Synthesis and Properties of 8-CN-flavin Nucleotide Analogs and Studies with Flavoproteins*

A high potential analog of riboflavin with a cyano function at the 8-position was synthesized by employing novel reaction conditions, starting from 8-amino-riboflavin. This was converted to the FAD level with FAD synthetase. The reduced 8-CN-riboflavin, unlike normal reduced flavin, has a distinctive absorption spectrum with two distinctive peaks in the near ultraviolet region. The oxidation-reduction potential of the new flavin was determined to be -50 mV, ∼160 mV more positive than that of normal riboflavin. The 8-CN-riboflavin and 8-CN-FMN were found to be photoreactive and need to be protected from exposure to light. However such complications were not encountered with protein-bound flavins. The apoproteins of flavodoxin and Old Yellow Enzyme (OYE) were reconstituted with the 8-CN-FMN and apoDAAO was reconstituted with 8-CN-FAD. Spectral properties of the enzyme-bound neutral and anionic semiquinones were determined from these reconstituted proteins. In the case of 8-CN-FMN-OYE I, it was shown that the comproportionation reaction of a mixture of reduced and oxidized enzyme bound flavin is very rapid, compared with the same reaction with native protein, resulting in ∼100% thermodynamically stable anionic semiquinone. In the case of 8-CN-OYE I, it was shown that the rate of reduction of the enzyme bound flavin by NADPH is ∼40 times faster, and the rate of reoxidation of reduced enzyme bound flavin by oxygen is an order of magnitude slower than with the normal FMN enzyme. This is in accord with the high oxidation-reduction potential of the flavin, which thermodynamically stabilizes the reduced enzyme.

A wide variety of redox transformations in biological systems are catalyzed by flavoenzymes. The oxidation-reduction potential of the flavin and the transfer of electrons by the semiquinone or the fully reduced forms to the acceptor are among the most significant features of the chemistry of flavoprotein catalysis, which is in general controlled by the potential of the flavin. Hence it is possible to regulate and manipulate electron flow during catalysis by altering the redox potential of the flavin. Various structurally modified flavins with altered redox potentials have been synthesized previously and employed as mechanistic probes (1)(2)(3)(4). While deazaflavin derivatives best represent the low potential probes, flavin analogs with various electron withdrawing groups at the 8-position constitute the high potential series. It is known that the introduction of elec-tronegative groups like chlorine (5), fluorine (6), methylsulfonyl (7,8) at the 8-position of isoalloxazines increases electrophilicity and shifts the potential to more positive values. However these functionalities are highly reactive and can undergo displacement reactions either with external nucleophiles or with nucleophilic amino acid residues like cysteine at the active site. Accordingly these properties have been exploited to advantage in determining the solvent accessibility of the 8-position of the flavin in various flavoproteins (9) and also for the covalent labeling of active site amino acid residues (10 -12). Elegant model studies by Bruice et al. (13) with 8-CN-isoalloxazines have demonstrated that cyanylation at the 8-position affords an isoalloxazine with one of the highest known potentials along with very interesting chemical and spectral properties. Although the CN functionality is strongly electronegative, it is resistant to displacement reactions, since these would involve carbon-carbon bond breaking. Also the relatively small size of the CN substitution meets the steric requirements for a good active site probe. Hence various laboratories have attempted the synthesis of 8-CN-flavin nucleotide derivatives by displacement reactions on flavins having leaving groups at the 8-position (e.g. chlorine, fluorine, methylsulfonyl, etc.,) and also by the Sandmeyer reaction on 8-NH 2 -riboflavin, but without success. Attempts to synthesize the aromatic building block with CN substitution for the construction of the flavin were also a failure. In view of the relevance of this analog as a high potential active site probe for flavoproteins, we decided to reinvestigate the above approaches. The aromatic building block, 3-nitro-4-chloro-6-CN-toluene, which was successfully synthesized, failed to react with ribitylamine. 1 However, we discovered that 8-NH 2 -riboflavin can be converted to 8-CN-riboflavin under novel reaction conditions as described in this paper. The riboflavin derivative is readily converted to the FAD and FMN derivatives by the FAD synthetase of Brevibacterium ammoniagenes (14).

MATERIALS AND METHODS
8-NH 2 -riboflavin, which was synthesized as described previously (15), was a kind gift from Dr. S. Ghisla, University of Konstanz. Copper(II) cyanide and sodium cyanide were from Aldrich. 8-NH 2 -FAD was prepared as reported previously (16).
Preparation of Flavoproteins and Their Corresponding Apoproteins-The holoproteins and apoproteins were prepared as described previously: riboflavin binding protein from hen egg white (17), flavodoxin from Megasphera elsdenii (18,19), and OYE I, 2 overexpressed in Escherichia coli containing the plasmid pET-3b (20). The apoOYE I was prepared from the recombinant enzyme by using the procedure reported for the enzyme from brewers' bottom yeast (21,22). The apoDAAO from pig kidney was obtained from Calzyme (San Luis Obisco, CA).
Reconstitution of Apoproteins-ApoRBP and apoflavodoxin concentrations were determined by titration with pure riboflavin and FMN, respectively. ApoDAAO concentration was determined by titrating pure FAD with apoprotein in the presence of benzoate (23). Reconstitution of the apoproteins with 8-CN-flavins was accomplished by mixing 1.5-fold excess of the flavin with apoprotein and incubating on ice for 3 h. Excess flavin was removed by a Centricon-30 microconcentrator (Amicon).
Synthesis of 8-CN-riboflavin-10 mg of 8-amino-riboflavin was suspended in 3 ml of water in a test tube. To this suspension, 6 N HCl was added until a clear solution was obtained. This solution was cooled to 0°C on ice, and three aliquots each of 40 l of saturated sodium nitrite solution were added with continuous shaking of the test tube. After 5 min, 300 l of saturated urea solution was added to destroy the excess sodium nitrite. The cold diazo salt solution was then added with a glass transfer pipette to a 10-ml saturated solution of NaCN ϩ CuCN (70:30) in a 50-ml glass beaker with vigorous stirring at room temperature. After 20 min, the reaction mixture was loaded on a 20-cc C-18 Sep-Pak (Millipore, Milford, CT) cartridge. The cartridge was prewashed thoroughly with excess water, methanol, and again with water before loading the reaction mixture. The Sep-Pak cartridge was eluted with water followed by 5% acetonitrile to remove salts and a red band of unknown structure. Elution with 15% acetonitrile gave the 8-CN-riboflavin and with 20% acetonitrile gave 8-chlororiboflavin. Evaporation of the flavin solutions with a Speed Vac concentrator gave 6 mg of the 8-CN-riboflavin as a yellow powder.
Conversion of 8-CN-riboflavin to FAD and FMN-8-CN-riboflavin was converted to the FAD level with partially purified FAD synthetase from B. ammoniagenes by incubating the flavin in 0.002 M potassium P i , pH 7.5 at 25°C, following the procedure of Spencer et al. (14). After 14 h, HPLC analysis of the incubation mixture showed 100% conversion to the FAD form. The reaction mixture was loaded on to the prewashed (as described in the case of 8-CN-riboflavin) 20-cc C-18 Sep-Pak cartridge and eluted with 100 ml of water to wash off salts and breakdown products from ATP. Elution with 5% acetonitrile in water gave pure flavin, which was concentrated on a Speed Vac concentrator to obtain 8-CN-FAD as a yellow powder. 8-CN-FMN was obtained by hydrolysis of the FAD in 0.05 M potassium P i , pH 7, with snake venom phosphodiesterase (Naja naja venom).
Cyanylation of 8-NH 2 -FAD-To 500 l of 8-NH 2 -FAD, 100 l of 6 N HCl was added at ice temperature. Then two 20-l aliquots of saturated NaNO 2 were added with mixing. After 2 min, 100 l of saturated urea solution was added. When effervescence stopped, 500 l of a saturated solution of 70:30 NaCN ϩ CuCN solution was added into the diazo-FAD solution at room temperature. After 3-5 min, the solution was passed through a prewashed small Sep-Pak cartridge, and the salts were washed off with water. The flavin was then eluted with 5% acetonitrile. HPLC analysis of the 5% acetonitrile fraction showed both 8-CN-FAD (ϳ80%) and 8-CN-FMN (ϳ20%) when eluted with 80% 0.01 M potassium P i , pH 6, and 20% methanol on a reverse phase C-18 column. Interestingly, no formation of 8-chloro-FAD or 8-chloro-FMN was observed.
Enzyme Assays-Old Yellow Enzyme activities were measured in 0.1 M phosphate, pH 7.0 at 25°, in a stopped flow spectrophotometer (Kinetic Instruments, Ann Arbor, MI) either under anaerobic conditions (when cyclohexenone was employed as electron acceptor) or under controlled oxygen concentrations when oxygen was electron acceptor, monitoring both the consumption of NADPH at 340 nm and in the same experiments the level of flavin oxidation/reduction at 470 nm. The concentrations of both NADPH and the acceptor (cyclohexenone or O 2 ) were varied systematically to determine true k cat and K m values.

RESULTS AND DISCUSSION
Attempts to convert the 8-diazo salts of flavins to 8-CNflavins by the Sandmeyer reaction were unsuccessful. The choice of the cyanylating reagent has a critical effect on the course of the reaction (24,25). For the 8-diazo-flavins, in our hands, either sodium cyanide or cuprous cyanide alone were ineffective. Lately several copper-cyano complexes of the kind Na 3 (Cu(CN) 4 ), K 3 (Cu(CN) 4 ), K 2 (Cu(CN) 4 ⅐NH 3 ) were shown to have large advantages in cyanylation reactions affording high yields of nitriles (26). Before trying these complexes for the present cyanylation, we tried treating the diazoflavin with a saturated solution of an approximately 3:1 mixture NaCN ϩ CuCN and found 8-CN-riboflavin in about 60% yield (Scheme 1). The 8-chlororiboflavin was obtained as a side product in about 20% yield, presumably because of the CuCl formed from CuCN and HCl. It was found that the ratio between NaCN and CuCN is crucial for both the reaction to occur as well as to obtain isolatable yields of the cyanoflavin. When the CuCN ratio was increased, an insoluble solid was obtained with no isolation of any flavin.
The max of the UV-visible spectrum for 8-CN-riboflavin is slightly shifted from 445 nm for the normal flavin spectrum to 456 nm. The near-UV peak is shifted from 375 nm for the normal flavin to 338 nm for 8-CN-riboflavin (Fig. 1A). The cyanoflavin is fluorescent with an emission maximum at 530 nm and excitation maxima at 338 and 456 nm, identical with the absorption spectrum. Further characterization of the new flavin was made by recording the positive ion fast atom bombardment mass spectrum which showed M ϩ1 at 388 (Fig. 1B) (molecular weight of the 8-CN-riboflavin is 387) and also by the 1 H NMR spectrum (Fig. 1C). In the proton NMR spectrum, the electron-deficient nature of the flavin is well evident from the fact that the C6 and C9 protons are down-shifted and well separated due to the CN substitution in the benzene ring. The aromatic protons are at 8.0 and 8.4 ppm compared with 7.9 and 8.0 ppm in the normal flavin. The lone methyl group on the C7 is seen as a singlet at 2.6 ppm, down-shifted from 2.3 ppm in normal flavin. The rest of protons from the ribityl side chain are recorded between 3.6 and 4.8 ppm.
The 8-CN-riboflavin was converted to the FAD level by reacting with FAD synthetase from B. ammoniagenes in 100% yield without any complications. This again reaffirms the fact that the small cyano substitution meets the steric requirements of the 8-position of the flavin. The 8-CN-FMN was obtained by hydrolysis of the HPLC-pure 8-CN-FAD with the phosphodiesterase from snake venom. This reaction is accompanied by an increase in extinction for the FMN at 456 nm of 8.8% and with a 10.25-fold increase in fluorescence. The cyanylation reaction also worked with the 8-NH 2 -FAD, except that about 20% of the FAD got hydrolyzed to 8-CN-FMN because of the drastic acidic conditions of the diazotization reaction.

Reduction of 8-CN-riboflavin
It was found that the reduction of 8-CN-riboflavin to its dihydro form can be accomplished with several reducing agents such as dithiothreitol, dithionite, NaBH 4 , photochemically with EDTA as photodonor, NADPH, and NADH. The reduction is fully reversible with oxygen. However the spectrum for reduced 8-CN-riboflavin is different from that of normal flavin, as reported previously for 8-CN-7-nor-3-methyl-lumiflavin by Bruice et al. (13). Reduced 8-CN-riboflavin has two well defined peaks above the 300 nm region with maxima at 312 and 372 nm at pH 7 and at 316 and 362 nm under acidic conditions (Fig. 1A).

The pK a of Reduced Cyanoriboflavin
The pK a value of the reduced cyanoflavin was determined to be at pH 5.6 from the change in absorption spectrum with pH ( Fig. 1A, inset). Hence the electronegative 8-CN-group de-SCHEME 1

8-CN-flavins and Flavoproteins
creases the pK a of the reduced flavin from the usual value (27) of ϳ6.7 to 5.6 in accord with the electron-deficient nature of the flavin. The spectra of the neutral and anionic reduced forms are shown in Fig. 1A.

Determination of Reduction-Oxidation Potential
The reduction-oxidation potential for 8-CN-FAD was measured by using the xanthine/xanthine oxidase system and indigo tetrasulfonate as the reference dye (28). Reduction of the 8-CN-FAD with the xanthine/xanthine oxidase system proceeded with isosbestic points at 352 and 406 nm and the reduction of the dye with isosbestic points at 338 and 502 nm. These wavelengths were used to monitor the reduction of the components in a mixture of the two. A plot of log(ox/red) of the dye against log(ox/red) of the flavin gave a midpoint potential for 8-CN-FAD of Ϫ50 mV, which is 158 mV more positive than that of normal flavin. This is in accord with the strongly electron withdrawing nature of the CN substituent, making it a more electron deficient system, shifting the potential to more positive values.

Reaction with Sulfite
Native flavin is known to form an N-5 adduct with sulfite, but the reaction is only half completed at saturating sulfite concentrations because of the high dissociation constant (ϳ2.5 M) (29). The absorption spectrum of the flavin adduct is similar to that of reduced flavin with a new peak maximal at 320 nm. With a series of artificial flavins a good correlation was found to exist between the dissociation constant of the flavin-sulfite adduct and the two-electron redox potential of the flavin (29). The straight line obtained by plotting the logarithm of the dissociation constants of the complexes and the redox potentials of the flavins shows that the more positive the potential, the tighter the sulfite binding. It is reasonable that flavins with more positive potentials are the most electron deficient and prone for the addition of two electrons through adduct formation. As anticipated, the cyanoflavin formed a tight complex with sulfite, with a K d of 0.73 mM (Fig. 2, inset). The absorption spectrum of the adduct has a peak in the near-UV region at 310 nm (Fig. 2), similar to that with normal flavin and distinctively different from that of reduced 8-CN-flavin. From the previously observed correlation of the redox potential and K d for the complex (29), the oxidation-reduction potential at pH 7 would be predicted to be Ϫ50 mV, in perfect agreement with direct measurement.

Hydrolysis of 8-CN-riboflavin
The hydrolysis of 8-CN-riboflavin in 0.5 M carbonate, pH 10.25, was studied. The spectral changes that were observed are similar to those of the cyanoisoalloxazine observed by Bruice et al. (13) and compatible with formation of a spirohydantoin (30). As is the case with the cyanoisoalloxazine derivative, repetitive spectral scans during the hydrolysis failed to provide any evidence for the 10a-or 4a-hydroxyl adducts, which were considered as possible intermediate species during hydrolysis.

Photostability of 8-CN-flavin Derivatives
It is well known that flavins are photoreactive and their spectroscopic and photochemical properties have been the subject of intense studies (31). It was found that 8-CN-riboflavin is highly photoreactive. When a sample of the flavin was left exposed to room light, it was transformed over a period of a few days into a stable product with absorbance maxima at 320 and 388 nm. A similar reaction was also observed with the 8-CN-FMN derivative. However when 8-CN-FAD was exposed under the same conditions, it was found to be extremely stable toward photoreaction. Only about 5% of the reaction was observed in the same time scale where 8-CN-riboflavin and 8-CN-FMN reacted completely. It is known that FAD forms an intramolecular complex between the adenine moiety and the isoalloxazine ring (32). The increased stability of FAD toward photolysis is attributed to this intramolecular stacking (33). The photoreaction involves the ribityl side chain, since 8-CN-lumiflavin is completely stable under the same conditions. The nature of the product and the chemistry of the photoreaction is under investigation and will be reported separately.

Riboflavin-binding Protein
Although this protein has no known catalytic activity, binding experiments of artificial flavins with this protein are useful in interpreting the structural characteristics of the flavin. Riboflavin binds to apoRBP with a dissociation constant of 1.3 nM and shows considerable spectral shifts and resolution on binding (17). 8-CN-riboflavin showed similar spectral changes to those of normal flavin upon binding to apoRBP. The flavin absorption peaks are red shifted from 340 and 454 nm to 344 and 470 nm with a decrease in the extinction for both peaks. The binding of 8-CN-riboflavin to apoRBP resulted in the complete quenching of the fluorescence. The dissociation constant was calculated to be ϳ0.27 M and by standardization of apoprotein with native riboflavin, the extinction coefficient of the 8-CN-riboflavin was determined as 10,400 M Ϫ1 cm Ϫ1 . The 5-deazaflavin-catalyzed photoreduction (34) of 8-CN-riboflavin-RBP proceeded to the fully reduced form without stabilizing any semiquinone. The spectrum for the reduced protein-bound flavin is almost identical to that of the free reduced flavin except that the max is slightly red shifted by 4 nm to 376 nm. The initial oxidized flavin spectrum is regained rapidly on admission of air.  (Fig. 3). The blue neutral semiquinone shows maxima at 598 and 644 nm. Previous studies with native flavodoxin showed essentially quantitative formation of the neutral semiquinone at the midpoint of dithionite reduction (36). However, in case of the 8-CN-FMN-protein the plot of A 458 versus A 644 (see the inset in Fig. 3) shows that only ϳ75% of the radical species was formed. This suggests that the potentials of the EFl ox /EFlH ⅐ and EFlH ⅐ / EFl red (where Fl is flavin) couples are closer than those of the native flavoprotein. The reduced flavoprotein showed two well resolved peaks in the absorption spectrum similar to the unbound reduced flavin. However the peak in the near-UV region shifted further to the lower wavelength region and the band at 370 nm in the free flavin showed a bathochromic shift to 388 nm. Interestingly, 25% of the absorption at the 460 nm region is retained with a shifted absorption band at 485 nm (Fig. 3). There is no adduct formation with sulfite in accordance with studies of the native protein (37).

Old Yellow Enzyme
Old Yellow Enzyme is the first discovered flavoprotein and was isolated from brewers' bottom yeast. It is a mixture of homodimers and a heterodimer with one FMN per subunit, products of separate genes (38,39). As the physiological role of this protein is yet to be determined, structural and chemical reactivity studies in the direction of elucidating its function are the subject of extensive studies in this laboratory. Saito et al. (20) cloned a gene encoding an isoform of OYE (OYE I) from Saccharomyces carlsbergenesis, and its crystal structure has been determined (40). For the present investigations, this recombinant enzyme (OYE I) was used. The binding of apoOYE with 8-CN-FMN was followed by measuring the flavin fluorescence as well as changes in absorption spectrum. At the end point, the fluorescence was quenched almost completely with a residual intensity of ϳ6% of that of the free flavin. The titration plots suggested a binding affinity of the 8-CN-FMN to apoOYE similar to that with FMN (K d ϳ 10 Ϫ9 M) (41). The extinction coefficient of the cyano-FMN was determined as 11,000 M Ϫ1 cm Ϫ1 by standardizing the apoprotein with pure native FMN. Binding of Phenols-The oxidized form of OYE binds to a variety of ligands forming spectroscopically distinct complexes (38). Phenols are the most striking and well studied as they result in long wavelength charge transfer bands in the region 500 -800 nm accompanied by strong perturbation in the flavinvisible absorption bands (21,22). It was shown by a positive correlation between the energy of the charge transfer transition and the Hammett para constant that the phenol is the charge transfer donor and the oxidized flavin of the enzyme is the acceptor (22). A good correlation was also shown to exist between the redox potential of the flavin and the maximum of the long wavelength transition, from studies where the native flavin was replaced by a variety of synthetic flavin analogs of different oxidation-reduction potential (22). Native enzyme binds p-chlorophenol with a K d of 1 M and has a maximum of 645 nm for the long wavelength charge transfer band. The 8-CN-FMN enzyme binds p-chlorophenol with a K d of 6.5 M and the charge transfer maximum is shifted to 763 nm, but with less than the usual perturbations of the spectrum in the 450 nm region (Fig. 4). As the oxidation-reduction potential of the flavin becomes more positive, the wavelength maximum should shift to longer wavelengths, since it is expected to require less energy to form the charge transfer transition. As will be discussed later, the 115 nm red shift observed in case of 8-CN-FMN enzyme falls in line with the more positive potential compared with that of the native flavin.
Reduction of 8-CN-FMN-OYE-The reduction of 8-CN-FMN enzyme with the xanthine/xanthine oxidase system with benzylviologen as mediator was carried out at 25°C in 100 mM potassium P i , pH 7 (Fig. 5). It was found to proceed through the formation of the anionic semiquinone to the two-electron reduced form. The plot of A 470 versus A 416 suggest that ϳ 98% of the semiquinone radical species was formed (Fig. 5, inset). The reduced 8-CN-FMN enzyme showed a peak at 360 nm, a blue shift of 10 nm from that of the free cyanoflavin. Photoreduction of the 8-CN-FMN enzyme in the presence of EDTA resulted in ϳ100% anionic semiquinone without any further reduction to the fully reduced form.
Stabilization of Semiquinone-With normal flavin semiquinone can be obtained either by one-electron reduction of the oxidized form or by one-electron oxidation of the reduced flavin. But in solution only about 2% of the total flavin is stabilized as semiquinone at pH 7 (42). This is because of the fact that in the following equation, the equilibrium always favors the disproportionation (right to left) of the semiquinone.
Fl ox ϩ Fl red º 2 Fl°(Eq. 1) However, several apoproteins upon binding the flavin stabilize larger amounts of the semiquinone form through electrostatic interactions between the flavin and the protein (43). When native OYE I was reduced with ϳ0.5 equivalents of NADPH, a mixture of reduced and oxidized forms of the enzyme was obtained. This mixture underwent a slow comproportionation reaction over a period of 4 h to form ϳ60 -65% stable anionic semiquinone, establishing the thermodynamic stabilization of the semiquinone by the protein. When this reaction was carried out with 8-CN-FMN enzyme under identical conditions, it resulted in ϳ100% formation of stable anionic semiquinone. The reaction was found to be very fast when compared with the native enzyme, being complete within a few minutes, with a rate constant of 3 ϫ 10 4 M Ϫ1 s Ϫ1 for this comproportionation reaction. These results show that the potential for the EFl ox / EFl SQ couple is much higher than that of the EFl SQ /EFl red couple. The potential of the EFl ox/ /EFl SQ couple for 8CN-FMN OYE I was determined by equilibration versus toluidine blue using the xanthine/xanthine oxidase reduction method (28) and found to be 125 Ϯ 3 mV. The potential of the second electron reduction, EFl SQ /EFl red , was determined similarly by equilibration versus phenosafranin to be Ϫ205 Ϯ 13 mV. These values define the two electron potential of the 8-CN-FMN enzyme to be Ϫ40 Ϯ 18 mV, similar to that of the free flavin. A linear relationship is expected for the energy of the long wavelength transition of the charge transfer complex between oxidized flavin and phenolate anion and the first 1-electron reduction potential of the flavin (43). As can be seen in Fig. 6, the 115 nm red shift of the transition with p-chlorophenol fits well with the previously determined values with other artificial flavins and that with the 8-CN-FMN enzyme.
Steady State Turnover of 8-CN-FMN-OYE with ␤-NADPH and Oxygen-Enzyme-monitored turnover experiments were carried out by using a stopped flow spectrophotometer. A known concentration of enzyme (6.2 M) in 100 mM potassium P i , pH 7, was reacted with a range of concentrations of NADPH in the same buffer, equilibrated with different concentrations of oxygen at 25°C. The concentrations of oxygen used were 189, 256, 433, and 743 M. The reaction was followed at both 470 nm, from approach to steady state until final reoxidation by excess oxygen, and also at 340 nm (Fig. 7).
The rate constant for the reductive-half reaction was obtained by following the initial reduction of the enzyme bound flavin at 470 nm. A rate constant of 245 s Ϫ1 was obtained from a double-reciprocal plot of k obs versus NADPH concentration (results not shown). The k obs values for this phase were found to be independent of the oxygen concentration used, indicating that k red Ͼ Ͼ k ox . The value of k red for native OYE I by NADPH was determined 3 as 5.9 s Ϫ1 . Thus the reduction of 8-CN-FMN-OYE I by NADPH occurs about 40 times faster than with the recombinant enzyme. This is consistent with the high oxidation-reduction potential of the flavin, although the rate enhancement is smaller than might be expected from the large increase in thermodynamic driving force associated with the ϳ160 mV increase in redox potential. The electron-withdrawing nature of the cyano substituent might be expected to destabilize the oxidized isoalloxazine and stabilize the reduced (anionic) flavin, leading to an earlier transition state with less development of negative charge and a correspondingly relatively small effect on the reduction rate constant.
The rate constant for the oxidation of reduced enzyme by oxygen was estimated from the final part of turnover traces such as shown in Fig. 7, which show that flavin reoxidation occurs only on complete oxidation of NADPH. The final phase of these traces fit perfectly to a single exponential process. The values of k obs obtained from such analyses were independent of the initial NADPH concentration and were directly proportional to the oxygen concentration, yielding a second order rate constant of 3.5 ϫ 10 2 M Ϫ1 s Ϫ1 . Reduced native enzyme reacts with oxygen with a rate constant of 3.8 ϫ 10 3 M Ϫ1 s Ϫ1 . 3 Thus the reduction of oxygen is an order of a magnitude slower than with normal enzyme, in general accord with the high redox potential of the CN-flavin. From the model studies of Bruice and colleagues (44)

8-CN-flavins and Flavoproteins
the measured potential was that of the anionic semiquinone, the only thermodynamically stabilized radical species found with Old Yellow Enzyme.
The reaction was also followed at 340 nm in the above experiments at varied concentrations of NADPH and oxygen. From the results (Fig. 7) it is clear that the disappearance of NADPH is essentially linear with time, indicating a low K m for NADPH. The plots of reciprocal of turnover numbers versus 1/[NADPH] showed parallel line kinetics, similar to the situation with native enzyme (38). However, the determination of kinetic constants was not possible, since the secondary plot of intercepts was almost directly proportional to the reciprocal of the oxygen concentration (results not shown). All of the results are thus consistent with a kinetic mechanism in which the reduction rate constant (k red ) is high and the reaction rate constant with O 2 is low (45).
Steady State Turnover of 8-CN-FMN-OYE with ␤-NADPH and Cyclohex-2-en-1-one-It has been reported recently that cyclohexenone is an efficient electron acceptor wherein the carbon-carbon double bond is reduced to form cyclohexanone (39). Turnover experiments were carried out by reacting enzyme with mixtures of NADPH and cyclohexenone under anaerobic conditions in the stopped flow spectrophotometer. Oxidation of NADPH by the 8-CN-FMN reconstituted enzyme was monitored at 340 nm. It was found that the turnover numbers were independent of cyclohexenone concentration over the range of 50 M to 1 mM. For 8-CN-FMN enzyme a turnover number of 7.5 min Ϫ1 was obtained with a K m for NADPH of 0.2 M, whereas the native OYE I has a turnover number 3 of ϳ300 min Ϫ1 . These results show that cyclohexenone is a poor electron acceptor for the 8-CN-FMN enzyme and the reoxidation of the reduced enzyme-bound cyanoflavin by cyclohexenone, unlike that of the native enzyme, is very slow, again a reflection of the high oxidation-reduction potential of the flavin. Attempts to measure directly the rate constant for oxidation of the reduced enzyme by cyclohexenone were unsuccessful, since the slowly produced oxidized enzyme reacted with remaining reduced enzyme to form anionic semiquinone (see earlier section). However, if the 8-CN-FMN enzyme functions like the normal enzyme, where it has been shown that the NADPHcyclohexenone reductase reaction functions by a ping-pong kinetic mechanism, where k cat ϭ k red ⅐k ox /(k red ϩ k ox ), it may be concluded that k ox for cyclohexenone with the 8-CN-FMN enzyme ϭ k cat , i.e. 0.125 s Ϫ1 , an 870-fold decrease from the value of 109 s Ϫ1 found for native enzyme. 3 This large decrease in reaction rate constant compared with that with native enyzme is again consistent with the electron-withdrawing nature of the cyano substituent, which by stabilization of the anionic reduced flavin would make reduction of the olefinic bond of cyclohexenone thermodynamically less favorable. The magnitude of the decreased reactivity (ϳ900-fold) should be compared with the 40-fold increase in the reduction reaction rate constant with NADPH and implies a late transition state in the reaction.

D-Amino Acid Oxidase
The binding of 8-CN-FAD to apoDAAO is accompanied by around 80% quenching of the fluorescence but by only small spectral shifts from free flavin to the bound flavin. The maxima of the free flavin shifted from 340 and 456 nm to 344 and 458 nm on binding to the apoprotein (Fig. 8). As in the case of the binding of native FAD, no vibronic resolution of peaks is observed on the binding of 8-CN-FAD. When the apoDAAO was titrated with FAD in presence of benzoate, the titration plots suggest tight binding similar to that found with native FAD (23).
Photoreduction of 8-CN-FAD-DAAO-When the 8-CN-FAD-DAAO was irradiated with light in presence of EDTA, it formed ϳ100% of anionic semiquinone (Fig. 8) with no further reduction to the two-electron reduced form on prolonged light irradiation. These results are similar to those found with native enzyme, where the anionic flavin semiquinone is also stabilized thermodynamically (35). The semiquinone has a sharp absorption maximum at 418 nm with an extinction coefficient of ϳ20,000 M Ϫ1 cm Ϫ1 , somewhat lower than that found with SCN FMN OYE. However, when the photoreduction was carried out with the 8-CN-FAD enzyme-benzoate complex, the reaction resulted in essentially two-electron reduced enzyme with only ϳ5% intermediate semiquinone. These results are consistent with the fact that the binding of benzoate modulates the oxidation-reduction potential of the protein-bound flavin in such a way that two-electron reduction is facilitated and making the formation of the anionic radical more difficult. It was shown that in case of the native protein, the redox potentials for the EFl ox /EFl SQ and EFl SQ /EFl red couples are Ϫ98 and Ϫ204 mV for the holoenzyme and Ϫ260 and Ϫ140 mV for the benzoate complex (46). Although we have not attempted to determine the redox potentials of 8-CN-FAD DAAO, it is clear that a similar situation exists as with the native enzyme.
Reduction of 8-CN-FAD-DAAO with D-Alanine-When 8-CN-FAD-DAAO was reacted anaerobically with 50-fold excess Dalanine, the spectrum for the two-electron reduced enzyme was formed rapidly (Fig. 8). The reduced enzyme has a peak at 370 nm. On opening to air, the oxidized enzyme spectrum was regained quickly once turnover was complete.
Benzoate Binding-Pronounced spectral changes were observed in the region of 480 and 520 nm upon binding of benzoate to the oxidized form of D-amino acid oxidase (47,48). The binding of sodium benzoate to 8-CN-FAD-D-amino acid oxidase resulted in similar spectral changes with marked resolution of the peaks as in the case of the benzoate-FAD-enzyme. The K d for benzoate with the native enzyme is 3 ϫ 10 Ϫ6 M at pH 8.5 with 1:1 stoichiometry (23). Titration plots with the 8-CN-FAD enzyme give a K d for benzoate of ϳ120 M, showing that benzoate binds with less affinity to the 8-CN-FAD enzyme. It is interesting to recall that DAAO from the yeast Rhodotorula gracilis also binds benzoate less tightly, with a K d of 245 M (49).
Sulfite Addition-As discussed earlier, sulfite adds to the electrophilic N-5 of the oxidized form of flavins to afford adducts with similar but not identical spectral properties to those of the reduced flavins. However with the 8-CN-flavin, the adduct spectrum is completely different from its dihydroflavin spectrum and resembles more the reduced normal flavin spectrum. Native D-amino acid oxidase forms the adduct slowly (37) with a dissociation constant of 3.5 ϫ 10 Ϫ3 M. The 8-CN-FAD enzyme forms the adduct with complete bleaching of both the 450 and 340 nm peaks, with the formation of a new peak at 324 nm as in the case of the unbound 8-CN-flavin. The addition was also found to be slow, as in the case of native enzyme, with a K d of 9 M. The tighter binding of sulfite to 8-CN-FAD enzyme than to normal enzyme is again consistent with the higher redox potential of the 8-CN-FAD enzyme.
In conclusion, we have accomplished the synthesis of a high potential flavin analog, 8-CN-riboflavin, suitable for use as a flavoprotein active site probe, which has long been seen as an attractive flavin analog following from the model studies of Bruice et al. (13) with the corresponding isoalloxazine derivative. It was further taken to the FAD level with FAD synthetase. The reduced 8-CN-flavin, unlike normal reduced flavin, has a distinctive absorption spectrum with two peaks in the UV-visible region with maxima at 320 and 370 nm. The oxidation-reduction potential of the flavin was determined as Ϫ50 mV, about 160 mV more positive than that of normal flavin. Significantly tighter binding of sulfite to cyanoflavin with a K d of 0.73 mM is in accord with its high redox potential. The apoproteins of RBP, flavodoxin, OYE I, and DAAO have been reconstituted with the appropriate cyanoflavin form, and the properties of the reconstituted flavoproteins indicate that the relatively small cyano functionality meets the steric requirements for a good active site probe. Although the oxidationreduction potential of 8-cyanoflavin was found to be the highest among all the known flavin analogs, it is extremely stable toward nucleophilic displacement reactions unlike some of the other known derivatives (6 -8). Hence this flavin allows the adjustment of the potential of protein-bound flavin to extreme positive values without resulting in any chemical modification of the protein. This permits the study of how the oxidation-reduction potential controls catalysis, and it is possible that the 160 mV positive potential might reverse the redox chemistry of flavoproteins in some of their reactions. 4 However, as enzymefree 8-CN-riboflavin and 8-CN-FMN were found to be photolabile, it is important to take care in avoiding exposure to light during the holoenzyme reconstitution process.