzymatic cocaine hydrolysis; a mechanistic study (20) indicates there is no formation of a covalent bond or breaking between the substrate and CAb during the reaction process. We aimed to design a long-acting CocH form that has both the long biological t1/2 of an antibody and the high catalytic activity of a CocH against (-)-cocaine. For this purpose, starting from human IgG1 (a dimer), which has both the Fc region (constant) and Fab region (variable), as seen in Fig. 1C, we may replace the Fab region by a CocH for each subunit of the dimeric IgG1 to construct a catalytic antibody analog. Technically, the C terminus of CocH3 [the A199S/F227A/S287G/A328W/Y332G mutant (12) of human BChE, full-length or truncated after amino acid 529 to delete the tetramerization domain] was fused with the N terminus of the hinge region linked with Fc. CocH3 has a significantly higher catalytic efficiency against (-)-cocaine (12) than does CocH1 or TV-1380. The obtained dimeric CocH3-Fc fusion enzyme depicted in Fig. 1D is highly efficient for cocaine hydrolysis. Obviously, the CocH3-Fc is different from but is structurally similar to antibody IgG1, containing Fc that can bind with FcRn in the acidic environment to prolong the biological t1/2. Further possible mutations on the Fc region of CocH3-Fc also were examined so that the CocH3-Fc enzyme would have the longest possible biological t1/2. Various CocH3-Fc forms, including Fc mutants, were proposed, prepared, and tested for their actual in vitro and in vivo profiles. In addition, Albu-CocH1 or TV-1380 also was prepared and tested for comparison. As shown below, the optimized CocH3-Fc indeed has both a significantly higher catalytic activity against cocaine and a much longer biological t1/2 than TV-1380. Results Optimization of the CocH3-Fc Entity. The key in vivo data obtained are depicted in Figs. 2–4. All pharmacokinetic (PK) data depicted in Fig. 2 were based on i.v. injection of the enzymes in rats. The PK data reveal that a CocH3-Fc form with the truncated CocH3 (without the tetramerization domain) has a much longer biological t1/2 (up to ∼107 h) than that (∼18 h) of the corresponding CocH3-Fc containing the full-length CocH3, denoted as “CocH3(FL)” in Fig. 2A, with a given Fc region. Notably, the only difference between CocH3(FL) and CocH3 is that CocH3(FL) has the extra 45 amino acid residues, 530–574, an α-helix (21), on the C terminus. Chen et al.

Dose Dependence of the Enzyme Activity in Comparison with TV-1380.

To characterize further the most promising CocH3-Fc form, CocH3-Fc(M3), we developed a stable cell line using a lentiviral vector (25) for large-scale production of CocH3-Fc(M3). The purified CocH3-Fc(M3) is indeed a dimer, as expected (SI Appendix, Fig. S1). The in vitro kinetic parameters of the fused CocHs are essentially the same as the corresponding unfused CocHs (see Fig. 1 B and D and SI Appendix, Fig. S2 for the determined kcat and Km values). As shown in Fig. 2C, the biological t1/2 of CocH3-Fc(M3) injected i.v. in rats is independent of the dose, and therefore the CocH3-Fc(M3) concentration at a given time point is linearly proportional to the dose. Fig. 2D shows the time-dependent maximum CocH activity against cocaine: Vmax ≡ kcat[E] where [E] represents the concentration of the enzyme [CocH3-Fc(M3) or Albu-CocH1] in rat plasma. Vmax = 1 U/L means that the enzyme is capable of hydrolyzing up to 1 μmol cocaine in 1 L of plasma per minute (i.e., 1 μM/min). The data in Fig. 2D indicate that CocH3-Fc(M3) is far superior to Albu-CocH1 (TV-1380). In particular, after injection of 5 mg/kg Albu-CocH1 (TV-1380), the rat plasma had a Vmax ≥ 50 U/L within ∼48 h. In comparison, the Vmax was 50 U/L at ∼100 h after the injection of 0.2 mg/kg CocH3-Fc(M3) and at ∼275 h after the injection of 0.6 mg/kg CocH3-Fc(M3). Even a tiny dose (0.06 mg/kg) of CocH3-Fc(M3) produced a larger Vmax value in plasma than produced by 5 mg/kg Albu-CocH1 (TV1380) starting at ∼80 h after the injection of CocH [CocH3-Fc (M3) or Albu-CocH1]; 0.2 mg/kg CocH3-Fc(M3) generated a larger Vmax value in the plasma than generated by 5 mg/kg AlbuCocH1 (TV-1380) starting at ∼38 h after the injection of CocH; and 0.6 mg/kg CocH3-Fc(M3) led to a larger Vmax value in the plasma than generated by 5 mg/kg Albu-CocH1 (TV-1380) starting at ∼18 h after the CocH injection. PNAS | January 12, 2016 | vol. 113 | no. 2 | 423

NEUROSCIENCE

Fig. 1. Protein structures and their catalytic parameters for cocaine hydrolysis. (A) Human BChE. (B) Albu-CocH1 (or TV-1380). (C) CAb 15A10. (D) CocH3-Fc.

Compared with the CocH3-Fc structure, CocH3(FL)-Fc has 45 extra amino acid residues between CocH3 and Fc. These 45 residues increase the length of the linker between CocH3 and Fc; thus, as shown in SI Appendix, Fig. S3, the CocH3 region is far away from the Fc region in CocH3(FL)-Fc. The longer linker may be proteolyzed more easily because it is flexible; therefore CocH3(FL)Fc has a shorter biological t1/2 than CocH3-Fc. The CocH3-Fc forms discussed below are those with the truncated CocH3 region. To optimize the Fc region of the CocH3-Fc construct, we tested the use of wild-type Fc, denoted “Fc(WT),” and a triple mutant (i.e., A1V/D142E/L144M), denoted “Fc(M3).” In addition, we also tested the use of known Fc mutants [A1Q/C6S/C12S/C15S/P24S mutant Fc (22) and A1V/E58Q/E69Q/E80Q/D98N/N101D/D142E/ L144M mutant Fc (23), for convenience here denoted “Fc(M1)” and “Fc(M2),” respectively; see Methods for details], that have been used to prolong significantly the half-lives of other types of therapeutic proteins, i.e., abatacept (22) and alefacept (23). As seen in Fig. 2B, within the CocH3-Fc forms examined, CocH3-Fc(M3) showed the longest biological t1/2 (∼107 h) in rats. The A1V/D142E/ L144M mutations on the Fc region prolonged the biological t1/2 of this CocH3-Fc by ∼21 h. The other two CocH3-Fc forms, i.e., CocH3-Fc(M1) and CocH3-Fc(M2), actually shortened the biological t1/2 of CocH3-Fc compared with Fc(WT), although Fc(M1) and Fc(M2) successfully prolonged the biological half-lives of other fusion proteins (22, 23) compared with Fc(WT). Further, we note that the enzyme fusion with Fc (or Fc mutant) does not necessarily prolong the biological t1/2 of an enzyme. In fact, we also tested fusing a thermally stable mutant of bacterial cocaine esterase (CocE) (24) with Fc(M3) and found no significant improvement in the biological t1/2 (within the measurement errors), with t1/2 still being only within a few minutes (SI Appendix, Fig. S4). The unusually large number of negative charges (−34e) possessed by a bacterial CocE protein could significantly affect the interaction of Fc in the CocE-Fc fusion protein with FcRn and, potentially, with other proteins in the body.

Fig. 2. Time-dependent normalized CocH activity (A–C) or Vmax (U/L) (D) of various CocH3-Fc forms and Albu-CocH1 (TV-1380). For convenience, “CocH3(FL)” represents full-length CocH3, and “CocH3” (without “FL”) refers to truncated CocH3 (amino acid residues 1–529). Rats (n = 3 for each dose condition) were injected i.v. with the enzyme [CocH3-Fc(M3) or Albu-CocH1] at a dose of 0.06 mg/kg body weight unless explicitly indicated otherwise. Blood samples were collected at various time points, and the separated plasma samples were analyzed by a sensitive radiometric assay using [3H](-)-cocaine.

Cocaine Clearance Is Accelerated by CocH3-Fc(M3). In a further in vivo test, rats were injected with a single dose of CocH3-Fc(M3) (0.2 mg/kg i.v. on day 0), followed by i.v. injection of cocaine (5 mg/kg) on days 8, 11, 14, and 20. After each cocaine injection, blood samples were collected at 2, 5, 10, 15, 30, and 60 min and were analyzed for the concentrations of cocaine and benzoic acid (cocaine metabolite). The control curves in Fig. 3 reflect the overall effects of all possible cocaine-elimination pathways (25–27). In the control rats, the average concentration of cocaine at the first time point (2 min) was ∼7.4 μM, whereas the average concentration of benzoic acid was ∼0.2 μM. In the presence of CocH3-Fc(M3) on day 8 after CocH3-Fc(M3) injection, the average concentration of cocaine at ∼2 min in the blood sample was below 1 μM (∼0.9 μM) (SI Appendix, Fig. S3A), whereas the average concentration of benzoic acid at the first time point (2 min) was ∼11 μM (SI Appendix, Fig. S3B). Thus CocH3-Fc(M3) hydrolyzed nearly all the cocaine molecules within ∼2 min after the cocaine injection on day 8. According to the data in Fig. 3B, the overall enzyme activity in rats decreased gradually from day 8 to days 11, 14, and 20, as expected. However, even at such a low dose (0.2 mg/kg), the CocH3-Fc(M3) activity in rats was still significant after 20 d. It should be noted that CocH3-Fc(M3) is constructed from human protein sequences with mutations on only a few residues. Because the human protein sequences associated with CocH3-Fc(M3) are different from the corresponding rodent protein sequences, we did not try to examine the potential antigenicity/immunogenicity of CocH3Fc(M3) in animals in this study. However, according to clinical data reported for TV-1380 (7), one can reasonably assume that CocH3-Fc (M3) would induce no, a weak, or a transient immune response in humans. The effectiveness of CocH3-Fc(M3) in these animals for 20 d supports this assumption. Effectiveness of CocH3-Fc(M3) in Blocking the Physiological Effects of Cocaine. In animal behavior studies, mice were injected i.v. with a

single dose (2, 0.2, or 0 mg/kg) of CocH3-Fc(M3) on day 0, followed by multiple sessions of i.p. injection of cocaine (15 mg/kg) with hyperactivity (measured by increased horizontal distance traveled) 424 | www.pnas.org/cgi/doi/10.1073/pnas.1517713113

recording after 1 and 24 h (day 1) and then every other day (days 3, 5, 7, 9, 11, 13, and 15). For each session, the hyperactivity recording began 30 min before cocaine injection. Data recorded for these sessions (Fig. 4 A–H) show that CocH3-Fc(M3) completely blocked cocaine-induced hyperactivity 1 d (24 h) after the 0.2 mg/kg CocH3Fc(M3) injection and still significantly attenuated cocaine-induced hyperactivity after 11 d. At a dose of 2 mg/kg, CocH3-Fc(M3) completely blocked cocaine-induced hyperactivity within 9 d and still significantly attenuated cocaine-induced hyperactivity after 15 d. Effectiveness of CocH3-Fc(M3) in Blocking the Toxic Effects of Cocaine.

It is well known that cocaine is lethal at high doses [LD50 = 14 mg/kg i.v. (28) or 95.1 mg/kg i.p. (29) in mice]. To examine how long CocH3-Fc(M3) can protect mice from the acute toxicity of a lethal dose of cocaine, mice were injected with a single dose of CocH3-Fc(M3) (2 mg/kg, i.v.) on day 0, followed by a lethal dose of cocaine (100 mg/kg, i.p.) every day (starting on day 4) until a convulsion occurred (the end point of the toxicity test for a given mouse). All mice in the control group [without CocH3-Fc(M3) injection] had a convulsion within minutes after the first cocaine challenge, and 60% of the mice died soon after the convulsion. For the test group receiving 2 mg/kg CocH3-Fc(M3), a convulsion occurred only during the final cocaine challenge, after 10.2 d in average; a convulsion did not occur during the daily cocaine challenge for 9.2 d on average. The average protection time, tp, provided by 2 mg/kg CocH3-Fc(M3) is 9.7 ± 1.7 d. Thus, injection of 2 mg/kg CocH3-Fc(M3) can protect mice effectively from the acute toxicity of the lethal dose of cocaine (100 mg/kg i.p.) for nearly 10 d. Discussion We have successfully designed a novel, long-acting cocainemetabolizing enzyme form, CocH3-Fc. The optimized CocH3-Fc entity, CocH3-Fc(M3), indeed has a long biological t1/2, like an antibody, but with an unprecedentedly high catalytic efficiency compared with all CAbs known to date. Compared with the most active anticocaine CAb, 15A10 (kcat = 2.3/min, Km = 220 μM, and Chen et al.

an Fc mutant) of human IgG1 may not necessarily prolong the biological t1/2 of the enzyme, because the Fc fusion strategy did not prolong the biological t1/2 of the bacterial CocE at all. To develop a truly long-acting enzyme form fused with Fc, one must optimize structurally both the enzyme and the Fc portions of the Fc-fused enzyme. Furthermore, even if the protein fusion with Fc really can prolong the biological t1/2 of a therapeutic protein, the data obtained in this investigation demonstrate that certain amino acid mutations on the Fc portion that can prolong the biological t1/2 of a given Fc-fused therapeutic protein may not necessarily prolong the biological t1/2 of another Fc-fused therapeutic protein. Methods

Fig. 3. Cocaine clearance accelerated by CocH3-Fc(M3). Saline or 0.2 mg/kg CocH3-Fc(M3) was injected i.v. in rats (n = 4), followed by i.v. injection of 5 mg/kg cocaine after 8, 11, 14, and 20 d. Blood samples were collected 2, 5, 10, 15, 30, and 60 min after each i.v. injection of 5 mg/kg cocaine and were analyzed for the concentrations of (A) cocaine and (B) benzoic acid (a metabolite of cocaine).

t1/2 < 24 h in rats) (30), CocH3-Fc(M3) has an even significantly longer biological t1/2 (kcat = 5,700/min, Km = 2.8 μM, and t1/2 = ∼107 h in rats) and ∼200,000-fold improved catalytic efficiency (kcat/Km) against cocaine. The long-acting CocH3-Fc(M3) is highly efficient in blocking the physiological and toxic effects of cocaine for a long period. Further, it is well known that a protein drug generally has a significantly longer biological t1/2 in humans than in rats [e.g., the, Fc(M1)-fused human cytotoxic lymphocyte-associated antigen abatacept has a t1/2 of 3–6 d in rats and 12–23 d in humans (22)]. Thus, one may reasonably expect that an appropriately chosen dose of CocH3-Fc(M3) could protect humans effectively against both relapse and overdose for 2–4 wk or longer during treatment for cocaine addiction. The general strategy of developing a long-acting drug-metabolizing enzyme for treatment of cocaine abuse also may be useful in the development of long-acting forms of other therapeutic enzymes. On the other hand, one must keep in mind that the simple fusion of an enzyme (such as bacterial CocE, mentioned above) with Fc (or with Chen et al.

Construction of Expression Plasmids. With specific base-pair alterations, sitedirected mutagenesis was performed to obtain the cDNAs for Fc(WT), Fc(M1) (i.e., the A1Q/C6S/C12S/C15S/P24S mutant Fc), Fc(M2) (i.e., the A1V/E58Q/ E69Q/E80Q/D98N/N101D/D142E/L144M mutant Fc), and Fc(M3) of human IgG1 using pFUSE-hIgG1-Fc2 plasmid as the template (31). Five expression plasmids, pCMV-CocH3-Fc(WT), pCMV-CocH3-Fc(M1), pCMV-CocH3-Fc(M2), pCMV-CocH3-Fc(M3), and pCMV-CocH3(FL)-Fc(M3), were constructed. Each plasmid contains a sequence encoding the native BChE signal peptide followed by CocH3 (truncated or full length) linked to the N-terminal of Fc (WT or mutant). To fuse CocH3 to Fc, overlap extension PCR with Phusion DNA polymerase was used. The four primers used for each plasmid construction are listed in SI Appendix, Table S1. Primers 1 and 2 were used to amplify CocH3 with the N-terminal signal peptide using pRc/CMV-CocH3 as template; primers 3 and 4 were used to amplify Fc using corresponding Fc cDNA as template. Once two DNA fragments were obtained, they were fused together by using another PCR with primers for the far ends (primers 1 and 4). Then the PCR product was digested with restriction endonucleases Hind III and Bgl II and was ligated to the pCMV-MCS expression vector using T4 DNA ligase. The same method also was used to construct plasmids pCSC-CocH3-Fc (M3) and pCSC-Albu-CocH1 for large-scale protein production. Small-Scale Protein Expression and Purification. CHO-S cells, incubated at 37 °C in a humidified atmosphere containing 8% CO2, were transfected with plasmids encoding the proteins using the TransIT-PRO Transfection Kit once cells had grown to a density of 1.0 × 106 cells/mL. The culture medium (FreeStyle CHO Expression Medium with 8 mM L-glutamine) was harvested 7 d after transfection. Enzyme secreted in the culture medium was purified by protein A affinity chromatography. Pre-equilibrated rmp Protein A Sepharose Fast Flow was added into cell-free medium and was incubated overnight with occasional stirring, and then the suspension was packed in a column. The column was washed with 20 mM Tris·HCl (pH 7.4) until an OD280 < 0.02 was achieved; then the protein was eluted by adjustment of the pH and salt concentration. The eluate was dialyzed in storage buffer (50 mM Hepes, 20% sorbitol, 1 M glycine, pH 7.4) by Millipore Centrifugal Filter Units. The entire purification process was carried out in a cold room at 6 °C. The purified protein was stored at −80 °C before enzyme activity tests and in vivo studies.

PNAS | January 12, 2016 | vol. 113 | no. 2 | 425

NEUROSCIENCE

Materials. The cDNAs for CocH1 (the A199S/S287G/A328W/Y332G mutant of human BChE) and CocH3 (the A199S/F227A/S287G/A328W/Y332G mutant of human BChE) containing N-terminal signal were generated in our previous studies (11, 12). pFUSE-hIgG1-Fc2 plasmid was purchased from InvivoGen. The cDNA (clone ID: HsCD00005810) for HSA was obtained from the Dana Farber/Harvard Cancer Center DNA Resource Core. The protein expression vector pCMV-MCS was ordered from Agilent, and pCSC-SP-PW lentiviral vector (plasmid 12335), pMDLg/pRRE (plasmid 12251), pRSV-Rev (plasmid 12253), and pCMV-VSV-G (plasmid 8454) were obtained from Addgene. Phusion DNA polymerase, restriction endonucleases, and T4 DNA ligase were ordered from New England BioLabs. DpnI endonuclease was obtained from Thermo Fisher Scientific. All oligonucleotides were synthesized by Eurofins MWG Operon. CHO-S cells, HEK-293FT cells, FreeStyle CHO expression medium, hypoxanthine/thymidine (HT) supplement, L-glutamine, DMEM, FBS, 4–12% Tris-glycine Mini Protein Gel, and SimpleBlue SafeStain were from Invitrogen. The TransIT-PRO Transfection Kit was obtained from Mirus. Reductionmodified protein (rmp) Protein A Sepharose Fast Flow was purchased from GE Healthcare Life Sciences. Centrifugal filter units were from Millipore. (−)-Cocaine was provided by the National Institute on Drug Abuse Drug Supply Program, and [3H](−)-Cocaine (50 Ci/mmol) was ordered from PerkinElmer. Sprague–Darley rats (200–250 g) and CD-1 mice (25–30 g) were ordered from Harlan. All other chemicals were purchased from Thermo Fisher Scientific or Sigma-Aldrich.

Fig. 4. Effects of CocH3-Fc(M3) on cocaine-induced hyperactivity and toxicity in mice. (A–H) Hyperactivity test (n = 8 for each dose condition): Saline or CocH3-Fc(M3) (enzyme) was injected i.v., followed by multiple sessions of locomotor activity tests with i.p. injection of 15 mg/kg cocaine or saline after 1 h and 24 h (1 d) and then every other day. A–H, respectively, report data from the locomotor activity sessions on days 1, 3, 5, 7, 9, 11, 13, and 15 after the CocH3-Fc (M3) injection. For each locomotor activity session, locomotor activity recording started 30 min before the cocaine injection. (I) Toxicity test (n = 5). A single dose of CocH3-Fc(M3) (2 mg/kg, i.v.) was followed by daily dosing of cocaine (100 mg/kg, i.p.) starting on day 4 until convulsion occurred (the end point of the toxicity test) for a given mouse). For a given mouse, if convulsion occurred during the final cocaine challenge on day m but did not occur during the cocaine challenge on and before day m − 1, the protection time (tp) for this mouse was considered to be between days m − 1 and m: m − 1 < tp > m or tp = m − 1/2 d. represents the average tp value for all mice in the group.

Large-Scale Protein Expression and Purification. A lentivirus-based method described in our previous report (25) was performed to generate a highly efficient stable cell line expressing CocH3-Fc(M3). Briefly, to package the lentivirus particles carrying the gene of CocH3-Fc(M3), lentivirus was produced by cotransfecting pCSC-CocH3-Fc(M3) plasmid with the packaging vectors (pMDLg/pRRE and pRSV-Rev) and envelope plasmid (pCMV-VSV-G) into HEK293FT cells by lipofection. The packaged lentivirus particles were transfected to CHO-S cells. After lentiviral transductions, infected cells were allowed to recover from the infection for 2 d or more and then were transferred to a shake flask for further culture. The obtained stable CHO-S cell pool was kept frozen before being used for large-scale protein production. Large-scale protein production was performed in an agitated bioreactor BioFlo/CelliGen 115 (Eppendorf). Before being transferred into the bioreactor, cells were grown at 37 °C in shake flasks to the designated volume and density. On the day of transfer, cells in shake flasks were centrifuged at a speed of 1,500 × g for 5 min at room temperature, resuspended in fresh culture medium, and transferred into the bioreactor. The CO2/air gas overlay was set so that the pH of the cell-culture medium was maintained at 7.0–8.0. The bioreactor was run in a batch model with a temperature of 32 °C. After 9 d the culture medium was harvested, and the protein was purified. CocH3-Fc(M3) was purified using the aforementioned protein A affinity chromatography, which was performed on an ÄKTA avant 150 system (GE Healthcare Life Sciences). The purified protein was dialyzed in storage buffer and stored at −80 °C. Electrophoresis. Purified CocH3-Fc(M3) was analyzed by SDS/PAGE electrophoresis. Two protein samples, protein in its native state and protein in a denatured state, were loaded in a 4–12% Tris-Glycine Mini Protein Gel. To break up the quaternary protein structure, the protein sample was mixed with SDS-loading buffer and was heated at 95 °C for 10 min in the presence of a reducing agent, DTT. Electrophoresis was run at 100 V for 2 h. The protein gel was stained with SimpleBlue SafeStain. In Vitro Activity Test Against (−)-Cocaine. A radiometric assay, which was used in our previous studies (12, 14), was used to test the catalytic activities of enzymes against (−)-cocaine. Briefly, 150 μL of the enzyme solution (100 mM phosphate buffer, pH 7.4) was added to varying concentrations of 50 μL of [3H]-labeled (−)-cocaine, denoted as “[3H](−)-cocaine” (with 3H labeled on

426 | www.pnas.org/cgi/doi/10.1073/pnas.1517713113

the benzene ring). The reaction was stopped by adding 200 μL 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. All measurements were performed at 25 °C. Substrate concentration-dependent data were analyzed using standard Michaelis–Menten kinetics so that the catalytic parameters (kcat and Km) were determined. Determination of Biological t1/2 in Rats. Rats were used in accordance with the Guide for the Care and Use of Laboratory Animals as adopted and promulgated by the National Institutes of Health (32). The experimental protocols were approved by the IACUC at the University of Kentucky. Rats were injected with purified protein via the tail vein [0.06, 0.2, or 0.6 mg/kg for CocH3-Fc(M3); 0.06 or 5 mg/kg for Albu-CocH1; and 0.06 mg/kg for all other CocH3-Fc protein forms]. Blood samples were taken by needle puncture of the saphenous vein. Approximately 50–100 μL of blood was collected into a heparin-treated capillary tube at various time points after enzyme administration. Collected blood samples were centrifuged for 15 min at a speed of 5,000 × g to separate the plasma, which was kept at 4 °C before analysis. Radiometric assay using 100 μM (−)-cocaine was used to measure the enzyme concentration in plasma. The time-dependent enzyme concentrations ([E]t) were fitted to a well-known double-exponential equation (33) by GraphPad Prism 5: (½Et = Ae−k1 t + Be−k2 t ), which accounts for both the enzyme distribution process (the fast phase, associated with k1) and the elimination process (the slow phase, associated with k2). The t1/2 associated with the enzyme elimination rate constant k2 is known as the elimination t1/2 or biological t1/2. Characterization of Cocaine Clearance Accelerated by CocH3-Fc(M3). Four rats received i.v. saline before the i.v. injection of 5 mg/kg (−)-cocaine, and four other rats received enzyme CocH3-Fc(M3) (0.2 mg/kg, i.v.) on day 0, followed by i.v. administration of 5 mg/kg (−)-cocaine on days 8, 11, 14, and 20 after the CocH3Fc(M3) injection. Blood samples (50–100 μL) were collected from the saphenous vein into a heparin-treated capillary tube 2, 5, 10, 15, 30, and 60 min after (−)-cocaine administration and were mixed immediately with 100 μL of 25 μM paraoxon (which inactivates the enzymes). Blood samples were stored at −80 °C until analysis by HPLC. To assay the (−)-cocaine and benzoic acid (the product of cocaine hydrolysis by the enzyme) concentrations in the blood samples, frozen

Chen et al.

Locomotor Activity Assay. The effects of CocH3-Fc(M3) on cocaine-induced hyperactivity were evaluated using a video-tracking system at the University of Kentucky’s Rodent Behavior Core. Mice received a single dose of CocH3-Fc(M3) (0.2 or 2 mg/kg, i.v.) or saline, followed by multiple doses of cocaine (15 mg/kg, i.v.) or saline after 1 h, 1, 3, 5, 7, 9, 11, 13, and 15 d. Before cocaine or saline administration, mice were allowed to acclimate to the test chambers for 1 h. The total distance traveled during the last 30 min, which was collected in 5-min bins, was used as the basal level. After cocaine or saline administration, mice

1. Landry DW, Zhao K, Yang GX, Glickman M, Georgiadis TM (1993) Antibody-catalyzed degradation of cocaine. Science 259(5103):1899–1901. 2. Milton AL, Everitt BJ (2012) Neuroscience. Wiping drug memories. Science 336(6078): 167–168. 3. Ersche KD, et al. (2012) Abnormal brain structure implicated in stimulant drug addiction. Science 335(6068):601–604. 4. Zheng F, Zhan C-G (2012) Are pharmacokinetic approaches feasible for treatment of cocaine addiction and overdose? Future Med Chem 4(2):125–128. 5. Gorelick DA (2008) Pharmacokinetic approaches to treatment of drug addiction. Expert Rev Clin Pharmacol 1(2):277–290. 6. Skolnick P, White D, Acri JB (2015) Editorial: Emerging targets for stimulant use disorders: Where to invest in an era of constrained resources? CNS Neurol Disord Drug Targets 14(6):691. 7. Cohen-Barak O, et al. (2015) Safety, pharmacokinetics, and pharmacodynamics of TV1380, a novel mutated butyrylcholinesterase treatment for cocaine addiction, after single and multiple intramuscular injections in healthy subjects. J Clin Pharmacol 55(5):573–583. 8. Shram MJ, et al. (2015) Assessment of pharmacokinetic and pharmacodynamic interactions between albumin-fused mutated butyrylcholinesterase and intravenously administered cocaine in recreational cocaine users. J Clin Psychopharmacol 35(4): 396–405. 9. Brimijoin S, et al. (2008) A cocaine hydrolase engineered from human butyrylcholinesterase selectively blocks cocaine toxicity and reinstatement of drug seeking in rats. Neuropsychopharmacology 33(11):2715–2725. 10. Sun H, Pang Y-P, Lockridge O, Brimijoin S (2002) Re-engineering butyrylcholinesterase as a cocaine hydrolase. Mol Pharmacol 62(2):220–224. 11. Pan Y, et al. (2005) Computational redesign of human butyrylcholinesterase for anticocaine medication. Proc Natl Acad Sci USA 102(46):16656–16661. 12. Zheng F, et al. (2008) Most efficient cocaine hydrolase designed by virtual screening of transition states. J Am Chem Soc 130(36):12148–12155. 13. Xue L, et al. (2011) Design, preparation, and characterization of high-activity mutants of human butyrylcholinesterase specific for detoxification of cocaine. Mol Pharmacol 79(2):290–297. 14. Zheng F, et al. (2014) A highly efficient cocaine-detoxifying enzyme obtained by computational design. Nature Commun 5:3457. 15. Yang W, Xue L, Fang L, Chen X, Zhan C-G (2010) Characterization of a high-activity mutant of human butyrylcholinesterase against (-)-cocaine. Chem Biol Interact 187(13):148–152. 16. Willyard C (2015) Pharmacotherapy: Quest for the quitting pill. Nature 522(7557): S53–S55. 17. Roopenian DC, Akilesh S (2007) FcRn: The neonatal Fc receptor comes of age. Nat Rev Immunol 7(9):715–725. 18. Larsen NA, et al. (2004) Crystallographic and biochemical analysis of cocaine-degrading antibody 15A10. Biochemistry 43(25):8067–8076.

Chen et al.

were returned immediately to the test chamber for the remaining 1 h of the session for activity monitoring. The locomotor tests were performed in highdensity, nonporous plastic chambers measuring 50 cm (L) × 50 cm (W) × 38 cm (H) in a light- and sound-attenuating behavioral test enclosure (San Diego Instruments). The cumulative distance traveled was recorded by an EthoVision XT video tracking system (Noldus Information Technology) to represent the locomotor activity. Protection Study in Mice. Cocaine-induced acute toxicity was characterized in this study by the occurrence of a convulsion. A cocaine-induced convulsion was defined as the loss of righting posture for at least 5 s with the simultaneous presence of clonic limb movements (35). Mice received a single dose of CocH3-Fc (M3) (2 mg/kg, i.v.) or saline (i.v.) on day 0, followed by daily dosing of cocaine (100 mg/kg, i.p.) starting on day 4 until the occurrence of a convulsion in a given mouse. Following cocaine administration, mice were placed immediately in containers for observation. The presence or absence of a convulsion was recorded for 60 min following cocaine administration (12). ACKNOWLEDGMENTS. This work was supported in part by NIH Grants UH2 DA041115, R01 DA035552, R01 DA032910, R01 DA013930, and R01 DA025100 and by National Science Foundation Grant CHE-1111761.

19. Yang G, et al. (1996) Anti-cocaine catalytic antibodies: A synthetic approach to improved antibody diversity. J Am Chem Soc 118(25):5881–5890. 20. Pan Y, Gao D, Zhan C-G (2008) Modeling the catalysis of anti-cocaine catalytic antibody: Competing reaction pathways and free energy barriers. J Am Chem Soc 130(15): 5140–5149. 21. Pan Y, Muzyka JL, Zhan C-G (2009) Model of human butyrylcholinesterase tetramer by homology modeling and dynamics simulation. J Phys Chem B 113(18):6543–6552. 22. Maxwell LJ, Singh JA (2010) Abatacept for rheumatoid arthritis: A Cochrane systematic review. J Rheumatol 37(2):234–245. 23. Rigby MR, et al. (2015) Alefacept provides sustained clinical and immunological effects in new-onset type 1 diabetes patients. J Clin Invest 125(8):3285–3296. 24. Fang L, et al. (2014) Rational design, preparation, and characterization of a therapeutic enzyme mutant with improved stability and function for cocaine detoxification. ACS Chem Biol 9(8):1764–1772. 25. Xue L, et al. (2013) Preparation and in vivo characterization of a cocaine hydrolase engineered from human butyrylcholinesterase for metabolizing cocaine. Biochem J 453(3):447–454. 26. Sun H, Shen ML, Pang Y-P, Lockridge O, Brimijoin S (2002) Cocaine metabolism accelerated by a re-engineered human butyrylcholinesterase. J Pharmacol Exp Ther 302(2):710–716. 27. Hou S, Zhan M, Zheng X, Zhan C-G, Zheng F (2014) Kinetic characterization of human butyrylcholinesterase mutants for the hydrolysis of cocaethylene. Biochem J 460(3): 447–457. 28. Smith M, Garner D, Niemann JT (1991) Pharmacologic interventions after an LD50 cocaine insult in a chronically instrumented rat model: Are beta-blockers contraindicated? Ann Emerg Med 20(7):768–771. 29. Bedford JA, Turner CE, Elsohly HN (1982) Comparative lethality of coca and cocaine. Pharmacol Biochem Behav 17(5):1087–1088. 30. Mets B, et al. (1998) A catalytic antibody against cocaine prevents cocaine’s reinforcing and toxic effects in rats. Proc Natl Acad Sci USA 95(17):10176–10181. 31. Braman J, Papworth C, Greener A (1996) Site-directed mutagenesis using doublestranded plasmid DNA templates. Methods Mol Biol 57:31–44. 32. Committee on Care and Use of Laboratory Animals (1996) Guide for the Care and Use of Laboratory Animals (Natl Inst Health, Bethesda), DHHS Publ No (NIH) 85-23. 33. Lockridge O, Schopfer LM, Winger G, Woods JH (2005) Large-scale purification of butyrylcholinesterase from human plasma suitable for injection into monkeys; a potential new therapeutic for protection against cocaine and nerve agent toxicity. J Med Chem Biol Radiol Defense 3:1–20. 34. Mercolini L, et al. (2008) Quantitative analysis of cocaine in human hair by HPLC with fluorescence detection. J Pharm Biomed Anal 48(2):456–461. 35. Ko MC, et al. (2007) Cocaine esterase: Interactions with cocaine and immune responses in mice. J Pharmacol Exp Ther 320(2):926–933.

PNAS | January 12, 2016 | vol. 113 | no. 2 | 427

NEUROSCIENCE

blood samples were thawed on ice for more than 3 h. Then 150 μL mobile phase (76% 0.1% trifluoroacetic acid and 24% acetonitrile) and 50 μL 7% HClO4 were added to each sample. The extraction mixture was vortex mixed for 1 min and then was centrifuged at 4 °C for 15 min at a speed of 13,200 × g. The supernatant was decanted into a 1.5-mL tube, labeled, and stored at –80 °C until analysis by HPLC. The chromatographic analysis was carried out on a Gemini 5-μm C18 column (Phenomenex), using a mobile phase at a flow rate of 1 mL/min. The (−)-cocaine fluorescence was monitored at 315 nm with excitation at 230 nm (25, 34), and benzoic acid absorbance was monitored at 230 nm. The chromatographic system used comprised a Waters 1525 binary HPLC pump, a 717 plus Autosampler, a 2487 dual λ absorbance detector, and a 2475 multi λ fluorescence detector (Waters).