Abstract
(+)-Cis-3,5-dimethyl-2-(3-pyridyl)thiazolidin-4-one hydrochloride (SM-12502) is a newly developed drug as a platelet-activating factor receptor antagonist. The disposition of SM-12502 was investigated in plasma from 28 healthy Japanese volunteers after a single i.v. administration of SM-12502. Three of 28 subjects were phenotyped as poor metabolizers (PMs). Genomic DNAs from three extensive metabolizers or three PMs of SM-12502 were analyzed by Southern blot analysis with CYP2A6 cDNA as a probe. DNAs from three PMs digested with SacI and SphI showed novel restriction fragment length polymorphisms (RFLPs); one type without 4.5- and 2.6-kb fragments and a weak density of a 6.4-kb fragment (E-type), and the other type without 7.1- and 5.5-kb restriction fragments (C′-type) as compared with three extensive metabolizers, respectively. The deletional restriction fragments specific to three PMs in SacI- and SphI-RFLPs were identified as CYP2A6. Using polymerase chain reaction-RFLP analyses of the gene from the three PMs, we found that the exon 1, exon 8, and exon 9 in CYP2A6 were absent. A new RFLP characterized by SacI and SphI was found to be due to the entire gene deletion of the three exons and was associated with the decreased metabolism of SM-12502. This study demonstrates a new deletional allele in the human CYP2A6gene responsible for the poor metabolic phenotype of SM-12502.
The disposition of (+)-cis-3,5-dimethyl-2-(3-pyridyl)thiazolidin-4-one hydrochloride (SM-12502) is a newly developed platelet-activating factor receptor antagonist that shows rapid absorption and a long duration of pharmacological activity after oral administration (Imanishi et al., 1994). It also has several other pharmacological actions, such as the induction of neutrophil aggregation, bronchoconstriction, hypotension, and an increase in vascular permeability.
We previously clarified that the cytochrome P-450 (CYP)2A6 isozyme was mainly involved in the S-oxidation of SM-12502. A close correlation (r = 0.908, p < .0001) was observed between the activities of SM-12502 S-oxidase and the coumarin 7-hydroxylase in human liver microsomes (Nunoya et al., 1996). It has been well known that CYP2A6 is responsible for coumarin 7-hydroxylation (Yamano et al., 1990; Miles et al., 1990; Yun et al., 1991; Maurice et al., 1991). In in vitro studies, the coumarin 7-hydroxylase activity in human liver microsomes showed marked interindividual variations in association with the levels of CYP2A6 expression (Kapitulnik et al., 1977; Pelkonen et al., 1985). This variability was also found in the levels of CYP2A6 mRNA and protein (Yamano et al., 1990; Miles et al., 1990; Pearce et al., 1992). In addition, it has been reported that there was a great interindividual variation in the capacity to hydroxylate coumarin at the 7 position in in vivo studies (Cholerton et al., 1992; Rautio et al., 1992; Iscan et al., 1994). However, the existence of polymorphism in the metabolic capacity was not clarified until Fernandez-Salguero et al. (1995)reported CYP2A6v1 and CYP2A6v2 allelic variants. These variants, however, were not necessarily thought to be the cause of the slow metabolizers of coumarin. Thus, it was needed to clarify the variant alleles that accounted for the large interindividual variations of CYP2A6 activity in humans. We found a newSacI-restriction fragment length polymorphism (RFLP), D-type, deleting a 2.6-kb restriction fragment in the CYP2Agene with Southern blot analysis using CYP2A6 cDNA as a probe (Nunoya et al., 1998). The D-type was involved in the reduced metabolic capacities of SM-12502 and coumarin in vitro. A 2.6-kb restriction fragment contained the region from intron 5 to exon 9 inCYP2A6.
The disposition of SM-12502 was investigated in plasma or urine from 28 healthy Japanese volunteers after a single i.v. administration of SM-12502 (T. Kainuma, unpublished data). Three of 28 Japanese were phenotyped as poor metabolizer (PM). In the present study, we found novel SacI- and SphI-RFLPs, E-type and C′-type, respectively, seen in common in the three PMs.
Experimental Procedures
Materials.
Restriction endonucleases were obtained from Toyobo (Osaka, Japan), Takara Shuzo (Kyoto, Japan), or New England Biolabs (Beverly, MA). Nylon membranes (Nytran NY13) were obtained from Schleicher & Schuell (Dassel, Germany); lambda ZAP II, lambda FIXII, and Gigapack III Gold Packaging Extracts were purchased from Stratagene (La Jolla, CA). T4 DNA polymerase were obtained from Takara Shuzo, Sequenase Version 2 was purchased from United States Biochemical (Cleveland, OH), and Ampli Taq DNA polymerase was obtained from Perkin-Elmer Cetus Instruments (Norwalk, CT). All other chemicals and solvents were of the highest grade commercially available.
Subjects.
Twenty-eight healthy Japanese volunteers (age, 33.2 ± 6.6 years) were male, unrelated, and from different geographic places. The disposition of SM-12502 was investigated with plasma or urine from 28 healthy Japanese volunteers after a single i.v. administration of SM-12502. The subjects given SM-12502 were classified into PMs and extensive metabolizers (EMs) on the basis ofCmax,T1/2(β), area under the plasma concentration-time curve (AUC), and urinary excretion of SM-12502 orS-oxide metabolite. Three of 28 Japanese were phenotyped as PM. In the three PMs, the disposition of SM-12502 was markedly lower; 13-, 14-, and 11-fold differences in AUC and 11-, 17-, and 11-fold differences in T1/2 were seen as compared with EMs, respectively. The differences were so clear that we did not define the PM and EM by the exact number of metabolic ratio. The S-oxide was detected as a sole metabolite in plasma and urine after the administration of SM-12502. One hundred twenty-four Japanese (80 men and 44 women; aged 47.2 ± 19.1 years), who were investigated for the frequency of CYP2A6 gene deletion, were unrelated, from different geographic places, and randomly chosen.
Southern Blot Analyses of Genomic DNAs Prepared from Peripheral Leukocytes.
Genomic DNA was extracted from peripheral lymphocytes by phenol-chloroform followed by ethanol precipitation (Sambrook et al., 1989). The DNA preparations (10 μg) were digested with restriction endonucleases. The digested DNA was subjected to electrophoresis with a 0.6% agarose gel. The gels were treated with 0.5 M NaOH containing 1.5 M NaCl, neutralized with 0.5 M Tris-HCl buffer (pH 8.0) containing 1.5 M NaCl, and equilibrated with 0.3 M sodium citrate containing 3 M NaCl before the transfer of the DNA to nylon membranes. The membranes were baked at 80°C for 2 h, prehybridized at 65°C for 2 h, and hybridized at 65°C for 8 h. The hybridization was performed in a reaction mixture containing the 1.6-kb fragment of a 32P-labeled human CYP2A6 cDNA, 50 mM Tris-HCl buffer (pH 8.0), 1 M NaCl, 11.5 mM EDTA, 0.1% SDS, 0.1% ficoll, 0.1% polyvinylpyrrolidone, 0.1% BSA, and 0.1 mg/ml of denatured salmon sperm DNA. The membranes were washed with 7.5 mM sodium citrate containing 75 mM NaCl and 0.1% SDS at 50°C for 20 min. Hybridized bands were visualized using an X-ray film. CYP2A6 cDNA was obtained by reverse transcription-polymerase chain reaction (PCR) as described previously (Nunoya et al., 1998).
Construction and Screening of Genomic DNA Libraries.
Genomic DNAs from human subjects PM1, PM3, and EM3 partially digested bySau 3AI were fractionated by the sucrose density gradient (10–38%) centrifugation. Fragments with mean lengths of 15 kb were partially filled in with dGTP and dATP and ligated to lambda FIXII vectors that had been digested with XhoI and partially filled in with dTTP and dCTP. The ligated products were packaged in vitro using Gigapack III Gold Packaging Extracts. These libraries were screened by the plaque hybridization method (Sambrook et al., 1989) using the entire coding region (1.6 kb) of human CYP2A6 cDNA as a probe. Phage DNAs were purified from positive plaques and digested with various restriction endonucleases. The DNAs were subcloned into pUC18, M13 mp18, or M13 mp19. The nucleotide sequences of positive clones were determined by the dideoxy method with Sequenase Version 2 using universal primer as well as synthesized oligonucleotides (17–23 mers) corresponding to the 5′ and 3′ side region of each 9 exon. Oligonucleotides used in this study were synthesized with a DNA synthesizer (model 381A; Applied Biosystems, Foster City, CA).
Genomic DNA (852) digested by SacI was fractionated by sucrose density gradient (10–38%) centrifugation. The fractions of the SacI fragments with 4.5 ± 0.5-kb length were ligated into a vector derived from lambda ZAP II that had been engineered to accept SacI fragments. The ligated product was packaged in vitro using Gigapack III Gold Packaging Extracts. The library was screened by the plaque hybridization method using the entire coding region (1.6 kb) of human CYP2A6 cDNA as a probe. Phage DNAs were purified from positive plaques. The inserted DNA fragments were subcloned into a plasmid pBluescript SK(−) by in vitro excision. These clones were sequenced as described above.
PCR Analysis for Human CYP2A6 Gene Deletion.
To confirm whether the E-type mutation occurred due to the deletion ofCYP2A6 gene, PCR amplification was performed. The primers used for PCR are shown in Table 1. The primers 2A6–8S, 2A7-B1, 2A6-B6, and 2A6-B1 were used in our previous study (Nunoya et al., 1998). The primers 2A6-F3 and 2A6–1AS-2 were used to amplify exon 1 in CYP2A6, the primers 2A6–8S, 2A7-B1, and 2A6-B6 were for exon 8, and the primers 2A6–9S and 2A6-B1 were for exon 9, respectively. PCR reactions were performed in a 50-μl reaction mixture containing 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCl2, 0.001% (w:v) gelatin, 50 μM each dNTP, 0.5 μM each primer, 0.5 μg genomic DNA and 1 U of Ampli Taq DNA Polymerase. Before the addition of 1 U of Ampli Taq DNA polymerase, the reaction mixtures were heated at 94°C for 5 min and cooled at 0°C for 3 min. After the addition of 50 μl of mineral oil, the mixtures were immediately cycled 32 times through a cycle consisting of denaturation at 94°C for 1 min, annealing for 2 min, and extension reaction at 72°C for 2 min for exons 1, 8, and 9, respectively. The annealing temperature was 60°C for exon 1, 47°C for exon 8, and 64°C for exon 9, respectively. The 184-bp PCR products in exon 1 were digested with HaeIII restriction endonuclease at 37°C for 2 h and separated in a 15% polyacrylamide gel. The 140-bp PCR products of exon 8 were digested with Bst NI restriction endonuclease. A clone for CYP2A6 cDNA was obtained by reverse transcription-PCR (Nunoya et al., 1998). Clones for CYP2A7 and CYP2A13 genes were obtained from the human genomic library.
Results
RFLPs of CYP2A Genes from Three PMs and Three EMs after Digestion with SacI and SphI.
When genomic DNAs from three PMs and three EMs were digested with SacI and SphI, RFLPs in the three PMs were distinct from those in the EMs (Fig. 1). In our previous report (Nunoya et al., 1998), we reported that SacI-RFLPs can be classified into four types (A, B, C, and D). As shown in Fig. 1B,SacI-RFLP specifically seen in the three PMs was newly found. The new allele named as the E-type was characterized by the absence of 4.5- and 2.6-kb fragments, and by a weak density of a 6.4-kb fragment. A SphI-RFLP specific in the three PMs was also newly found (Fig. 1D). This allele named as the C′-type was characterized by the presence of 6.3-kb and the absence of 7.1- and 5.5-kb fragments. SphI-RFLPs were classified into three types: one allele was characterized by the presence of 7.1-, 6.3-, and 5.5-kb fragments (A′-type), a second allele by the presence of 7.1- and 5.5-kb fragments and the absence of a 6.3-kb fragment (B′-type), and the third allele was the C′-type as described above. The three EMs in this study showed RFLP pattern of the B′-type. The A′-type was observed in other individuals in this study.
Analysis of CYP2A Genes Showing PM-SpecificSacI-RFLPs.
Lamda FiXII genomic DNA libraries were prepared from peripheral lymphocytes of subjects PM1, PM3, and EM3, respectively. Lamda ZAPII genomic DNA library was prepared from subject 852. The libraries were screened using the entire coding region (1.6 kb) of human CYP2A6 cDNA as a probe. By screening about 2.4 × 106 plaques from subject 852, two positive clones containing the CYP2A-related sequence were obtained. One of two clones, 2A6–35, was obtained with the insert lengths of 4.5 kb (Fig.2B). The sequence of the exon 9 and 3′-noncoding region was completely identical with that ofCYP2A6 cDNA reported by Yamano et al. (1989). The other clone, containing a SacI-digested 4.3-kb fragment, showed aCYP2A-related sequence (K.N., unpublished data). We had obtained a SacI-digested 2.6-kb fragment (2A6–5-1) that showed completely the same sequence in exons 6, 7, 8, and 9 of CYP2A6 cDNA, and a 2.55-kb fragment that showed 86% identity to the sequence in exon 4 (503–568) of CYP2A6 cDNA (Nunoya et al., 1998).
Approximately 5.6 × 105 and 3.0 × 105 plaques from PM1 and PM3 were screened, respectively. Eight positive clones were obtained with the mean insert lengths of 15 kb, although the clones did not contain theCYP2A6 gene. Two clones contained CYP2A7, three clones contained CYP2A13, and three clones contained another unknown CYP2A showing 82.4% identity to exon 3 of CYP2A6 cDNA (401–502). Lamda FiXII genomic DNA library from PM3 was prepared again and rescreened (approximately 7.1 × 105 phages) using the 1.6-kb fragment of human CYP2A6 cDNA (exons 1–9) and the SacI 2.6-kb fragment ofCYP2A6 gene (exons 6–9) as probes. We obtained severalCYP2A clones that did not contain the CYP2A6gene.
By screening about 1.1 × 106 plaques from EM3, fourteen positive clones were obtained with the mean insert length of 15.5 kb. Of these clones, three clones contained theCYP2A6 gene. The 2A6–7E (Fig. 2B) clone contained the region from the 5′-upstream to intron 6 of the CYP2A6 gene. As mentioned above, 2.6- and 4.5-kb fragments isolated from the human subject 852 (clones 2A6–5-1 and 2A6–35) contained a downstream region from intron 5 of the CYP2A6 gene (Fig. 2B). On the basis of the SacI restriction maps of the CYP2A6 gene in clones 2A6–7E, 2A6–5-1, and 2A6–35, the 6.3-, 2.6-, and 4.5-kb fragments were assumed to be contained in the regions from exons 1 to 5, from exons 5 to 9, and from exon 9 to a 3′-untranslated region, respectively (Fig. 2C). The 7.1- and 5.5-kb fragments obtained by digestion with SphI corresponded to the regions from exons 1 to 5 and from exon 5 to a 3′-untranslated region, respectively (Fig.2C). Thus, we could not obtain a genomic clone containing theCYP2A6 gene from a genomic library of PM3, whereas clones containing the same gene could be obtained. These results seemed to be in accordance with our previous results indicating the absence ofSacI and SphI fragments in Southern blot analysis of the gene from three PMs. Based on these results, we postulated the deletion of CYP2A6 genes in the three PMs. In theCYP2A6 gene, the consensus sequences for splicing junctions, GT and AG, were found at the boundaries of all eight introns (Breathnach and Chambon, 1981). The CYP2A6 gene was approximately 7 kb in length. The nucleotide sequence of all exons from ATG to TGA was completely identical with hIIA3 reported as CYP2A6 cDNA (Yamano et al., 1989, 1990). Of the other eleven positive clones, four clones contained CYP2A7, two clones containedCYP2A13, and five clones contained unknownCYP2A-related sequences, respectively, judging from restriction maps, Southern blot analyses, and nucleotide sequence analyses.
Frequency of SacI andSphI-RFLPs.
Frequency of alleles identified bySacI- or SphI-RFLPs was investigated (Table2). In SacI-RFLPs, the frequency distribution of the A-type (9.8-kb, 8.9-kb, 6.4-kb, 4.9-kb, 4.5-kb, 4.3-kb, 4.0-kb, 3.2-kb, 2.6-kb, and 2.5-kb fragments), B-type (9.8-kb, 8.9-kb, 6.4-kb, 4.5-kb, 4.3-kb, 3.2-kb, 2.6-kb, and 2.5-kb fragments), C-type (9.8-kb, 6.4-kb, 4.9-kb, 4.5-kb, 4.3-kb, 4.0-kb, 3.2-kb, 2.6-kb, and 2.5-kb fragments), D-type (9.8-kb, 8.9-kb, 6.4-kb, 4.9-kb, 4.5-kb, 4.3-kb, 4.0-kb, 3.2-kb, and 2.5-kb fragments), and E-type (9.8-kb, 8.9-kb, 6.4-kb, 4.3-kb, 3.2-kb, and 2.5-kb fragments) were 50.0%, 26.6%, 18.5%, 1.6%, and 3.2%, respectively. InSphI-RFLPs, the frequency of the A′-type (7.1-kb, 6.3-kb, 5.5-kb, 4.0-kb, and 2.8-kb fragments), B′-type (7.1-kb, 5.5-kb, 4.0-kb, and 2.8-kb fragments), and C′-type (6.3-kb, 4.0-kb, and 2.8-kb fragments) were 55.6%, 41.1%, and 3.2%, respectively.
Determination of Human CYP2A6 Deletional Allele by PCR Analysis.
We examined whether the exons 1, 8, and 9 in theCYP2A6 gene were deleted in the three PMs using the PCR-RFLP method. Thus, we amplified exon 1 in CYP2A6 andCYP2A13 with primers 2A6-F3 and 2A6–1AS-2. The sequences of these primers were identical with both genes (Fig.3). The PCR products were digested withHaeIII. The homozygous wild-type or heterozygous type of exon 1 in the CYP2A6 gene yields 151-bp, 142-bp, 30-bp, 21-bp, and 12-bp fragments (Fig. 4A). On the other hand, the homozygous deletion-type gene yields 142-bp, 30-bp, and 12-bp fragments (Fig. 4A). These three fragments must be derived from the CYP2A13 gene. The genotyping for the deletion of exon 1 in the CYP2A6 gene is shown in Fig. 4D. Three PMs shown as 142-bp, 30-bp, and 12-bp fragments were assumed to possess the homozygous deletion of exon 1 in the CYP2A6 gene. Three EMs shown as 151-bp, 142-bp, 30-bp, 21-bp, and 12-bp fragments were regarded as a homozygous or heterozygous wild gene. In Fig. 4, B and C, typical examples of the homozygous and heterozygous wild- and the homozygous deletion-types of RFLPs in exons 8 and 9 are shown. The genotypes of the three PMs indicated that the genes from these subjects could be classified as the homozygous deletion-type of exons 8 and 9 (Fig. 4, E and F).
Frequency of CYP2A6 Gene Deletion.
CYP2A6 gene deletion (exon 1) was examined by the PCR-RFLP method. The frequency of the homozygous deletion of the CYP2A6 gene was investigated with 124 Japanese subjects except the three PMs and the three EMs used for gene analysis as mentioned above. Six of 124 Japanese (4.8%) carried the alleles of the homozygous deletion (Table3). Four subjects were E-type and two subjects were D-type. The relationship between the homozygous deletion of exon 1 in the CYP2A6 gene and the SacI- orSphI-RFLPs of CYP2A genes was investigated in 130 Japanese including the three PMs and the three EMs (Table4). All seven deletional alleles of the E-type were classified to the C′-type. Two subjects (D-type and A′-type) who showed low activities of SM-12502 S-oxidase and coumarin 7-hydroxylase (Nunoya et al., 1998) were classified into the homozygous deletion-type (Table 4). All of the others (A-, B-, and C-types) were judged as homozygous wild-type or heterozygous type. These results suggest that this PCR-RFLP analysis is a reliable method to detect the homozygous deletion allele in the CYP2A6 gene.
Discussion
In the three PMs, AUC andT1/2(β) of SM-12502 were markedly higher than those in the 25 EMs. The AUC andCmax of the S-oxide metabolite in the three PMs were lower. Thus, we postulated that large interindividual differences in SM-12502 S-oxidase activity were the main factors for the variation of the pharmacokinetics of this drug.
The cause of low SM-12502 S-oxidase activity was investigated, focusing on CYP2A6 gene mutation. All three PMs judged by an in vivo phenotyping of the SM-12502S-oxidation were classified as E- and C′-type in theSacI- and SphI-RFLPs, respectively. On theSacI-RFLPs, it had been reported that 4.9-kb and 4.0-kb bands were derived from the fragments of 8.9-kb (Wainwright et al., 1985), whereas these RFLPs did not relate to the CYP2A6 activities (Rautio et al., 1994). They reported the fragments as a 9.3-kb fragment (A1) and 5.2-kb and 4.1-kb fragments (A2); A1 or A2 was supposed to correspond to the 8.9-kb fragment (B-type) or 4.9- and 4.0-kb fragments (C-type) in this study, respectively. Referring to these reports, we postulated that A-, B-, and C-types of SacI-RFLPs did not correlate with SM-12502 S-oxidase activity. We reported that the D-type was associated with the low S-oxidase activity in in vitro metabolism of SM-12502 using human liver microsomes (Nunoya et al., 1998). The D-type was identified as the CYP2A6 gene deletion lacking the region from intron 5 to exon 9. It was suggested that the D-type deletion was different from the E-type judging from the presence of 6.4-kb and 4.5-kb fragments and the absence of the 2.6-kb fragment. The 6.4-kb fragment of SacI-D type, which contained exon 1 to intron 5 in the CYP2A6 gene, showed about a half intensity as compared with other subjects in a gel electrophoresis (Nunoya et al., 1998). The D-type and the E-type were also characteristic of the absence of exon 1 in the CYP2A6gene. For the reasons mentioned above, it seems reasonable to suppose that the D-type lacks the region from exons 1 to 9 in theCYP2A6 gene. The frequency of CYP2A6 gene deletion was 1.6% (2/124) and 3.2% (4/124) for the D-type and E-type in 124 subjects, respectively (Table 2). In Table 3, the frequency of homozygous deletion-allele in 124 Japanese was 4.8%. We have developed a PCR-RFLP method that distinguishes the D- and E-types from other types (Table 4). Accordingly, it is necessary to examine whether the D- or E-type accounts for all PMs.
In the present study, we could not obtain the CYP2A6gene from genomic DNA libraries from PMs. CYP2A6v1 andCYP2A6v2 genes reported by Yamano et al. (1990) andFernandez-Salguero et al. (1995) were not contained in clones that we obtained as CYP2A6-related ones. In Fig. 2B, clone 2A6–35 contained the 4.5-kb region downstream of exon 9 of theCYP2A6 gene. This clone showed a higher identity withCYP2A6 reported by Yamano et al. (1989) than that later reported by Yamano et al. (1990; data not shown). Therefore, we assume that unreported CYP2A6-related genes might be present.
CYP2A6 is a primary enzyme that catalyzes (S)-nicotine to (S)-nicotine Δ1′, 5′-iminum ion, which is further converted to (S)-cotinine by aldehyde oxidase (Cashman et al., 1992). Berkman et al. (1995) indicated that there is a 26-fold variability in the formation of (S)-cotinine in human liver microsomes. Although not examined yet, CYP2A6gene deletion may contribute to the variability in nicotine metabolism. It is also well known that CYP2A6 is capable of metabolically activating aflatoxin B1 (Aoyama et al., 1990),N-nitrosodiethylamine (Crespi et al., 1990), and the tobacco-specific nitrosamine 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (Crespi et al., 1991). Therefore, genetic defects in the CYP2A6 gene may also affect susceptibility to the procarcinogens in the environment. The genotyping procedures described here will be useful in clinical studies investigating the importance of the CYP2A6 gene polymorphism in the metabolism of various drugs and susceptibility to various disease states, such as chemically induced cancer.
Footnotes
-
Send reprint requests to: Dr. Tetsuya Kamataki, Ph.D., Division of Drug Metabolism, Faculty of Pharmaceutical Sciences, Hokkaido University, N12W6, Sapporo 060 Japan. E-mail:kamataki{at}pharm.hokudai.ac.jp
-
↵1 This study was supported in part by a Grant-in-Aid from the Ministry of Education, Science, Sports and Culture of Japan.
- Abbreviations:
- AUC
- area under the plasma concentration-time curve
- CYP
- cytochrome P-450
- EM
- extensive metabolizer
- PCR
- polymerase chain reaction
- PM
- poor metabolizer
- RFLP
- restriction fragment length polymorphism
- SM-12502
- (+)-cis-3,5-dimethyl-2-(3-pyridyl)thiazolidin-4-one hydrochloride
- Received April 21, 1998.
- Accepted December 6, 1998.
- The American Society for Pharmacology and Experimental Therapeutics