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 co- factors, FADHz and 5,10-methenyltetrahydrofolate. The function of the folate has been characterized in greater detail in this series of experiments. Investiga- tions 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 polygluta- of the folate and that the exogenously added monoglutamate derivative is less tightly associated with the the remains of enzyme also demonstrates that the folate acts catalytically. The stimulation of turnover by ex- ogenous folate seen at low concentrations of photolyase is shown to be due to the lower affinity of photolyase for the monoglutamate derivative used in reconstitu- tion 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.

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.
Exposure of DNA to UV (200-300 nm) light causes the formation of pyrimidine dimers between adjacent pyrimidines in DNA. One cellular defense against these lesions is the enzyme DNA photolyase (deoxyribodipyrimidine photolyase, EC 4.1.99.3). DNA photolyase binds to cis-syn pyrimidine dimers in a light-independent step and then catalyzes the photoreversal of these dimers to monomers in a light-dependent (300-500 nm) step. Purified photolyase from Escherichia coli and Saccharomyces cerevisiae contains two prosthetic groups, FADH? and CH'-H, folate' (1,2), which facilitate the repair reaction. We have previously shown that the folate cofactor of the enzyme serves to transfer light energy to the flavin, which then catalyzes the (2 + 2) cycloreversion of the photodimer by electron transfer (3-5). However, it was unclear from these studies whether the folate lost its functionality after a single event or could participate in multiple turnovers.
Indeed, we have observed that at low concentrations of enzyme (5-10 nM)  Background counts of buffer were subtracted as well.
Gel Retardation Assays-All gel retardation assays and multiple turnover assays were performed under yellow light. Gel retardation assays of photolyase activity were performed as described previously (4). The substrate, a 48-mer duplex containing a single thymine dimer at a specific internal site, was prepared as described previously (12). concentrations, the repair kinetics shown in the upper gel in Fig. 1 were observed. However, if excess CH+-H4folate was added to the reaction mixture, the rate of repair was stimulated as shown in the gel in the lower panel of Fig. 1. This reproducible phenomenon was designated the "folate effect." Controls showed that DNA incubated with excess of CH'-H,folate and exposed to photoreactivating light in the absence of enzyme did not produce such an effect (data not shown). Preirradiation of the DNA in the presence of excess CH+-H4folate followed by addition of photolyase and examination by the gel shift assay also failed to produce any visible change in the ability of the DNA to be bound by photolyase (Fig. 1, bottom, lanes 9 and 10).
The observed folate effect might be attributed to one of two possibilities.
It was possible that at low ( contained DNA only. Lanm 2-9 contained DNA and a IO-fold excess of photolyase. The sample in lam 2 was not photoreactivated. The photoreactivating fluences for /nnes 3-9 were 'i.50, 1500, 2250, 3000, 6000. 9000, and 9000 erg/mm', respectively. Hottom, lanes l-8 received identical fluences to the comparable lanes in the fop ~4. Lane 9 contained DNA incubated with excess CH+-H.,folate and was given a fluence of 9000 erg/mm'. Lane IO contained an aliquot of the DNA from lone 9 that was then incubated with lo-fold excess photolyase for 20 min. The uppc~r bands on the gel represent the enzyme-DNA complexes (unrepaired), while the lowr bands represent free DNA (repaired). Photoreactivations were performed with a black lamp with maximum output at X5 nm. of photolyase the folate dissociated from the enzyme and that the presence of excess folate shifted the binding equilibrium in the direction of enzyme-bound folate. It also seemed possible that absorption of light by the folate could degrade the folate and cause its release from the enzyme. In this case, the excess folate in the reaction mixture would replace the modified folate and lead to a higher rate of repair. The observation that the folate cofactor is photodecomposed by prolonged exposure to high intensity light (3,4) might also be attributed to the lability of a high energy folate intermediate. mate nor the hexaglutamate derivatives are exchanged to any significant extent. The exchange of these folylpolyglutamates is not stimulated by the addition of photodimers and photoreactivation of these enzyme-substrate mixtures (data not shown). We interpret this as evidence that these polyglutamate derivatives of the folate cofactor are bound more tightly to the enzyme than is the monoglutamate derivative and that the polyglutamate moiety contributes substantially to the binding energy.
Catalytic Folate-Though the folate effect seemed likely to be attributable to reconstitution with the nonphysiological monoglutamate derivative and dissociation of this derivative at low concentrations, it was necessary to address the issue of whether the folate was catalytic. All of the assays so far had been performed under conditions involving a single turnover of the enzyme. In fact, quantum yield measurements of photolyase are typically conducted under single turnover conditions (13). It remained to be established whether the photoexcited folate might be released when the enzyme is allowed to undergo several cycles of reaction.
In order to address this question, photoreactivation exper-iments were conducted under multiple turnover conditione and at micromolar concentrations of photolyase so that the amount of CH'-H,folate that would freely dissociate from the enzyme would be negligible. Photolyase that had been treated with sodium borohydride and reconstituted with "H-labeled 3',5',7,9-CH+-H4folate was incubated in the presence and absence of single-stranded oligo(dT)15-containing thymine dimers such that dimers were in 20-fold excess over photolyase. Mixtures were exposed to photoreactivating light sufficient to allow 10 enzyme turnovers and then concentrated on Centricon 30 microconcentrators and analyzed for enzymebound 3H-labeled 3',5',7,9-CH+-H,folate. The data clearly showed that even under conditions of multiple turnover, the folate remains bound to enzyme. That the double-stranded 4%mer and single-stranded oligo(dT),, were repaired at approximately equal rates was verified by a spectrophotometric photolyase repair assay (14). Comparison of the repair 01 oligo(dT)ls by the spectrophotometric assay and repair of the 48-mer by T4 endonuclease V assay revealed that when all 48-mer had been repaired in the multiple turnover experiments, 10 enzyme turnovers had occurred (data not shown). Folate on DNA Photolyase Previous studies have demonstrated that the E. coli DNA photolyase has equal affinity for pyrimidine dimers in singlestranded DNA and double-stranded DNA (17) and that the dimer in the 4%mer double-stranded DNA substrate is repaired with the same quantum yield as dimers in oligo(dT) (18).
T4 Endonuclease V Assays-Further verification of the catalytic nature of the folate cofactor of E. coli DNA photolyase was provided by T4 endonuclease V assays. It has previously been shown that folate-free photolyase exhibits a decreased photolytic cross-section relative to photolyase-containing stoichiometric folate (4), indicating that the light absorbed by the folate is used for dimer repair. By utilizing the T4 endonuclease V assay under the multiple turnover conditions with reaction mixtures containing oligo(dT)ls and trace 48-mer with dimer, the rates of repair by borohydridetreated photolyase, fully supplemented photolyase, and fully supplemented photolyase with additional excess folate were measured. It is evident from Figs. 4 and 5 that the repair by the borohydride-treated enzyme is significantly lower than that by the fully supplemented enzyme or the fully supplemented enzyme with excess folate. These data further confirm the catalytic nature of the folate, since under the conditions of the assay (10 turnovers), if the folate were lost after a single turnover, the repair curve for fully supplemented photolyase in the absence of added folate should have resembled  Photoreactivation times: lanes I, 8, and 15, not photoreactivated; lanes 2, 9, and 16, 30 s; lanes 3, 10, and 17, 1 min; lanes 4, 11, and 18, 2 min; lanes 5, 12, and 19, 3 min; lanes 6, 13, and 20, 5 min; and lanes 7, 14, and 21, 10 min. Lanes 22 and 23 contained "2P-labeled dimer treated with T4 endonuclease V and not exposed to the endonuclease, respectively. The data points were obtained from the autoradiogram shown in Fig 4. The rates of repair for folate-free photolyase (M), photolyase containing stoichiometric CH'-H4folate (0). and photolyase containing stoichiometric CH+-H4folate plus excess CH'-H,foiate (A.) were obtained by gel scan of the autoradiogram in Fig. 4. iments performed with limiting light (4), that the presence of the folate increases the photolytic cross-section of the enzyme.

Assays of Photolyase Supplemented
with CH+-Hjolate and CH+-H,folylpolyglutamate-To test the hypothesis that the folate effect is in fact primarily a dissociation effect, samples of photolyase that had been reconstituted with either CH+-H4folate or heterogeneous CH+-H.,folylpolyglutamate were assayed by gel retardation assay under single turnover conditions and under limiting light for repair of the 48-mer containing a thymine dimer. The results of a representative experiment are shown in Fig. 6. As predicted, the folate effect was evident at low concentrations (6 nM), but the magnitude of the stimulation by excess folate was much greater for the photolyase samples that had been reconstituted with CH+-H4folate. This finding was consistent with our observation that dissociation of the monoglutamate form of the folate cofactor occurs in this range and that the presence of excess folate permits saturation of the folate site on the enzyme. The observation of a lesser magnitude of the folate effect for the folylpolyglutamate-reconstituted enzyme corroborates the results of binding studies showing tighter binding of the longer chain folylpolyglutamates to photolyase. The stimulation observed with the folylpolyglutamate-reconstituted enzyme is probably due to the dissociation of the triglutamate and tetraglutamate derivatives of CH+-H,folate. Of interest also was the HPLC profile of lo-CHO-folylpolyglutamates isolated from the CH'-H4folylpolyglutamate-reconstituted photolyase. Reconstitution of photolyase with the heterogeneous mixture of CH+-H4folylpolyglutamates yielded an enzyme preparation containing a higher proportion of the hexa-and pentaglutamate derivatives than were present in the reconstitution mixture (data not shown). This again demonstrates the higher affinity for the longer chain CH+-Hlfolylpolyglutamates inferred from the exchange experiments.

DISCUSSION
The goal of these studies was to determine whether the CH+-H,folate cofactor of E. coli DNA photolyase behaves catalytically or noncatalytically during the reaction catalyzed by the enzyme. Earlier observations that the enzyme-bound folate, but not the free species, was photodecomposed by prolonged exposure to high intensity light (3,4) suggested the possibility that during the light-harvesting reaction, photoexcited folate might be chemically altered and released from the enzyme. The observation that the catalytic efficiency of photolyase supplemented with stoichiometric folate and assayed at low enzyme concentrations under single turnover conditions was stimulated by excess folate was in accord with this Assays were performed with photolyase reconstituted with stoichiometric CH'-Hlfolate in the presence (m) or absence (Cl) of excess CH+-Hlfolate or with photolyase reconstituted with CH+-Hlfolylpolyglutamate in the presence (A) or absence (A) of excess CH'-H,folate. possibility.
The failure to observe this folate effect with limiting substrate but at higher concentrations of enzyme was explainable as being due to replacement of released folate (that must be turned over with a quantum yield of less than 1) with folate cofactor from the excess enzyme in solution. However, it also seemed feasible that this folate effect was not due to decomposition of the folate after each catalytic event but rather to its dissociation from the enzyme. Purified E. coli DNA photolyase contains a heterogenous mixture of CH+-H4folylpolyglutamates (2) but exhibits tighter binding to CH'-H,folate derivatives containing 5 or 6 glutamate residues. This was evident from the exchange experiments which demonstrate dissociation of mono-, tri-, and tetraglutamate derivatives but not of the penta-and hexaglutamate derivatives.
The greater affinity of photolyase for the longer side chain derivatives of CH'-H,folate is not unexpected, since these folylpolyglutamate derivatives are more prevalent in E. coli (15). The presence of subsaturating amounts of folate in purified photolyase has been attributed partly to an insufficiency of intracellular folate to meet the demand of the overproduced photolyase and partly to dissociation of this chromophore from the apoenzyme during purification (4). for the folate-binding site of photolyase is also evidenced by the fact that less of a folate effect is observed for enzyme reconstituted with folylpolyglutamates than with the monoglutamate as determined by gel retardation assay.
The issue of folate turnover was not resolved by these binding studies, though they did explain the folate effect satisfactorily.
To truly address the issue of folate turnover, it was necessary to use multiple turnover experiments.
By using photolyase reconstituted with 3H-labeled CH'-H,folate at micromolar concentrations to minimize dissociation of this monoglutamate derivative of the folate, it was determined after multiple turnovers that the 3H label remained bound to the enzyme. In addition, the use of coupled T4 endonuclease V assays revealed that photolyase at micromolar concentrations and containing stoichiometric folate exhibited the same rate of repair in the presence or absence of excess folate. This rate was greater than that exhibited by folate-free photolyase assayed under identical conditions. This was further verification that enzyme-bound CH+-H4folate behaves catalytically in the E. coli DNA photolyase.
The presence of a catalytic folate in photolyase is consistent with other observations. Payne (18) found that photolyase assayed under multiple turnover conditions did not exhibit any changes in absorbance at 384 nm. Other studies have revealed that the CH+-H,folylpolyglutamates of the yeast DNA photolyase are not photodecomposed by excess light (19).' Though we have demonstrated that the folate is a catalytic cofactor, it is still possible that the structure of the folate is reversibly altered during the light harvesting process; ' G. Sancar, personal communication. for instance, the energy transfer to the flavin could be driven by cleavage of the methenyl bridge of the CH+-H4folate to yield lo-CHO-Hlfolate.
The results obtained by folylpolyglutamate exchange assay with photoreactivation in the presence of excess dimers strongly suggest that there is no transient release of a lo-CHO-H,folate derivative, as CH+-H,folylpolyglutamate exchange with free monoglutamate is not stimulated during enzyme turnover. Therefore, if cleavage to the lo-CHO-Hlfolate derivative occurs, the latter must be rapidly converted to CH+-H,folate.
Scheme 1 illustrates our model for catalysis by the E. co11 DNA photolyase. The enzyme is shown to contain FADH, and folate, since the FADH' seen in the purified enzyme is apparently photoreduced before the catalytic cycle is initiated (7,20,21). Most of the light absorption is by the folate cofactor since the extinction of enzyme containing folate is 6-fold higher at 384 nm than that of folate-free enzyme (4). It has been postulated that the photoexcited folate can transfer energy to the FADH, (3)(4)(5). In the absence of the folate, apparently the FADH* itself can absorb sufficient light to become photoexcited, as enzyme containing only the flavin cofactor has been demonstrated to show activity, though with a reduced photolytic cross-section (3, 4). The photoexcited flavin is proposed to directly catalyze the dimer photoreversal by a mechanism involving electron donation/abstraction (5, 22). The reaction catalyzed by the E. coli DNA photolyase is the only known reaction which utilizes a tightly bound folate cofactor to harvest light. It is also the only known reaction which makes catalytic use of the folate rather than utilizing the carbon or hydrogen donor capabilities of this cofactor. T4 endonuclease V and irradiated oligo(dT)15 and Ywan Feng Li for her assistance in the preparation of DNA substrates. The technical assistance of Meleah Phillips in correlating the rates of repair of the DNA substrates, oligo(dT),s, and the 48.mer was much appreciated.