A stable tyrosyl radical in monoamine oxidase A.

We present spectroscopic evidence consistent with the presence of a stable tyrosyl radical in partially reduced human monoamine oxidase (MAO) A. The radical forms following single electron donation to MAO A and exists in equilibrium with the FAD flavosemiquinone. Oxidative formation of the tyrosyl radical in MAO is not reliant on neighboring metal centers and uniquely requires reduction of the active site flavin to facilitate oxidation of a tyrosyl side chain. The identified tyrosyl radical provides the key missing link in support of the single electron transfer mechanism for amine oxidation by MAO enzymes.

The mammalian monoamine oxidases (MAO) 1 A and B are flavoproteins localized to the outer mitochondrial membrane. MAO catalyzes the oxidative deamination of neurotransmitters and exogenous alkylamines. The human enzymes are important pharmaceutical targets for antidepressants, and inhibitors of MAO B are used synergistically with L-DOPA in the treatment of Parkinson disease (1). Elevated levels of MAO B induce apoptosis in kidney (2) and neuronal cells (3) and are also associated with plaque astrocytes in the brains of Alzheimer patients (4). The anti-apoptotic action of a MAO B inhibitor is important in novel Alzheimer treatments (5). MAO is also implicated in the onset of Parkinson syndrome through bioactivation of 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine, an impurity in many sources of synthetic heroin (6).
Despite the recent crystal structures of MAO B (7) and MAO A (8) and extensive literature on substrate and inhibitor specificities, the mechanism of substrate oxidation remains obscure (9). Much of the debate has centered on the possible existence of radical species, direct evidence for which has not been forthcoming. An early proposal invoked a polar nucleophilic mechanism involving attack of the deprotonated amine substrate at the flavin C4a to form a substrate-flavin C4a adduct and proton abstraction from the ␣-carbon of the adduct by an active site base (10). Support for this mechanism came from chemical model studies in reactions of amines with lumiflavins (11,12). Studies of quantitative structure-activity relationships with MAO B have been used to support a second mechanism in which substrate ␣-C-H bond cleavage is via direct hydrogen atom transfer to a protein-based non-flavin radical followed by electron transfer to the flavin (13,14). An organic radical species was originally reported in EPR spectra of resting bovine liver MAO B (15) but later was attributed to an artifact of purification of MAO B from bovine liver following EPR studies of highly purified recombinant sources of MAO A and MAO B (16). Edmondson and Miller (16) have proposed a concerted polar nucleophilic mechanism for MAO A involving a substrate-flavin C4a adduct and proton abstraction by the highly basic N-5 atom of the flavin. This mechanism is consistent with studies of quantitative structure activity relationships and kinetic isotope effects and with the apparent lack of an organic protein-based radical in EPR spectra of the resting form of the enzyme. The aminium cation radical mechanism for MAO proposed by Silverman et al. (17) involves single electron transfer from the substrate nitrogen lone pair to yield the substrate radical and flavin semiquinone. This mechanism is based on the susceptibility of amines to undergo single electron transfer chemistry during electrochemical and chemical oxidations (17) and is consistent with the results of mechanism-based inhibitor studies with a series of cyclopropyl inhibitors, which undergo rapid ring opening on formation of the cyclopropylaminyl radical (reviewed in Ref. 9). The aminyl radical cation mechanism is consistent with electronic effects observed in quantitative structure activity relationships studies with a series of substituted benzylamines (16), but the identity of the 1-electron oxidant required for formation of the aminyl radical cation remains a major concern (16). Edmondson and Miller (16) argue that the ground state of flavin (E m Ϸ Ϫ0.2 to 0 V, where E m is midpoint redox potential) is unlikely to oxidize a primary amine (E m ϭ ϩ1.5 V versus standard calomel electrode, although some oxidation occurs at much lower potentials), but based on a number of arguments, large apparent barriers to electron transfer might not prevent the reaction in an enzymeactive site (e.g. distortion of substrate by intrinsic binding to alter the geometry of the substrate and/or flavin (18), perturbation of reduction potential by substrate binding (19), and endergonic tunneling over short distances (20) followed by a thermodynamically favorable step in the reaction pathway). That a protein-bound oxidant is the 1-electron acceptor has been considered, but detectable EPR signals have not been found in purified recombinant MAO A or MAO B.

MATERIALS AND METHODS
Enzyme Purification-Purified human liver monoamine oxidase A (heterologously expressed in Saccharomyces cerevisiae) (21) 1 The abbreviations used are: MAO, monoamine oxidase; ENDOR, electron nuclear double resonance; W, watt. pH 7.2, containing 0.05% Brij-35, was made anaerobic in an anaerobic UV-visible cuvette by cycling with argon. The concentration of MAO A was calculated from the absorbance at 456 nm using an extinction coefficient of 12,800 M Ϫ1 cm Ϫ1 (22). Activity with 1 mM kynuramine was measured at 314 nm in 50 mM potassium phosphate, pH 7.2, 0.05% Triton X-100 (23). Samples prepared for EPR and ENDOR analysis were titrated in the absence of redox mediators.
EPR and ENDOR Spectroscopies-EPR and ENDOR experiments were performed on a Bruker ELEXSYS E500/560 spectrometer operating at X-band employing the Super High Q cylindrical cavity for EPR (Q factor ϳ16,000) and the EN801 cavity for ENDOR (Q factor ϳ2000). Temperature control was provided using an Oxford Instruments ESR900 helium flow cryostat linked to an ITC503 temperature controller (both supplied by Bruker).
Reductive Titration-Reductive titration of MAO A was performed in a Belle Technology glove box under a nitrogen atmosphere, essentially as described previously (24). All solutions were degassed under vacuum with argon. Oxygen levels were maintained at Ͻ2 ppm. The protein solution (ϳ50 M in 5 ml of the same buffer described for the purification) was progressively reduced by the addition of 0.5-and 1.0-ml aliquots of a sodium dithionite stock solution (ϳ100 mM) prepared anaerobically in 100 mM potassium phosphate, pH 7.0. Mediators (2 M phenazine methosulfate, 5 M 2-hydroxy-1,4-naphthoquinone, 0.5 M methyl viologen, and 1 M benzyl viologen) were included to mediate in the range between ϩ100 and Ϫ480 mV, expediting electronic equilibration at each point in the titration. Spectra (300 -700 nm) were recorded with a Cary UV-50 Bio UV-visible scanning spectrophotometer using a fiber optic probe immersed in the protein solution and connected externally to the spectrophotometer. Electrochemical potential of the solution was measured using a Hanna pH 211 meter coupled to a Pt/Calomel electrode (Thermo Russell Ltd.) at 25°C. The electrode was calibrated using the Fe 3ϩ /Fe 2ϩ EDTA couple as a standard (ϩ108 mV). A factor of ϩ244 mV was used to correct relative to the standard hydrogen electrode. Spectral data were imported into Origin (Microcal), and spectral subtractions were performed to correct for base-line drift during the titration (bringing absorption back to zero at 700 nm, where there is no significant absorption from the flavin cofactor in oxidized or reduced states).

RESULTS AND DISCUSSION
The studies reported herein provide the first evidence for the presence of a protein radical, in the form of a tyrosyl radical, which is formed in partially reduced MAO. Reduction of an anaerobic solution of MAO A using sodium dithionite gives rise to the EPR spectrum shown in Fig. 1A, spectrum a. This spectrum exhibits features not observed for typical anionic flavosemiquinones (25). Typical anionic flavosemiquinone EPR spectra include those formed by electron-transferring flavoprotein or glucose oxidase at pH 9. When Fig. 1A, spectrum b, an example of these spectra, is subtracted, the result is a residual spectrum shown in spectrum c. 2 This spectrum exhibits a g av of 2.0042, which is too high to arise from the flavosemiquinone FIG. 1. A, EPR spectra of the following: a, human MAO A reduced using sodium dithionite; b, the stable anionic flavosemiquinone formed in Methylophilus methylotrophus electron-transferring flavoprotein; c, the difference spectrum (a Ϫ b); d, the neutral tyrosyl radical Y D ⅐ formed in P. laminosum photosystem 2; and e, simulation of the MAO A tyrosyl radical using the hyperfine coupling constants from Table I. Experimental conditions for a and b were microwave power 2 W, modulation frequency 100 kHz, modulation amplitude 1.2 G, at a temperature of 70 K. For d, the conditions were microwave power 10 W, modulation frequency 100 kHz, modulation amplitude 1.6 G, at a temperature of 30 K. For e, simulation employed the program POWFUN (courtesy of the late Prof. Gerry Babcock) and the G values and Euler angles for tyrosyl radicals given in Ref. 20. B, EPR spectra of the MAO A tyrosyl radical formed by subtracting spectra of the electron-transferring flavoprotein anionic flavosemiquinone from spectra of sodium dithionite-reduced MAO at the temperatures indicated in a-f. Experimental conditions were microwave power 500 nW, modulation frequency 100 kHz, modulation amplitude 1.2 G. radical (2.0032) (25) but is consistent with assignment to a neutral tyrosine (tyrosyl) radical (26). This radical accounts for 18% of the total number of unpaired electrons observed. The EPR spectrum c of Fig. 1A also shows the partially resolved hyperfine splitting that is typical of such neutral tyrosine radicals as Y D ⅐ of photosystem 2 (27) , Fig. 1A, spectrum d, but the MAO radical has a smaller line width than reported previously for a tyrosyl radical. Such narrowing of the EPR spectrum can occur when the unpaired electron is exchanging between two sites or species on a time scale that is rapid compared with the spectrometer frequency (28). Rapid exchange might occur between spatially close tyrosine residues in the active site of MAO (6). The rate of electron exchange between the two sites will be temperature-dependent, and thus the EPR spectrum of such a species should also be temperature dependent. The EPR spectra of the presumptive tyrosyl radical (formed by subtraction of an anionic flavosemiquinone spectrum recorded under the same conditions as shown in Fig. 1, except temperature), recorded at various temperatures, are shown in Fig. 1B. These spectra show clear evidence of temperature-dependent broadening, which is not seen for other known tyrosyl radicals (e.g. Y D ⅐ ). Therefore, a possible explanation for the narrow line width of the presumptive tyrosyl radical is exchange narrow-ing. However, the relative proportions of flavosemiquinone and tyrosyl radical are the same at all the temperatures employed, so there is no temperature-dependent redistribution of the unpaired electron between these species. ENDOR spectroscopy allows for the measurement of the hyperfine couplings to the protons of the tyrosyl radical, as shown in the spectrum of Fig. 2A, spectrum a, and tabulated in Table I. Only those features to high frequency of the Larmor frequency are shown, as the symmetrically displaced features to low field of the Larmor frequency are typically weak and difficult to discern (27). The experimental conditions employed were chosen to emphasize the contribution of the tyrosyl radical to the ENDOR spectrum at the expense of the radical of the overlapped flavosemiquinone anion. Fig. 2A, spectrum b, shows the ENDOR spectrum of Phormidium laminosum Y D ⅐ , the  Following the addition of mediators and a minimal amount of dithionite reductant, a small amount of the tyrosyl radical is formed (green spectrum). With progressive addition of sodium dithionite, the flavin is reduced further (absorption decrease at 456 nm, indicated by black arrow; selected spectra are shown as black dashed lines), whereas the intensity of the tyrosyl radical increases at 412 nm (blue arrows), reaching a maximum at ϳ0 mV (spectrum shown as solid red line). The red anionic form of the MAO A semiquinone also accumulates in this range, with a distinctive absorption at ϳ365 nm (red arrows). Further reduction by dithionite leads to progressive bleaching of the flavin and to gradual diminution of the radical signal (red dotted lines). The spectrum for the fully reduced MAO A (blue solid line; recorded at ϳϪ200 mV) indicates that the spectral contribution from both flavin and tyrosyl radicals are removed. known tyrosyl radical that exhibits spectra most similar to that of MAO A (27). Simulation of the EPR spectrum using these hyperfine coupling constants and G values for tyrosyl radicals from the literature produces a spectrum similar in shape to Fig.  1A, spectrum c, but without the exchange narrowing, as shown in Fig. 1A, spectrum e. The rotameric angle of the tyrosyl radical ring relative to the ␤-CH 2 group (see Fig. 2B) can be determined from the ␤-CH 2 hyperfine coupling constants using the Heller-McConnell equation (29), where A iso is the isotropic hyperfine coupling constant for the ␤ proton, B is a constant (162 MHz), is the unpaired electron spin density at C(1) of the tyrosyl radical, and is the rotameric angle between the C(␤)-H bond and the normal to the tyrosine ring plane. The angles calculated for the two ␤-CH 2 protons are 54.6°and 61.1°. The structure of MAO B shows that tyrosine residues Tyr-60, Tyr-398, and Tyr-435 (equivalent to Tyr-69, Tyr-407, and Tyr-444 in MAO A) are in the vicinity of the active site (7). Of these three, Tyr-398 (Tyr-407 in MAO A) has values closest to those measured for the tyrosyl radical studied here (ϳ50°and 70°), and the orientation looks similar in the MAO A structure. Mutation of Tyr-69 to alanine, serine, or phenylalanine had no effect on activity. Mutation of tyrosines 407 and 444 to serine (Y407S, Y444S) in MAO A leads to loss of activity (30); however, mutation to phenylalanine has a much larger effect at position 444 (Y444F) than 407 (Y407F). Al-though these data demonstrate a role for tyrosine residues, given the results of the temperature dependence studies (Fig.  1B), it may not be useful to think of the radical as being located on one identifiable tyrosine residue but rather as delocalized over a number of residues. Thus, mutation of individual tyrosine residues may not have a large impact on activity.
Previous EPR studies have not revealed the presence of the tyrosyl radical reported here (15,31). However, the conditions employed in these studies were not optimized for the detection of the additional features shown in Fig. 1A, spectrum a, that provided our first evidence for the existence of the tyrosyl radical. Optical spectra of MAO reduced under anaerobic conditions using sodium dithionite (22,31,32) consistently show formation of the anionic flavosemiquinone, indicated by a band at ϳ365 nm and a narrow feature at ϳ412 nm (Fig. 3). Although the latter could arise from an anionic flavosemiquinone (33), it seems that the extinction coefficient of this feature in MAO A is not in keeping with those reported for other anionic flavosemiquinones (33). An alternative source of the 412 nm band is the identified tyrosyl radical, as such radicals are known to give rise to sharp bands at 410 -415 nm (34). A similar 412 nm absorption band is also observed during equilibrium titration of human MAO B with sodium dithionite, consistent also with the presence of a tyrosyl radical as part of a similar redox equilibrium to that seen in MAO A (31). In MAO A, the optical data indicate that the 412 nm absorption is maximal at ϳ0 mV and is populated with the flavin semiquinone (Fig. 3). The absorption at 412 nm disappears on further reduction of the enzyme, indicating that the tyrosyl and flavin semiquinone form a redox equilibrium in partially reduced MAO A.
The formation of a tyrosyl radical in sodium dithionite-reduced MAO A reported here provides the key missing link in support of a single electron transfer mechanism for amine oxidation. A variation on the aminoalkyl (cation) radical mechanism for amine oxidation employing a tyrosyl radical (Fig. 4) that was proposed by Silverman et al. (17), similar to that proposed previously by Edmondson (35), would account for our observations. This scheme requires that a redox equilibrium exists between the flavosemiquinone and a tyrosine radical in the active site. Formation of flavosemiquinone by titration with sodium dithionite establishes this equilibrium leading to the observed formation of the tyrosyl radical (Fig. 4A), and this accounts for why previous workers have not obtained evidence for an organic radical in the resting form of the enzyme. During catalytic turnover, reduction of the enzyme by single electron transfer from the substrate to flavin also generates a redox equilibrium with appearance of the tyrosyl radical and the aminyl radical cation intermediate (Fig. 4B). Upon establishing the tyrosyl radical, the reaction could proceed either by direct hydrogen transfer (as shown in Fig. 4B) or alternatively by proton transfer and formation of a radical centered on the ␣-carbon route. The latter route is consistent with the electronic effects seen in quantitative structure activity relationships analysis (16) of MAO A and with studies of mechanismbased inhibitors (36). We infer that the radical is short lived in reactions with substrate, thus accounting for the lack of optical signature for the tyrosyl radical species in stopped-flow studies with substrate (13). This is consistent with (i) reversible electron transfer from substrate to flavin with the reverse reaction being fast and (ii) rapid establishment of an equilibrium with FADH 2 and the tyrosyl radical species. The transient appearance of the tyrosyl radical by electron transfer from substrate to flavin prevents adventitious reaction of a highly reactive species in the resting form of the enzyme and thus ensures that radical chemistry occurs only in the presence of substrate.