Role of Enzyme-bound 5, IO-Methenyltetrahydropteroylpolyglutamate in Catalysis by Escherichia coli DNA Photolyase”

DNA photolyase catalyzes the photoreversal of py- rimidine dimers. The enzymes from Escherichia coli and yeast contain a flavin chromophore and a folate cofactor, 5,lO-methenyltetrahydropteroylpolygluta-mate. E. coli DNA photolyase contains about 0.3 mol of folate/mol flavin, whereas the yeast photolyase contains the full complement of folate. E. coli DNA photolyase is reconstituted to a full complement of the folate by addition of 5,lO-methenyltetrahydrofolate to cell lysates or purified enzyme samples. The reconsti- tuted enzyme displays a higher photolytic cross section under limiting light.

Irradiation of DNA with UV light causes the formation of dimers between adjacent pyrimidines. One cellular defense against these mutagenic lesions is the enzyme DNA photolyase (deoxyribodipyrimidine photolyase, EC 4.1.99.3). DNA photolyases bind to cis-syn pyrimidine dimers in a lightindependent step and then catalyze the reversal of these dimers to pyrimidine monomers in a light-dependent step driven by near-UV or visible light. DNA photolyases have been purified from many different organisms and shown to contain intrinsic chromophores. The action spectrum of each enzyme varies according to the nature of the chromophores, thus implicating these chromophores in catalysis.
Escherichia coli DNA photolyase is a monomeric protein of molecular mass 54,000 daltons (1). The purified protein con-* This work was supported in part by National Institutes of Health Grant GM31082. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked ''advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Supported by National Research Scientist Award Grant
(1 To whom correspondence and reprint requests should be ad-  dressed. tains FAD as a neutral blue flavin radical (2) as well as the recently identified 5,10-methenyltetrahydropteroylpolyglutamate, CH+-H,PteGlu,' (3), which was previously referred to as the "second chromophore." The flavin radical displays a series of absorption bands above 400 nm and contributes about one-third of the absorbance at 384 nm in purified enzyme. Free CH'-H,folate has an absorption maximum at 355 nm at pH 2 with t = 25,000 M" cm" (4). The proteinbound chromophore has an absorption maximum a t 384 nm at neutral pH. The folate is the major absorbing component at 384 nm, accounting for at least two-thirds of the extinction (5). The semiquinone nature of the flavin appears to be a purification artifact, because it is not detected by EPR in vivo and photochemical or dithionite reduction of the radical increases the photolytic cross-section and the quantum yield to levels comparable to the in vivo value (6).
The action spectrum of the purified blue E. coli enzyme corresponds to the absorption spectrum of the enzyme with a maximum at 384 nm (7), indicating that the bound CH+-H,PteGlu, participates in the repair reaction. One possible mechanism has been proposed in which light is absorbed by this chromophore and transferred as energy to the reduced flavin (9). Reversible electron transfer could then occur directly from the flavin to the dimer, producing a dimer radical anion which could spontaneously monomerize and transfer its extra electron back to the flavin. The role of the flavin as the electron donor is supported by photochemical studies of free flavins and dimers (9, 10).
Experimental procedures for selective modification of the second chromophore have been described. Exposure of the E. coli photolyase to repeated camera flashing (11) or excess sodium borohydride (12) results in elimination of the folate absorption band a t 384 nm. While no quantitative data are available on the borohydride-reduced enzyme, analysis of camera-flashed enzyme indicates that although &, the photolytic cross-section (t = molar extinction coefficient and 4 = quantum yield), decreases, C $ remains constant (11). Because the flavin appears to remain intact after each treatment, it seems to be capable of sustaining activity in the absence of the folate absorption band.
In this study of the role of the folate cofactor of E. coli photolyase, the structural changes in the folate cofactor that are brought about by camera flashing and sodium borohydride are characterized. The amount and chemical state of the Assay of Photolyase-E. coli DNA photolyase was assayed by a gel retardation assay. A 48-base pair duplex containing a single dimer in a specified site synthesized by the method of Husain et al. (22) was kindly provided by Dr. Intisar Husain. The 48-mer was end-labeled with 3'P using T4 polynucleotide kinase. Excess photolyase was mixed with substrate so that at equilibrium all dimers were bound by the enzyme, and the mixture was irradiated with light of limiting intensity. Under these conditions, repair can be quantitated as a function of fluence. The photoreactivation cross-section, e$, is obtained using the following equation (23): where k, is the photolytic rate constant and is obtained from the plot of fraction of dimers remaining uersus fluence (Rupert plot), as described by Sancar et al. (7). The k, for the photolyase reaction is defined in the following equation: All manipulations were performed under illumination from General Electric gold fluorescent lamps. No photoreactivation occurs under these conditions. Reaction mixtures containing 50 mM Tris-HC1, pH 7.4, 100 mM NaCl, 1 mM EDTA, 20 mM DTT, 100 gg/ml bovine serum albumin in addition to 32P-labeled substrate (1 nM) and excess (10-40 times) enzyme were incubated at room temperature in the dark for 30-60 min. After the incubation, the sample was photoreactivated at 384 nm with a Quantacount monochromator-actinometer from Photon Technology International (Princeton, NJ) calibrated by potassium ferrioxalate actinometry (24).
The Quantacount was equipped with a Fisher circulating water bath which maintained a constant sample temperature of 23°C. All 384 nm irradiations were performed through a Balzers 10.2% filter and a BP370 filter from Schott Glass Technologies Incorporated.
Aliquots of 50 g1 were removed at various fluences during the photoreactivation and run on a 5% acrylamide gel as described previously (25). Under these assay conditions all dimers are complexed by photolyase unless repaired and released during photoreactivation (26). The free (repaired) DNA and enzyme-DNA complexes migrate differently and are located by autoradiography. The bands due to repaired DNA are more compact and are typically used for quantitation instead of the more diffuse enzyme-DNA bands. By excision and quantitation by Cerenkov counting of the bands due to repaired DNA, both the extent of repair and the fraction of dimers remaining at any given fluence can be determined. Because some of the enzyme-DNA complexes dissociate during electrophoresis, some "free" DNA is present even when the photolyase is in excess and all substrate is enzyme bound (as determined by flash photolysis) prior to electrophoresis. Therefore a correction is made by assuming 100% retardation for nonphotoreactivated samples. When it was necessary, individual assays were corrected for nonspecific binding by determining the amount of protein in the same excess that was bound to unmodified 48-mer.

RESULTS
Folate Content of the E. coli DNA Photolyase-We previously reported that E. coli DNA photolyase contains tightly bound CH'-H,PteGlu, but did not determine its stoichiometry relative to the flavin cofactor. Quantitative analysis of CH'-H,PteGlu, bound to photolyase revealed that the amount of the folate cofactor varied from one preparation to another, supporting an earlier observation that the ratio of the absorbance at 384 nm (due primarily to the folate) to the absorbance at 580 nm (due entirely to radical flavin) is not constant (6). Photolyase purified by the standard procedure The finding that purified enzyme does not have a stoichiometric complement of folate is not surprising. CH'-H,folate has an extinction of 25,000 "' cm-' at 355 nm at pH 2. In contrast, the extinction of a typical photolyase preparation a t 384 nm is only 18,100 M" cm" and this extinction is due to the absorption of both the folate and the radical flavin (5). This discovery also explains the variability in the absorbance ratios at 384 and 580 nm that was previously reported (6). In contrast, yeast DNA photolyase purified from an overexpressing E. coli strain has a 377 nm peak which is considerably larger than the corresponding 384 nm peak for the E. coli enzyme even after subtraction of the contributions of the respective flavin chromophores (5,27). This suggested that the yeast enzyme might have a higher folate content. HPLC analysis of the CH+-H,PteGlu, from yeast photolyase indicated that the CH'-H,PteGlu, was present in a 1:l stoichiometry with flavin. Quantitation of total folate verified that all bound folate was accounted for by CH'-H,PteGlu,.
Supplementation of E. coli DNA Photolyase with CH'-HJolate-Reconstitution of photolyase to its full complement of folate required the use of conditions under which both CH'-H,folate and enzyme were stable. It was found that the use of 50 mM citrate buffer, pH 6.0, with 50 mM NaC1, 1 mM EDTA, 10 mM DTT met these requirements. The results of a reconstitution experiment using CH+-H,folate are presented in Fig. 1. The 384 nm peak of the supplemented enzyme increased dramatically, indicating that the added CH+-H,folate was bound by the enzyme and exhibited the same bathochromic shift from 355 to 384 nm as the native chromophore. The bound CH'-H,folate was not removed from the enzyme by chromatography on Sephadex G-50 as evidenced by the retention of the strong 384 nm absorption band on the enzyme after chromatography. In addition, when I4CH+-H,folate was used to supplement the enzyme, the enzyme retained the radioactivity after chromatography on Sephadex G-50 (data not shown). Analysis of supplemented enzyme indicated that it contained a full complement of the CH+-H,folate derivative. From these binding experiments, it is also apparent that the polyglutamate side chain is not critical for the stable binding of CH'-H,folate to photolyase.
In order to understand the basis for the folate-deficient state of purified photolyase, the purification procedure was modified slightly by lysing the E. coli cells in a citrate buffer at pH 6.0 which contained additional CH'-H,folate. The absorption band at 384 nm of photolyase purified using this procedure clearly showed that the CH'-H,folate added to the crude extract had supplemented the enzyme in that extract (data not shown). HPLC analysis of this supplemented photolyase indicated that the enzyme contained a stoichiometric complement of CH'-H,PteGlu,. It seems, then, that the primary cause of the lack of stoichiometric folate in standard purified enzyme is an insufficiency of the appropriate folate species in uiuo, caused by the gross overproduction (15% of the total cellular protein) of the apoenzyme.
Borohydride Treatment of Free and Enzyme-bound CH'-Hdolate-Jorns et al. (12) have shown that the addition of sodium borohydride to E. coli photolyase results in a dramatic decrease in the 384 nm absorption, suggesting that the second chromophore (folate) was in fact eliminated by this treatment. The borohydride reacted selectively, as evidenced by the lack of change in the spectrum of the enzyme-bound radical flavin. We have examined the effect of sodium borohydride on the enzyme-bound CH'-H,PteGlu, in greater detail by analyzing the chemical nature of the products of the reaction and establishing whether these products were still enzyme-bound. Reaction of free CH'-H,folate with sodium borohydride has been demonstrated to result in conversion to CH,-H,folate or CH3-H,folate, depending upon the amount of borohydride used in the reaction (4). The observed effect of borohydride on the enzyme, i.e. elimination of the folate absorption band at 384 nm, was consistent with reduction of the enzymebound CH+-H,folate to one of these reduced derivatives with absorption maxima below 300 nm.
To facilitate analysis of the products of borohydride reduction of photolyase, the enzyme was reconstituted with I4CH+-H,folate such that it contained approximately 70% of the bound folate as the l4C-labeled monoglutamate. Fig. 2 shows the absorption spectrum of photolyase supplemented with 14CH'-H,folate before and after treatment with sodium borohydride. The elimination of the 384 nm absorption band of the bound CH'-H,PteGlu, is apparent. The results from HPLC analysis of the acid-released products of the borohydride reaction in an acid/methanol gradient are shown in Fig.  3. The folate from supplemented photolyase showed a single peak of CH+-H,folate as judged spectrally and by recovery of radioactivity. After 2 h of borohydride treatment, the CH'-H,folate peak had almost completely disappeared and two new peaks were observed. These peaks were identified as CH,-H,folate and CH,-H,folate by their spectra and by HPLC coelution with the appropriate standards. The yields of CH3-H,folate and CH,-H,folate were about 60 and 30%, respectively. The remaining counts appeared in the excluded volume of the column and were probably due to free [14C]formaldehyde formed from the acid-catalyzed degradation of the CH,-H,folate to H,folate and formaldehyde (28). All of the initial counts from the enzyme-bound CH'-H,folate were recovered in these three products. These conclusions were confirmed by HPLC analysis of the products in tetrabutylammonium phosphate/methanol. Chromatography in the second solvent revealed the same products in the same distribution.
To determine whether CH3-H,folate and CH,-H,folate remained enzyme-bound, photolyase that had been supplemented with I4CH'-H,PteGlu and treated with borohydride was centrifuged through Penefsky columns containing Sephadex G-50. The included and excluded column volumes were  recovered and analyzed for radioactivity. All of the counts in the untreated sample were recovered in the excluded volume, whereas 85% of the counts for the borohydride-reduced sample were present in the included volume. These findings indicate that the affinity of photolyase for the CH,-H,folate and CH,-H,folate derivatives produced by the borohydride reaction is much less than for the CH'-H,folate derivative.
Modification of Free and Enzyme-bound CH+-Hdolate by Camera Flashing-Prolonged flashing of E. coli photolyase causes a reversible reduction of the enzyme-bound flavin radical (29) but an irreversible decline in absorption at 384 nm attributable to the "selective decomposition" of the second chromophore (11). To examine the nature of the chemical modification of the folate chromophore caused by flashing, photolyase supplemented with l4CH+-H,folate was flashed repeatedly until no further decrease in the absorbance at 384 nm was observed. The spectra in Fig. 4 demonstrate the decrease in the 384 nm band that occurred with exposure to camera flashes. The photoreduction of the flavin radical with the concomitant appearance of a peak at 360 nm due to FADH, was also evident. Supplemented/flashed photolyase was examined for its ability to be supplemented again with CH+-H,folate and was found to be reconstituted to a full complement of the CH+-H,folate under standard conditions. This indicates that the folate-binding site on the enzyme remains intact during this treatment. HPLC analysis in an acid/methanol gradient of the acidreleased products from unflashed enzyme demonstrated that all of the I4C counts were associated with the CH+-H,folate. Analysis of the flashed sample by HPLC in an acid/methanol gradient indicated that no CH2-H4folate or CH3-H,folate were present. Chromatography in tetrabutylammonium phosphate/methanol at pH 7.4 (Fig. 5)   sample was flashed and analyzed on Penefsky columns it was found that 83% of the initial counts were lost from the excluded fraction. Because the 3H label was present in both the pterin andp-aminobenzoic acid moieties, it is evident that both of these components of the folate, in addition to the formate, were released by flashing. Although the exact nature of the folate products of flashed enzyme remains to be established, it is clear that camera-flashed enzyme has lost most of its folate. Activity of Supplemented, Supplemented/Borohydridetreated, and Supplemented/Flashed Enzyme-With the availability of fully supplemented enzyme it was possible to examine further the role of the folate in the repair reaction. Supplemented photolyase was assayed by a gel retardation assay. The autoradiogram of such a gel is shown in Fig. 6, and a Rupert plot of data from several experiments is presented in Fig. 7. The photolytic cross-section for supplemented enzyme was calculated by extrapolating the initial phase of the repair curve to find the kp value. The biphasicity observed here is characteristic of the reaction with supplemented enzyme but its cause is unknown. The photolytic cross-section for unsupplemented photolyase was reported previously (7) and is shown here for comparison by the dashed and dotted line in Fig. 7. It is evident that supplemented photolyase exhibits a higher photolytic cross-section under limiting light than does unsupplemented photolyase. This increase is not due to an increase in the affinity of supplemented enzyme for the substrate, because the same amount of DNA is bound by supplemented and unsupplemented enzyme (data not shown). The increased catalytic efficiency must therefore be due to an increase in the overall efficiency of the photochemical reaction per incident quantum, leading to the conclusion that the externally added CH+-H,folate enhances the photochemical reaction of the enzyme.
The decrease in absorbance of the enzyme at 384 nm due to formation and release of the CH,-H,PteGlu, and CH,-H4PteGlu, upon treatment with borohydride should then be reflected in a lower photolytic cross-section at 384 nm than that of supplemented or unsupplemented photolyase under these conditions. To test this proposal, supplemented/borohydride-treated photolyase was also assayed by the gel retardation technique under enzyme excess conditions. Fig. 6 shows the autoradiogram of a typical gel assay, and the results of several assays are shown in the Rupert plot in Fig. 8. It is apparent from these results that the removal of CH+-H,PteGlu, by reduction with borohydride does dramatically decrease the photolytic cross-section of the enzyme.
Supplemented/flashed enzyme showed an attenuated efficiency under limiting light (Fig. 6, gel, and Fig. 8, plot). Also of interest was the finding that enzyme that had been supplemented, flashed, and resupplemented with folate had a catalytic efficiency comparable to that of supplemented enzyme, indicating that the decrease observed for the flashed sample was not due to damage to the enzyme from the flashing procedure. Table I

TABLE I
Relative photolytic cross-seetiom at 384 nm c3@ is 37,000 M" cm" for the supplemented and supplemented/ flashed/supplemented photolyase preparations used in these experiments. This value is derived from the c580 for the flavin semiquinone of the photolyase preparations. Exact values depend upon knowledge of the total flavin content of each preparation. c3% is 6000 M" cm" for supplemented/borohydride-treated photolyase, 11,000 M" cm" for supplemented/flashed photolyase, and 18,100 M" cm" for unsupplemented photolyase. coli photolyase preparations occurs specifically, as evidenced by the long term stability of the supplemented complex at neutral p H where the free compound is unstable. The bathochromic shift of the absorption maximum of the folate from 355 to 384 nm is further evidence for specific binding. The dramatically higher photolytic cross-section of supplemented enzyme under conditions of limiting light at 384 nm attests to the productive nature of the binding of the added chromophore.
As summarized in Fig. 9, reduction of supplemented photolyase with excess sodium borohydride for 2 h results in the formation of CH2-H,folate and CH,-H,folate, both of which are released from the enzyme. The distribution of these two products is probably a function of the incubation time with borohydride, with longer incubations of photolyase with borohydride producing a greater percentage of CH,-H,folate. The photolytic cross-section for borohydride-treated photolyase under limiting light at 384 nm is considerably lower than that of supplemented and unsupplemented photolyase. Jorns et al. (12) have reported that borohydride-treated enzyme shows virtually no change in catalytic activity, in contrast to the results presented here. It is to be noted that under high intensity illumination where the light-independent step, ES complex formation, becomes rate limiting, photolyases with different amounts of folate will have the same activities. The assay conditions utilized here, however, are designed specifically to assess differences in light-harvesting capabili- ties. Camera flashing of supplemented photolyase also results in release of the counts associated with the enzyme-bound CH+-H,folate, indicating that virtually no bound folate remains. Camera-flashed enzyme is also active, but its photolytic cross-section is lower than that of supplemented or unsupplemented photolyase. This finding is consistent with the data reported by Heelis et al. (11) that the photolytic cross-section of flashed enzyme decreases. As the data summarized in Table I indicate, it is evident that in the absence of folate the enzyme is able to catalyze dimer repair. However, it is obvious that both the photolytic cross-sections and catalytic efficiencies increase with increasing contents of folate.
The presence of less than stoichiometric amounts of folate in the purified enzyme is surprising, considering the ease and rapidity of supplementation with CH+-H,folate and 10-CHO-H,folate. The finding that addition of CH+-H,folate to crude cell lysates as a prelude to the purification procedure yields a fully supplemented preparation indicates that overproduction of photolyase in the genetically engineered strain is the primary cause of the insufficiency of folate in the enzyme. The broad distribution of folylpolyglutamates bound to photolyase (3) might also be attributed to the overproduction of photolyase.
It has been proposed that the role of the second chromophore is that of a light harvester which could garner more light energy than reduced flavin because of its high extinction (8). The folate does have a light harvesting function because the increase in the extinction at 384 nm results in a corresponding increase in the photolytic cross-section of the enzyme. It is conceivable that the excited CH+-H,PteGlu, transfers energy to reduced flavin and that excited reduced flavin reversibly transfers an electron to the dimer to complete the repair reaction. Energy transfer from the CH+-H,PteGlu, might occur through orbital overlap. Nonradiative energy transfer, or Forster energy transfer, is known to occur in biological systems. This type of energy transfer requires overlap by the fluorescence emission spectrum of the donor with the absorption spectrum of the acceptor (31). Because the fluorescence emission of the donor (folate) is approximately at 470 nm (2) and the lowest energy absorption band of the acceptor (FADH,) is at 360 nm, this type of mechanism does not seem to be feasible.
Assays of borohydride-treated and camera-flashed photolyase demonstrate that enzyme containing only flavin can catalyze the repair reaction. The CH+-H,PteGlu, might also be able to catalyze repair independently of the flavin either by electron donation or abstraction to form a stable folate radical and an unstable dimer radical that would collapse to monomers. This might occur by excitation of the enzymebound CH+-H,PteGlu, with cleavage at the bridging carbon and production of 10-CHO-H,PteGlu, with the bond cleavage providing energy to drive radical formation. The enzyme could then spontaneously recyclize the 10-CHO-H,PteGlu, that would eventually be regenerated to CH+-H,PteGlu,.
All of the possibilities mentioned so far invoke a catalytic function for the folate. However, folates typically function noncatalytically. The E. coli and yeast DNA photolyases belong to a small subset of enzymes that not only use folate but are actually purified with tightly bound folate. Two other enzymes, the rat liver sarcosine dehydrogenase and dimethylglycine dehydrogenase (32,33) are purified with tightly bound H,PteGlu,. A reaction mechanism for these two dehydrogenases has been proposed in which the bound H,PteGlu, serves to trap the carbon unit produced during the oxidative demethylation of dimethylglycine and sarcosine (33, 34, 35). The CH,-H,PteGlu, that is formed from this reaction is released from the enzyme and more H,PteGlu, then binds. It is conceivable that the biphasicity of the rate that is observed with the supplemented preparations is due to formation of a folate species on the enzyme that requires replacement by free folate. However, there is no evidence for folate release from photolyase during catalysis.
Until now, CH+-H,PteGlu, has not been identified as a participant in a folate-requiring reaction. Our results demonstrate definitively that light energy absorbed by the enzyme-bound CH'-H,PteGlu, is used to drive dimer repair. The mechanism of the reaction involving this folate cofactor is still unclear but further investigations along the lines described should enable us to elucidate the mechanism of this intriguing reaction.