Flavinylation of monoamine oxidase B.

Monoamine oxidase B (MAO B) catalyzes the oxidative deamination of biogenic and xenobiotic amines. The oxidative step is coupled to the reduction of an obligatory cofactor, FAD, which is covalently linked to the enzyme at Cys. In this study, we developed a novel riboflavin-depleted (Rib) COS-7 cell line to study the flavinylation of MAO B. ApoMAO B can be obtained by expressing MAO B cDNA in these cells. We found that MAO B is expressed equally in the presence or absence of FAD and that apoMAO B can be inserted into the outer mitochondrial membrane. Flavinylation of MAO B was achieved by introducing MAO B cDNA and different flavin derivatives simultaneously into Rib COS-7 cells via electroporation. Since the addition of riboflavin, FMN, or FAD resulted in equal levels of MAO B activity, we conclude that the flavin which initially binds to apoMAO B is FAD. In our previous work, we used site-directed mutagenesis to show that Glu in the dinucleotide-binding motif of MAO B is essential for MAO B activity, and we postulated that this residue is involved in FAD binding. In this study, we tested the role of residue 34 in flavin binding by expressing wild-type or mutant MAO B cDNA in Rib COS-7 cells with the addition of [14C]FAD. We found that Glu is essential for both FAD binding and catalytic activity. Thus, FAD binds to MAO B in a dual manner at Glu noncovalently and Cys covalently. We conclude that Glu is critical for the initial non-covalent binding of FAD and is instrumental in delivering FAD to the covalent attachment site at Cys.

Monoamine oxidase B (MAO B) catalyzes the oxidative deamination of biogenic and xenobiotic amines. The oxidative step is coupled to the reduction of an obligatory cofactor, FAD, which is covalently linked to the enzyme at Cys 397 . In this study, we developed a novel riboflavindepleted (Rib ؊ ) COS-7 cell line to study the flavinylation of MAO B. ApoMAO B can be obtained by expressing MAO B cDNA in these cells. We found that MAO B is expressed equally in the presence or absence of FAD and that apoMAO B can be inserted into the outer mitochondrial membrane. Flavinylation of MAO B was achieved by introducing MAO B cDNA and different flavin derivatives simultaneously into Rib ؊ COS-7 cells via electroporation. Since the addition of riboflavin, FMN, or FAD resulted in equal levels of MAO B activity, we conclude that the flavin which initially binds to apoMAO B is FAD. In our previous work, we used site-directed mutagenesis to show that Glu 34

in the dinucleotide-binding motif of MAO B is essential for MAO B activity, and we postulated that this residue is involved in FAD binding.
In this study, we tested the role of residue 34 in flavin binding by expressing wild-type or mutant MAO B cDNA in Rib ؊ COS-7 cells with the addition of [ 14 C]FAD. We found that Glu 34 is essential for both FAD binding and catalytic activity. Thus, FAD binds to MAO B in a dual manner at Glu 34 noncovalently and Cys 397 covalently. We conclude that Glu 34 is critical for the initial noncovalent binding of FAD and is instrumental in delivering FAD to the covalent attachment site at Cys 397 .
The major amine-degrading enzymes in the central nervous system and peripheral tissues of mammals are monoamine oxidase A and B (MAO 1 A and B, amine:oxygen, oxidoreductase (deaminating, flavin-containing), EC 1.4.3.4). These isozymes are integral proteins of the outer mitochondrial membrane (1) and can be distinguished by differences in substrate preference (2), inhibitory specificity (3), tissue and cell distribution (4 -6), and immunological properties (7)(8)(9). Furthermore, comparison of their nucleotide and deduced amino acid sequences show that human MAO A and B are two distinct proteins encoded by different genes (10).
Oxidation of amines by MAO is coupled to the reduction of an obligatory cofactor, FAD, which is covalently linked to the enzyme. Five types of bonds are generally found in the covalent linkage of flavins to their respective apoproteins (11). These include a histidine residue which can be attached through its N-1 or N-3 atom to the 8␣-methyl group of the isoalloxazine ring to form a tertiary amine; a cysteine residue which forms a thioether linkage with either the 8␣-methyl group or the C-6 of the xylene ring of the flavin molecule; or a tyrosine residue can become linked to the 8␣-methyl group to form an (O)-8␣-flavin bond. In MAO A and B, the 8␣-methyl group of FAD is bound covalently to cysteine through a thioether linkage in the pentapeptide SGGCY (12,13). Comparison of this segment with the complete deduced amino acid sequences of MAO A and B indicated that FAD is covalently bound to Cys 406 in MAO A and Cys 397 in MAO B, respectively (10). In addition, site-directed mutagenesis studies of MAO B, where Cys 397 was substituted with serine or histidine, showed that this cysteine residue is essential for catalytic activity (14,15). Although the amino acid sequences surrounding the FAD covalent attachment site in different flavoproteins bear little homology, a distinct non-covalent FAD-binding site displays high sequence identity in many FAD-containing enzymes of diverse function (16,17). This non-covalent FAD binding region is commonly referred to as the dinucleotide-binding site or motif due to its interaction with the AMP moiety of FAD. This motif consists of a ␤ 1 -sheet-␣-helix-␤ 2 -sheet beginning with a highly conserved Gly-X-Gly-X-X-Gly sequence between the first ␤-sheet and the ␣-helix. The second ␤-sheet usually ends with a glutamate residue in which the ␥-carboxylate group is thought to interact through a hydrogen bond with the 2Ј-hydroxyl group of ribose in the AMP moiety of FAD. In MAO A and B, this motif is located at the N terminus of MAO A (residues 15-43) and MAO B (residues 6 -34) and ends in Glu 43 and Glu 34 , respectively. Site-directed mutagenesis studies, where Glu 34 was replaced with aspartate, glutamine, or alanine, resulted in near complete or total loss of catalytic activity in MAO B (18).
A fundamental process in the intracellular generation of functional flavoenzymes is the molecular mechanism which generates holoenzyme from apoenzyme and its cofactor. Following the discovery of the first known enzyme with covalently linked FAD (succinate dehydrogenase, 19), extensive research in many laboratories has been conducted to elucidate how FAD is coupled to its respective proteins. The precise steps involved remain unknown. In this study, we developed a novel riboflavin-depleted (Rib Ϫ ) COS-7 cell line to investigate the flavinylation of MAO B. ApoMAO B was obtained by expressing MAO B cDNA in these cells. We show that the expression of MAO B apoenzyme is independent of FAD and that apoMAO B can be inserted into the outer mitochondrial membrane. Coupling of flavin to the apoenzyme was studied using FAD, flavin derivatives, or [ 14 C]FAD. We also examined the role of a critical glutamate residue (Glu 34 ) in flavinylation of MAO B using site-directed mutants. We find that Glu 34 plays an essential role in flavin coupling to the apoenzyme. We propose that the dinucleotide-binding site at the N terminus of MAO B provides a topological dock for the inital binding of FAD, and then FAD is delivered to the covalent attachment site at Cys 397 . (20). The reaction mixture (530 l) contained 15 mM MgCl 2 , 6.5 mM ATP, 0.12 mM [ 14 C]riboflavin (Amersham Corp., 50 mCi/mmol) and 200 g of FAD synthetase (purified from Brevibacterium ammoniagenes). After incubation at 37°C for 20 h, the mixture was filtered through a 100,000 molecular weight cut-off spin filter (Millipore) to remove the insoluble components. The clear yellow solution was loaded on a C-18 Semi-Prep HPLC column (Beckman), and eluted with a linear gradient from 100% A/0% B (A ϭ 10 mM (NH 4 ) 2 HPO 4 , pH 6.8, B ϭ acetonitrile) to 60% A/40% B in 20 min at a flow rate of 4 ml/min using a Beckman HPLC (System Gold). The peak corresponding to [ 14 C]FAD eluted at a retention time identical to a FAD standard (Sigma). [ 14 C]FAD was collected in sterilized silicone-coated glass tubes, dried in a Beckman speedvac, and stored at Ϫ20°C in powder form.

Synthesis of [ 14 C]FAD-[ 14 C]FAD was prepared by a modified method of Manstein and Pai
Synthesis of 8␣-Hydroxyriboflavin-Synthesis of 8␣-hydroxyriboflavin was carried out by the method of McCormick (21). Briefly, riboflavin was added to a solution of acetic acid/acetic anhydride (1:1) and the yellow solution was stirred at room temperature for 24 h. Tetraacetylriboflavin (TAR) was extracted from the aqueous reaction mixture with CHCl 3 , followed by extraction with water and evaporation to give a yellow residue of essentially pure TAR. Dibenzoyl peroxide and dioxane dibromide in dioxane were added to a solution of TAR in dioxane, and the solution was refluxed. The crude bromo-TAR was separated from the reaction mixture on a C-18 Semi-Prep HPLC column (Beckman). The bromo-TAR was hydrolyzed to yield 8␣-hydroxyriboflavin, which was separated by HPLC (System Gold, Beckman) through a linear gradient from 0.5% trifluoroacetic acid in water to 0.5% trifluoroacetic acid in acetonitrile for 50 min at a flow rate of 4 ml/min.
Cell Culture-Mammalian COS-7 cells were selected for transient expression of MAO B cDNA because they were found to contain no endogenous MAO B, as determined by ELISA, Western blot, and radiometric activity assays in initial experiments. Mammalian COS-7 cells were grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum (Life Technologies, Inc.) at 37°C with 5% CO 2 . Since this medium (Dulbecco's modified Eagle's medium ϩ 10% fetal bovine serum) contains riboflavin, COS-7 cells grown in this medium were defined as riboflavin-containing COS-7 cells (Rib ϩ COS-7 cells). MAO B holoenzyme was obtained by expressing MAO B cDNA in Rib ϩ COS-7 cells. Mammalian COS-7 cells were also grown in riboflavin-free medium (riboflavin-free Dulbecco's modified Eagle's medium ϩ 10% dialyzed fetal bovine serum, Life Technologies, Inc.) at 37°C with 5% CO 2 , with fresh riboflavin-free medium changed every 48 -72 h. COS-7 cells grown in riboflavin-free medium for longer than 100 days were defined as riboflavin-depleted (Rib Ϫ ) COS-7 cells. ApoMAOB was obtained by expressing MAO B cDNA in Rib Ϫ COS-7 cells. COS-7 cells were grown in riboflavin-free medium for greater than 5 months without any detectable change in morphology.
Preparation of Mutant MAO B cDNA-Mutagenesis was carried out by the method of Deng and Nickoloff (22) using a Transformer Sitedirected Mutagenesis kit (Clontech). Glu in position 34 was replaced with Asp (E34D), Gln (E34Q), or Ala (E34A), and Val in position 10 was replaced with Ile (V10I) as described by Kwan et al. (18).
FAD Coupling in Intact Cells-Wild-type or mutant MAO B cDNAs were transiently transfected into COS-7 cells by electroporation (23) as described previously (18). Briefly, Rib ϩ -or Rib Ϫ COS-7 cells were harvested during late log phase growth and resuspended to a concentration of 3.1 ϫ 10 6 cells/ml in either riboflavin-containing or riboflavin-free medium, respectively. Wild-type or mutant MAO B cDNAs (15 g) were added to 0.8 ml of cell suspension. In experiments where flavinylation of wild-type and variant MAO B enzymes were studied, 20 l of 0.8 mM unlabeled FAD, [ 14 C]FAD or other flavin derivatives were also added to the Rib Ϫ COS-7 cell suspension. Electroporation was carried out in a Bio-Rad Gene Pulser with a setting of 250 V and 500 microfarads. Cells were resuspended in 15 ml of riboflavin-containing or riboflavin-free medium and incubated at 37°C with 5% CO 2 . Transfected COS-7 cells were harvested at 48 h and homogenized in a lysis solution containing 20 mM Tris-HCl, 1 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, pH 8.0. The homogenate was diluted with an equal volume of the same buffer supplemented with 300 mM NaCl to obtain a 150 mM final NaCl concentration. Triton X-100 (Pierce) was added to the lysate to give a final concentration of 0.25%, and the samples were allowed to stir for 50 min at 4°C to extract MAO B from the outer mitochondrial membrane. After Triton extraction, the lysate was centrifuged at 1300 ϫ g for 5 min at 4°C to remove insoluble cell debris. The supernatant was then analyzed for protein concentration, MAO B concentration, enzymatic activity, and FAD coupling.
FAD Coupling in Vitro-ApoMAO B was obtained by expressing MAO B cDNA in Rib Ϫ COS-7 cells. The cells were then harvested and homogenized as described above. Half of the lysate was stirred in the presence of 0.25% Triton X-100 at 4°C for 50 min to extract apoMAO B from the outer mitochondrial membrane. The second half of the lysate was not extracted with Triton X-100 to permit MAO B to remain in the membrane. FAD coupling assays were carried out for both fractions in reaction vials (200 l) containing 10 l of cell lysate, 50 mM phosphate buffer, and FAD (1.5 nmol). Assays were also carried out in the presence of an energy mixture (10 mM ATP, 32 mM P-enolpyruvate, and 2.4 g of pyruvate kinase) and 25% glycerol in the reaction vials. Each sample was run in triplicate. After 1, 2, or 3 h of incubation at 30°C, each sample was assayed for MAO B activity as described below.
Subcellular Fractionation of COS-7 Cells-COS-7 subcellular fractionation was carried out with a modified method of Clark and Waterman (24). Transfected Rib ϩ -or Rib Ϫ COS-7 cells were harvested, washed twice with ice-cold phosphate-buffered saline, and pelleted by microcentrifugation at 500 ϫ g for 5 min. The cells were then homogenized in a Dounce homogenizer. Greater than 95% of the cells were lysed, as determined by trypan blue staining. The homogenate was diluted with an equal volume of 10 mM Tris-HCl, 1 mM EDTA, pH 8.0, to obtain a 0.25 M final sucrose concentration and layered over 0.5 volume of 0.5 M sucrose pad (0.5 M sucrose in 10 mM Tris-HCl, 1 mM EDTA, pH 8.0). The solution was centrifuged at 5,000 revolutions/min for 3 min using a swinging bucket rotor (TLS.55, Beckman TL-100) to remove cell debris and nuclei (P 1 fraction). The supernatant plus the interface of the 0.5 M sucrose pad was again layered over another 0.5 M sucrose pad and centrifuged as above at 17,000 revolutions/min for 20 min to isolate mitochondria (P 2 fraction). The resulting supernatant was centrifuged at 70,000 revolutions/min (TL100.3, Beckman TL-100) for 30 min to sediment microsomes (P 3 fraction). The final supernatant was the cytosol (S). After subcellular fractionation, P 1 , P 2 , and P 3 were resuspended in 20 mM Tris-HCl, 1 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, 150 mM NaCl, pH 8.0. All samples were treated with Triton X-100 to extract MAO B, and then protein concentrations, MAO B concentrations, and enzymatic activities were determined.
Quantitation of MAO B Protein-Total protein concentrations of samples containing MAO B holoenzyme, apoenzyme, or variant MAO Bs were determined by a MicroBCA kit (Pierce). All samples were then adjusted to equal protein concentrations and assayed for MAO B protein by ELISA using a modification of the method of Yeomanson and Billett (25) as described previously (18).
Enzyme Activity Determination-MAO B activity was assayed radiometrically by a modification of the method of Wurtman and Axelrod (26) as described previously (18).
Immunoprecipitation of Holo-, Apo-or Variant MAO B-Transfected Rib ϩ -or Rib Ϫ COS-7 cells were homogenized and extracted with 0.25% Triton X-100 for 50 min at 4°C. The cell lysates were centrifuged at 1300 ϫ g for 5 min, and an aliquot of each supernatant was analyzed for MAO B concentration by ELISA. All supernatants were then adjusted to equal MAO B concentration, and immunoprecipitation was carried out using polyclonal goat anti-MAO B antibody (18). The immunocomplex was then dissolved in SDS-PAGE sample buffer and analyzed by Western blot or fluorography.
Western Blot Analysis-The immunoprecipitated proteins (obtained as described above) were subjected to electrophoresis in a 10% SDSpolyacrylamide gel and examined by Western blotting as described previously (18).
Fluorography-The immunoprecipitated proteins (obtained as described above) were subjected to electrophoresis in a 10% SDS-polyacrylamide gel. The SDS-PAGE gel was fixed by soaking in 7% acetic acid, 10% methanol, 83% H 2 O for 1 h, and then processed for fluorography as described by Bonner and Laskey (27). The dried gel was exposed to Kodak X-OMAT AR film at Ϫ80°C for 2 weeks.

Synthesis of [ 14 C]FAD and 8␣-Hydroxyriboflavin-Retention
times of riboflavin, FMN, FAD, and ATP standards (Sigma) were determined (Fig. 1A). FAD was observed to have a retention time of 10.0 min. Synthetic [ 14 C]FAD had an identical retention time of 10.0 min (Fig. 1B) and eluted as a large sharp single peak. The radioactivity of the [ 14 C]FAD-containing fraction was determined in a scintillation counter, and the sample was dried to obtain a fine yellow powder.
Synthesis of 8␣-hydroxyriboflavin was carried out by the method of McCormick (21). Synthetic 8␣-hydroxyriboflavin was resolved on HPLC to yield a major sharp peak on the chromatogram (Fig. 1C). The authenticity of 8␣-hydroxyriboflavin was further confirmed by spectroscopic analysis (UV and mass spectrometry).
MAO B Expression Is Independent of FAD Cofactor-In order to study the covalent binding of FAD to human MAO B, it was necessary to develop a method for obtaining apoMAO B. To accomplish this, mammalian COS-7 (Rib ϩ ) cells were grown in riboflavin-free medium to deplete the endogenous riboflavin. MAO B cDNA was expressed sequentially at different time intervals in these cells during this riboflavin depletion process (Fig. 2). Each data point in Fig. 2 represents an individual expression assay. For each assay, a sample of COS-7 cells grown in riboflavin-free medium were transfected with MAO B cDNA via electroporation. Concurrently, Rib ϩ COS-7 cells were transfected with MAO B cDNA to serve as a control. Following incubation for 48 h, the cells were homogenized and assays were performed to determine protein concentration, MAO B concentration by ELISA using polyclonal antibodies, and MAO B activity using [ 14 C]benzylamine. Table I shows one set of analyses performed on these COS-7 cells that had been grown in riboflavin-free medium for 76 days (point 4 in Fig. 2). The enzymatic activity of MAO B expressed in these cells was 12.7% of the control, while the level of expression was essentially identical to the control (0.90 g MAO B/mg protein versus 0.86 g MAO B/mg protein). As seen in Fig. 2 (Table II). The distribution of total protein in Rib ϩ -or Rib Ϫ COS-7 cells was essentially identical, with the largest amount of protein found in the cytosolic fraction. Approximately 80% of the holo-or apoMAO B enzymes were found in the mitochondrial fraction of the Rib ϩ -or Rib Ϫ COS-7 cells, respectively. The activity of expressed holo-or apoMAO B in various fractions was also determined. The activity distribution of holo-MAO B corresponded closely with the distribution of the enzyme with the majority of activity (about 80%) located in the mitochondrial fraction. Although the total activity of apoMAO B expressed in Rib Ϫ COS-7 cells was dramatically reduced, the small amount of remaining activity was also found mainly in the mitochondrial fraction (about 83%).
FAD Coupling in Intact Cells-When exogenous FAD was added simultaneously with MAO B cDNA in Rib Ϫ COS-7 cells during electroporation, restoration of MAO B activity was observed (Table III) However, only 40% of MAO B holoenzyme activity was obtained by the addition of 8␣-hydroxyriboflavin. As expected, the addition of NAD ϩ along with MAO B cDNA during electroporation did not yield active MAO B. Expressed MAO B enzymes (with or without cofactor additions) were further analyzed by Western blot using our MAO B-specific monoclonal antibody MAO B-1C2 (Fig. 3). A band at approximately 59 kDa was observed in all lanes that contained apo-or holoMAO B.
FAD Coupling in Vitro-Expressed MAO B holoenzyme, which served as a control, remained fully active in a cell lysate for up to 3 h at 30°C. (Fig. 4). When exogenous FAD was added to Triton-extracted or -non-extracted lysates, which contained mitochondrial membrane-free or mitochondrial membranebound apoMAO B, respectively, no MAO B catalytic activity was observed. Flavinylation of apoMAO B in vitro was also attempted in the presence of an energy mixture and glycerol, but no MAO B activity was obtained.

TABLE I Comparison of MAO B enzymatic activity and MAO B expression between Rib ϩ COS-7 cells and COS-7 cells which had been grown in
riboflavin-free medium for 76 days An equal amount of MAO B cDNA (15 g) was expressed in both types of cells (Rib ϩ and Rib Ϫ ), and the cells were then incubated at 37°C with 5% CO 2 for 48 h. Both transfected cell samples were homogenized, and the cell lysates were extracted with Triton X-100. After the protein concentration was equalized in both samples, MAO B quantitation (by ELISA) and activity measurements were performed.

DISCUSSION
Flavinylation of MAO B has been difficult to study in the past because FAD is covalently attached to Cys 397 , and this cofactor cannot be removed without sacrificing MAO B activity (28). For mammalian flavoproteins, the conventional approach has been to study flavinylation in animals. Rabbits or mice were fed riboflavin-free diets to deplete the endogenous riboflavin, and the animals were sacrificed to obtain the organs or tissues for analysis (29). This method is time-consuming, tedious, and subject to variation due to individual differences in animals. We have now developed a convenient and rapid method to study flavinylation of eucaryotic proteins in Rib Ϫ COS-7 cells. Since COS-7 cells are not capable of synthesizing riboflavin, enzymes expressed in these cells lack flavin cofactors. Rib Ϫ COS-7 cells were used to produce apoMAO B to study the steps involved in flavinylation. In other related studies, Nishikimi et al. (30) produced the apoenzyme of L-gulono-␥lactone oxidase in a baculovirus expression system in which riboflavin levels were reduced. Enzymatic activity was observed upon addition of FAD, but no covalently bound FAD could be obtained using this system.
The expression level (0.95 Ϯ 0.04 g/mg protein) of MAO B in transfected Rib Ϫ COS-7 cells remained unchanged in sequential transfections during the process of riboflavin depletion (Fig. 2). This observation indicates that MAO B expression is not dependent upon riboflavin or FAD concentrations in the cell. The level of expressed MAO B was determined by ELISA, which is based upon epitope recognition by antibodies and is susceptible to major conformational changes. Both MAO B-1C2 monoclonal antibody and goat anti-MAO B polyclonal antibodies were capable of recognizing apoMAO B. In another study, the apoenzyme of bacterial 6-hydroxy-D-nicotine oxidase, which contains covalently bound FAD in its holoenzyme, is not recognized by a molecular chaperone as aberrant (31). We suggest that the conformation of the apoMAO B, like apo-6hydroxy-D-nicotine oxidase, may be similar to that of the native holoenzyme.
Mitoma and Ito (32) found that the mitochondria targeting sequence of MAO B is located on the C terminus of the molecule. Deletion of the C-terminal 28 amino acids of MAO B abolished transfer of the enzyme to the mitochondria, while deletion of the N-terminal 55 amino acids had no effect on mitochondrial targeting. Furthermore, an expressed hybrid protein, in which the C-terminal 29 amino acids of MAO B was fused to the hydrophilic portion of cytochrome b 5 , was localized in the mitochondria. In our work, we found that apoMAO B expressed in Rib Ϫ COS-7 cells was localized in the mitochondrial fraction of cell lysates (Table II), indicating that bound FAD is not necessary for MAO B insertion into the mitochondria membrane. These results are consistent with those of Mitoma and Ito (32) and support the notion that the target C-terminal sequence alone is sufficient for insertion into the membrane.
One advantage of using Rib Ϫ COS-7 cells to study flavinylation is that exogenous FAD or its derivatives can be introduced with MAO B cDNA into the cells during the transfection process. The enzymatic activity of MAO B with the addition of   (Table III). Addition of FAD resulted in the restoration of about 75% of holoMAO B activity. Interestingly, approximately 75% of holoMAO B activity was also achieved by the addition of riboflavin or FMN to transfected Rib Ϫ COS-7 cells, suggesting the presence of abundant levels of cellular FAD synthetase. The addition of 8␣hydroxyriboflavin gave an enzyme with 40% activity of the control, which raises the possibility that this flavin may represent an intermediate in the activation of FAD (discussed below). Full recovery of MAO B enzymatic activity obtained in Rib ϩ COS-7 cells was not achieved for reasons that remain unknown. In related studies, however, Brandsch and Bichler (33) found that the covalent flavinylation of 6-hydroxy-D-nicotine oxidase in vitro required specific effectors (phosphorylated three carbon compounds), such as glycerol 3-phosphate, glyceraldehyde 3-phosphate, or glycerate 3-phosphate. Effectors that could enhance the activity of MAO B have not been identified. We speculate that the achievement of only 75% of activity may be due to a slight change in metabolism of Rib Ϫ COS-7 cells which have been adapted to grow in riboflavin-free medium for more than 100 days.
Although it is known that FAD is covalently attached to active MAO B molecules, the form of the flavin which initially binds to MAO B in vivo has not previously been established. Theoretically, riboflavin or FMN could first bind to apoMAO B followed by phosphorylation and adenylation, respectively, to form FAD. If riboflavin or FMN is the form that initially binds to apoMAO B, we would expect FAD binding to be much less effective than riboflavin or FMN. Since MAO B activity was recovered to approximately the same extent (75%) using FAD, FMN, or riboflavin, we conclude that the flavin moiety which initially binds to apoMAO B is FAD. Apparently, FAD synthetase in these cells rapidly converted riboflavin and FMN to FAD by phosphorylation and adenylation, respectively, prior to incorporation. The presence of FAD was confirmed by measuring the covalent binding of [ 14 C]FAD to MAO B.
The covalent attachment of FAD to Cys 397 could be autocatalytic or catalyzed by an as yet uncharacterized enzyme. In either case, one of the participants, the 8␣-methyl group of the flavin moiety or Cys 397 of MAO B, must be activated prior to coupling. Although the nucleophilicity of the Cys 397 residue may be influenced by surrounding amino acid residues, it is difficult to envision that a cysteine derivative would react with the inert 8␣-methyl group of the flavin moiety. From a chemical point of view, activation of the 8␣-methyl group appears essential for coupling of FAD to apoMAO B. An enzymatically facilitated pathway for the incorporation of FAD into flavoproteins has been proposed by Decker (11) in which a flavin cofactor may be enzymatically activated by hydroxylation of the 8␣methyl group, followed by (pyro)phosphorylation (Fig. 6). Since the (pyro)phosphate is a good leaving group, a simple S N 2 reaction could facilitate the formation of the thioether between the flavin moiety and MAO B. To test this hypothesis, we synthesized the putative activated intermediate 8␣-hydroxyriboflavin and determined in Rib Ϫ COS-7 cells its ability to generate MAO B enzymatic activity (synthesis of 8␣-phosphate-riboflavin was also attempted, but was unsuccessful because the highly reactive hydroxyl groups on the ribityl moiety were also phosphorylated). If the flavin derivative is truly an intermediate, we assumed that it would be capable of entering the flavinylation pathway to produce active MAO B. MAO B activity was obtained, but the level was only about half of that obtained with the addition of riboflavin (Table III). One possible explanation for the low activity is that a flavinylating enzyme binds the flavin substrate and catalyzes hydroxylation and phosphorylation sequentially without release of the 8␣hydroxy intermediate. Thus, the 8␣-hydroxy intermediate may not be recognized as efficiently as riboflavin during the initial binding step. Alternatively, the covalent flavinylation of MAO B may be autocatalytic, since the unactivated form of the flavins (riboflavin, FMN, and FAD) has higher efficiency of incorporation into apoMAO B than the putative activated form. FIG. 4. In vitro flavinylation assays (see "Materials and Methods"). Triton extracted MAO B holoenzyme from a transfected Rib ϩ COS-7 cell lysate, which served as a positive control, remained fully active during 3 h incubation at 30°C (ϫ). However, no MAO B catalytic activity was observed when FAD was added after apoMAO B had been synthesized. Triton extracted (f) or nonextracted (å) MAO B apoenzyme from transfected Rib Ϫ COS-7 cell lysate were incubated at 30°C with exogenous FAD. Triton-extracted MAO B apoenzyme from transfected Rib Ϫ COS-7 cell lysate were also incubated at 30°C with exogenous FAD, an energy mixture and with or without 25% glycerol (ࡗ, with glycerol; E, without glycerol). The enzymatic activity of each sample was determined at 1-h time intervals using [ 14 C]benzylamine as substrate. Studies were also conducted to determine whether flavinylation occurs as a co-translational or post-translational process. When FAD and MAO B cDNA were added simultaneously to Rib Ϫ COS-7 cells, active MAO B (containing FAD) was obtained. However, when FAD was added in vitro to whole cell lysates after apoMAO B was synthesized, MAO B activity could not be regenerated (Fig. 4). Furthermore, when apoMAO B was extracted from the mitochondrial membrane, attempts to regenerate active flavinylated MAO B were unsuccessful, even in the presence of various energy mixtures and glycerol (Fig. 4). The inability to couple FAD to apoMAO B in vitro may indicate that flavinylation occurs as a cotranslational process during elongation of nascent chains to form functionally competent MAO B molecules.
Our previous work demonstrated that Glu 34 in the dinucleotide binding motif was critical for MAO B catalytic activity (18). Two variants at Glu 34 (E34A and E34Q) were devoid of enzymatic activity, and another conservative variant, E34D, had only 7% of the wild-type activity. It was not known, however, whether the role of Glu 34 is confined to alignment of FAD for participation in the oxidation-reduction cycle of catalysis, or is involved in FAD incorporation. In this study, we show that the loss of activity in Glu 34 variants is linked to the inability to bind FAD covalently (Fig. 5).
Since FAD binds to two regions of MAO B (noncovalently at Glu 34 and covalently at Cys 397 ), the absence or low levels of FAD incorporation into Glu 34 variants reveals an important feature of the flavinylation process. If FAD coupling occurred by initial covalent attachment to Cys 397 , Glu 34 variants would contain covalently bound FAD, but would be inactive because FAD could not interact properly at the dinucleotide-binding site. Since we find little or no covalent binding of FAD in the Glu 34 variants, we conclude that FAD binds to Glu 34 first. We propose that the dinucleotide-binding site (including Glu 34 ) provides a topological dock for the initial binding of FAD and is instrumental in the delivery of FAD to Cys 397 in MAO B. The incoming flavin cofactor, which is initially bound to the dinucleotide-binding site of MAO B, could be held for a finite time in a position which places the 8␣-methyl group of FAD in exact and close proximity to Cys 397 to facilitate covalent flavinylation.
The dinucleotide-binding sites in various flavoproteins contain high sequence identity (17). However, the location within the primary structure varies from protein to protein, indicating that this site performs an autonomous function of cofactor binding within a heterologous group of flavoproteins. Furthermore, in many flavoproteins containing dinucleotide-binding sites, FAD is not covalently bound (17). Our finding that the dinucleotide-binding site in MAO B plays a role in initial FAD binding indicates that this site alone is sufficient for a flavoprotein to bind a flavin cofactor. The significance of covalent linkage between FAD and its flavoenzyme remains unresolved, but covalent binding could play a role in enzyme integrity and stability, substrate stereospecificity, cofactor economy, or redox potentials. Understanding the MAO flavinylation process may lead to the design of MAO enzymes with high redox potentials for better catalysis and to the rational design of MAO inhibitors. Since MAO inhibitors have long been used for the treatment of various psychiatric and neurological disorders, including depression (35) and Parkinson's disease (36), our studies on flavinylation may lead to the development of therapeutic drugs (analogs of FAD) for these disorders.