Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The number of clandestine methamphetamine laboratory seizures has increased in the 1990s (U.S. Department of Justice, Drug Enforcement Administration, 1998). Depending on the capabilities of the chemist involved in the synthesis and the procedure(s) used, the methamphetamine on the street can be impure, containing synthetic intermediates and by-products (Verweij, 1989; Fester, 1999). It would be expected that individuals using illicit methamphetamine would be exposed to these impurities, but police, firemen, and even neighbors are unwittingly exposed to these impurities when illegal laboratories contaminate local buildings, water sources, and/or soil due to careless or intentional dumping of chemicals and chemical waste inside and outside of the laboratory (U.S. Department of Justice, Drug Enforcement Administration, 1998). Forensic chemists have characterized the synthetic impurities associated with the various synthetic methods used in the clandestine synthesis of methamphetamine (Verweij, 1989), but there is minimal information concerning the occurrence, pharmacological effects, or potential toxicity of these impurities (Noggle et al., 1985; Ketema et al., 1990; Moore et al., 1995; Moore et al., 1996; Varner et al., 2001).

Due to its ready availability, pseudoephedrine (e.g., Sudafed) is currently a common precursor for the synthesis of illicit methamphetamine (U.S. Department of Justice, Drug Enforcement Administration, 1998). An approach used in the clandestine synthesis of methamphetamine from pseudoephedrine is shown in Fig. 1 and indicates that both diastereomeric -halogenated intermediates are formed. These intermediates can cyclize to form cis- or trans-1,2-dimethyl-3-phenylaziridine (Allen and Kiser, 1987; Lekskulchai et al., 2000). The occurrence of these intermediates in clandestinely synthesized methamphetamine has been documented (Noggle et al., 1986; Allen and Kiser, 1987; Cantrell et al., 1988; Skinner, 1990; Tanaka et al., 1992). The extent to which these impurities are present in illicit methamphetamine has not been determined, however, in one report the chromatogram of a forensic sample showed that similar amounts of methamphetamine and chloroephedrine were present (Noggle et al., 1986). Since amphetamine related compounds are substrates for human CYP2D6 present in the liver (Wu et al., 1997) and central nervous system (Tyndale et al., 1999), it is possible that haloephedrines and their respective aziridines can bind to CYP2D6. The goal of this study is to determine what effect, if any, a haloephedrine and/or its respective aziridine could have on human CYP2D6. (+)- and ()-Chloroephedrine HCl were chosen since they have been shown to be present in clandestinely synthesized methamphetamine, are stable as the HCl salt, and are capable of cyclizing to cis- and trans-1,2-dimethyl-3-phenylaziridine when present as the free base (Taguchi and Kojima, 1959; Noggle et al., 1986; Allen and Kiser, 1987; Lekskulchai et al., 2000). A computational approach using a 3D¹ molecular model of a xenobiotic binding site of a CYP2D6 (Modi et al., 1996) was used initially to evaluate the feasibility of the interaction. The favorable interactions observed in the model warranted in vitro studies with the chloroephedrines using human microsomes. The in vitro studies suggest that ()-chloroephedrine, but not (+)-chloroephedrine, could irreversibly inhibit CYP2D6.

View larger version (16K): [in this window] [in a new window]   Fig. 1.   Reported synthesis of S-methamphetamine from (+)-pseudoephedrine and isomers formed upon protonation of cis- or trans-1,2-dimethyl-3-phenylaziridine.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Molecular Modeling. The human CYP2D6 molecular model was kindly provided by S. Modi (Modi et al., 1996). The energy of binding for the isomers was evaluated using FlexiDock within SYBYL (version 6.7; Tripos Inc., St. Louis, MO). Binding interactions were also evaluated with HINT (Kellogg and Abraham, 2000). The protonated structure of (+)- or ()-chloroephedrine and cis- or trans-1,2-dimethyl-3-phenylaziridine were initially placed in the CYP2D6 binding site and the structure was minimized around the binding region using the Tripos force field within SYBYL to a gradient of 0.05 kcal/mol- Å with the constraint that nitrogen be within 2.8 to 4.0 Å of Asp301 carbonyl, a normal hydrogen bonding distance. A localized annealing minimization of the active site was performed as follows: atoms in amino acids that reside within 3.0 Å of the ligand and the target Asp301, Ser304, and heme cofactor were allowed to freely move; atoms in the adjacent 3 to 6 Å shell were fixed but contributed to the energy evaluation.

Reagents. The (+)- and ()-chloroephedrine HCl were synthesized as previously reported (Lekskulchai et al., 2000) and were 98% a single diastereomer. The S-methamphetamine HCl was a kind gift from Dr. Richard A. Glennon (Virginia Commonwealth University, Richmond, VA). Dextromethorphan (DM), dextrorphan, 3-methoxymorphinan, 7-ethoxyresorufin, and resorufin were obtained from Sigma-Aldrich (St. Louis, MO). Individual human liver microsomes from two subjects were prepared by differential centrifugation using standard procedures and stored at 80°C until thawed prior to using. Protein concentration was determined using the Bicinchoninic Acid Protein Assay kit from Pierce Chemical Co. (Rockford, IL).

CYP1A2, CYP2D6, and CYP3A4/5 Incubations with (+)- or ()-Chloroephedrine. CYP1A2 activity was measured by spectrofluorometric measurement of formation of resorufin from 7-ethoxyresorufin (Burke and Mayer, 1974) in the presence of 50 µM (+)- or ()-chloroephedrine HCl. CYP2D6 and CYP3A4/5 activity could be simultaneously monitored by measuring formation of dextrorphan and 3-methoxymorphinan, respectively, during incubations with DM (Yu and Haining, 2001). High performance liquid chromatography analyses (Rege et al., 2002) were carried out on a Waters Alliance 2690 Workstation (Waters Corporatiion, Milford, MA) with a Waters 474 Scanning Fluorescence Detector. The protein precipitates from the incubation mixture were removed by centrifugation at 6000g for 10 min. The aqueous supernatant was transferred to Nanosep centrifugal concentrator (VWR Scientific, West Chester, PA), which contained a 10 kDa molecular mass cut-off centrifugal filter. The incubation mixture was centrifuged at 6000g for 20 min, and after centrifugation, 50 µl of the ultrafiltrate was injected. A reversed phase Phenyl Spherisorb column (4.6 × 250 mm, 5 µm packing; Waters Corporation) protected by a Phenyl Bondapack Guard-Pak column (10 µm packing; Waters Corporation) was used for the separation. The mobile phase was 25% acetonitrile/75% 0.01 M phosphate buffer, pH 2.9, containing 0.01% v/v triethylamine maintained at ambient temperature. The flow rate was 1 ml/min. The excitation and emission wavelengths of the fluorescence detector were 278 and 312 nm, respectively. Under these conditions, dextrorphan, 3-hydroxymorphinan, levallorphan (internal standard), 3-methoxymorphinan, and dextromethorphan eluted at 7.5, 8.6, 10.6, 13.6, and 15.5 min, respectively. Before performing the kinetic experiments in both the CYP2D6 and CYP3A4/5 assays the formation of product was determined to be linear up to 10 min at 0.5 mg/ml microsomal protein. Each enzyme exhibited Michaelis-Menten kinetics under these assay conditions. The experimentally determined enzyme constants for CYP2D6 DM O-demethylation are as follows: subject 1, K_m = 6.8 ± 0.9 µM, V_max = 198 ± 11 pmol/min/mg of microsomal protein; subject 2, K_m = 5.0 ± 0.4 µM, V_max = 252 ± 8 pmol/min/mg of microsomal protein; and for CYP3A4/5 N-demethylation: subject 2, K_m = 523 ± 120 µM, V_max = 804 ± 85 pmol/min/mg of microsomal protein. The formation of 3-hydroxymorphinan could be monitored during the assay. At the completion of the 10 min assay, 3-hydroxymorphinan accounted for less than 1% of the metabolites formed. The apparent K_i (single point) measured for S-methamphetamine for CYP2D6 using microsomes from subject 1 and 2 was 26 and 13 µM, respectively. These values were consistent with a prior report for S-methamphetamine (25 µM) (Wu et al., 1997). The initial incubations were done by preincubating 50 µM of (+)- or ()-chloroephedrine HCl with 0.5 mg/ml microsomal protein and 1 mM NADPH in pH 7.4 phosphate buffer for 10 min. Then 25 µM DM was added to initiate the reaction and to make a total volume of 0.5 ml. The incubation was continued for 10 additional minutes and was stopped by addition of 10 µl of 70% perchloric acid. The effect of preincubation of various reagents on CYP2D6 and CYP3A4/5 activity as measured with DM was done as described above by varying the times of preincubation (0, 5, 10, 30, and 60 min), (+)- or ()-chloroephedrine HCl (50 or 2000 uM), and NADPH (0 and 1 mM). Once the effect of preincubation of ()-chloroephedrine HCl with the microsomes was recognized, the remaining studies were done by sequential addition of NADPH, DM, and (+)- or ()-chloroephedrine HCl over 15 s to initiate the reaction. The remainder of the assay was run as described above. The initial investigation into the mechanism of inhibition for CYP2D6 and CYP3A4/5 was determined by incubation of varying concentrations of ()-chloroephedrine HCl (50, 100, 500, 1000 µM) and DM (10, 25, 100, 250 µM). To determine whether CYP2D6 activity could be regenerated after preincubation with ()-chloroephedrine HCl, duplicate samples of 100 µM ()-chloroephedrine HCl were preincubated with the microsomes for 60 min at 37°C. Duplicate samples without ()-chloroephedrine HCl were run in parallel to serve as controls. After the preincubation period, the microsomes were centrifuged, supernatant removed, resuspended in phosphate buffer, vortexed, centrifuged, supernatant removed, and the microsomes were then resuspended in pH 7.4 phosphate buffer. Then NADPH and DM were added, and CYP2D6 activity was determined. Measurement of time and concentration-dependent inactivation (Kitz and Wilson, 1962) was done by incubation of ()-chloroephedrine HCl (0, 100, 250, 500 µM) with 5 mg/ml microsomal protein at 37°C for the times indicated above. Then 50 µl was diluted with 338 µl of buffer followed by addition of NADPH and DM for a total volume of 0.5 ml and final concentrations of 1 mM NADPH and 25 µM DM. The rest of the assay was run as previously described. Regression lines for all kinetic data were based on a least-squares fit. The first order inactivation constant (k) at 50 µM ()-chloroephedrine HCl was obtained from linear regression of log percentage remaining activity versus time plot (Palamanda et al., 2001). An apparent K_i value was estimated by nonlinear fitting of the Michaelis-Menten equation for a competitive inhibitor using SigmaPlot, version 1.0 (SPSS Science, Chicago, IL) and by linear regression of the double reciprocal plot of the rates of inactivation of DM O-demethylation activity as a function of inhibitor concentration.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Molecular Modeling. After energy minimization of molecular models for the putative substrates (+)- and ()-chloroephedrines and protonated cis- or trans-1,2-dimethyl-3-phenylaziridine, the distance between the nitrogen and para position on the aromatic ring was the following: R-(+)-chloroephedrine, 5.639 Å; S-()-chloroephedrine, 5.639 Å; cis-1,2-dimethyl-3-phenylaziridine, 5.211 Å; trans-1,2-dimethyl-3-phenylaziridine, 5.268 Å. The docking calculations for protonated (+)- and ()-chloroephedrine, 1R,2S,3R- and 1S,2S,3R-cis-1,2-dimethyl-3-phenylaziridine, and 1R,2S,3S- and 1S,2S,3S-trans-1,2-dimethyl-3-phenylaziridine were done by placing the structure and optimizing as described in the CYP2D6 molecular model and the results are given in Table 1. The structures of the protonated aziridines are depicted in Fig. 1.

Microsomal Studies. No inhibition of CYP1A2 was observed following preincubation of (+)- or ()-chloroephedrines for 30 min (measurements were taken at 0, 5, 10, and 30 min) with microsomes (8.8 ± 8.0% and 9.0 ± 11.5% respectively). After preincubation of 50 µM (+)-chloroephedrine with microsomes for 30 min (sampled at 0, 5, 10, and 30 min) prior to the addition of 25 µM DM, no inhibition of CYP2D6 activity was observed (8.1 ± 6.5%). In the same sample inhibition of CYP3A4/5, activity was observed (32.6 ± 2.2%, p < 0.05). In incubations of 2000 µM (+)-chloroephedrine with microsomes prior to the addition of 1000 µM DM, no inhibition of CYP2D6 activity was observed for the first 5 min of preincubation, then a 19% inhibition of CYP2D6 activity was observed after 30 min. Under these same conditions inhibition of CYP3A4/5, activity was initially 36% and increased to 59% after 30 min of preincubation.

1

View larger version (11K): [in this window] [in a new window]   Fig. 2.   Effect of preincubation of ()-chloroephedrine HCl and NADPH on the metabolism of DM are as follows: first order inactivation constant (k) for 0 µM ()-chloroephedrine HCl, 0 mM NADPH (), k = 0.004 (±0.001, 95% confidence interval) min1; 0 µM ()-chloroephedrine HCl, 1 mM NADPH (), k = 0.019 (±0.004) min1; 50 µM ()-chloroephedrine HCl, 0 mM NADPH (), k = 0.047 (±0.010) min1; 50 µM ()-chloroephedrine HCl, 1 mM NADPH (), k = 0.054 (±0.020) min1.

View larger version (16K): [in this window] [in a new window]   Fig. 3.   Dixon plot for effect on O-demethylation and N-demethylation by ()-chloroephedrine HCl (50, 100, 500, 1000 uM) as an inhibitor of DM metabolism at 10 (), 25 (), 100 (), and 250 () µM. A, CYP2D6, apparent Ki = 98 ± 6 (S.E.M.) µM; the inset shows the replot of the slopes of the Dixon plot as a function of the reciprocal of DM concentration (yint = 3.0 ± 6.7 × 108, 95% confidence interval); B, CYP3A4/5.

View larger version (16K): [in this window] [in a new window]   Fig. 4.   Time and concentration-dependent inactivation of DM O-demethylation activity of CYP2D6 by ()-chloroephedrine in human liver microsomes. 0 (), 100 (), 250 (), and 500 () µM ()-chloroephedrine HCl at 0, 5, 10, 30, and 60 min. Reaction conditions are described under Materials and Methods. The inset shows the double reciprocal plot of the apparent rate constants of inhibition of DM O-demethylation versus ()-chloroephedrine concentration.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The occurrence of impurities in clandestinely synthesized methamphetamine has been well documented. Although numerous drugs/chemicals have been observed to be used to modify, "cut", or dilute methamphetamine, the occurrence of impurities associated with a specific synthetic method is limited. These impurities can be categorized as synthetic precursors (starting material, solvents, reagents), intermediates, and by-products. These impurities may have unique pharmacological or toxicological properties that may or may not parallel those observed for the drug being synthesized. A good example of this would be 1-methyl-4-phenyl-1,2,5,6-tetrahydropyridine (a potent dopaminergic neurotoxin) found in 1-methyl-4-phenyl-4-propionoxy piperidine (a meperidine related mu analgesic) (Langston et al., 1983). A better understanding of the pharmacological effects associated with the impurities in the drugs may provide a better understanding of the unusual effects or toxicity associated with clandestinely synthesized drugs.

With the increasing availability of 3D molecular models for studying ligand-protein interactions, structures can be screened for potential binding within an enzyme or receptor prior to initiating in vitro or in vivo experiments. Numerous 3D models for CYP2D6 have been developed and are being continually refined (Koymans et al., 1993; Ellis et al., 1996; Modi et al., 1996; Lewis et al., 1997; deGroot et al., 1999). In the CYP2D6 models, small molecules such as amphetamine are believed to orient through interaction of the protonated amine with Asp301 in the CYP2D6 binding site (Koymans et al., 1993; Ellis et al., 1995; Lewis et al., 1997; deGroot et al., 1999). Additional common characteristics for many CYP2D6 compounds that bind or are substrates requires the presence of one basic nitrogen, a distance of 5 to 7 Å between the basic nitrogen and a potential site of oxidation and a flat hydrophobic area near the site of oxidation (Islam et al., 1991; Koymans et al., 1993; Strobl et al., 1993; deGroot et al., 1997). All of the ligand models built for this study demonstrated these common structural characteristics.

The electrostatic interaction of the protonated amine with the Asp301 of the CYP2D6 was of primary importance for all of the ligands in this study, accounting for approximately 70% of the binding energy. If the chloroephedrine isomers were the primary structure present in solution, then the model suggested they would both bind equally well, comparable with methamphetamine, and better than either the cis- or trans-1,2-dimethyl-3-phenylaziridine. However, studies with structurally related compounds have shown that formation of the aziridines in water occur with a half-life (t_1/2, 37°C) ranging from 0.61 min for phenoxybenzamine (Henkel et al., 1976) to 5.7 min for N-(2-chloroethyl)-4-piperidinyl diphenylacetate (Thomas et al., 1992). Under the conditions used in this study, formation of the 1,2-dimethyl-3-phenylaziridines would be expected to occur during the incubations. When cyclization occurs, protonation of the aziridine nitrogen introduces a new chiral center as shown in Fig. 1. Based on steric interactions (comparative internal energies) the 1R,2S,3R-cis-1,2-dimethyl-3-phenylaziridine and the 1S,2S,3S-trans-1,2-dimethyl-3-phenylaziridine are predicted to be of lowest energy for the respective pairs and would be the predominate structure in solution. After docking to CYP2D6 and comparing the energy of binding for the two isomers, a difference of 12.4 (20%) kcal/mol occurs between the cis-and the trans-1,2-dimethyl-3-phenylaziridine with binding of the trans-1,2-dimethyl-3-phenylaziridine being favored.

The orientation of trans-1,2-dimethyl-3-phenylaziridine in the proposed binding site places the phenyl ring over the porphyrin ring positioned for hydroxylation. The values in Table 1 indicate that the angle between the aromatic ring in the phenethylamine and the porphyrin ring of the CYP2D6 are comparable to that observed for methamphetamine and similar to that observed in the binding of DM (de Groot et al., 1997). In contrast, in the minimized structure containing the 1R,2S,3R-cis-1,2-dimethyl-3-phenylaziridine, the porphyrin ring showed signs of distortion, and the phenyl ring was not oriented over the porphyrin ring. Instead, the ring face was directed toward the carboxyl side chain of the porphyrin ring. If the aziridines predominate in solution, then trans-1,2-dimethyl-3-phenylaziridine (formed from ()-chloroephedrine) would be predicted to bind better than cis-1,2-dimethyl-3-phenylaziridine (formed from (+)-chloroephedrine). Thus, while we did not learn much from modeling the chloroephedrines, because of their expected cyclization to aziridines, the modeling of the aziridines indirectly predicts that the ()-chloroephedrine should more readily bind to CYP2D6 than the (+)-chloroephedrine.

The initial in vitro study with the human liver microsomes indicates that only ()-chloroephedrine was an inhibitor of CYP2D6 at micromolar concentrations, suggesting that the active species was the trans-1,2-dimethyl-3-phenylaziridine. However, since these compounds are aziridines, the inhibition could have been due to nonspecific alkylation and protein denaturation. Studies under comparable conditions in which CYP1A2 and CYP3A4/5 activity was measured, no inhibition by ()-chloroephedrine was observed indicating that no significant nonspecific protein denaturation was occurring under these assay conditions. The (+)-chloroephedrine exhibited no inhibition of CYP1A2 and weak inhibition (35%) of CYP3A4/5. Further studies would be needed to characterize this inhibition of CYP3A4/5 by (+)-chloroephedrine in human liver microsomes, however, prior studies with rat microsomes on the metabolism of a mixture of cis- and trans-1,2-dimethyl-3-phenylaziridine reported their metabolism to cis- and trans-1-phenylpropene and nitrosomethane (Hata et al., 1976; Hata and Watanabe, 1994). The chemical instability of ()-chloroephedrine as its free base and of the trans-1,2-dimethyl-3-phenylaziridine as the free base or in protonated form in water or organic solvents has been reported (Taguchi and Kojima, 1959; Allen and Kiser, 1987; Lekskulchai et al., 2000). Consistent with this instability was that in pH 7.4 buffer, ()-chloroephedrine underwent total decomposition in 5 min, and addition of these decomposition products to the microsomes had no effect on the CYP2D6 activity. Unexpectedly, in the presence of microsomes, it can be seen in Fig. 4 that ()-chloroephedrine causes increasing inhibition of CYP2D6 microsomes with time and a half-life for inactivation of 23.2 min. The microsomes appear to stabilize ()-chloroephedrine and/or trans-1,2-dimethyl-3-phenylaziridine, giving it time to diffuse to the binding site of CYP2D6. A comparable phenomenon was observed for a structurally related compound, (4-acetoxyphenyl)-2-chloro-N-methyl-ethylammonium chloride, which appeared to be stabilized by binding of the -chloro amine precursor to serum proteins (Louw et al., 1997). Whether the increased stability is due to binding of the ()-chloroephedrine or trans-1,2-dimethyl-3-phenylaziridine cannot be determined at this time. Information is lacking concerning the interrelationships between the physicochemical properties of these impurities and their ability to undergo biodistribution and biotransformation. In general with this class of compounds, the rate of hydrolysis of the aziridinium ion once formed is slower than the rate of cyclization (Henkel et al., 1976; Thomas et al., 1992).

The inhibition of CYP2D6 over time in the presence or absence of NADPH was comparable, indicating inhibition by ()-chloroephedrine was not dependent on metabolic activation. The ()-chloroephedrine appears to be interacting within the CYP2D6 binding pocket as a competitive inhibitor based on the Dixon plot and a replot of the slopes of the Dixon plot. Finally, the inhibition appears to be irreversible. Usually, the kinetics of a competitive irreversible inhibitor will appear as a mixed-type of inhibition due to loss of active enzyme during the assay. In addition, assays of irreversible inhibitors require dilution of the inhibitor (usually a 100-fold) prior to doing the assay to minimize competitive inhibition by the inhibitor during measurement of the probe substrate. Due to the limitation of assay sensitivity, the maximal dilution we could obtain prior to carrying out time and concentration-dependent inactivation studies was only 10-fold, not great enough to completely eliminate the presence of inhibitor during the assay with DM. Allowing for the above limitation, the kinetics describing the inactivation of CYP2D6 over time at different concentrations of ()-chloroephedrine (Kitz-Wilson replot) are consistent with a reversible complex being formed in the concentration range studied that progresses to an irreversible inhibition (k_inact). In this preliminary characterization of this kinetically complex situation, no evidence of a mixed or more complicated type of enzyme kinetics was observed, and the values obtained for the apparent K_i by either analysis were comparable (approximately 2-fold). A possible explanation is that inactivation of the enzyme during the assay (10 min) is insignificant due to the higher affinity of substrate (dextromethorphan, K_m ~5 µM) than apparent affinity of inhibitor [()-chloroephedrine]. All of the kinetic data are consistent with the trans-1,2-dimethyl-3-phenylaziridine acting as an alkylating agent within the binding pocket of CYP2D6. In the 3D models of CYP2D6, the following amino acids are reported to be exposed in the xenobiotic binding site with potential to undergo alkylation: Asp301, Ser116, Ser289, Ser304, and Thr309. In addition, the carboxyls on the porphyrin ring side chains are exposed in the binding site (Modi et al., 1996). The site of alkylation on the trans-1,2-dimethyl-3-phenylaziridine could occur at either the 2 or 3 position with the 3 position being favored electronically (benzylic) and the 2 position being favored sterically. The model as shown in Fig. 5 suggests that if binding occurs prior to alkylation the Asp301, Ser304, and the carboxyl on the proprionic side chain of the porphyrin ring are in closest proximity (ranging from 2.9 to 5.8 Å). Further speculation on the site of addition is not possible.

View larger version (17K): [in this window] [in a new window]   Fig. 5.   3D molecular model of Heme binding region of 1S,2S,3S-trans-1,2-dimethyl-3-phenylaziridine as docked and energy minimized with CYP2D6.

The dangers associated with these impurities when present in methamphetamine or as waste materials from a clandestine laboratory are difficult to assess at this time. Since ()-chloroephedrine is a weaker inhibitor of CYP2D6 than methamphetamine, it is unclear to what extent it would be able to interact with CYP2D6 unless a sample is badly contaminated with the impurity. However, since illicit use of methamphetamine includes oral, intravenous, "snorting", and smoking, almost any organ, including the liver or central nervous system, could be exposed to ()-chloroephedrine present in a contaminated methamphetamine sample.

This initial study has shown that a methamphetamine impurity has the potential to contribute to the pharmacological profile of clandestinely synthesized methamphetamine. These studies have focused on the CYP2D6 enzyme in the liver; however, the potential for these impurities to interact selectively and irreversibly with other amphetamine binding sites (e.g., receptors or transporters) is possible.