Abstract
Sequence polymorphisms in enzymes involved in drug metabolism have been widely implicated in the differences observed in the sensitivity to various xenobiotics. The flavin-containing monooxygenase (FMO) gene family in humans catalyzes the monooxygenation of numerous N-, P- and S-containing drugs, pesticides, and environmental toxicants. Six genes (FMO1–6) have been identified so far, but the major alleles of FMO2 and FMO6 encode nonfunctional proteins due to a nonsense mutation and splice-site abnormalities, respectively. Data on structural variants exist for humanFMO2 and 3, whereas very little is known about the other FMO genes. FMO1–6 were scanned in 50 individuals of African-American descent using the method, detection of virtually all mutations-single-strand conformational polymorphism. A total of 49 sequence variants were identified in a total 1.35 megabases of scanned sequence, of which 29 were variants affecting protein structure or expression. Some of these are expected to affect the activity of the protein, including a nonsense mutation in FMO1 (R502X) and missense mutations in FMO1 (I303T), FMO4 (E339Q), and FMO5 (P457L) that occur in highly conserved amino acids. Additional deleterious substitutions in FMO2 (del337G) and FMO6 (Q105X) were also identified. Multiple structural variants in the FMO gene family were observed in this African-American sample. Some of the substitutions identified in this study might be useful markers in future association studies assessing sensitivity to environmental toxicants and common disease.
The flavin-containing monooxygenase gene family in humans catalyzes the monooxygenation of numerous N-, P- and S-containing xenobiotics (Cashman, 1995). Most of the xenobiotics are transformed into more soluble compounds that are readily excreted or further metabolized by phase II enzymes, but in some cases, FMOs1 can also catalyze the formation of reactive metabolites.
There are currently six FMO genes identified. The FMO gene family probably arose as a duplication of an ancestral gene. Human FMOs share 55 to 60% amino acid identity, with the exception of FMO3 and FMO6, which share 71% identity. The genes are located on the long arm of chromosome 1 (Shephard et al., 1993; McCombie et al., 1996) and the translation products are 532 to 558 amino acids in length. Highly conserved amino acid residues, corresponding to FAD and NADPH binding domains, are located in residues 4 to 32 and 186 to 213, respectively. Each isoform is expressed in a highly tissue- and ontogenic-fashion, and significant interindividual levels of expression (2–20X) are also observed (Overby et al., 1997; Koukouritaki et al., 2002).
Although cytochrome P450 (P450) is the primary oxidative phase I pathway in humans, increasing attention has been directed at determining the role of FMOs in xenobiotic metabolism given the prevalence of N-and S-containing drugs and environmental toxicants. It is likely that the role of FMOs in drug metabolism is underestimated. Some of the drugs identified as FMO substrates include frequently prescribed drugs such as cimetidine, ketoconazole, ranitidine, perazine, and tamoxifen (Cashman et al., 1993; Chung et al., 2000; Hodgson et al., 2000). Environmental toxicants of concern due to potential bioactivation by FMO include numerous thioether-containing pesticides (Hajjar and Hodgson, 1980;Ziegler-Skylakakis et al., 1998). A few endogenous compounds for FMOs have also been identified, but the precise physiological role of each enzyme of the family is unclear (Elfarra, 1995).
Single nucleotide polymorphisms (SNPs) in genes encoding both phase I and II xenobiotic-metabolizing enzymes are responsible for a different clinical response in individuals harboring these variations, following exposure to certain drugs and chemicals. This may be due to a change in the expression level, activity, or substrate specificity of the enzyme. P450 is the primary oxidative phase I pathway for the metabolism of drugs in humans, and certain polymorphisms in the CYP2D6 isoform are associated with adverse drug reactions (Agundez et al., 2001). Polymorphisms in phase II enzymes associated with adverse effects also have been described (Burchell and Hume, 1999). Many of these variants differ with ethnicity (Weber, 1999). Frequencies of certain polymorphisms in FMO2 and FMO3 are also found to vary dramatically with ethnicity (Whetstine et al., 2000; Cashman et al., 2001). A premature stop codon (1414T) in FMO2, removing the last 64 amino acids from the C-terminus, renders the protein inactive in Asians and white persons, but 26% of the African-American population have the allele (1414C) encoding the full-length protein (Dolphin et al., 1998; Whetstine et al., 2000). Current data on FMO6 indicates that it is nonfunctional in most of the population due to alternative splicing events leading to skipping of exons and a protein that is not being translated (Hines et al., 2002).
The goal of the current investigation is to explore the variability found in the FMO gene family in African Americans. With the exception of FMO2 and FMO3, little is known about sequence variants in the FMO genes. Such knowledge may be important in future efforts to predict the response to pharmaceuticals and environmental toxicants. Mutation scanning using a robotically enhanced scanning method that detects virtually all mutations (Liu et al., 1999; Buzin et al., 2000) has been employed to detect sequence variation in the complete coding and splicing-regions of the six genes in the human FMO gene family.
Materials and Methods
Patient Samples.
DNA from blood of fifty unrelated African-American individuals (AA1-AA50) were extracted with phenol chloroform (Gustafson et al., 1987). The work was approved by the City of Hope's Human Subjects Use Committee.
PCR Amplification and DOVAM-S.
DOVAM-S was used to scan for sequence variation. DOVAM-S is a robotically enhanced, highly redundant modification of SSCP that detected 240 of 240 sequence changes in three blinded analyses (180 different mutations) (Liu et al., 1999; Buzin et al., 2000).
Genomic DNA at a concentration of 40 ng/μl were used to generate 61 separate PCR segments labeled with 33P, which included all coding exons of FMO1–6, including the splice junctions. Promoter or 5′ and 3′ untranslated regions were not scanned. The PCR segments ranged from 180 to 373 nucleotides and were verified individually by agarose gel electrophoresis. The PCR primers were designed from Genbank accession number NT_029874 (Tables1 and 2). All PCR-cycling conditions were 94°C for 10 min, 35 cycles, 94°C for 15 s, 55°C for 30 s, 72°C for 60 s, and a final step of 72°C for 10 min. Amplification by PCR, and pooling of segments, were carried out on an ABI Prism 877 robotically controlled thermocycler (Applied Biosystems, Foster City, CA). The pooled segments were first electrophoresed on a denaturing PAGE/sequencing gel to identify unamplified segments or detect any small deletions or insertions. Reamplification was attempted on missing bands, and the nondenaturing gels were electrophoresed with five different conditions: I:Pageplus/tricine/triethanolamine, II:HR1000/tricine/triethanolamine/4°C, III:Pageplus/TBE/5% glycerol, IV:HR1000/TBE/2.5% glycerol/4°C, V:Pageplus/CAPSO/4°C/no light. Electrophoresis was carried out on a 50 cm × 38 cm × 0.4 mm-sequencing gel at a constant power of 15W for 12 to 16 h.
Control lanes containing two to four labeled segments were included to ease the identification of bands and band shifts. Following electrophoresis, gels were dried and exposed to BioMax MR film (Eastman Kodak, Rochester, NY) for at least 2 days. Segments with altered mobility were re-amplified, purified, and sequenced in both directions with the ABI model 377 DNA sequencer (Applied Biosystems). The sequence-chromatograms were analyzed using the Gene Codes Sequencher software package from (Gene Codes Corp., Ann Arbor, MI). Sequence changes were confirmed by an independent PCR and sequencing in the opposite direction. When a common polymorphism was observed, a minimum of three individuals and all individuals without the polymorphism in at least one of the two flanking lanes were sequenced.
Evolutionary Conservation Analysis.
The software package ClustalW v1.7 and Blast (http://www.ncbi.nlm.nih.gov/BLAST/) were used to analyze the degree of amino acid residue conservation. The following species and Genbank files (in parenthesis) were used for evolutionary relationship studies:Mus musculus (FMO1: U87456, FMO2: AF184981, FMO3: U87147, FMO4: AF461145, FMO5: U90535), Rattus norvegicus (FMO2:AF458414, FMO3: AF286595, FMO4: AF458416, FMO5: AF458413), Macaqa mulatta (FMO2: U59453, FMO3: AY063498), Canis familiaris (FMO1: AF384053, FMO3: AF384054), Cavia porcellus (FMO2: L10037, FMO5: L37081), Gallus gallus(FMO3: AJ431390), Oryctolagus cuniculus (FMO1: M32030, FMO2:M32029, FMO3: L10391, FMO4: L10392), Drosophila melanogaster(FMO-1: AF405243, FMO-2: AF432229), Arabidopsis thaliana(Putative FMO: AY113858), Caulobacter crescentus (Putative FMO: NC_002696), Schizosaccharomyces pombe (Putative FMO: NC_003423).
Results
In this study, the coding sequences and splice junctions ofFMO1–6 were analyzed in 50 individuals of African-American origin using DOVAM-S. In FMO1-6, 2740, 2830, 2784, 2650, 2558, and 2986 nucleotides were scanned, respectively. One hundred alleles per gene were scanned for a total of 1.35 Mb of DNA. Of the 49 sequence variants identified, 29 were VAPSEs, including 2 nonsense, 1 deletion, 3 insertion, and 22 missense changes (Tables3-8). With the exception of four frequent changes in FMO3 (E158K and V257M) and FMO2 (X472Q and Ins1589T), all the VAPSEs were novel. Mutations disrupting the highly conserved NADPH- or FAD-binding sites were not observed. The remaining variants were either silent codon changes or intronic sequence changes not expected to affect the splicing of the mRNA species. On average, almost eight SNPs were identified per gene and one SNP per 290 base pairs.
Discussion
This study confirms the relatively large genetic variation often found in enzymes involved in the metabolism of xenobiotic compounds (Board et al., 1998; Bertilsson et al., 2002). Interindividual variation in the activity of drug-metabolizing enzymes due to single-nucleotide substitutions, or insertions/deletions, is well documented (Hassett et al., 1994; Dolphin et al., 1997; Tang et al., 2001). In many cases, altered clinical responses or altered susceptibility to various chemicals due to these sequence variants have also been observed (Agundez et al., 2001). The nonsense mutations are highly likely to eliminate function and some of the 20 novel missense mutations may alter function. Expression studies may be helpful in identifying absent or altered enzymatic activities. Some of the novel changes may produce biological effects not detected by the simple in vitro assays. Ultimately, large epidemiological studies or analysis of transgenic mouse models are the most powerful ways to determine genotype to phenotype relationships. The FMO family is believed to contribute importantly to the oxygenation of various N-andS-containing chemicals, including drugs and pesticides, but FMOs may be one of the least characterized of the major xenobiotic-metabolizing enzymes in terms of sequence variants and corresponding phenotypes and has yet to be associated with any xenobiotic-related adverse effects.
FMO1 is the prevalent hepatic FMO in most animal species, except humans. Expression of FMO1 in humans is observed mainly in the fetal liver, adult kidney, and to a lesser extent, the intestine (Dolphin et al., 1996; Yeung et al., 2000). This distinct expression pattern and levels of FMO1 present in the fetal liver suggest that it may play an important role in pediatric drug disposition. It is speculated that FMO1 may play a role in development of the fetus (Koukouritaki et al., 2002).
One individual had a premature stop codon in FMO1. The nonsense mutation, R502X, occurs near the end of the protein, but it is almost certain to be deleterious because R502 is conserved among all 26 mammalian FMO sequences as well as chicken (Table9). In rabbit FMO2, the last 26 amino acids are not critical for catalytic activity nor membrane association (Lawton and Philpot, 1993), but the human truncated FMO2 (64 a.a.) is catalytically inactive (Dolphin et al., 1998). Catalytic activity of the truncated FMO1 awaits further expression and characterization studies of the protein. The pattern of FMO1 expression and proposed role in the fetus makes it an important target for further analysis.
Since most drugs freely cross the placenta into the fetal circulation, the fetus is exposed to maternally administered drugs, often in high concentrations (Mihaly and Morgan, 1983). The ability of the fetus to tolerate this exposure is poorly understood. This also applies to maternal exposure to environmental compounds. The total fetal P450 content is dominated by CYP3A7 (up to 50%) (Wrighton and Vandenbranden, 1989), but surprisingly little is known about the presence and activity of the other phase I enzymes in human fetal liver. Activity data suggests that human FMO1 has a wide substrate specificity and substantial catalytic activity (Yeung et al., 2000) and thus may be important in pediatric drug disposition and fetal xenobiotic clearance in general. Whether deleterious mutations affect basic drug or endogenous metabolism awaits further elucidation of the function of FMO1 in vivo. Deleterious mutations in FMO1 will most likely have different effects in the human fetus and adult due to the expression pattern observed.
Two different missense alterations occurred at a frequency of 4% each at I303. This residue is isoleucine in the four other mammalian FMO1 sequences, valine in the remaining 21 mammalian sequences of FMO2 through 6 (Table 9), and isoleucine or valine in chicken,Drosophila, Arabidopsis, Schizosaccharomyces, or Caulobacter. Thus I303T occurs at a residue that is either I or V for more than three billion years of evolutionary divergence and thus is likely to be deleterious. I303V may be a neutral polymorphism or it may affect activity if isoleucine is specifically required for the integrity of FMO1. To assess the latter possibility, it would be helpful to haveFMO1 sequences from more species, especially for nonmammals.
Liver FMO3 is the major human adult isoform and thought to be responsible for most of the FMO-mediated xenobiotic metabolism in the liver (Lomri et al., 1992; Phillips et al., 1995). Expression of FMO3 occurs postpartum. Sequence variation in FMO3 has been studied extensively because some of the substitutions identified are associated with abnormal metabolism (Treacy et al., 1998). Trimethylaminuria, or “fish-odor” syndrome is a disorder in which the individual excrete massive amounts of trimethylamine, and it is associated with lack of FMO3 activity; FMO3 is responsible forN-oxidation of trimethylamine to the nonodorousN-oxide. FMO3 has also been associated with the metabolism of commonly used pharmaceuticals (Cashman et al., 1993; Chung et al., 2000), but modulation of therapeutic efficacy has not been observed consistently in individuals with reduced FMO3 activity. None of the rare FMO3 genotypes that are associated with trimethylaminuria were observed in this investigation.
The major brain isoform, FMO4, is distinguished from the other FMO genes by the location of its stop codon, which makes it 20 to 25 amino acids longer than the other FMO proteins. Expression studies of the extended human protein (558 a.a.) have been difficult because of problems with translation due to the extended 3′-region, but the activity of the bacterially-expressed truncated protein demonstrates a limited activity toward the prototypic FMO substrate, methimazole (Itagaki et al., 1996). E339Q occurs at a residue that is either E or D in all mammalian FMO sequences (Table 9), suggesting that this variant may affect FMO4 activity.
FMO5 demonstrates the most limited substrate specificity of the gene family. Methimazole is only metabolized to a moderate extent (1/5000 the activity of FMO3) (Overby et al., 1995), and its involvement in drug-metabolism is probably not significant. The low affinity of FMO5 for xenobiotics is also species-independent. FMO5 is currently the only human FMO that has been demonstrated to be under hormonal regulation (Miller et al., 1997); FMO5 mRNA sequences are up-regulated in breast cancer cells expressing progesterone receptor of the B-isoform, treated with the synthetic progestin R5020. Gender-specific regulation of other FMO genes has frequently been observed in other species (Falls et al., 1997) but not in humans. One male displayed a Pro > Leu to substitution at position 457 in FMO5, which is a highly conserved residue in every FMO (regardless of isoform) found in mammals (Table9), suggesting it might be deleterious.
The FMO2 1414C allele was found in 9 (18%) of the 50 individuals, but additional variants were also found. A deletion of the 337G nucleotide in FMO2 is present in 27% of the alleles examined. The following changes in conserved amino acids were also found: D36G (5%), S195L (51%), and R238Q (2%). Of the nine individuals who would likely express a full-length protein, only three had alleles that did not include additional substitutions. Thus, the projected 26% of African Americans with functional FMO2 may be lower due to these additional substitutions. In FMO6, a nonsense variation, Q105X, was identified in 66% of the alleles examined, potentially leading to a truncated protein if spliced correctly. None of the deleterious variants in FMO2 or FMO6 were found in either chimpanzee or gorilla (data not shown).
Finally, nondeleterious genetic variants may combine to affect the activity of the enzymes even though they are individually not in conserved residues, e.g., one individual may be a compound heterozygote or contain a protein with both the FMO1 H97Q and I303V (AA2), and another individual has both the FMO3 E158K and E362Q alleles (AA42). Unfortunately, the family members are unavailable for examining segregation of the alleles. Further analysis is required to determine whether combinations of these mutations are associated with modulating activity of the protein and if there are any phenotypic effects of these variants. The individuals with the highly conserved VAPSEs in FMO1, FMO3, FMO4, and FMO5 are all heterozygotes. This is in contrast to FMO2 and FMO6 where a high percentage of individuals are homozygous for the deleterious variants. The importance of having one versus two VAPSEs depends on which allele is expressed and whether the FMO isoform is involved in bioactivation or detoxification of a particular xenobiotic. The clinical relevance of the polymorphisms observed depends on a number of factors including whether the parent compound or the metabolite(s) are formed or metabolized by the FMO, and whether the parent compound or metabolite are reactive and detrimental to the cell. Considering the highly overlapping substrate specificity for many of the P450s and FMOs, the overall contribution of polymorphic enzymes to the metabolic pathway of xenobiotics must also be carefully investigated. The variants described may contribute to a more precise understanding of possible correlations between genotypes and disease-susceptibility phenotypes or risk for side effects from drugs.
In summary, 45 novel VAPSEs were identified in FMO1–6. Some of the substitutions identified in this investigation may be used as specific markers for further association studies involving diseases and sensitivity to pharmaceuticals and environmental toxicants.
Acknowledgments
We thank Kathleen Hill and Qiang Liu for helpful comments.
Footnotes
- Abbreviations used are::
- FMO
- flavin-containing monooxygenase
- P450
- cytochrome P450
- SNP
- single nucleotide polymorphism
- DOVAM-S
- detection of virtually all mutations-single-strand conformational polymorphism
- PCR
- polymerase chain reaction
- PAGE
- polyacrylamide gel electrophoresis
- VAPSE
- variants affecting protein structure or expression
- TBE
- Tris borate-EDTA
- CAPSO
- 3-(cyclohexylamino)-2-hydroxy-1-propanesulfonic acid
- a.a.
- amino acids
- Received August 20, 2002.
- Accepted November 5, 2002.
- The American Society for Pharmacology and Experimental Therapeutics