Energy transduction during catalysis by Escherichia coli DNA photolyase.

Native DNA photolyase from Escherichia coli contains 1,5-dihydroFAD (FADH2) plus 5,10-methenyltetrahydropteroylpolyglutamate. Quantum yield and action spectral data for thymine dimer repair were obtained by using a novel multiple turnover approach under aerobic conditions. This method assumes that catalysis proceeds via a (rapid-equilibrium) ordered mechanism with light as the second substrate, as verified in steady state kinetic studies. The action spectrum observed with native enzyme matched its absorption spectrum and an action spectrum simulated based on an energy transfer mechanism where dimer repair is initiated either by direct excitation of FADH2 or by pterin excitation followed by singlet-singlet energy transfer to FADH2. The quantum yield observed for dimer repair with native enzyme (phi Native = 0.722 +/- 0.0414) is similar to that observed with enzyme containing only FADH2 (phi EFADH2 = 0.655 +/- 0.0256), as expected owing to the high efficiency of energy transfer from the natural pterin to FADH2 [EET = 0.92]. The quantum yield observed for dimer repair decreased (2.1-fold) when the natural pterin was partially (68.8%) replaced with 5,10-CH(+)-H4folate (phi obs = 0.342 +/- 0.0149). This is consistent with the energy transfer mechanism (phi calc = 0.411 +/- 0.0118) since a 2-fold lower energy transfer efficiency is observed when the natural pterin is replaced with 5,10-CH(+)-H4folate (EET = 0.46) (Lipman & Jorns, 1992). The action spectrum observed for 5,10-CH(+)-H4folate-supplemented enzyme matched a simulated action spectrum which exhibited a small (5 nm) hypsochromic shift as compared with the absorption spectrum (lambda max = 385 nm).(ABSTRACT TRUNCATED AT 250 WORDS)     

Action mechanism of Escherichia coli DNA photolyase. II. Role of the chromophores in catalysis

DNA photolyase repairs pyrimidine dimers in DNA in a reaction that requires visible light. Photolyase from Escherichia coli is normally isolated as a blue protein and contains 2 chromophores: a blue FAD radical plus a second chromophore that exhibits an absorption maximum at 360 nm when free in solution. Oxidation of the FAD radical is accompanied by a reversible loss of activity which is proportional to the fraction of the enzyme flavin converted to FADox. Quantitative reduction of the radical to fully reduced FAD causes a 3-fold increase in activity. The results show that a reduced flavin is required for activity and suggest that flavin may act as an electron donor in catalysis. Comparison of the absorption spectrum calculated for the protein-bound second chromophore (lambda max = 390 nm) with fluorescence data and with the relative action spectrum for dimer repair indicates that the second chromophore is the fluorophore in photolyase and that it does act as a sensitizer in catalysis. On the other hand, enzyme preparations containing diminished amounts of the second chromophore do not exhibit correspondingly lower activity. This suggests that reduced flavin may also act as a sensitizer in catalysis. The blue color of the enzyme is lost upon reduction of the FAD radical. The fully reduced E. coli enzyme exhibits absorption and fluorescence properties very similar to yeast photolyase. This indicates that the two enzymes probably contain similar chromophores but are isolated in different forms with respect to the redox state of the flavin.     

Effect of 5-deazaflavin on energy transduction during catalysis by Escherichia coli DNA photolyase.

DNA photolyase from Escherichia coli contains 1,5-dihydroFAD (FADH2) plus 5,10-methenyltetrahydropteroylpolyglutamate. The action spectrum observed for apoenzyme reconstituted with 5-deazaFADH2 (EdFADH2) matched its absorption spectrum after correction for the presence of a small amount of inactive 5-deazaFADox. The quantum yield for dimer repair with EdFADH2 (phi EdFADH2 = 0.110) was 6-fold lower than that observed with apoenzyme reconstituted with FADH2. Excited-state redox potential calculations indicate that 5-deazaFADH2 singlet is a better one-electron donor (E = -3.5 V) than FADH2 singlet (E = -2.7 V). Other studies indicate that the quantum yield for electron transfer from reduced flavin singlet to pyrimidine dimer (0.88) is unaffected when FADH2 is replaced by 5-deazaFADH2. Enhanced back electron transfer from pyrimidine dimer radical to flavin radical may account for the decreased quantum yield observed with EdFADH2 since, in the ground state, 5-deazaFADH. is a better oxidant than FADH.. The action spectrum observed for apoenzyme reconstituted with 5-deazaFADH2 plus 5,10-CH(+)-H4folate (EPtedFADH2) matched the absorption spectrum determined for enzyme-bound 5-deazaFADH2, indicating that the pterin chromophore was inactive as a sensitizer. This differs from results obtained with native enzyme, where pterin acts as a sensitizer via efficient singlet-singlet energy transfer to FADH2. The quantum yield for dimer repair by 5-deazaFADH2 bound to EPtedFADH2 (phi EPtedFADH2 = 0.0318) was 28.9% of that observed for EdFADH2. Spectroscopic studies indicate that singlet-singlet energy transfer in EPtedFADH2 is very efficient but only occurs in the "wrong" direction, i.e., from excited 5-deazaFADH2 to pterin.(ABSTRACT TRUNCATED AT 250 WORDS)     

Purification of Escherichia coli DNA photolyase

Escherichia coli photolyase is a DNA repair enzyme which monomerizes pyrimidine dimers, the major UV photoproducts in DNA, to pyrimidines in a light-dependent reaction. We recently described the construction of a tac-phr plasmid that greatly overproduces the enzyme (Sancar, G. B., Smith, F. W., and Sancar, A. (1983) Nucleic Acids Res. 11, 6667-6678). Using a strain carrying the overproducing plasmid as the starting material, we have developed a purification procedure that yields several milligrams of apparently homogeneous enzyme. The purified protein is a single polypeptide that has an apparent Mr of 49,000 under both denaturing and nondenaturing conditions. The enzyme has no requirement for divalent cations and it restores the biological activity of irradiated DNA only in the presence of photoreactivating light. The purified photolyase has a turnover number of 2.4 dimers/molecule/min; this value agrees well with the in vivo rate of photoreactivation in E. coli.     

Mechanism of damage recognition by Escherichia coli DNA photolyase

Escherichia coli DNA photolyase binds to DNA containing pyrimidine dimers with high affinity and then breaks the cyclobutane ring joining the two pyrimidines of the dimer in a light- (300-500 nm) dependent reaction. In order to determine the structural features important for this level of specificity, we have constructed a 43 base pair (bp) long DNA substrate that contains a thymine dimer at a unique location and studied its interaction with photolyase. We find that the enzyme protects a 12-16-bp region around the dimer from DNase I digestion and only a 6-bp region from methidium propyl-EDTA-Fe (II) digestion. Chemical footprinting experiments reveal that photolyase contacts the phosphodiester bond immediately 5' and the 3 phosphodiester bonds immediately 3' to the dimer but not the phosphodiester bond between the two thymines that make up the dimer. Methylation protection and interference experiments indicate that the enzyme makes major groove contacts with the first base 5' and the second base 3' to the dimer. These data are consistent with photolyase binding in the major groove over a 4-6-bp region. However, major groove contacts cannot be of major significance in substrate recognition as the enzyme binds equally well to a thymine dimer in a 44-base long single strand DNA and protects a 10-nucleotide long region around the dimer from DNase I digestion. It is therefore concluded that the unique configuration of the phosphodiester backbone in the strand containing the pyrimidine dimer, as well as the cyclobutane ring of the dimer itself are the important structural determinants of the substrate for recognition by photolyase.     

Escherichia coli DNA photolyase is a flavoprotein

Escherichia coli DNA photolyase (photoreactivating enzyme) was purified to homogeneity from a strain that greatly overproduces the protein. The purified enzyme has absorption peaks at 280 and 380 nm, a fluorescence emission peak at 480 nm and, upon denaturation, releases a chromophore that has the spectroscopic properties of flavin adenine dinucleotide (FAD), indicating that FAD is an intrinsic chromophore of the enzyme.     

Binding of Escherichia coli DNA photolyase to UV-irradiated DNA

Escherichia coli DNA photolyase is a flavoprotein which catalyzes the photomonomerization of pyrimidine dimers produced in DNA by UV irradiation. In vivo, the enzyme acts by a two-step mechanism: it binds to dimer-containing DNA in a light-independent reaction and upon exposure to 300-500-nm light breaks the cyclobutane ring and dissociates from the substrate. Using photolyase purified to homogeneity, we have investigated in vitro the first step of the reaction, DNA binding; enzyme-DNA complex formation was quantitated by the nitrocellulose filter binding assay. We find that the enzyme binds specifically to UV-irradiated DNA regardless of whether the DNA is in the superhelical, open circular, or linear form or whether the DNA is single or double stranded. The binding reaction is optimum at a NaCl concentration of 125 mM and at pH 7.5. Although photolyase is retained by the nitrocellulose filters with near 100% efficiency, the binding efficiency of a single enzyme-substrate complex is about 0.34. The complexes can be dissociated by exposing them to photoreactivating light either in solution or on the filter.     

Effect of flavin structure and redox state on catalysis by and flavin-pterin energy transfer in Escherichia coli DNA photolyase

5-DeazaFAD bound to a hydrophobic site in apophotolyase and formed a stable reconstituted enzyme, similar to that observed with FAD. Although stoichiometric incorporation was observed, the flavin ring modification in 1-deazaFAD interfered with normal binding, decreased protein stability, and prevented formation of a stable flavin radical, unlike that observed with FAD. The results suggest that an important hydrogen bond is formed between the protein and N (1) in FAD, but not N (5), and that there is sufficient space at the normal flavin binding site near N (5) to accommodate an additional hydrogen but not near N (1). Catalytic activity was observed with enzyme containing 5-deazaFADH2 (42% of native enzyme) or 1-deazaFADH2 (11% of native enzyme) as its only chromophore, but no activity was observed with the corresponding oxidized flavins, similar to that observed with FAD and consistent with a mechanism where dimer cleavage is initiated by electron donation from excited reduced flavin to substrate. The protein environment in photolyase selectively enhanced photochemical reactivity in the fully reduced state, as evidenced by comparison with results obtained in model studies with the corresponding free flavins. Phosphorescence was observed with free or photolyase-bound 5-deazaFADH2, providing the first example of a flavin that exhibits phosphorescence in the fully reduced state. Formation of an enzyme-substrate complex resulted in a nearly identical extent of quenching of 5-deazaFADH2 phosphorescence (85.1%) and fluorescence (87.5%). The data are consistent with a mechanism involving exclusive reaction of substrate with the excited singlet state of 5-deazaFADH2, analogous to that proposed for FADH2 in native enzyme. Direct evidence for singlet-singlet energy transfer from enzyme-bound 5-deazaFADH2 to 5,10-CH(+)-H4folate was provided by the fact that pterin fluorescence was observed upon excitation of 5-deazaFADH2, accompanied by a decrease in 5-deazaFADH2 fluorescence. On the other hand, the fluorescence of enzyme-bound pterin was quenched by 5-deazaFADox, consistent with energy transfer from pterin to 5-deazaFADox. In each case, the spectral properties of the chromophores were consistent with the observed direction of energy transfer and indicated that transfer in the opposite direction was energetically unlikely. Unlike 5-deazaFAD, energy transfer from pterin to FAD is energetically feasible with FADH2 or FADox. The results indicate that the direction of flavin-pterin energy transfer at the active site of photolyase can be manipulated by changes in the flavin ring or redox state which alter the energy level of the flavin singlet.     

Identification of a pterin derivative in Escherichia coli DNA photolyase

DNA photolyase from Escherichia coli contains reduced flavin adenine dinucleotide plus a second chromophore, partially characterized in previous studies. Both chromophores function as sensitizers in catalysis. The second chromophore has been identified as a 6-substituted pterin derivative. The compound is oxidized with permanganate to yield 6-carboxypterin or reduced with sodium cyanoborohydride to yield a 5,6,7,8-tetrahydropterin derivative. The second chromophore exhibits spectral properties (lambda max = 360, 255 nm, pH 2) similar to that observed for 7,8-dihydropterin cations. The compound does not exhibit a spectrally detectable pKa around 4 but is converted to a dication (lambda max = 346, 255 nm) in strong acid (pKa approximately 1). Similar ionization behavior is observed with 7,8-dihydropterin derivatives that are alkylated at N(5). The instability of the second chromophore in weakly alkaline solution is due to a fully reversible conversion to a labile bleached form. As compared with other pterin derivatives, the hydrolytic instability is unusual but is very similar to that observed for 5,6-dialkyl-7,8-dihydropterinium salts. It is proposed that the second chromophore is a 7,8-dihydropterin with substituents at positions 5 and 6. The discovery that a pterin derivative functions as a photosensitizer in DNA repair is apparently the first example of a photobiological function for pterins.     

Dissection of functional domains in Escherichia coli DNA photolyase by linker-insertion mutagenesis.

Summary The phr gene, which encodes protein of 472 amino acid residues, is required for light-dependent photoreactivation and enhances light-independent excision repair of ultraviolet light (UV)-induced DNA damage. In this study, dodecamer HindIII linker insertions were introduced into the cloned phr gene and the functional effects of the resulting mutations on photoreactivation and light-independent dark repair in vivo were studied. Among 22 mutants obtained, 7 showed no photoreactivation as well as no enhancement of light-independent repair. Four of these were located in amino acid residues between Gln333 and Leu371 near the 3 end of the gene, two were located in a small region at Glu275 to Glu280 near the middle of the gene and the remaining one was between Pro49 and Arg50. Three mutants that had insertions located in the 42 by segment from 399 to 441 by of the phr coding sequence (corresponding to amino acid residues Ile134 to Lys149) lost the light-independent repair effect but retained photoreactivation. These results suggest that (i) Escherichia coli DNA photolyase contains several critical sites that are distributed over much of the enzyme molecule, and (ii) a functional domain required for the effect on light-independent repair is at least in part distinct from that necessary for light-dependent photoreactivation.     

Femtochemistry in enzyme catalysis: DNA photolyase

Photolyase uses light energy to split UV-induced cyclobutane pyrimidine dimers in damaged DNA. This photoenzyme encompasses a series of elementary dynamical processes during repair function from early photoinitiation by a photoantenna molecule to enhance repair efficiency, to in vitro photoreduction through aromatic residues to reconvert the cofactor to the active form, and to final photorepair to fix damaged DNA. The corresponding series of dynamics include resonance energy transfer, intraprotein electron transfer, and intermolecular electron transfer, bond breaking-making rearrangements and back electron return, respectively. We review here our recent direct studies of these dynamical processes in real time, which showed that all these elementary reactions in the enzyme occur within subnanosecond timescale. Active-site solvation was observed to play a critical role in the continuous modulation of catalytic reactions. As a model system for enzyme catalysis, we isolated the enzyme–substrate complex in the transition-state region and mapped out the entire evolution of unmasked catalytic reactions of DNA repair. These observed synergistic motions in the active site reveal a perfect correlation of structural integrity and dynamical locality to ensure maximum repair efficiency on the ultrafast time scale.     

The folate cofactor of Escherichia coli DNA photolyase acts catalytically

Escherichia coli DNA photolyase catalyzes the light-driven (300-500 nm) repair of pyrimidine dimers formed between adjacent pyrimidine bases in DNA exposed to UV light (200-300 nm). The light-driven repair process is facilitated by two enzyme-bound cofactors, FADH2 and 5,10-methenyltetrahydrofolate. The function of the folate has been characterized in greater detail in this series of experiments. Investigations of the relative binding affinities of photolyase for the monoglutamate and polyglutamate forms of 5,10-methenyltetrahydrofolate show that the enzyme has a greater affinity for the naturally occurring polyglutamate forms of the folate and that the exogenously added monoglutamate derivative is less tightly associated with the protein. Multiple turnover experiments reveal that the folate remains bound to photolyase even after 10 turnovers of the enzyme. Examination of the rates of repair by photolyase containing stoichiometric folate in the presence or absence of free folate under multiple turnover conditions and at micromolar concentrations of enzyme also demonstrates that the folate acts catalytically. The stimulation of turnover by exogenous folate seen at low concentrations of photolyase is shown to be due to the lower affinity of photolyase for the monoglutamate derivative used in reconstitution procedures. These results demonstrate that the folate of E. coli DNA photolyase is a bona fide cofactor and does not decompose or dissociate during multiple turnovers of the enzyme.     

Reconstitution of Escherichia coli DNA photolyase with various folate derivatives

DNA photolyase from Escherichia coli contains both flavin and pterin. However, the isolated enzyme is depleted with respect to the pterin chromophore (0.5 mol of pterin/mol of flavin). The extinction coefficient of the pterin chromophore at 360 nm is underestimated by a method used in earlier studies which assumes stoichiometric amounts of pterin and flavin. The extinction coefficient of the pterin chromophore, determined on the basis of its (p-aminobenzoyl)polyglutamate content (epsilon 360 = 25.7 x 10(3) M-1 cm-1), is in good agreement with that expected for a 5,10-methenyltetrahydrofolate derivative. Also consistent with this structure, the pterin chromophore could be reversibly hydrolyzed to yield a 10-formyltetrahydrofolate derivative or reduced to yield a 5-methyltetrahydrofolate derivative. The isolated enzyme could be reconstituted with various folate derivatives to yield enzyme that contained equimolar amounts of pterin and flavin. Similar results were obtained in reconstitution studies with the natural pterin chromophore, with 5,10-methenyltetrahydrofolate, and with 10-formyltetrahydrofolate. The results show that the polyglutamate moiety, previously identified in the natural chromophore, is not critical for binding. Reconstitution with the natural pterin chromophore did not affect catalytic activity. The latter is consistent with our previous studies which show that, although the pterin chromophore acts as a sensitizer in native enzyme, it is not essential for dimer repair which can occur at the same rate under saturating light with flavin (1,5-dihydro-FAD) as the only chromophore.     

Light-induced reactions of Escherichia coli DNA photolyase monitored by Fourier transform infrared spectroscopy

Cyclobutane-type pyrimidine dimers generated by ultraviolet irradiation of DNA can be cleaved by DNA photolyase. The enzyme-catalysed reaction is believed to be initiated by the light-induced transfer of an electron from the anionic FADH- chromophore of the enzyme to the pyrimidine dimer. In this contribution, first infrared experiments using a novel E109A mutant of Escherichia coli DNA photolyase, which is catalytically active but unable to bind the second cofactor methenyltetrahydrofolate, are described. A stable blue-coloured form of the enzyme carrying a neutral FADH radical cofactor can be interpreted as an intermediate analogue of the light-driven DNA repair reaction and can be reduced to the enzymatically active FADH- form by red-light irradiation. Difference Fourier transform infrared (FT-IR) spectroscopy was used to monitor vibronic bands of the blue radical form and of the fully reduced FADH- form of the enzyme. Preliminary band assignments are based on experiments with 15N-labelled enzyme and on experiments with D2O as solvent. Difference FT-IR measurements were also used to observe the formation of thymidine dimers by ultraviolet irradiation and their repair by light-driven photolyase catalysis. This study provides the basis for future time-resolved FT-IR studies which are aimed at an elucidation of a detailed molecular picture of the light-driven DNA repair process.     

Direct evidence for singlet-singlet energy transfer in Escherichia coli DNA photolyase.

The active form of native Escherichia coli DNA photolyase contains 1,5-dihydro-FAD (FADH2) plus 5,10-methenyltetrahydropteroylpolyglutamate [5,10-CH(+)-H4Pte(Glu)n]. Enzyme containing FADH2 and/or 5,10-methyltetrahydrofolate (5,10-CH(+)-H4folate) can be prepared in reconstitution experiments. Fluorescence quantum yield measurements at various wavelengths with native or reconstituted enzyme provide a simple method for detecting singlet-singlet energy transfer from pterin to FADH2, a key step in the proposed catalytic mechanism. The data satisfy the following criteria: (1) Wavelength-independent quantum yield values are observed for 5,10-CH(+)-H4folate in the absence (0.434) or presence (3.57 X 10(-2)) of FADH2, for 5,10-CH(+)-H4Pte(Glu)n in the presence of FADH2 (5.58 X 10(-2)) and for FADH2 in the absence of pterin (5.34 X 10(-3)); (2) The observed decrease in pterin fluorescence quantum yield in the presence of FADH2 can be used to estimate the efficiency of pterin fluorescence quenching (EQ = 0.918 or 0.871 with 5,10-CH(+)-H4folate or 5,10-CH(+)-H4Pte(Glu)n, respectively); (3) The fluorescence quantum yield of FADH2 is increased in the presence of pterin and varies depending on the excitation wavelength, in agreement with the predicted effect of energy transfer on acceptor fluorescence quantum yield [phi acceptor (+ donor)/phi acceptor (alone) = 1 + EET(epsilon donor/epsilon acceptor), where EET is the efficiency of the energy transfer process]. With 5,10-CH(+)-H4Pte(Glu)n in native enzyme the value obtained for EET (0.92) is similar to EQ, whereas with 5,10-CH(+)-H4folate in reconstituted enzyme the value obtained for EET (0.46) is 2-fold smaller than EQ.(ABSTRACT TRUNCATED AT 250 WORDS)     

Photochemical properties of Escherichia coli DNA photolyase: a flash photolysis study

Escherichia coli DNA photolyase contains a stable flavin neutral blue radical that is involved in photosensitized repair of pyrimidine dimers in DNA. We have investigated the effect of illumination on the radical using light of lambda greater than 520 nm from either a camera flash or laser. We find that both types of irradiations result in the photoreduction of the flavin radical with a quantum yield of 0.10 +/- 0.02. While photoreduction with the camera flash is minimal in the absence of an electron donor (dithiothreitol), laser flash photolysis at 532 nm reduces the flavin to the same extent in the presence or absence or an electron donor. Thus, it is concluded that the primary step in photoreduction involves an electron donor that is a constituent of the enzyme itself. Laser flash photolysis produces a transient absorption band at 420 nm that probably represents the absorption of the lowest excited doublet state (2(1)IIII*) of the radical and decays with first-order kinetics with k1 = 0.8 X 10(6) s-1. The photoreduction data combined with the results of recent studies on the activity of dithionite-reduced enzyme suggest that electron donation by excited states of E-FADH2 is the mechanism of flavin photosensitized dimer repair by E. coli DNA photolyase.     

Reversible resolution of flavin and pterin cofactors of His-tagged Escherichia coli DNA photolyase

Escherichia coli photolyase catalyzes the repair of cyclobutane pyrimidine dimers (CPD) in DNA under near UV/blue-light irradiation. The enzyme contains flavin adenine dinucleotide (FAD) and methenyltetrahydrofolate (MTHF) as noncovalently bound light sensing cofactors. To study the apoprotein-chromophore interactions we developed a new procedure to prepare apo-photolyase. MTHF-free photolyase was obtained by binding the C-terminal His-tagged holoenzyme to a metal-affinity column at neutral pH and washing the column with deionized water. Under these conditions the flavin remains bound and the defolated enzyme can be released from the column with 0.5 M imidazole pH 7.2. The MTHF-free protein was still capable of DNA repair, showing 70% activity of native enzyme. Fluorescence polarization experiments confirmed that MTHF binding is weakened at low ionic strength. Apo-photolyase was obtained by treating the His-tagged holoenzyme with 0.5 M imidazole pH 10.0. The apo-photolyase thus obtained was highly reconstitutable and bound nearly stoichiometric amounts of FAD(ox). Photolyase reconstituted with FAD(ox) had about 34% activity of native enzyme, which increased to 83% when FAD(ox) was reduced to FADH(-). Reconstitution kinetics performed at 20 degrees C showed that apo-photolyase associates with FADH(-) much faster (k(obs) approximately 3,000 M(-1) s(-1)) than with FAD(ox) (k(obs)=16 [corrected] M(-1) s(-1)). The dissociation constant of the photolyase-FAD(ox) complex is about 2.3 microM and that of E-FADH(-) is not higher than 20 nM (pH 7.2).     

Pathways of electron transfer in Escherichia coli DNA photolyase: Trp306 to FADH

We describe the results of a series of theoretical calculations of electron transfer pathways between Trp306 and *FADH. in the Escherichia coli DNA photolyase molecule, using the method of interatomic tunneling currents. It is found that there are two conformationally orthogonal tryptophans, Trp359 and Trp382, between donor and acceptor that play a crucial role in the pathways of the electron transfer process. The pathways depend vitally on the aromaticity of tryptophans and the flavin molecule. The results of this calculation suggest that the major pathway of the electron transfer is due to a set of overlapping orthogonal pi-rings, which starts from the donor Trp306, runs through Trp359 and Trp382, and finally reaches the flavin group of the acceptor complex, FADH.     

Excited-state properties of Escherichia coli DNA photolyase in the picosecond to millisecond time scale

Escherichia coli DNA photolyase contains a stable flavin radical that is readily photoreduced in the presence of added electron donors. Picosecond, nanosecond, and conventional flash photolysis technique have been employed to investigate the events leading to photoreduction from 40 ps to tens of milliseconds following flash excitation. Direct light absorption by the flavin radical produces the first excited doublet state which undergoes rapid (within 100 ps) intersystem crossing to yield the lowest excited quartet (n pi*) state. In contrast, light absorption by the folate chromophore produces a new intermediate state via interaction of the folate excited singlet state with the ground-state flavin radical, leading to an enhanced yield of the excited radical doublet state and hence quartet state. Subsequent reaction of the excited quartet state involves hydrogen atom abstraction from a tryptophan residue. Secondary electron transfer from added electron donors occurs to the oxidized tryptophan radical with rate constants ranging from 10(4) (dithiothreitol) to 4 x 10(6) M-1 s-1 (n-propyl gallate). The low value of the latter rate compared to reduction of the tryptophan radical in lysozyme suggests that the reactive tryptophan is highly buried in photolyase. A redox potential diagram has been constructed for the ground and excited states involved. It is concluded that the one-electron reduction potential of the excited quartet state of the flavin radical must be at least 1.23 V more positive than the ground state, in agreement with the value of delta E greater than 1.77 V calculated from spectroscopic data.