Demonstration of Isoleucine 199 as a Structural Determinant for the Selective Inhibition of Human Monoamine Oxidase B by Specific Reversible Inhibitors*

Several reversible inhibitors selective for human monoamine oxidase B (MAO B) that do not inhibit MAO A have been described in the literature. The following compounds: 8-(3-chlorostyryl)caffeine, 1,4-diphenyl-2-butene, and trans,trans-farnesol are shown to inhibit competitively human, horse, rat, and mouse MAO B with Ki values in the low micromolar range but are without effect on either bovine or sheep MAO B or human MAO A. In contrast, the reversible competitive inhibitor isatin binds to all known MAO B and MAO A with similar affinities. Sequence alignments and the crystal structures of human MAO B in complex with 1,4-diphenyl-2-butene or with trans,trans-farnesol provide molecular insights into these specificities. These inhibitors span the substrate and entrance cavities with the side chain of Ile-199 rotated out of its normal conformation suggesting that Ile-199 is gating the substrate cavity. Ile-199 is conserved in all known MAO B sequences except bovine MAO B, which has Phe in this position (the sequence of sheep MAO B is unknown). Phe is conserved in the analogous position in MAO A sequences. The human MAO B I199F mutant protein of MAO B binds to isatin (Ki = 3 μm) but not to the three inhibitors listed above. The crystal structure of this mutant demonstrates that the side chain of Phe-199 interferes with the binding of those compounds. This suggests that the Ile-199 “gate” is a determinant for the specificity of these MAO B inhibitors and provides a molecular basis for the development of MAO B-specific reversible inhibitors without interference with MAO A function in neurotransmitter metabolism.

Two isoforms of monoamine oxidase (MAO), 1 MAO A and MAO B, exist in humans and are both ϳ60-kDa outer-mito-chondrial membrane-bound flavoenzymes that share ϳ70% sequence identities (1). Because these enzymes have distinct and overlapping specificities in the oxidative deamination of neurotransmitters and dietary amines, the development of specific reversible inhibitors has been a long sought goal. Expression levels of MAO B in neuronal tissue increase ϳ4-fold with age (2), resulting in an increased level of dopamine metabolism and the production of higher levels of hydrogen peroxide, which are thought to play a role in the etiology of neurodegenerative diseases such as Parkinson's and Alzheimer's diseases (3). Thus, the development of specific, reversible MAO B inhibitors could lead to clinically useful neuroprotective agents.
Recent studies in the literature have shown that 8-(3chlorostyryl)caffeine (CSC), an A 2A adenosine receptor antagonist, is also a potent and selective inhibitor of mouse brain MAO B (K i ϭ 100 nM) but not MAO A (4). In addition, trans,trans-farnesol, a component of tobacco smoke, is a potent, reversible inhibitor specific for MAO B. 2 Another study has established that 1,4-diphenyl-2-butene (K i ϭ 35 M), a contaminant of polystyrene bridges used for MAO B crystallization, and 1,4-diphenyl-1,3-butadiene (K i ϭ 7 M) are potent, competitive MAO B-specific reversible inhibitors (6,7). Because none of these compounds (see Scheme 1 for their respective structures) inhibit MAO A, a more detailed study of the molecular basis for their specificities could provide opportunities to develop MAO B-selective inhibitors with potential neuroprotective properties.
Recent crystal structures of human MAO B in complex with several pharmacologically important inhibitors have been solved to 1.6-Å resolution (6,8). The access channel from the surface of the protein to the active site of the enzyme consists of two cavities, the entrance cavity and the active site cavity (9). Depending on the nature of the inhibitory species, these cavities are either separate or fused depending on the conformation of the Ile-199 "gate" residue. The analogous position in MAO A is known from sequence comparisons to be Phe-208. The recently reported structure of MAO A (10) is at a resolution (3.2 Å) that precludes a detailed molecular description of the properties of the active site that would allow a comparison with that of human MAO B. Previous studies on the catalytic and inhibitor properties of rat (11) and of human (12) I199F MAO B mutant proteins show clear differences with the respective native enzymes which could not be readily interpreted in molecular detail.
In this report we compare the functional differences of inhibitor interactions between human MAO A and human, mouse, rat, horse, cow, and sheep MAO B. The crystal structures of human MAO B in complex with trans,trans-farnesol and that of the I199F MAO B mutant protein are also reported. These data identify structural determinants in MAO B that provide a molecular interpretation for the specificity of these reversible inhibitors.

MATERIALS AND METHODS
All reagents used in this study were purchased from commercial sources unless noted otherwise. Human recombinant MAO A and MAO B were expressed in Pichia pastoris and purified as described previously with the modification of the MAO B purification protocol by replacement of the polymer fractionation and differential centrifugation steps with a single chromatographic step using a Bio-Rad High Q anion exchange column and elution with a phosphate gradient (13,14). Mitochondria from bovine, sheep, human, horse, rat, and mouse livers were isolated using a procedure described previously (15).

Construction of MAO A F208I and MAO B I199F Mutant Proteins-
The site-directed mutations were introduced into full-length MAO A and MAO B cDNA-containing pPIC3.5k vectors using the QuikChange XL kit (Invitrogen). The instructions from the kit were followed, and a single base mismatch containing pairs of complementary oligonucleotide primers (5Ј-CCACTCGGATAATCTCTGTCACC-3Ј and 5Ј-GGTGA-CAGAGATTATCCGAGTGG-3Ј for MAO A F208I and 5Ј-CAACAAGA-ATCTTCTCGACAAC-3Ј and 5Ј-GTTGTCGAGAAGATTCTTGTTG-3Ј for MAO B I199F) were used. The presence of desired mutations in the constructs was confirmed by DNA sequencing. P. pastoris spheroplasts were prepared and transformed according to the Invitrogen protocol, and clones with the highest resistance to G-418 antibiotic were selected for fermentation as described earlier (13,14).
MAO A and MAO B Activity Measurements-MAO A activity was determined spectrophotometrically (316 nm) using kynuramine as a substrate in 50 mM potassium phosphate buffer, pH 7.5, containing 0.5% (w/v) reduced Triton X-100 at 25°C. MAO B activity was also determined spectrophotometrically (250 nm) using benzylamine as a substrate in 50 mM HEPES buffer, pH 7.5, containing 0.5% (w/v) reduced Triton X-100 at 25°C. MAO B activity in mitochondrial preparations was measured spectrophotometrically (420 nm) using 1-methyl-4-(1-methylpyrrol-2-yl)-1,2,3,6-tetrahydropyridine (MMTP) as a substrate in 100 mM sodium phosphate buffer, pH 7.4, as described earlier (16). Any MAO A activity in these mitochondrial preparations was eliminated by a 30-min pretreatment with 1 M clorgyline. Kinetic isotope effects for the MAO B mutant were determined by comparing K m and k cat values obtained for benzylamine and ␣,␣-dideuterobenzylamine under the conditions described above. Oxygen-saturated buffers were prepared by slowly bubbling oxygen gas through 50 mM HEPES buffer, pH 7.5, containing 0.5% (w/v) reduced Triton X-100 at 25°C for 2 h.
K i Determinations-Competitive K i values were determined by measuring initial rates of substrate oxidation in the presence of varying concentrations of inhibitor (all assays were performed in duplicate). The hydrophobic nature of the inhibitors used in this study required their solubilization in dimethylformamide and the addition of aliquots of these solutions to the assay. Control experiments demonstrated no deleterious effect of small amounts of dimethylformamide (Ͻ10%) on MAO B or MAO A activities. Apparent K m values for each inhibitor concentration (slopes of double reciprocal plots) were plotted as a function of inhibitor concentration and the K i values determined (K i ϭ intercept/slope). Inactivation rates of MAO B I199F with rasagiline were determined as described previously (17).
X-ray Crystallography-Wild-type MAO B was incubated with 0.1 mM trans,trans-farnesol, and the I199F mutant was incubated with 5 mM inhibitor (isatin or rasagiline). Crystallization experiments were carried out by the sitting-drop vapor diffusion method at 4°C (9). The protein solution contained 2 mg/ml inhibited-MAO B, 8.5 mM Zwittergent 3-12, and 25 mM potassium phosphate buffer, pH 7.5. The reservoir solution consisted of 12% (w/v) polyethylene glycol 4000, 70 mM lithium sulfate, and 100 mM N-(2-acetamido)-2-iminodiacetic acid, pH 6.5. Xray diffraction data were collected at 100 K at the Swiss Light Source in Villigen and at beam lines of the European Synchrotron Radiation Facility in Grenoble, France. For data collection, crystals were transferred into a mother liquor solution containing 18% glycerol and flashcooled in a stream of gaseous nitrogen at 100 K. Data processing and scaling (Table I) were carried out using MOSFLM (18) and programs of the CCP4 package (19). The structures of wild-type MAO B in complex with isatin and rasagiline were used as starting models (6,8) for crystallographic refinement, which was performed with the programs REFMAC5 (20) and O (21) as described previously (6). Refinement statistics are listed in Table I. Cavities were identified with the program Voidoo (22). Structural representations were produced using Molscript (23) and Raster3d (24). These results point to interesting differences in MAO B from various species with regards to their sensitivities to these inhibitors and reflect differences in their respective active site architectures.

Comparison of MAO B Inhibition by CSC, DPB, and trans,trans-Farnesol in Liver Mitochondria
Structure of the MAO B-trans,trans-Farnesol Complex-The structure of the MAO B-trans,trans-farnesol complex was determined to 1.8-Å resolution ( Fig. 1 and Table I). The bound trans,trans-farnesol isoprenoid chain traverses both the entrance and the substrate cavities ( Fig. 2) with Ile-199 in its "open" conformation as is observed in the 1,4-diphenyl-2butene structure (6). No structural changes in the enzyme are observed relative to previous structures (root mean square ϭ 0.2 Å for 974 C␣ atoms) (6). The OH moiety of the bound trans,trans-farnesol is located 3.4 Å from the C(4a) position of the flavin, and the 1-methylene carbon is 3.4 Å from the N(5) of the flavin. Additional enzyme-trans,trans-farnesol interactions are hydrophobic between the isoprenoid chain and the amino acid residues defining the cavities.
Amino Acid Sequence Alignments-Amino acid sequence alignments of the regions of the available MAO A and B sequences known to interact with 1,4-diphenyl-2-butene in human MAO B (6)   Interestingly, all known MAO A sequences contain Phe in the analogous position (residue 208 in the human MAO A sequence). Ile-199 was previously identified as the "gate" residue in human MAO B, because it can exist in two different conformations; a closed conformation separating the substrate and entrance cavities or opened conformation fusing the two cavities (6). Because bovine MAO B as well as human MAO A are not inhibited by CSC, DPB, or trans,trans-farnesol, the substitution of Phe in the position of Ile-199 is predicted to abolish the binding of these inhibitors but not isatin binding in the active site of human MAO B. To demonstrate this prediction, the properties of the I199F human MAO B mutant protein were investigated.
Human MAO B I199F Catalytic Properties-A comparison of the kinetic parameters for the oxidation of benzylamine, 2-phenylethylamine, tyramine, and MMTP oxidation for wild type human MAO B and the I199F mutant protein is shown in Table  III. The K m values for these substrates are higher for the I199F mutant as compared with wild type enzyme, and the mutant enzyme's turnover numbers are ϳ50% those for wild type MAO B. Catalytic efficiencies (k cat /K m ) for the substrates tested with the I199F mutant protein are in the range of 30 -50% of those for wild type human MAO B. The k cat values for the purified mutant enzyme differ from those reported by Geha et al. (12) for the membrane-bound form of this mutant enzyme expressed in a baculovirus system. The rate-determining step in benzylamine oxidation by MAO B I199F is the C-H bond cleavage step (as in WT MAO B), because the kinetic isotope effect (the ratio of k cat /K m for oxidation of benzylamine and ␣,␣dideuterobenzylamine) is found to be 3.7, which agrees with the kinetic isotope effect determined for the wild type MAO B (14). MAO B I199F exhibits a similar affinity for the acetylenic inhibitor rasagiline (K i ϭ 0.5 M) and for isatin (K i ϭ 0.5 M).    (8). In both mutant structures, the side chain of Phe-199 extends into the entrance cavity with essentially identical conformations (Fig. 4) rather than the "open/closed" conformations seen with Ile-199.
The increased size of the Phe aromatic ring relative to Ile (28.6 Å 3 ) (25) prevents its occupation of the alternate conformations observed with Ile-199. The only available space for it to occupy is the entrance cavity of the enzyme. The presence of this bulky residue in the entrance cavity does not appear to alter the kinetic properties of the enzyme or the binding affinity of a small competitive inhibitor, but it does prevent the binding of molecules that must traverse the two cavities in their bound form. One question that arises is how this information relates to the functional and structural properties of human MAO A.
Human MAO A F208I Mutant Enzyme-To test whether the analogous mutation in MAO A would generate an enzyme capable of binding CSC, trans,trans-farnesol or DPB, the hu-man MAO A F208I mutant enzyme was expressed in P. pastoris as described for MAO B I199F mutant protein. All data reported on human MAO A F208I were obtained using mitochondrial preparations isolated from P. pastoris, because the catalytic activity of the mutant was rapidly lost on solubilization from the membrane and subsequent purification. This mutant enzyme is capable of oxidizing MMTP (K m ϭ 100 M), and its activity was completely abolished on incubation with clorgyline in agreement with earlier results (12). No inhibition was observed on incubations of the membrane-bound form with CSC, trans,trans-farnesol, or DPB, which demonstrates that mutating Phe-208 to Ile does not allow the binding of these compounds. Although the mutant enzyme was catalytically active, it was no longer capable of binding isatin at the level of affinity observed with WT MAO A. DISCUSSION Due to their pharmacological importance, the molecular basis for understanding the respective substrate and inhibitor specificities of MAO A and MAO B has been under investigation for some time. Once it became possible to express recombinant MAO A and MAO B as functional enzymes, chimeras could be used (26,27) as an approach to identify the sites for their functional differences. The high resolution crystal structures of human MAO B (6,8) have provided needed structural information that suggests a molecular rationale for further investigation of the relative inhibitor specificities of MAO B. The structural and functional data presented in this study demonstrate that reversible inhibitors that occupy both the entrance and substrate cavities of MAO B exhibit a specificity for this isozyme and do not bind to MAO A. These data also point out that MAO B from different species does not exhibit the same inhibitor specificities. Therefore, prudence is advised when extrapolating the conclusions of studies on MAO B from differing species sources to the human enzyme.
Although the structure of bovine MAO B has not been determined, this source of enzyme has served as the standard for numerous MAO B studies, because it is readily available and can be isolated without contamination with MAO A. Although it does exhibit other functional properties in common with MAO B from other sources, the question arises as to why the bovine enzyme contains the Ile/Phe substitution that results in its loss of sensitivity to inhibitors such as trans,trans-farnesol. The similar behavior of the sheep enzyme suggests it also carries this substitution, although the sequence of sheep MAO B is yet undetermined. One common trait between bovine and sheep is that they are both ruminant animals and therefore may be exposed to higher levels of trans,trans-farnesol (isoprenoid levels vary considerably among plant species (28)). Therefore, their MAO B sequences might have evolved differently (both MAO A and MAO B are thought to have evolved from a common ancestral gene) than in other mammals to provide a protective mechanism against MAO B inhibition by components of their diets. This suggestion raises unanswered questions, because other animals (such as the horse), which also subsist on plant materials in their diets (although horses are not ruminants) have MAO Bs exhibiting properties similar to other mammals. Answers to these questions await further genome sequence determinations and further biochemical studies.
The structure of the trans,trans-farnesol-human MAO B complex is of interest from a mechanistic viewpoint. The polar OH group of trans,trans-farnesol is situated 3.4 Å from the flavin C(4a) position, and the 1-CH 2 is positioned 3.4 Å from the N(5) of the flavin in the binding site of MAO B (Fig. 2B). The polar nucleophilic mechanism for MAO catalysis (29) predicts substrate amine attack on the C(4a) position of the flavin as the initial step in the catalytic mechanism with a possible concerted ␣-CH bond cleavage occurring with the flavin N(5) functioning as the base. If the bound trans,trans-farnesol is considered as a substrate mimic, these structural data provide support for the polar nucleophilic mechanism. The lowered nucleophilicity of a hydroxyl group relative to that of an aminyl group would not lead to catalytic oxidation. Preliminary data show the aminyl analogue of trans,trans-farnesol does function as a substrate for MAO B.
An interesting aspect of this work is the finding that the reciprocal mutation in MAO A (F208I) does not lead to binding of this class of inhibitors. This mutation in human MAO A does not abolish catalytic activity but does alter its catalytic properties (12). In wild type MAO A, the gate between the entrance and substrate cavities is presumably formed by Ile-335 and Phe-208, which would be altered to a gate formed by two aliphatic side chains in MAO A F208I. Although MAO A F208I is catalytically competent, its active site is altered (with respect to wild type MAO A) as shown by almost complete loss of isatin binding capability. Thus, other secondary structural effects need to be considered in addition to alterations that would occur as a result of lowering the steric bulk of the amino acid side chain at position 208. The fact that clorgyline still inactivates mutant MAO A is in agreement with previous reports (30) that alterations in the structure of the catalytic site are subtle.
Bovine MAO B exhibits a higher catalytic turnover number (31) than the human I199F mutant enzyme, which merits some discussion. Although the mutation of Ile-199 to Phe results in a "gate" consisting of two aromatic residues (Phe-199 and Tyr-326) that would exhibit a higher rigidity, this cannot be the explanation for the observed differences in k cat , because the bovine enzyme also contains a Tyr residue at this position. It is likely that the bovine enzyme has evolved structural differences through other amino acid substitutions to achieve a catalytically more efficient enzyme. It is of interest that other studies have shown that the Y326I mutation in MAO B leads to an enzyme with substrate and deprenyl/clorgyline sensitivities more similar to MAO A and that the I335Y mutation of human MAO A results in a mutant enzyme with closer properties to MAO B (30). Structural data of human MAO B show that Tyr-326 is located near the junction of the entrance and substrate cavities and that the phenolic side chain of this residue forms one of the walls of the substrate cavity. Therefore, replacement of this bulky aromatic ring with an aliphatic side chain may relieve some of the steric constraints in the substrate cavity that have been documented with MAO B and found to be less constraining in MAO A by previous QSAR studies (5,32). Although single amino acid substitutions can influence the respective individual properties of MAO A and of MAO B, their differences in substrate and inhibitor specificities are due to more complex structural alterations that probably result from contributions of multiple sequence changes. providing rasagiline and their continuous interest in the project. Milagros Aldeco provided excellent technical assistance with this project.