Abstract
Nicotine is primarily metabolized to cotinine by cytochromes P450 (CYPs). The degree of variation in the metabolism of nicotine to cotinine and the relative roles of the polymorphic enzymes CYP2A6 and CYP2D6 in this metabolism were investigated. The apparentKm and V max values (mean ± S.D.) for cotinine formation in human liver microsomes (n = 31) were 64.9 ± 32.7 μM and 28.1 ± 28.7 nmol/mg of protein/hr, respectively. A 30-fold difference was seen among the individual V max values, with four livers showing significantly higher rates of cotinine formation. CYP2D6 is unimportant in nicotine metabolism because quinidine (a CYP2D6 inhibitor) had little effect on inhibition of cotinine formation;V max values for dextromethorphan (CYP2D6 probe substrate) and nicotine (n = 9) did not correlate (r = .49, P = .18), and a cDNA CYP2D6 expression system failed to metabolize nicotine to cotinine. CYP2A6 appears to be the major P450 involved in human nicotine metabolism to cotinine. Coumarin, a specific and selective CYP2A6 substrate, competitively inhibited cotinine formation by 85 ± 11% (mean ± S.D.) in 31 human livers. The Ki value for this inhibition ranged from 1 to 5 μM, and a CYP2A6 monoclonal antibody inhibited cotinine formation by >75%. Immunochemically determined CYP2A6 correlated significantly with nicotine-to-cotinine V max values (r = .90, n = 30, P < .001) and to inhibition of nicotine metabolism by coumarin (r = .94, n = 30, P < .001). These data indicate that nicotine metabolism is highly variable among individual livers and that this is due to variable expression of CYP2A6, not CYP2D6.
Nicotine is the primary compound present in tobacco, and it plays the crucial role in establishing and maintaining tobacco dependence (Henningfieldet al., 1985). Understanding the pattern of nicotine metabolism and the sources of variation of this metabolism in humans is important because of the key role of nicotine in producing tobacco dependence. Nicotine is primarily metabolized in humans to cotinine (70%) through a two-step process (Murphy, 1973) (fig.1). The first step is catalization by the CYP system to produce the nicotine-Δ-1′(5′) iminium ion (Williams et al., 1990a). This intermediate is further oxidized through a cytosolic aldehyde oxidase reaction (Petersonet al., 1987; Brandage and Lindblom, 1979; Gorrod and Hibberd, 1982). A 3-fold variation in the rate of nicotine metabolism between individuals (n = 14) has been reported (Benowitz et al., 1982). This variability in nicotine metabolism could be an important determinant of smoking behavior.
CYP2A6 is responsible for coumarin-7-hydroxylase activity in humans (Yamano et al., 1990; Pearce et al., 1992). In addition to coumarin, CYP2A6 has been shown to metabolize several procarcinogens, such as 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (Crespi et al., 1991), aflaxtoxin B1 (Yun et al., 1991), hexamethylphosphoramide (Ding and Coon, 1988) and nitrosodimethylamine (Davies et al., 1989;Fernandez-Salguero et al., 1995). Marked interindividual differences in CYP2A6 activity have been detected in human liver microsomes through coumarin-7-hydroxylase activity (Kapitulnik et al., 1977; Pelkonen et al., 1985). Variability in human livers was also found in levels of CYP2A6 mRNA (Miles et al., 1990; Yamano et al., 1990) and CYP2A6 protein (Yunet al., 1991). Some of the CYP2A6 variation may be due to induction of CYP2A6 by environmental compounds such as phenobarbital (Pearce et al., 1992). The CYP2A6 gene also displays a genetic polymorphism (Yamano et al., 1990;Fernandez-Salguero et al., 1995). Because cDNA studies implicate CYP2A6 in nicotine metabolism (Flammang et al., 1992; McCracken et al., 1992) and CYP2A6 expression is genetically regulated, variation in CYP2A6 activity may contribute to interindividual variation in nicotine metabolism.
CYP2D6 is an well-characterized enzyme that is involved in the metabolism of >40 clinically used drugs, including dextromethorphan (Coutts, 1994). CYP2D6 displays a genetic polymorphism in which ∼5% to 10% of caucasian populations (Kalow, 1987; Lennard, 1990; Veronese and McLean, 1991) 0% to 1% of Oriental populations (Lou et al., 1987; Wanwimolruk et al., 1990; Bertilsson et al., 1992) show impaired CYP2D6 activity. There remains some controversy concerning the importance of CYP2D6 in nicotine metabolism. McCracken et al. (1992) implicated CYP2D6 in nicotine metabolism in cDNA studies but this was contradicted by Flammang et al. (1992). In addition, Cholerton et al. (1994) reported, in in vivo studies, that all poor metabolizers of nicotine were genotypically poor metabolizers of CYP2D6-mediated reactions. However,Benowitz et al. (1996) found no association between the poor metabolism of nicotine with the poor metabolism of dextromethorphan, a CYP2D6 substrate.
To date, research on identification of the human CYPs involved in nicotine metabolism to cotinine has suggested several enzymes, including CYP2A6, CYP2B6, CYP2C8, CYP2C9, CYP2D6, CYP2E1, CYP2F1 and CYP4B1 (Flammang et al., 1992; McCracken et al., 1992). Because CYP2A6 and CYP2D6 display genetic polymorphisms, involvement of these cytochromes suggests that some individuals who lack these enzymes may be poor metabolizers of nicotine. We conducted the following study to determine the contribution of CYP2D6 and CYP2A6 to nicotine metabolism.
Materials and Methods
Drugs and chemicals.
(S)-Nicotine, (S)-cotinine, dextromethorphan hydrobromide, quinidine, NADPH, Tris · HCl, cumene hydroperoxide, octanesulfonic acid, troleandomycin, orphenadrine and ketamine were obtained from Sigma Chemical (St. Louis, MO). Coumarin was obtained from Caledon (Ontario, Canada). Potassium phosphate was purchased from Mallinckrodt (Ontario, Canada). Antibodies against CYP2A6, CYP2B1, CYP2E1 and CYP3A2 were purchased from Gentest Corp. (Woburn, MA). CYP2D6 was generously provided by Alastair Cribb and Merck Research Laboratories (West Point, PA). Dextrorphan, methoxymorphinan and hydroxymorphinan were kindly provided by Hoffman-La Roche (Nutley, NJ). Budipine was obtained from Byk Gulden Pharmazeutika (Konstanz, Germany). Microsomal preparations of CYP2D6 expressed in yeast (aH22/pelt1 cells) and control yeast (AH22/pMA91 cells) were provided by Dr. M. S. Lennard (University of Sheffield, UK) (Ching et al., 1995). Lymphoblastoid cells expressing either h2A3 (CYP2A6) or h2D6-Val (CYP2D6) cDNA and their respective control parent vector lines were purchased from Gentest Corp.
Human liver microsomes.
The characteristics and sources of the K series livers used in this study have been previously described (Campbell et al., 1987; Tyndale et al., 1989), whereas the L series livers were obtained from organ donors. These liver samples were generously provided by Drs. T. Inaba and E. Roberts. Table 1 summarizes the known sex and age of the donors of the livers. Microsomes were prepared and stored according to established techniques (Tyndale et al., 1989). Briefly, livers were thawed on ice, minced and combined with 2 ml of cold 1.15% KCl/mg of liver. Samples were homogenized and subjected to a 20-min centrifugation at 9000 × g at 4°C. The supernatant was then centrifuged for 60 min at 100,000 ×g at 4°C. The microsomal pellets were washed with cold 1.15% KCl and recentrifuged. Washed pellets were resuspended in a volume of 1.15% KCl equal to the mass of the original liver sample. Microsomal protein concentrations were determined using the BCA protein assay kit (Pearce Chemical, Rockford, IL). Cytosolic fractions from the livers of four male Wistar rats were used as a source of aldehyde oxidase.
Nicotine assay.
Nicotine metabolism was assayed by incubating microsomal protein with (S)-nicotine in 1 ml of 0.04 M potassium phosphate buffer, pH 7.4. The choices of buffer and concentration were chosen as optimal on the basis of the study of Pierce et al. (1992). Incubation mixtures also contained 1 mM NADPH and 20 μl of rat liver cytosol (as the aldehyde oxidase source). Excess aldehyde oxidase was added, so the CYP oxidation was rate limiting (Cashman et al., 1992). Incubations were carried out at 37°C and stopped with the addition of 100 μl of 20% Na2CO3. Ketamine (10 μl of 0.25 mg/ml) was added as the internal standard. Samples were extracted with 3 ml of ethyl acetate and back-extracted into 400 μl of 0.01 N HCl. Samples were partially dried under nitrogen for 25 min to remove excess ethyl acetate; then, 30 μl of each sample was subjected to HPLC analysis series with an UV detector (set at 210 nm). Separation of nicotine and metabolites was achieved using a CSC-Spherisorb-Hexyl column (15 × 0.46 cm) and a mobile phase consisting of 20% acetonitrile and 80% of 20 mM potassium phosphate, pH 4.6, containing 1 mM octanesulfonic acid. The separation was performed with isocratic elution at a flow rate of 1 ml/min. The retention times for cotinine, nicotine and ketamine were 3.5, 4.2 and 7.0 min, respectively. Nicotine-to-cotinine kinetic studies were performed by incubating 1, 5, 10, 50, 100 and 200 μM (S)-nicotine with 0.5 mg/ml microsomal protein for 45 min. Standard curves were created for cotinine (1.25–10 μM) with 10 μl of ketamine (0.25 mg/ml) as the internal standard. Cotinine was measured as a peak height ratio and compared with the standard curve, enabling peak height ratios for a given sample to be converted to cotinine concentrations (nmol/ml). The detection limit of our system was 300 pmol of cotinine/ml of incubation mixture. For cotinine concentrations of 2.5 and 5.0 nmol/ml, the within-day coefficients of variations were 3.1% and 2.3% and between-day variations were 7.2% and 8.4%, respectively.
Nicotine was incubated with yeast microsomes and lysed lymphoblastoid cells expressing CYP2D6 with similar incubation conditions. Control experiments consisted of incubation of nicotine with microsomes from yeast or cells expressing just the vector.
Dextromethorphan assay.
Incubation conditions of this assay were essentially those of Otton et al. (1983). Briefly, the incubation mixture consisted of 125 μl of phosphate buffer (0.2 M, pH 7.4), 50 μl of microsomal protein (0.3 mg of protein/ml), 50 μl of dextromethorphan (1, 2.5, 5, 10, 50 and 75 μM) and 25 μl of NADPH (0.8 mM) for a total volume of 250 μl. Incubations were carried out at 37°C for 30 min in a shaking water bath and terminated by the addition of 10 μl of 70% perchloric acid. Budipine was used as an internal standard. Samples were then centrifuged at 3000 rpm for 5 min, and 30 μl of the supernatant was analyzed by HPLC with a CSC-Spherisorb-Phenyl (5 μm, 4.6 mm × 25 cm) column and a mobile phase consisting of 10 mM potassium phosphate buffer containing 1 mM heptanesulfonic acid, pH 3.8, and acetonitrile (80:20 v/v); the flow rate was set at 1.7 ml/min. Dextromethorphan and various metabolites were detected as described by Broley et al.(1989), except excitation and emission wavelengths were set at 195 and 280 nm, respectively, for a higher sensitivity. Dextrorphan calibration curves were linear from 0 to 120 pmol, with the lowest detectable level of 5 pmol for dextrorphan. The coefficient of within-day variation was 2.7% and 2.0% (n = 5) for 0.25 and 0.5 nmol/ml injections of dextrorphan, respectively. The coefficient of between-day variation was 6.5% and 9.6% (n = 6) for 0.25 and 0.5 nmol/ml concentrations of dextrorphan, respectively.
Nicotine inhibition assays.
Chemical inhibition studies consisted of incubation of 100 μM (S)-nicotine with 150 μM concentrations of coumarin (a CYP2A6 substrate), orphenadrine (a CYP2B6 inhibitor), troleandomycin (a CYP3A inhibitor) or coumarin with orphenadrine in combination (30 human livers). Orphenadrine has been used as a specific inhibitor of cDNA/CYP2B6-mediated reactions (Changet al., 1993). Quinidine (a CYP2D6 inhibitor) at 0.1, 1, 10 and 100 μM concentrations was incubated with 50 μM nicotine using human liver microsomes (K12). Incubation conditions were as described above.
Immunoinhibition experiments consisted of incubation of 0.5 mg/ml K12 liver microsomes with CYP2A6 (monoclonal), CYP2B1 (polyclonal), CYP2E1 (polyclonal), CYP2D6-peptide (polyclonal) and CYP3A2 (polyclonal) antibodies. BSA and rabbit and goat antisera were used as negative controls. Antibodies and microsomes were preincubated on ice for 30 min, followed by the addition of 100 μM nicotine, 1 mM NADPH and 20 μl of rat cytosol in 0.04 M phosphate buffer, pH 7.4. Subsequent incubations were for 45 min at 37°C.
Western blot analysis.
Liver microsomal protein (30 μg) were resolved on 10% SDS-PAGE gels and transferred to nitrocellulose (120 V for 18 hr at room temperature) by wet electroblotting. Blots were blocked for 1 hr at room temperature with 2% (w/v) BSA dissolved in TBST. Incubations with primary and secondary antibodies were performed for 1 hr in TBST. The primary, monoclonal, CYP2A6 antibody (1:2000 dilution, Gentest) was added and incubated at room temperature for 1 hr in TBST. Blots were then washed three times with TBST every 10 min. The secondary antibody, an anti-mouse IgG horseradish peroxidase conjugate (1:2000 dilution; Amersham, Arlington Heights, IL), was then incubated for 1 hr in TBST. After a second wash, blots were visualized using the chemiluminescent ECL reagent (Amersham). The densities of the visualized bands were quantified using a MCID video-imaging system (Imaging Co., St. Catharines, Ontario, Canada). After determining the range in which there was linear detection of the immunoreactive CYP2A6 bands, a concentration of 30 μg of microsomal protein was used for comparisons of CYP2A6 immunoreactivity among the livers.
Statistical analysis.
The Student’s t test was used when comparing sex differences in nicotine metabolic kinetic values. A value of P ≤ .05 was considered significant. Correlation coefficient P values for CYP2A6 immunoactivity and nicotine metabolism were calculated using EasyStat Version 1.0.Km and V max values for the kinetic data were obtained using the software program ENZFITTER (Elsevier-BIOSOFT, Cambridge, UK).
Results
Nicotine-to-cotinine kinetics.
Km andV max values for nicotine-to-cotinine kinetics were calculated in 31 human liver microsomes from men and women (fig.2). The mean (± S.D.)Km value was 64.9 ± 32.7 μM, with a range of 13 to 162 μM. The V max results revealed marked interindividual variations in cotinine formation, with a mean (± SD) of 28.1 ± 28.7 nmol/mg of protein/hr and a range of 4.2 to 120 nmol/mg of protein/hr. Four human livers from women showed significantly higher (>5 S.D. above the meanV max values for women) rates of cotinine formation. There is a ∼30-fold difference in theV max values and a >50-fold difference inV max/Km values. The differences in V max values for men and women approached significance (P = .07; Student’s t test) but not after the four high V max values for women were removed (P = .78; Student’s t test).
Nicotine metabolism and CYP2D6.
CYP2D6 catalytic activity was assessed by measuring the metabolism of the probe drug dextromethorphan to dextrorphan (Evans and Relling, 1991). V max values for nicotine and dextromethorphan metabolism were not correlated (r = .49, P = .18, n = 9 different livers), nor were V max/Km values (r = .37, P = .33, n = 9 pairs).
Quinidine, a specific CYP2D6 inhibitor, was tested at 0.1, 1, 10 and 100 μM with 50 μM nicotine to evaluate the contribution of CYP2D6 activity to nicotine metabolism (K12 human liver microsomes). At 0.1 and 1 μM quinidine, no inhibition of cotinine formation was observed. At 10 and 100 μM quinidine, 100- to 1000-fold theKi value for CYP2D6 inhibition (Ki ∼ 100 nM; Kerry et al., 1994), 8% and 25%, respectively, inhibition of cotinine formation was observed. In nicotine incubations with either lysed lymphoblast cells expressing CYP2D6 or yeast microsomes expressing CYP2D6 cDNA, both failed to metabolize nicotine to cotinine but were able to metabolize the CYP2D6 substrate dextromethorphan (5 μM). In addition, a CYP2D6 polyclonal antibody showed little, if any, inhibition of nicotine metabolism by a liver (K12) genotyped and phenotyped as a CYP2D6 extensive metabolizer liver (fig.3).
Nicotine metabolism and CYP2A6.
Coumarin (100 μM), a CYP2A6 substrate (Pearce et al., 1992), inhibited cotinine formation by >80% (K27), with little evidence of augmentation when quinidine was added in combination with coumarin (data not shown). Coumarin (150 μM) significantly inhibited cotinine formation with a mean (± S.D.) inhibition of 85 ± 11% (n = 31) (table 2). The apparentKi value, in three separate trials, ranged from 1 to 5 μM (K27), as estimated from Dixon plot analysis (fig.4). Orphenadrine (150 μM), a CYP2B inhibitor, showed moderate inhibition (20 ± 16%; mean ± S.D.), whereas troleandomycin, a CYP3A inhibitor, had no effect on inhibition of cotinine formation (3 ± 11%; mean ± S.D.).
Lymphoblastoid cells expressing CYP2A6 were able to metabolize nicotine to cotinine with Km andV max values of 47 μM and 1.5 nmol/mg of protein/hr, respectively (n = 2). The differences in velocity of CYP2A6 activity between the cell line and human liver microsome can be partially explained by the difference in protein content. The cell expression incubations involved the use of whole lysed cells, whereas the human liver incubations involved the use of purified microsomes. We have measured the protein content in homogenized liver cells and liver microsomes and found that homogenized cells contain ∼20 times more protein. With this taken into consideration, the V max values for cotinine formation using the cell expression system is comparable to the human liver V max values.
Immunoinhibition of nicotine metabolism with a specific CYP2A6 monoclonal antibody showed >75% inhibition of cotinine formation, whereas CYP2E1, CYP2B1, CYP2D6 or CYP3A2 antibodies; BSA; or preimmune antibodies demonstrated only modest, nonspecific inhibition (fig. 3). These results are similar to the findings of Nakajima et al.(1996), who demonstrated a 70% inhibition of CYP2A6-catalyzed cotinine metabolism.
Immunoreactive CYP2A6 was measured in each of the 30 human liver microsomes. The densities of each band were used to compare the relative amounts of CYP2A6 between livers (fig.5). Band densities from different blots were normalized by division on the basis of the 30-μg L64 band density of each blot. This was done so that individual band densities can be compared between blots. Western blot analysis was repeated for 3- and 10-μg amounts for livers that were outside the linear range as determined by the L64 standard curves. A summary table of the values used in CYP2A6 and nicotine correlation studies can be found in table1. A high correlation was observed between immunoreactive CYP2A6 levels and V max/Km values for cotinine formation (r = .94, n = 30, P < .001; fig. 6) and withV max values (r = .90,n = 30, P < .001). A high correlation was also seen when CYP2A6 immunoactivity was plotted against the amount of cotinine inhibited in the presence of coumarin (150 μM;r = .94, n = 30, P < .001; fig.6).
Discussion
The results of this study have established that human CYP2D6 is not important in nicotine metabolism in human liver microsomes, which is in agreement with the results of Flammang et al. (1992)and Benowitz et al. (1996). In addition, our data indicate (1) CYP2A6 is the principal cytochrome P450 involved in nicotine metabolism and (2) variation in CYP2A6 is the principal reason for interindividual differences in nicotine kinetics. Specifically, our results revealed a >30-fold variation in nicotine-to-cotinineV max values (fig. 2).
The studies were conducted to identify whether CYP2D6 plays an important role in nicotine metabolism and whether the CYP2D6 polymorphism alters nicotine disposition (McCracken et al., 1992; Cholerton et al., 1994). McCracken et al.(1992) reported that CYP2D6 cDNA expressed in human cell lines was able to oxidize nicotine, and an in vivo study showed that all poor metabolizers of nicotine were also homozygous for CYP2D6 mutations (Cholerton et al., 1994). If CYP2D6 is involved in nicotine metabolism, our data indicate the role is very minor.
Some reports have suggested an overrepresentation of the CYP2D6 extensive metabolizer phenotype in patients with lung cancer (Agundez and Benitez, 1993; Hirvonen et al., 1993; Benitez et al., 1991; Ayesh et al., 1984), although others do not agree (Wolf et al., 1992; Speirs et al., 1990). If CYP2D6 played a major role in nicotine metabolism, then higher nicotine metabolism would be associated with higher smoke exposure because smokers regulate nicotine intake by adjusting inhalation patterns and smoking behavior (McMorrow and Foxx, 1983; Russel, 1987). Thus, it could be argued that CYP2D6 extensive metabolizers might require larger doses of nicotine from tobacco products to satisfy individual craving, thereby having higher exposure to tobacco smoke and carcinogens. Because CYP2D6 does not appear to be involved in nicotine metabolism, the association of CYP2D6 extensive metabolizers with lung cancer might be explained by activation by CYP2D6 of procarcinogens from cigarette smoke such as 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (Crespi et al., 1991). Alternatively, the CYP2D6 wild-type allele could be in linkage disequilibrium with a gene that increases risk for lung cancer.
We demonstrated that CYP2A6 plays an important role in nicotine metabolism and that variations in CYP2A6 expression are responsible for the high interindividual variation that we observed in vitroin cotinine formation. Genetic variation in CYP2A6 may contribute to the 3-fold variation observed in human subjects with respect toin vivo nicotine metabolism (Benowitz et al., 1982). Benowitz et al. (1995) described a 57-year-old woman who had a very low capacity to form cotinine from nicotine. The authors argue that this observation is the result of a genetic polymorphism in forming the iminium ion intermediate. They discovered that the patient had normal CYP2D6 activity. Another possibility was that this person was deficient in CYP2A6 activity.
In our study, the four livers with exceptionally high cotinine formation were from women, causing the mean rate of cotinine formation by livers from women to approach significance compared with livers from men (P = .07) Previous in vivo studies, however, showed that nicotine metabolism was more rapid in men than in women (Beckettet al., 1971; Benowitz and Jacob, 1984). We also found that there was no correlation between cotinine formation and age (r = −.15, P = .44). Phenobarbital can increase CYP2A6-mediated reactions in primates (Pearce et al., 1992) and has been shown to increase nicotine metabolism in rats (Rudellet al., 1987). Perfused rat livers pretreated in vivo with phenobarbital showed a 14-fold increase in nicotine elimination compared with saline-treated controls (Rudell et al., 1987). Human hepatocytes from individuals treated in vivo with phenobarbital showed higher-than-normal nicotine oxidation rates on hepatocyte harvest (Williams et al., 1990b). This study supports the argument that exposure to environmental inducers may induce CYP2A6. This, in turn, would affect the overall metabolism of nicotine from the body. A detailed history of individual drug use was not available but would have been helpful because the four livers from women that showed high rates of cotinine formation may have been exposed to barbiturates, which may increase CYP2A6 activity. As previously mentioned, because smokers adjust their smoking behavior to maintain nicotine body levels (McMorrow and Foxx, 1983; Russel, 1987), individuals with high CYP2A6 activity would rapidly metabolize nicotine. Therefore, rapid metabolizers of nicotine may smoke more cigarettes to maintain nicotine levels and, hence, are exposed to more toxic compounds. Conversely, slower metabolizers of nicotine may smoke less and might be at higher risk for nicotine-related adverse effects.
These results confirmed a major role for CYP2A6, not CYP2D6, in nicotine metabolism. We have also shown that nicotine metabolism is quite variable among individual human liver microsomes. We postulate that this variation will be evident in variations seen in smoking behavior and could affect the efficacy of nicotine-replacement treatments of tobacco addiction (e.g., nicotine patch and nasal spray). The identification of potent inhibitors of CYP2A6 could lead to new treatment approaches for tobacco dependence.
Acknowledgments
We thank Dr. E. Roberts and Dr. T. Inaba for generously providing the human liver samples and Dr. M. S. Lennard for providing CYP2D6-expressing yeast. We also thank Siu Cheung, Ewa Hoffmann and Mae Kwan for technical support in the laboratory.
Footnotes
-
Send reprint requests to: Dr. Edward M. Sellers, University of Toronto, Medical Sciences Building, 1 King’s College Circle, Rm. 4334, Toronto, Canada, M5S 1A8.
-
↵1 This work was supported in part by the National Institute of Drug Addiction (NIDA Grant DA06889), University of Toronto and Addiction Research Foundation.
- Abbreviations:
- CYP
- cytochrome P-450
- TBST
- 150 mM NaCl, 50 mM Tris · HCl and 0.05% Tween 20
- SDS
- sodium dodecyl sulfate
- PAGE
- polyacrylamide gel electrophoresis
- BSA
- bovine serum albumin
- Received December 18, 1996.
- Accepted May 16, 1997.
- The American Society for Pharmacology and Experimental Therapeutics