Abstract
To characterize the contribution of the human flavin-containing monooxygenase form 3 (FMO3) in the metabolism and disposition of drugs and xenobiotics, we determined the single nucleotide polymorphisms in the coding region and adjacent splice junctions of FMO3 in 134 African Americans and 120 Caucasians from the United States. In the regions examined, DNA resequencing or high throughput MassEXTEND studies coupled with mass spectrometric genotyping showed that 12 sites of variation were present. Three variants encoding synonymous mutations and four polymorphisms were observed in the noncoding region. Another three variants, Lys158-FMO3, Met257-FMO3 and Gly308-FMO3, previously reported in similar populations, were prominent polymorphisms. Two new polymorphisms, His132-FMO3 and Pro360-FMO3, were identified in this study. Both variants were found only in African Americans. To evaluate the effect of the amino acid substitutions on the function of FMO3, each amino acid substitution was introduced by site-directed mutagenesis into a wild-type FMO3 cDNA. Selective functional activity was studied with methimazole, trimethylamine, and 10-(N,N-dimethylaminopentyl)-2-(trifluoromethyl) phenothiazine. Both His132-FMO3 and Pro360-FMO3 variants were able to metabolize the substrates examined. Compared with wild-type FMO3, the His132-FMO3 was less catalytically efficient. The His132-FMO3 variant moderately altered the catalytic efficiency of FMO3 (decrease of 30%, 60% and 6% with methimazole, trimethylamine and 10-(N,N-dimethylaminopentyl)-2-(trifluoromethyl)phenothiazine, respectively). The Pro360-FMO3 variant was more catalytically efficient than wild-type FMO3. Pro360-FMO3 oxygenated methimazole, trimethylamine and 10-(N,N-dimethylaminopentyl)-2-(trifluoromethyl)phenothiazine, respectively, 3-, 5- and 2-fold more efficiently than wild-type FMO3. Based on the functional activity of the variant FMO3 enzymes, it is likely that population differences exist for compounds primarily metabolized by FMO3.
The flavin-containing monooxygenases (FMOs; E.C. 1.14.13.8) constitute a family of FAD-, NADPH-, and O2-dependent microsomal enzymes that catalyze the oxygenation of a large number of compounds, including drugs, pesticides, and industrial chemicals containing a soft nucleophil (Ziegler, 1988; Cashman, 1995). The FMOs oxygenate many nucleophilic xenobiotics producing more polar substances as a prelude to excretion. In a few cases, the FMOs have been shown to catalyze the formation of reactive metabolites capable of binding to cellular macromolecules (Mani et al., 1993; Cashman, 1995). In general, FMO can be considered a detoxication catalyst (Ziegler, 1988; Chiba et al., 1990). Several endogenous compounds such as trimethylamine (Lang et al., 1998), tyramine (Lin and Cashman, 1997), or cysteamine (Ziegler, 1988) have been identified as substrates for FMOs.
Currently, six forms of the FMO gene are known. But only evidence for five functional forms of human FMO exists (Hines et al., 2002). Homologous FMOs share 52 to 57% amino acid sequence identity. In the adult human liver, FMO3 is the prominent form and other FMOs are not expressed or are expressed at very low levels (Lomri et al., 1992). Human FMO3 is associated with the majority of FMO-mediated adult hepatic metabolism. Recent genetic studies have shown that polymorphisms of FMO3 gene loci may play a role in variation of the N-oxygenation of certain amines including drugs, dietary agents, and other xenobiotics (Kang et al., 2000; Cashman et al., 2001; Cashman, 2002; Park et al., 2002). Although not exhaustively established, a few examples showed the possible clinical implications of common FMO3 single nucleotide polymorphisms (SNPs). Because the frequency of some common FMO3 polymorphisms differ widely across ethnic groups, it is possible that interethnic variation contributes to therapeutic drug variability for drugs metabolized by FMO3. In addition to common polymorphisms, other very rare variants have been described that can significantly affect enzyme activity (Dolphin et al., 1997; Cashman, 2002). Some FMO3 mutations have been associated with trimethylaminuria, a rare metabolic disorder caused by the inability to N-oxygenate dietary-derived trimethylamine (TMA). The best understood form of trimethylaminuria is the primary genetic form associated with dysfunctional FMO3 arising from genetic mutations (Cashman et al., 2003). In addition, an acquired form possibly arising from hepatitis (Mitchell and Smith, 2001), childhood forms (Koukouritaki et al., 2002), a transient form associated with menstruation (Ayesh et al., 1993), and substrate or precursor overload or other impaired disease states can lead to trimethylaminuria. The work reported herein is largely related to the form of trimethylaminuria in which abnormal FMO3 enzyme activity is found when a coding region mutation or combination of mutations is present.
Because of the possible clinical involvement of SNPs in FMO3 function, it is necessary to identify the SNPs in the FMO3 gene, determine their frequencies, and characterize the effect of SNPs on functional FMO3 activity. In an attempt to predict the individual response to drugs and environmental toxicants, the major common FMO3 genetic variants in non-Hispanic Caucasians and non-Hispanic African Americans were examined in the present study. The relative frequencies of the major common polymorphisms of human FMO3 were examined in DNA samples from 120 Caucasians and 134 African Americans of the United States. SNPs were quantified by resequencing or by using a high-throughput mutation detection method combining MassEXTEND coupled with mass spectrometric measurement (Cashman et al., 2001). The prominent FMO3 variants observed were expressed as maltose-binding fusion proteins. Based on the activity of the variant FMO3 enzymes, variation in oxygenation of selective functional substrates suggests that population differences may exist for drugs or xenobiotics metabolized by FMO3.
Materials and Methods
Chemicals. All chemicals and reagents were purchased from Aldrich Chemical Co. (Milwaukee, WI) in the highest purity commercially available. The components of the NADPH generating system were obtained from Sigma-Aldrich (St. Louis, MO). Buffers and other agents were purchased from VWR Scientific, Inc. (San Diego, CA). The phenothiazine analog 10-(N,N-dimethylaminopentyl)-2-(trifluoromethyl)phenothiazine (5-DPT) was synthesized by the method previously described (Brunelle et al., 1997).
Genomic DNA and Study Populations. Genomic DNA was obtained from donors from several states in the United States. After institutional review board ethics approval, the frequencies of human FMO3 SNPs were experimentally determined from DNA obtained from independent individuals. The ethnicity of individuals was defined by a self-report questionnaire that indicated the race of both biological parents. Mouth swab samples were obtained from 63 Non-Hispanic African American individuals living in South Carolina (South Carolina group). Blood samples were obtained from 16 Non-Hispanic African American individuals visiting the Tutwiler Clinic in Tutwiler, Mississipi (Mississippi group). Five Non-Hispanic African American individuals and 28 Non-Hispanic Caucasian individuals were from California (HBRI group). Ninety-two Non-Hispanic Caucasians were from the National Institute of General Medical Sciences Human Genetic Mutant Cell Repository (Coriell Institute, Camden, NJ) and were comprised of individuals from the state of Utah (83), as well as French (4), Venezuelan (3), and Amish (2) individuals (Coriell group).
FMO3 Phenotyping. TMA is a selective functional substrate for human FMO3. Determination of urinary TMA and TMA N-oxide concentration is a convenient means to characterize the FMO3 phenotype in humans. In this study, two methods were used for determining TMA and TMA N-oxide: an NMR method and a mass spectrometric method. For the mass spectrometric determination of TMA and TMA N-oxide, the procedure previously described was used (Cashman et al., 2001). For the NMR method, the following protocol was used. Two milliliters of urine was placed into a sealed vial and refrigerated at 4°C. Preliminary studies indicated that no loss of TMA or TMA N-oxide occurred within 72 h postcollection under these conditions. TMA and TMA N-oxide were determined by measuring the concentrations in urine samples that were normalized against creatinine by proton NMR. This method has been used in earlier studies examining TMA and TMA N-oxide concentration in various biological matrices including urine (Hirama et al., 1993; Bedford et al., 1998; Murphy et al., 2000). Following transport to the laboratory, samples were acidified with HCl to pH 2.5. An internal standard of 2,2-dimethyl-2-silapentane-5-sulfonate (1 mmol) was added and the sample was analyzed by proton NMR using a 400-MHz Bruker instrument. Water suppression was required for samples, and 32 scans per sample were used to detect TMA and TMA N-oxide (detection limit = 0.2 nmol) in all individuals. Under these conditions, TMA N-oxide and TMA were identified as singlets at 3.2 and 2.9 ppm, respectively. Because each resonance is a singlet, integration of each peak was used quantitatively. Similar concentrations of 2,2-dimethyl-2-silapentane-5-sulfonate (that also possessed 9 protons at 0 ppm) were used to determine percentage of recovery. In all cases examined, the mean percentage of recovery was near 100%.
FMO3 Genotyping. Genomic DNA from various groups (i.e., Mississipi, South Carolina, and California) were extracted from blood or buccal brushes under standard conditions using a QIAGEN kit (QIAGEN, Valencia, CA). Individual exons and immediate flanking intronic regions were amplified from genomic DNA (50 ng) in the presence of specific primers (50 pmol) (Table 1) and Hotstart mix (QIAGEN). The amplification was done at 94°C for 5 min, 35 cycles of 94°C for 30 s, 58°C for 30 s, and 72°C for 2 min, followed by a final extension at 72°C for 5 min. The polymerase chain reaction products were fractionated, gel purified, and sequenced as described previously (Cashman et al., 2001). Sequencing was done for both forward and reverse strands and analyzed with Sequencher software (Gene Codes Corporation, Ann Arbor, MI) by procedures that could resolve heterozygotes under reliable quality-control conditions (Cashman et al., 2000). DNA from the Coriell group was examined for human FMO3 by high throughput chip-based matrix-assisted laser desorption/ionization time-of-flight mass spectrometry as previously described (Cashman et al., 2001).
Cloning and Expression. Wild-type human FMO3 and several variants including Lys158-FMO3, His132-FMO3, His132/Lys158-FMO3, Gly308-FMO3, Lys158/Gly308-FMO3, and Pro360-FMO3 were expressed as maltose-binding fusion proteins (i.e., MBP-FMO3). The wild-type human FMO3 and the variant Lys158-FMO3 were subcloned into the expression vector pMAL-c2 and used as templates to construct the other variants by directed mutagenesis, as described previously (Brunelle et al., 1997). Escherichia coli JM109 cells were transformed with pMAL-FMO3 plasmid and grown at 37°C in modified Hanahan's Broth (2% Bacto Tryptone, 0.5% yeast extract, 8 mM NaCl, 10 mM MgCl2, 2.5 mM KCl, 20 mM glucose) to an absorbance of 0.5 to 0.6 at 600 nm. Isopropyl β-thio galactopyranoside (0.2 mM), riboflavin (0.05 mM), and 100 μg/ml ampicillin were then added. The cells were incubated overnight at 30°C. Cells were harvested by centrifugation at 6000g for 10 min and resuspended in lysis buffer (50 mM Na2HPO4, pH 8.4, 0.5% Triton X-100) containing 0.2% l-α-phosphatidylcholine, 0.5 mM phenylmethylsulfonyl fluoride, and 40 μM flavin adenine dinucleotide. After an incubation for 30 min at 4°C, the resuspended cells were disrupted by sonication (i.e., three 2-min bursts separated by periods of cooling). The solution was centrifuged at 18,000g for 30 min at 4°C. The resulting supernatant was placed onto an amylose column (New England Biolabs, Beverly, MA). The pellets were extracted one additional time as described above. The resulting supernatant was loaded onto the same amylose column.
Purification of MBP-FMO3 Proteins. All purification procedures were carried out at 4°C. The resulting supernatants were applied (0.5 ml/min) to an amylose column (35 ml) equilibrated with buffer A (50 mM Na2HPO4, pH 8.4, 0.5% Triton X-100). The column was washed with five volumes of buffer A containing 40 μM flavin adenine dinucleotide. Bound proteins were eluted with 10 mM maltose in buffer A containing 40 μM flavin adenine dinucleotide. Eluted fractions were analyzed by SDS-polyacrylamide gel electrophoresis. The fractions containing the fusion protein were pooled and concentrated with Centriplus YM-30 (Millipore Corporation, Bedford, MA) for 4 to 5 h at 4000g at 4°C. The total protein concentration was measured by bicinchoninic acid assay (Pierce Chemical, Rockford, IL) using serum bovine albumin as a standard.
Determination of MBP-FMO3 Concentration by Immunoquantification. SDS-polyacrylamide gel electrophoresis and Western blots were performed as described by Laemmli (1970) and Towbin et al. (1979), respectively. MBPFMO3 was quantified by immunoquantification and compared with human FMO3 Supersomes (BD Gentest, Woburn, MA) as a standard. Briefly, MBP-FMO3 proteins and different quantities of standard FMO3 Supersomes (i.e., 250, 200, 150, 100, and 50 fmol) were resolved by electrophoresis on a 10% polyacrylamide gel under denaturing conditions and transferred to polyvinylidene difluoride membranes (Millipore Corporation). Immunoblots were incubated for 1 h with rabbit polyclonal anti-human FMO3 (1:5000) (BD Gentest). After washing, immunoblots were incubated with horseradish peroxidase-conjugated anti-rabbit IgG (1: 10,000). FMO3 was visualized by enhanced chemiluminescence on X-ray film. Quantification was done by densitometry analysis using a numeric camera employing Scion Image software (public domain, http://www.scioncorp.com/). The relationship between the relative intensity of the detected signal and FMO3 Supersome protein was linear (r2 = 0.97).
Enzyme Assays. The oxygenation of trimethylamine and methimazole was determined by monitoring the oxidation of NADPH associated with the N- or S-oxygenation of these substrates, respectively. The assay contained 50 mM sodium phosphate buffer (pH 8.4), 0.5 mM diethylenetriaminepentaacetic acid, 0.2 mM NADPH, and 0 to 0.1 mg of MBP-FMO3. Reactions were initiated by the addition of substrate and monitored at 340 nm after obtaining a stable baseline. Kinetic parameters for methimazole and TMA oxygenation were obtained from individual experiments after the addition of decreasing amounts of substrates (1000, 250, 50, 25, 12.5, 6.25, and 3.125 μM final concentration) to the sample cuvette.
FMO-dependent methimazole S-oxygenase activity was also determined according to the method of Dixit and Roche (1984). The S-oxygenation of methimazole was determined at 37°C in the presence of an NADPH-regenerating system (100 mM Tris-HCl buffer, pH 8.3, 0.1% Triton X-100, 0.5 mM NADP+, 5 mM glucose 6-phosphate, 1.5 U glucose-6-phosphate dehydrogenase) containing 0.06 mM 5,5′-dithiobis(nitrobenzoic acid), 0.02 mM dithiothreitol, and 0 to 0.1 mg of MBP-FMO3. The difference in absorbance between identical assay mixtures with and without methimazole was monitored at 412 nm for 3 min. Kinetic parameters for methimazole S-oxygenation were obtained from individual experiments after the addition of decreasing amounts of methimazole (1000, 250, 50, 25, 12.5, 6.25, and 3.125 μM final concentration) to the sample cuvette.
The N-oxygenation of 5-DPT was also determined as previously described (Brunelle et al., 1997). Briefly, a typical incubation mixture contained 10 mM potassium phosphate buffer (pH 8.4), 0.4 mM NADP+, 0.4 mM glucose 6-phosphate, 1 U of glucose-6-phosphate dehydrogenase, 0.8 mM diethylenetriaminepentaacetic acid, and 0 to 0.1 mg of MBP-FMO3. Reactions were initiated by the addition of 5-DPT, incubated at 37°C for 20 min, and stopped by addition of 4 volumes of cold dichloromethane. After addition of 20 mg of Na2CO3, the incubations were mixed and centrifuged to partition metabolites and the remaining 5-DPT into the organic fraction. The organic fraction was evaporated, dissolved in methanol, mixed thoroughly, centrifuged, and analyzed by HPLC as described previously (Brunelle et al., 1997). Kinetic parameters for 5-DPT N-oxygenation were obtained from individual experiments after the addition of decreasing amounts of 5-DPT (600, 300, 200, 150, and 127 μM final concentration).
Data Analysis. The estimation of kinetic parameters was achieved by the incubation of at least five different substrate concentrations with MBP-FMO3. Incubations were done in duplicate or triplicate. The ordinary least-squares criterion was used to fit the Michaelis-Menten model to the data, taking velocity as the dependent variable. The minimum values of the sum of squared residuals were computed using NAG foundation library routine EO4FDF (Numerical Algorithms Group Inc., Downers Grove, IL), a combined Gauss-Newton and modified Newton algorithm, using function values only. Data are presented as the mean ± standard deviation obtained from at least two separate experiments. Statistical analysis was done using StatView 5.0 software. Comparison of allelic frequencies between Non-Hispanic African-American and Non-Hispanic Caucasian populations was analyzed by using Fisher's exact test. Comparison between the catalytic efficiencies for two FMO3 variants was analyzed by using the Mann-Whitney U test.
Results
The human FMO3 genotype and allele frequency of the major common polymorphisms were determined in DNA samples obtained from populations of male and female non-Hispanic Caucasians and non-Hispanic African Americans. To identify novel SNPs, the coding sequences and splice junctions of FMO3 were analyzed in 134 Non-Hispanic African-American individuals and 120 Non-Hispanic Caucasian-American individuals, from various geographic locations. Two new SNPs, His132-FMO3 and Pro360-FMO3, were identified through this effort (Cashman and Zhang, 2002). The common polymorphisms at alleles 132, 158, 257, 308, and 360 were examined by either MassEXTEND (a high-throughput chip-based genotype variation detection method) with matrix-assisted laser desorption ionization time-of-flight mass spectrometry, previously described (Cashman et al., 2001), or by DNA resequencing.
Table 2 lists the relative allele frequencies for sites 132, 158, 257, 308, and 360 of the human FMO3. For the Lys158-FMO3 variant, Caucasians and African Americans had similar allelic frequencies (i.e., 0.53 and 0.50, respectively), and these values were in good agreement with previously reported values (i.e., 0.61 and 0.66, respectively) (Cashman et al., 2001). For Met257-FMO3, Caucasians and African Americans had distinct allele frequencies (i.e., 0.22 and 0.03, respectively, p < 0.001) compared with the previously reported values (i.e., 0.14 and 0.13, respectively). The common human FMO3 polymorphism Gly308-FMO3 had distinct relative frequency in Caucasians and African Americans (i.e., 0.22 and 0.06, respectively, p < 0.001) that was similar to the allele frequency reported before (i.e., 0.29 and 0.08, respectively).
We examined the phenotype of several individuals with variation at alleles 158 and 308. Generally, the urinary TMA to TMA N-oxide ratios for individuals with the Lys158-FMO3 variant were similar to that of individuals with wild-type Glu158-FMO3 (unpublished data). However, one heterozygotic Lys158/Gly308-FMO3 individual was available for phenotyping and gave a urinary TMA/TMA N-oxide ratio of 45:55 as determined by mass spectrometry.
Two new variants, His132-FMO3 and Pro360-FMO3, were identified in this study. The His132-FMO3 variant (i.e., change of an aspartic acid to a histidine) resulted from a G to T transition at nucleotide position 394. The allelic frequency of His132 in the African-American populations examined was 0.026 (Table 2). The His132 allelic frequency varied from 0 to 0.063 depending on the geographic location of the population. No homozygote for His132-FMO3 was found in this study (Table 3). Seven African-American individuals (from a total of 134 African Americans) were heterozygous for His132-FMO3. For two heterozygotic individuals, urinary concentration of TMA N-oxide and TMA was available. For one individual, the percentage of unmetabolized TMA in the urine (determined by NMR) was 0.1% and for the other it was 14%. The His132 substitution was not found among the Caucasian-American populations examined (Table 2).
The Pro360-FMO3 variant (i.e., change of a leucine to a proline) resulted from a T to A transition at nucleotide position 1079. The Pro360-FMO3 variant was observed only in individuals of African-American descent. The observed allelic frequency of Pro360-FMO3 in various African-American populations was presented in Table 2. The allelic frequency of Pro360-FMO3 among the 134 African-American individuals was 0.019. However, this frequency varied from 0 to 0.026 according to the geographic location of the population. Three heterozygotes and one homozygote for Pro360-FMO3 were found in this study among the 134 African-American samples examined. Unfortunately, no information about the phenotype of these individuals was available (Table 3).
Other previously reported human FMO3 variants were also observed in the populations examined (Cashman and Zhang, 2002). Thus, three silent mutations, S147S, N285N, and N302N, and other substitutions, deletions, and additions within introns were identified (results not shown). To date, no correlations of phenotype to genotype have been identified for these variants. Larger sample analysis will be required to understand the possible contribution of these variants to FMO3 pharmacogenetics.
To evaluate the effect of the amino acid substitutions, His132, Gly308, and Pro360, on the function of FMO3, each variant was introduced into a wild-type FMO3 or Lys158-FMO3 cDNA fused with MBP cDNA in the vector pMAL-2c. These constructs, along with a wild-type MBP-FMO3 construct as control, were over-expressed in E. coli JM-109, as described above. After affinity purification of the fusion-proteins, detailed kinetic studies were done. Preliminary studies showed that human MBP-FMO3 supplemented with NADPH catalyzed the rapid oxygenation of methimazole, TMA, and 5-DPT to their corresponding S- or N-oxide, respectively. The formation of the product was a linear function of protein concentration (0–0.1 mg of protein), with incubation time for at least 3 min. Kinetic constants for the oxygenation of substrates catalyzed by MBP-FMO3 and its variants were calculated from the rate of product formation at variable substrate concentration by the procedures described under Materials and Methods. The Km and Vmax values obtained for each variant are summarized in Tables 4, 5, 6, 7. As shown by the kinetic constants listed in Table 4, methimazole was an excellent substrate for the human wild-type MBP-FMO3. The apparent kinetic parameters were 11.6 μM and 58.6 nmol/min/nmol of enzyme for Km and Vmax, respectively (Table 4). To investigate the effect of allelic variation, the ability of each variant to S-oxygenate methimazole was examined by two different methods. Methimazole S-oxygenation was estimated by monitoring the oxidation of NADPH associated with the S-oxygenation of methimazole (Table 4) or by using the method of Dixit and Roche (1984) (Table 5). Two assays were used to verify that the methimazole S-oxide formed was not perturbing FMO3 function. The results obtained by both methods were in good agreement and indicated that each variant examined in this study was able to catalyze the S-oxygenation of methimazole.
Compared with wild-type FMO3, the prevalent polymorphic form of human FMO3, Lys158-FMO3 S-oxygenated methimazole with a similar Km, but a lower Vmax. The other polymorphic variant of human FMO3, Gly308-FMO3, showed significant differences in the kinetic parameters for methimazole S-oxygenation, with higher Km values and a lower Vmax. For Gly308-FMO3, the Vmax/Km ratio was 9-fold lower than the ratio obtained for wild-type FMO3. The double mutant Lys158/Gly308-FMO3 showed results similar (i.e., 8.5-fold lower for the Vmax/Km ratio) to those obtained for the single variant Gly308-FMO3.
The variant His132-FMO3 identified in this study was also able to catalyze the S-oxygenation of methimazole. The apparent kinetic parameters, 15.2 μM and 49.7 nmol/min/nmol of enzyme for Km and Vmax, respectively, were similar to those obtained for wild-type FMO3. However, the catalytic efficiency, determined by calculating Vmax/Km, showed that His132-FMO3 lost approximately 58% of the catalytic efficiency. The double mutant His132/Lys158-FMO3 afforded an enzyme with a lower Vmax (3-fold), and a loss of 75% of its catalytic efficiency (Tables 4 and 5).
In contrast to His132-FMO3, the Pro360-FMO3 variant increased (+215%) the apparent catalytic efficiency of MBP-FMO3 (Table 4). Whereas the Km values for Pro360-FMO3 were similar to those observed for wild-type FMO3, the Vmax for Pro360-FMO3 was 3-fold higher than that of the wild-type MBP-FMO3. The results obtained for the S-oxygenation of methimazole by Pro360-FMO3 (using the method of Dixit and Roche, 1984) were in good agreement and showed that the catalytic efficiency was increased 184% (Table 5).
TMA (Table 6) and 5-DPT (Table 7) were used to investigate the ability of each variant to catalyze the N-oxygenation of the tertiary amine to its N-oxide. TMA is an excellent substrate for human wild-type MBP-FMO3. The apparent kinetic parameters were 8.2 μM and 39.8 nmol/min/nmol of enzyme for Km and Vmax, respectively (Table 6). Compared with TMA, 5-DPT was less efficiently N-oxygenated by wild-type MBP-FMO3. The apparent kinetic parameters were 929 μM and 64.6 nmol/min/nmol of enzyme for Km and Vmax, respectively (Table 7). Each variant examined in this study was also able to catalyze the N-oxygenation of TMA and 5-DPT. Compared to wild-type MBP-FMO3, the catalytic efficiency to N-oxygenate TMA and 5-DPT by Lys158-FMO3 was significantly decreased (–60%). The catalytic efficiency to N-oxygenate TMA (–94%) and 5-DPT (–80%) of Gly308-FMO3 was severely decreased. For both TMA and 5-DPT, the double mutant Lys158/Gly308-FMO3 afforded a loss of catalytic efficiency similar to that of the Gly308-FMO3.
The catalytic efficiency of the His132-FMO3 to N-oxygenate 5-DPT was similar to that of wild-type MBP-FMO3 (Table 7) despite a lower Vmax. For TMA N-oxygenation (Table 6), the Vmax/Km ratio for His132-FMO3 was 3-fold lower than the Vmax/Km ratio for wild-type FMO3. The decrease was due to an increase in the Km values (i.e., 2-fold increase), suggesting a decrease in the affinity of the enzyme for TMA. However, the double mutant His132/Lys158-FMO3 gave a catalytic efficiency similar to that of the single mutant His132-FMO3 for the N-oxygenation of TMA or 5-DPT compared with the Lys158-FMO3.
In agreement with the S-oxygenation of methimazole (Tables 4 and 5), Pro360-FMO3 increased the catalytic efficiency of the enzyme for N-oxygenation of TMA (+348%) and 5-DPT (+88%). Although the affinity of the enzyme for TMA and 5-DPT was unchanged, the Vmax for Pro360-FMO3 was 6- and 2-fold higher, respectively, than the Vmax obtained for wild-type FMO3.
Discussion
FMO3 is the prominent form of FMO present in adult human liver and contributes to the oxygenation of nucleophilic heteroatom-containing chemicals, drugs, and endogenous substances. The dietary material TMA, arising from choline and other sources, is exclusively N-oxygenated by human FMO3 and provides a selective functional biomarker of FMO3 enzyme activity. In humans, it was noted that abnormal TMA metabolism was directly correlated with genetic variation of the FMO3 gene (Treacy et al., 1998; Akerman et al., 1999; Dolphin et al., 2000). Variation in the human FMO3 gene arises from SNPs that can lead to alteration of enzyme function (Cashman, 2002). In unaffected individuals, TMA is efficiently N-oxygenated (i.e., typically at least 97%) to the polar and nonodorous metabolite TMA N-oxide. Individuals with defective TMA metabolism are afflicted with trimethylaminuria and excrete large amounts of unmetabolized and odorous TMA. Human FMO3 mutations that cause severe trimethylaminuria are rare and affect very few individuals. On the other hand, more common allelic variation could also influence metabolism of TMA and other chemicals and affect a much larger population. Thus, it is of fundamental interest to identify and characterize the common allelic variations of the FMO3 gene. Because the frequencies of some common FMO3 SNPs have already been identified and vary according to ethnicity, we analyzed the FMO3 gene from non-Hispanic African-American and non-Hispanic Caucasian individuals. From these populations, the distribution of five variants (i.e., Lys158-FMO3, Met257-FMO3, Gly308-FMO3, His132-FMO3, and Pro360-FMO3) was characterized.
The common allelic variants previously described, Lys158-FMO3 (Treacy et al., 1996), and Met257-FMO3 and Gly308-FMO3 (Treacy et al., 1998), were found in both ethnic groups. But the presence of a large ethnic difference in the frequency distribution of Met257-FMO3 and Gly308-FMO3 was found. Met257-FMO3 and Gly308-FMO3 variants were more frequent in Caucasians than in African Americans (p < 0.001). These results are in agreement with previous studies (Cashman, 2002; Park et al., 2002). In the study herein, the two new variants His132-FMO3 and Pro360-FMO3, which were previously reported but not completely characterized (Cashman and Zhang, 2002), were found only in African Americans and not in Caucasians. However, it is possible that the screening of the FMO3 gene from much larger populations might show the presence of these variants in Caucasian individuals, but we estimate this frequency at less than 0.001. This result complements previous studies that showed the frequencies of human FMO3 SNPs varied widely across ethnic groups (Cashman et al., 2000, 2001). Ethnic group-related differences in human FMO3 SNPs is notable because it may suggest that FMO3-dependent drug or chemical metabolism may vary as a function of ethnic group.
TMA, methimazole, and 5-DPT are efficiently oxygenated by human FMO3, as shown by the kinetic constants presented in Tables 4, 5, 6, 7. One approach to investigating the contribution of individual (or combinations of) amino acid residues in enzyme function is to compare the catalytic efficiency (Vmax/Km) of wild-type FMO3 with common variants.
The Lys158-FMO3 and Gly308-FMO3 variants were already characterized previously (Cashman et al., 1997; Treacy et al., 1998), but the effect of the combination of both amino acid changes on the metabolism of methimazole, TMA, and 5-DPT has not been previously reported. In the study herein, we observe moderate to severe effects on catalytic efficiency with both single variants, Lys158-FMO3 and Gly308-FMO3, as well as the double variant Lys158/Gly308-FMO3. The Lys158-FMO3, Gly308-FMO3, and Lys158/Gly308-FMO3 variants have been previously characterized using ranitidine and thiobenzamide as substrates (Park et al., 2002), and the report indicated that a significant reduction in FMO activity was only observed when both amino acid variations occur in cis. It is possible that the effect of variation at codon 158 and 308 on FMO activity is substrate-dependent.
The His132-FMO3 variant is a less efficient FMO3, as shown by the results of functional substrate studies (Tables 4, 5, 6, 7). The His132-FMO3 variant results from an aspartic acid (containing a negatively charged side group) to a histidine (containing a positively charged side group) substitution. This change might lead to a different conformation of the FMO3. The aspartic acid in position 132 of the amino acid sequence is not a conserved residue between FMO3 of different species and between the different FMOs (Fig. 1). Indeed, a histidine in this same position is present in mouse FMO3 (Falls et al., 1997). Moreover, it was reported that mouse FMO3 as well as other FMO3s from different species are able to metabolize methimazole and TMA (Falls et al., 1997). The result for mouse FMO3 is in agreement with the results obtained herein, showing a limited loss of catalytic efficiency (i.e., 30, 60, and 6% with methimazole, TMA, and 5-DPT, respectively).
Recently, two African-American individuals were identified as possessing the His132-FMO3 variant (Cashman et al., 2003). Neither individual complained of trimethylaminuria. One homozygous individual was available for phenotyping and, after choline challenge, showed a decreased ability to metabolize TMA (i.e., 27% unmetabolized TMA). Another individual who was homozygous for the His132-FMO3 variant and also had a Gly308-FMO3 polymorphism revealed moderate to severe trimethylaminuria (i.e., 53% unmetabolized TMA) after choline challenge (Cashman et al., 2003). Apparently, His132-FMO3 alteration decreases but does not abrogate human FMO3 function.
The Pro360-FMO3 variant is more catalytically efficient than wild-type FMO3. The Pro360-FMO3 is able to oxygenate methimazole, TMA, and 5-DPT 3-, 5-, and 2-fold more efficiently, respectively, than wild-type FMO3. The Pro360-FMO3 results from a leucine to a proline substitution in position 360 of the amino acid sequence. Proline is often associated with structural changes in proteins, and the structure of Pro360-FMO3 might be modified. We examined the conservation of this residue across species and isoforms (Fig. 1). Leucine, in position 360 of the FMO3 amino acid sequence, is not a conserved residue among FMO3 of different species and between the FMO isoforms, even though the region (amino acid 352–363) is fairly conserved. For all the substrates examined, Pro360-FMO3 provided Km values similar to those obtained for wild-type FMO3, but the observed Vmax values were always higher (i.e., 3-, 5-, and 2-fold for methimazole, TMA, and 5-DPT, respectively). Amino acid residue 360 could be facilitating the dehydration of the FMO3 hydroxyflavin or desorption of product or NADP+ (thought to be rate-limiting steps) consistent with an accelerating effect on Vmax but not on Km. One interpretation is that the FMO3 substrate binding domain may be distinct and functionally different from the catalytic region. A similar conclusion was reached in studies of the stereoselective action of FMO when it was proposed that a secondary site of substrate interaction within the protein was an important determinant for substrate oxygenation (Cashman, 1995). The Pro360-FMO3 variant is the first FMO3 variant reported to have markedly enhanced catalytic efficiency. The relevance of this FMO3 variant in drug metabolism and ethnic group-related drug efficacy needs to be further explored.
In summary, the use of DNA resequencing and high-throughput MassEXTEND methodology to genotype populations of Caucasian and African-American descent has provided insight into the relative allelic frequencies of human FMO3. African Americans possess a novel fast metabolizer phenotype (i.e., Pro360-FMO3) and a slow metabolizer phenotype (i.e., His132-FMO3) that is absent in Caucasians. The genetic variation described herein may contribute to the ethnic group-dependent metabolism of chemicals and dietary materials that are FMO3 substrates.
Acknowledgments
We thank Katrin Matschke and Dennis Schade for technical assistance. We also thank George Preti and Steven Fakharzadeh (University of Pennsylvania) for describing their data concerning the His132-FMO3 variant (referenced in Cashman et al., 2003). We also thank Dr. Ann Brooks and Kristi Johnson of the Tutwiler Clinic for their help with obtaining clinical samples. We thank Andreas Braun of Sequenom Inc., (San Diego, CA) for the MassEXTEND data.
Footnotes
-
↵1 Abbreviations used are: FMO, flavin-containing monooxygenase; SNP, single nucleotide polymorphism; TMA, trimethylamine; 5-DPT, 10-(N,N-dimethylaminopentyl)-2-(trifluoromethyl) phenothiazine; MBP, maltose-binding protein.
-
The financial support of the National Institutes of Health (Grant DK59618) and the California State Tobacco Related Disease Program (Grant 9RT-0196) is gratefully acknowledged.
- Received January 13, 2003.
- Accepted March 24, 2003.
- The American Society for Pharmacology and Experimental Therapeutics