Oxidation-Reduction Properties of the 8&ubstituted Flavins*

SUMMARY The oxidation-reduction potentials of a variety of riboflavin derivatives substituted at the 8or position have been determined by anaerobic titration with dithionite in the presence of suitable dyes and spectrophotometric determination of the ratio of oxidized to reduced flavin. The potentials of 8a-substituted flavins, including the histidyl-8ar-riboflavin component of succinate dehydrogenase and the cysteinyl-8ol-riboflavin component of monoamine oxidase, were found to be 0.02 to 0.03 volt higher than the potential of riboflavin. In accord with this, the affinity of the 8ac-substituted flavins for sulfite was 10 to 20 times higher than that of riboflavin. Reduction by dithionite to the flavin hydroquinone proceeded with a 2 electron uptake and was completely reversible on admission of O2 in each case, except for the 8&5’-cysteinyl-sulfone of riboflavin. On reduction of the latter by equimolar dithionite, cysteinesulfinate was spontaneously eliminated and riboflavin was formed. This process follows first order kinetics with a pH optimum of 6.0 at 25”.


Oxidation-Reduction
Properties of the 8&ubstituted Flavins* (Received for publication, July 16, 1973) DALE E. EDMONDSONI~AND THOMAS P.SINGER From the Department of Biochemistry and Biophysics, University of California, San Francisco, California, 94143, and Molecular Biology Divisicjn, Veterans Administration Hospital, San Francisco, California 94121

SUMMARY
The oxidation-reduction potentials of a variety of riboflavin derivatives substituted at the 8or position have been determined by anaerobic titration with dithionite in the presence of suitable dyes and spectrophotometric determination of the ratio of oxidized to reduced flavin. The potentials of 8a-substituted flavins, including the histidyl-8ar-riboflavin component of succinate dehydrogenase and the cysteinyl-8ol-riboflavin component of monoamine oxidase, were found to be 0.02 to 0.03 volt higher than the potential of riboflavin. In accord with this, the affinity of the 8ac-substituted flavins for sulfite was 10 to 20 times higher than that of riboflavin. Reduction by dithionite to the flavin hydroquinone proceeded with a 2 electron uptake and was completely reversible on admission of O2 in each case, except for the 8&5'-cysteinylsulfone of riboflavin.
On reduction of the latter by equimolar dithionite, cysteinesulfinate was spontaneously eliminated and riboflavin was formed. This process follows first order kinetics with a pH optimum of 6.0 at 25".
During the past few years the chemical nature of the linkage of covalently bound flavin to the protein components of several enzymes has been elucidated.
In each case studied, the peptide chain is attached at the 8cr position of riboflavin.
In succinate dehydrogenase (l-3) and in n-&hydroxynicotine oxidase (4)) attachment of the flavin is to the N(3) position of the imidazole ring of histidine.
The flavin site of liver monoamine oxidase contains 8oc-cysteinyl FAD, and the cysteine is in a thioether linkage to 8o 6). Chromatium cytochrome cs52 also contains cysteine substituted at the 8a position of FAD, but the linkage in this case is a thiohemiacetal (7-9). Current studies in this laboratory1 show that the covalently bound FAD prosthetic groups of thiamine dehydrogenase (10) and @-cyclopiazonate oxidocyclase (11) are also linked by way of the 8a-CHZ group to their apoenzymes.
Although in the course of these studies a considerable amount of information has accrued on the chemical reactivity, fluorescence characteristics, absorption and electron spin resonance spectra, photochemistry, and chromatographic properties of 8asubstituted flavins, their oxidation-reduction potentials have not been measured.
Knowledge of the oxidation-reduction potentials of naturally occurring 8oc-substituted flavins was of interest, since it seemed conceivable that covalent linkage to the protein increases the potential of the flavin and thereby facilitates catalysis.
A case in point is succinate dehydrogenase, where the E m,7 of the substrate pair is some 0.250 volt higher (15) than that of the FAD-FADH2 couple. This paper deals with the electron affinity of the isoalloxazine ring system in various S(Ysubstituted flavins, as measured by oxidation-reduction potentials and sulfite affinities (16). In addition, we describe a reductive 80( elimination reaction, which may be of importance in the catabolism of covalently linked flavins.

EXPERIMENTAL PROCEDURE
Materials-Flavins containing various 8cr substituents were synthesized utilizing 8a-BrTARF2 (3) as described by Kenney and Walker (14) and by Ghisla and Hemmerich (13). The 8formylriboflavin compound was a gift from Dr. G. Blankenhorn (The University of California, Davis).
8ol-S-Cysteinylsulfone-TARF was synthesized by incubating 10 mg of cysteinesulfinic acid (65 PM) with 20 mg of 8a-BrTARF (40 pM) in 0.1 ml of dimethylformamide for 7 days at room temperature under anaerobic conditions. The product was then isolated by preparative high voltage electrophoresis at pH 1.6 with a final yield of 33%. The flavin sulfone prepared in this manner was identical in spectral properties with the material prepared by treating 8a-cysteinylriboflavin with oxidizing agents (6, 13).
Purity of the various substituted flavins were monitored by high voltage electrophoresis, by thin layer chromatography, and by descending paper chromatography under conditions previously described (2,3,14). The 8c+sulfur-containing flavins were unstable over prolonged storage in solution, and thus experiments were performed immediat.ely after purification.
The flavins thus synthesized and purified were identical in spectral properties (absorption and fluorescence) as well as electrophoretie mobility with those previously reported (12)(13)(14).
Anaerobic Titrations-All anaerobic experiments were performed in an atmosphere of helium that had been purified over hot copper and water-saturated in a gas train made of glass and butyl rubber tubing. Dithionite solutions were prepared anaerobically in a glass vessel and stored in a gas-tight Hamilton syringe.
Anaerobic titrations were performed in a glass cuvette similar to that reported by Burleigh et al. (17) but modified in that no serum stoppers were used. Dithionite solutions were standardized by anaerobic titration of a solution of 3-methyllumiflavin (a gift of Dr. S. Ghisla, The University of Michigan). Flavin reduction was monitored by measuring visible absorption spectra with a Cary 14 spectrophotometer thermostated at 25" after the addition of each aliquot of dithionite.
The oxidation-reduction potentials of the various 8cr-substituted flavins were measured by anaerobic dithionite titration of a mixture of the appropriate flavin and of indigo disulfonate (Ern,~ = -0.116 volt) or, in some cases, of anthraquinone-2,6disulfonate (E m,r = -0.184 volt) (15). The indigo dye is advantageous in that its reduction is easily monitored at 610 nm where there is no flavin absorption and has an isosbestic point at 460 nm, near the absorption maximum of oxidized flavin.
The spectrum of the reduced anthranquinone dye overlaps considerably with those of oxidized and reduced flavins. Dye reduction could be monitored by the decrease in absorbance at the flavin isosbestic point (around 330 nm), and flavin reduction could be monitored at the dye isosbestic of 352 nm. The oxidation-reduction potential for the system at equilibrium at various stages of reduction was calculated using the equation: The amounts of reduced and oxidized flavin were determined as described and the oxidation-reduction potential (E,,,) was determined from a plot of Eh (16)  Carboxy. absorption coefficients of the flavin-sulfite complexes at 450 nm N-3-Histidyl were estimated from Renasi-Hildebrand plots (18) and were in S-Cysteinyl.. . the range of 500 to 1000 M-I cm-l.
Formyl . Reversibility of the reaction was tested by acidification (pH ~0) of a solution of flavin plus excess sulfite, bubbling with helium for 1 hour to remove the resulting SOZ, neutralization, and measurement of the spectrum of the resulting solution.
In all cases the resulting absorption spectrum (after correcting for dilution) was identical with that of the untreated flavin. The presence of an isosbestic point during reduction and subsequent reoxidation by 02, with return of the original absorption spectrum, establishes the oxidation-reduction reversibility of the flavins. 8cu-Histidylriboflavin is reduced by 1 M equivalent of dithionite to its hydroquinone form, with an isosbestic point at 327 nm (Fig. 1). Subsequent reoxidation by air gave an absorption spectrum identical with that of unreduced flavin.
This behavior was observed with all of the 8cr-substituted flavins tested (Table I), with the exception of 8a-Scysteinylsulfone-TARF, which was unstable to reduction (see below). a Taken from Reference 15. * Taken from Reference 16.
of flavin-sulfite adducts are similar to, but not identical with, those of flavin hydroquinone.
It has also been demonstrated (16) that the oxidation-reduction potentials of various flavins correlate well with their respective sulfite affinities.
The degree of sulfite affinity for the various &-substituted flavins is, therefore, an indication of their oxidation-reduction potentials.

Interaction of SulJite with 8cA3ubstituted
Flcwins-Miiller and Massey (16) have shown that flavins form adducts with sulfite at the N(5) position, which may be considered a sulfamic acid In a manner analogous to unsubstituted flavins (16), the 450 nm band is bleached with increasing sulfite concentration and an isosbestic point is observed, denoting the presence of only two components in solution (oxidized flavin and the sul-   (16), as regards spectral properties, shows that the sulfite complex is at the N(5) position.
Flavin-sulfite complex formation followed a pseudo-first order rate, which was dependent on sulfite concentration.
As may be seen in Table I, with one exception, substitution in the Sty position brings about a 2-to 3.6-fold increase in the reaction rate of flavins with sulfite, but the values for the dissociation constant decreased lo-to 20-fold.
Substitution at the 8a position thus makes the isoalloxazine ring more electron deficient and the decline in KD values is mainly due to a decrease in koff rather than an increase in the "on" kinetic rate constant.
Of particular interest is that the nature of the &Y substituent has little influence on the sulfite affinity of the N(5) position.
Oxidation-Reduction Potential dleasurements-From the data of Miiller and Massey (16)  to -0.170 volt for 8hydroxyriboflavin (Table I), in agreement with the values suggested by the dissociation constants of the sulfite complexes. The plots in Fig. 3 show the linear relation between potential and the logarithm of the ratio of oxidized to reduced flavin with slopes as expected for a 2-electron reduction.
The agreement of the individual points with the theoretical line show the reliability of the oxidation-reduction measurements using this experimental method.
In accord with the sulfite data, the nature of the 8a substituent had little effect on the midpoint potentials, although they were 0.03 to 0.04 volt higher than for the corresponding unsubstituted flavins.
Reductive Cleavage of &-S-CysteinylsulfoneriboJlavin-Dithionite reduction of 8cr-S-cysteinylsulfoneriboflavin or its tetra- acetyl form and subsequent reoxidation by air gave rise to a flavin compound with a near ultraviolet absorption maximum at 372 nm, as compared with a maximum at 352 nm in the untreat,ed material.
A similar behavior is observed upon reduction of this flavin by light EDTA.
High voltage electrophoresis of the reduced flavin (by either dithionite or light EDTA) showed no migration at pH values of 1.6, 3.5, and 6.2, conditions under which the untreated material does migrate. The untreated flavin was ninhydrin-positive, while the flavin, treated as above, become ninhydrin-negative, showing the loss of the a-NH2 group. A ninhydrin-positive spot was observed, however, which migrated differently from the flavin, with RF value identical with that of cysteic acid. Formation of cysteic acid in this experiment was probably due to the oxidation of cysteinesulfinate by the HIOz formed during reoxidation of the reduced flavin by oxygen.
Thin layer chromatography of the modified flavin identified it as riboflavin.
To investigate the mechanism of elimination of the 80( substituent, a dithionite titration was performed on 8oc-S-cysteinylsulfone-TARF.
Upon the addition of the first aliquot of dithionite, 0.15 mole per mole of flavin, the 450 maxima was immediately bleached partially.
Subsequently, however, it slowly increased in intensity to a value identical with that observed before the addition of dithionite.
The absorption spectra showed a shift in the near ultraviolet band to the red (Fig. 4A). Subsequent additions of dithionite up to 1 mole per mole of flavin exhibited the same behavior with shift of the near ultraviolet band to 372 nm (Fig. 4A).
The isosbestic point at 363 nm indicates the presence of two species of oxidized flavin in solution.
Subsequent additions of dithionite up to 2 moles per mole of flavin bleached the flavin spectrum with an isosbestic point at 330 nm (Fig. 4B). The titration plot in Fig. 5 shows that 2 electrons are required for reductive cleavage of the 801 substituent and 2 electrons for the reduction of the resulting flavin compound.
To provide additional evidence that this observation is indeed a unimolecular cleavage reaction, the kinetics was measured as a function of pH. To an anaerobic solution of 8cr-S-cysteinylsulfone-TARF, 1 M equivalent of dithionite was added and the rate of increase in absorbance at 450 nm was observed.
The re-  4. A, spectral changes during the anaerobic addition of the first molar equivalent of dithionite to &-S-cysteinylsulfone-TARF in 0.1 M Pi, pH 7.0. Upon each addition of dithionite, the spectrum was recorded after no further spectral changes were apparent. The molar equivalents of dithionite for the respective curves are: Curve 1, none; Curve 2, 0.15; Curve S, 0.3; Curve 4, 0.6; Curve 5, 0.75; Curve 6, 1.0. The spectra are uncorrected for dilution. B, spectral changes during the anaerobic addition of a second molar equivalent of dithionite to the flavin solution in A. The total molar equivalents of dithionite added for the respective curves are: Curve 1, 1.1; Curve 2, 1.26; Curve 3, 1.43; Curve 4, 1.63; Curve 5, 2.40. The spectra are uncorrected for dilution. pH range of 4.5 to 8.8. Increasing the flavin concentration lofold had no significant effect on the rate constant, as expected for a first order reaction.
The rate of the reaction was strongly pH-dependent, with a maximal rate at pH 6.0 (see inset, Fig. 6), a rapid decrease at lower pH values, and a slightly slower decrease as the pH was raised. Analysis of the rate data was complex in that values on each side of the maximum were not consistent with a single proton ionization.
A possible explanation for this will be elaborated upon under "Discussion." It was of interest to see whether N(5) substitution by sulfite could also lead to the cleavage of the Sor-cysteinylsulfone.
The sulfite titration data (Table I)  This experiment was repeated by incubation of the cysteinylsulfone flavin with 1.8 M sulfite for 4 hours at room temperature.
Removal of the bound sulfite and spectral analysis showed properties identical with those of untreated material.
Thus, N(5) substitution by sulfite does not, under these conditions, lead to cleavage of the 8ar substituent.

DISCUSSION
The data presented in Table I show that the incorporation of a substituent in the 80~ position in place of hydrogen increases the electron affinity of the isoalloxazine ring in varying degrees, depending on the nature of the substituent.
As shown earlier for other flavin analogs (16), the sulfite affinities of the 8a-sub-8148 stituted flavins correlate quite well with their respective oxidation-reduction potentials.
Although there is a shift of approximately +0.03 volt in flavin oxidation-reduction potential upon 80( substitution, the nature of the substituent does not seem to drastically influence the observed values. The Chromatium cytochrome csb2 peptic tetrapeptide (FAD level) has a potential about 0.02 volt more positive than FAD, although the observed flavin-tyrosine interaction (9) could very well have an influence.
Indeed, the close proximity of an aromatic group (e.g. tyrosine or tryptophan) has been shown to raise the flavin potential 0.05 to 0.06 volt for normal flavins (20). The corresponding tryptic-chymotryptic flavin tripeptide in which the COOH-terminal tyrosine is removed was not reversibly oxidized on dithionite reduction, in agreement with an earlier observation (8). This disparity in properties between the two flavin peptides further points out the stabilizing influence of the COOH-terminal tyrosine on the 8a-thiohemiacetal bond, which is also seen on performic acid oxidation studies (9).
A development from this study which serves as a model system for the conversion of &-substituted flavins to "normal" flavins is the cleavage of the 8cr-cysteinylsulfone from the reduced flavin.
Previous studies have shown that histidine and cysteine are liberated from the 8a! position by strong acid hydrolysis or by treatment with zinc in acid media (3,5). Mild procedures, such as flavin reduction by a stoichiometric amount of dithionite, do not cause 80( elimination with any of the other analogs tested. The scheme in Fig. 7 depicts a mechanism for this elimination which is most consistent with the data. The oxidized flavin (I) is reduced to its hydroquinone form (ZZ). The electron rearrangement to the electron-deficient sulfur of the sulfone causes elimination of the cysteinylsulfinate, leaving the flavin in a "quinhydrone" form (III), which then tautomerizes (through IV as a possible structure) to form the unsubstituted flavin (V).
Structure ZZZ is a flavin intermediate which has been used to explain flavin dimerization under basic conditions (21) and to explain the deuterium-hydrogen exchange in the 8-methyl group when flavin is heated at neutral pH (22).
The first order nature of the reaction (Fig. 6) eliminates any participation of oxidized flavin at later stages of the reaction. Conceivably, oxidized flavin could be reduced to an equilibrium concentration.
However, the midpoint potentials are well separated (see sulfite KD values, The observation that sulfite addition to the N(5) position does not lead to 8cr elimination is of considerable interest, in view of its isoelectronic similarity with the hydroquinone form. The reason for this may lie in the geometry of the respective reduced flavin species.
The reduced form of flavin was first postulated to be noncoplanar, on the basis of its absorption spectra (23) and later confirmed by x-ray crystallography (24) to be in a bent "butterfly" configuration, with the two planes intersecting on the N(5), -N(lO) axis. To achieve maximal orbital overlap for electron transfer to the 8crsulfone, the nitrogen at the 5 position must be in the plane of the benzenoid ring. This may be achieved through the all-coplanar form (a conformation which is energetically unfavorable) but would not arise from other more energetically favorable mechanisms, such as ring inversion or nitrogen inversion (25). The population of molecules in the allcoplanar form, in accord with the Boltzmann distribution law, would be proportional to the rate of elimination.
In the case of flavin hydroquinone, where hydrogen is the N(5) substituent, an appreciable population may be in the all-coplanar form, while steric considerations would suggest that the flavin-sulfite adduct has a negligible population in this configuration.
The pH rate profile for the elimination suggests two effects on the rate of the reaction (inset, Fig. 6). The increase in rate as the pH is lowered from 8.8 to 6.0 may reflect the protonation of the flavin hydroquinone.
Spectral studies suggest the anionic flavin hydroquinone is in a more bent configuration than the neutral form (23) and thus one would expect an increase in rate as the population of molecules in which the N(5) is planar with the benzenoid ring increases.
Below pH 6, the difficulty in removing the N(5) proton increases, hence the decrease in rate. The opposing effects of these two processes could be responsible for the difficulty in analyzing the data in terms of a single ionization.
The observation of an elimination from the 8~ position by simply reducing the flavin presents a possible mechanism for the physiological conversion of 8cusubstituted flavins to normal flavin, by simply modifying the substituent to make it electrondeficient.
Furthermore, this process may have implications in the catalytic process of flavoenzymes which do not contain covalently bound flavin.
Efforts are under way to study further these new aspects of flavin chemistry.