DNA photoreactivating enzyme from the cyanobacterium Anacystis nidulans.

Photoreactivating enzyme, which specifically monomerizes pyrimidine dimers in UV-irradiated DNA, was purified 21,000-fold from the cyanobacterium Anacystis nidulans to apparent homogeneity with 41% overall yield. The enzyme consists of a single protein chain with 53,000 molecular weight. Maximal activity was found at pH 6.2 and 0.1 M NaCl. Purified photoreactivating enzyme exhibits a marked absorption spectrum with a main band in the blue region (maximum 437 nm), a protein band (maximum 266 nm), and a low intensity band above 500 nm. The molar extinction coefficient of native enzyme was estimated 53,000 at 437 nm. The action spectrum for photoreactivation shows maximal activity at 440 nm and correlates closely with the 437-nm absorption band. The enzyme contains two different intrinsic chromophores in equimolar amounts, which were identified as 7,8-didemethyl-8-hydroxy-5-deazariboflavin (FO) and (reduced) FAD. The low intensity absorption band of native photoreactivating enzyme exhibits a shoulder at 498 and maxima at 588 and 634 nm. This band is attributed to a neutral FAD semiquinone radical which accounts for the major part of the FAD present in dark equilibrated enzyme. Preillumination at 585 nm bleaches the semiquinone spectrum due to formation of fully reduced FAD, but exposure to air in the dark restores the spectrum completely. On preillumination at 437 nm the disappearance of FAD semiquinone is more rapid, indicating that the photoreduction is sensitized by the 8-hydroxy-5-deazaflavin chromophore. The 8-hydroxy-5-deazaflavin and possibly also the reduced FAD chromophore appear to act as a primary photon acceptor in the photoreactivation process.

Photoreactivating enzyme, which specifically monomerizes pyrimidine dimers in UV-irradiated DNA, was purified 2 1 ,OOO-fold from the cyanobacterium Anacystis nidulans to apparent homogeneity with 41% overall yield. The enzyme consists of a single protein chain with 53,000 molecular weight. Maximal activity was found at pH 6.2 and 0.1 M NaCl. Purified photoreactivating enzyme exhibits a marked absorption spectrum with a main band in the blue region (maximum 437 nm), a protein band (maximum 266 nm), and a low intensity band above 500 nm. The molar extinction coefficient of native enzyme was estimated 53,000 at 437 nm. The action spectrum for photoreactivation shows maximal activity at 440 nm and correlates closely with the 437-nm absorption band. The enzyme contains two different intrinsic chromophores in equimolar amounts, which were identified as 7&didemethyl-S-hydroxy-5-deazariboflavin (FO) and (reduced) FAD. The low intensity absorption band of native photoreactivating enzyme exhibits a shoulder at 498 and maxima at 588 and 634 nm. This band is attributed to a neutral FAD semiquinone radical which accounts for the major part of the FAD present in dark equilibrated enzyme. Preillumination at 585 nm bleaches the semiquinone spectrum due to formation of fully reduced FAD, but exposure to air in the dark restores the spectrum completely.
On preillumination at 437 nm the disappearance of FAD semiquinone is more rapid, indicating that the photoreduction is sensitized by the 8-hydroxy-5-deazaflavin chromophore. The Shydroxy-5-deazaflavin and possibly also the reduced FAD chromophore appear to act as a primary photon acceptor in the photoreactivation process.
Cyanobacteria are phototrophic prokaryotes capable of oxygenic photosynthesis.
They are mainly found in fresh and marine surface waters. In that habitat the UV component of sunlight will induce lesions in DNA and eventually damage other cell components, causing mutagenesis and cell death. It is expected therefore that cyanobacteria will possess repair systems to reverse the detrimental effects of UV radiation. Werbin and co-workers (1, 2) described the occurrence of photoreactivation in cyanobacteria.
In this repair process pyrimidine dimers in UV-irradiated DNA are split into the constituent pyrimidines by photoreactivating enzyme (pho-  (3,4). Photoreactivating enzyme (PRE)' was partially purified from the unicellular cyanobacterium Anucystis niduluns (5). Using this preparation the in uitro action spectrum of photoreactivation was determined which had its main band in the blue region (maximum 436 nm), in agreement with the in uiuo action spectrum of Agmenellum quadruplicatum, another cyanobacterium (6). Although evidence was gained for the presence of an intrinsic fluorescent chromophore in A. niduluns PRE (7), the structure of this chromophore was not elucidated.
In addition to photoreactivation cyanobacteria show dark repair, presumably excision repair (4,8,9), which gives, however, only a relative partial recovery (4, 9) while up to 100% survival can be obtained by photoreactivation even after extensive UV inactivation (8,(10)(11)(12). Photoreactivation also reduces the number of UV-induced mutations to the level of spontaneous mutations (13). Therefore, it seems to be the major repair pathway in cyanobacteria.
At present a few photoreactivating enzymes have been purified and characterized, each possessing two different intrinsic chromophores.
Two types of PRE can be recognized with respect to the nature of these chromophores.
PREs from Saccharomyces cerevisiae (14,15) and Escherichia coli (16,17) both contain reduced FAD and a reduced pterin (pterin-type) while PREs from Streptomyces griseus (18,19)  by centrifugation at 15,000 X g for 20 min at 4 "C. The pellets were resuspended in buffer B and passed once more through the sonic oscillator.
After centrifugation the supernatants were combined (yielding 1500 ml of crude extract) and loaded to a porous silica column (7.6 X 9.5 cm, containing 180 g of spherosil-type D, 100-200 grn, pore size 40-80 nm (Serva), flow rate 520 ml/h). The column was washed with buffer B and eluted with buffer B containing 0.1 M NaCl and 6% poly (ethylene glycol) 6000. The eluate (750 ml) was diluted with an equal volume of buffer C (10 mM NaCl, 10 mM potassium phosphate, 5 mM 2-mercaptoethanol, pH 7.0) and loaded overnight in the dark on a UV-DNA-cellulose column (4.4 X 10 cm, containing 50 g of UV-DNA-cellulose, flow rate 96 ml/h). The column was washed with buffer D (which is buffer B containing 5 mM P-mercaptoethanol) and eluted with buffer D containing 1.9 M NaCl (flow rate 170 ml/h).

Purification of Photoreactivating
Enzyme-Photoreactivating enzyme was isolated from A. nidulans cells and purified 21,000-fold with 41% overall yield by consecutive column chromatography on porous silica beads, UV-DNA-cellulose, heparin-Sepharose, DEAE-cellulose, and DEAE-Sepharose (Table I). The final preparation was apparently homogeneous as judged from SDS-electrophoresis (not shown). Moreover, gel electrophoresis under nondenaturing conditions revealed a single protein band after Coomassie Brilliant Blue staining. This band coincided with the photoreactivating activity profile obtained from an identical gel which was sliced, extracted, and assayed instead of stained (Fig. l) dicates that PRE consists of a single protein chain. The influence of pH on enzymatical activity gave a rather broad profile with a plateau ranging from pH 5.2 to 7.9 (80% activity). Maximal activity was found at pH 6.2 and 0.1 M NaCl, while there was hardly any activity above 0.3 M NaCl (not shown).
Purified PRE was used to determine the N-terminal amino acid sequence: For AA-23 an Ala signal was obtained, which could be due however to the two preceding Ala residues. For AA-25 both a Trp and an Ala signal were found.
The absorption spectrum of purified PRE is shown in Fig.  2. In addition to a protein band (maximum at 266 nm), a band is present in the visible region (maximum at 437 nm) as well as a low intensity absorption band above 475 nm. At enhanced sensitivity this band is resolved into a shoulder at 498 and two maxima at 588 and 634 nm. The high resolution action spectrum for photoreactivation (Fig. 2) shows a large band in the blue region (maximum at 440 nm) which coincides almost exactly with the 437-nm absorption band, indicating the presence of an intrinsic photochemically active chromophore. A second region of high photoreactivating activity was found in the UV-B region (maximum at 290 nm), likewise in good correspondence with the absorption spectrum. No photoreactivating activity, even at lo-fold enhanced light intensity, was found with wavelengths greater than 500 nm, indicating that absorption of light by the long wavelength absorption band does not give rise to enzymatic photoreactivation. azine (I Scheme 1 legend). Denaturation of PRE, which will release noncovalently bound cofactors, induced a relatively strong fluorescence in contrast with native PRE which is almost non-fluorescent.
The fluorescence excitation and emission spectra are nearly completely identical with the fluorescence spectra of 7,8-didemethyl-8-OH-5-deazariboflavin (II) (Fig. 3). The 8-OH-5-deazaflavin structure was corroborated by reaction with 8-OH-5-deazaflavin:NADPH oxidoreductase which reduces the chromophore to the non-fluorescent dihydro form (Fig. 3). Since this oxidoreductase is very specific, requiring the presence of both the 5-deazaflavin structure and the 8-OH substituent in its substrate (24), this enzymatical conversion together with the fluorescence measurements establishes the 8-OH-5-deazaflavin structure of the PRE chromophore, leaving the structure of the N(lO)-side chain unknown. Some information on the N( lo)-substituent can be obtained from the pK,, of the 8-OH group. At low ionic strength the pK, increases from 5.91 for 7,8-didemethyl-8-OH-5-deazariboflavin to 6.56 for F+ and 6.59 for SF420, apparently due to the presence of the charged phosphate in the N( lO)-side chain. We calculated a value of 5.92 from fluorescence emission spectra of the released A. nidulans PRE chromophore at different pH (Fig. 4), indicating the absence of such a charged group. Fluo-. rescence emission (X., = 418 nm) and excitation (X,, = 463 nm) spectra of denatured A. nidulans PRE (curves A, -) are identical with the fluorescence spectra of FO (curves A, ---). Also shown are the fluorescence spectra of denatured PRE after addition of NADPH and 8HDF:NADPH oxidoreductase (curues B), which converts the released 8HDF chromophore into the nonfluorescent reduced form. The high fluorescence below 380 nm in curve B is due to the added NADPH. A. nidulans PRE was denatured by heating at 65°C and fluorescence excitation spectra (X,, = 463 nm) were measured at pH 8.65 (highest intensity), 6.49, 6.14, 5.96, 5.75, 5.49, and 3.68. From these spectra a PK., of 5.92 was calculated for the PRE chromophore.
During fluorescence measurements a small but significant difference was found between the emission spectra of denatured PRE and 7,8-didemethyl-&OH-5-deazariboflavin on excitation at 450 nm (Fig. 5, curves A and B). The difference spectrum (curve C) has a maximum at 530 nm and resembles the fluorescence emission spectrum of normal flavins, pointing to the presence of a second flavin chromophore in PRE. In order to differentiate between FAD and riboflavin or FMN, denatured PRE was incubated with snake venom phosphodiesterase, which hydrolyzes the phosphodiester bond in FAD. This resulted in a large increase of fluorescence intensity (curve D), while there is no effect for riboflavin, FMN, or 7,8didemethyl-8OH-5-deazariboflavin (not shown). The difference spectrum (curve E) coincided with the fluorescence spectrum of FMN.
The structure of the second chromophore was further corroborated by reconstitution experiments with apo-D-aminoacid oxidase which specifically uses FAD as coenzyme. We found that the enzymatical activity of apo-D-aminoacid oxidase was restored on incubation with denatured A. nidulans PRE (results not shown).
Finally, the identity of both chromophores was confirmed  (Fig. 6) due to the release of intrinsic cofactors. The shape also changed from a single band to a more complicated pattern due to the presence of the two (released) chromophores.
From the absorption spectrum of denatured PRE, it can be calculated that this preparation contained 8.7 PM FAD and 8.4 PM 7,8-didemethyl-8-OH-5-deazariboflavin, using molar extinction coefficients of 11,400 (445 nm) and 37,500 (418 nm, pH 7.3), respectively. We conclude therefore that in A. nidulans PRE the chromophores are present in equimolar amounts. This is also indicated in Fig. 6 where in the 350-500 nm region the absorption spectrum of denatured PRE closely resembles, both at pH 7.3 and 2.8, the spectrum of an equimolar mixture of 7,8-didemethyl-8-OH-5-deazariboflavin and FAD. Below 350 nm there is an increasing divergence due to absorption and scatter of denatured protein.
The protein concentration was estimated 12 pM assuming M, = 53,000, indicating that approximately 28% of the PRE molecules lack the chromophores. From Fig. 6 it is also estimated that the molar extinction coefficient of native PRE is 53,000 at 437 nm.
Effect of Preillumination-The long wavelength absorption spectrum of purified PRE (Fig. 2) has a close resemblance with the absorption spectra of neutral (blue) flavin semiquinone radicals in polar solvents (33), suggesting that part of the FAD chromophore is present as semiquinone (FADH'). Assuming a molar extinction coefficient of 4500-5600 for the 580-nm absorption band of flavin semiquinones in flavoproteins (33), it can be calculated that 75-95% of the FAD chromophore is in the semiquinone form in dark equilibrated PRE.
When native PRE was preilluminated at 585 nm in the presence of 2-mercaptoethanol, the semiquinone band bleached whereas there were no large changes in the rest of the absorption spectrum (Fig. 7). On standing in the dark in air the original spectrum was slowly restored (see inset of Fig.  7). Even after several cycles of preillumination and dark recovery the final absorption spectrum was not altered (not shown). Photoreactivating activity did not change signifi-cantly by preillumination or dark incubation when measured with the standard assay at 425 nm. The same bleaching, but approximately 9 times faster, was observed when preillumination was done at 437 nm (not shown). This is in agreement with the absorption spectrum of native PRE as from Fig. 2 A437/A585 = 11.5 is obtained. Since the molar extinction coefficient of flavin semiquinones at 437 nm is lower than at 585 nm (34), the fast bleaching cannot be explained by light absorption of this compound. It is far more probable that the 8-OH-5-deazaflavin chromophore acts here as the photon acceptor for FAD semiquinone bleaching, indicating that these chromophores are photochemically coupled. DISCUSSION We have purified photoreactivating enzyme from the cyanobacterium A. nidulans till apparent homogeneity, enabling the elucidation of the structure of the two intrinsic PRE chromophores as well as the N-terminal amino acid sequence. Meanwhile, this sequence appeared to be sufficiently informative to permit the cloning of the Anacystis phr gene using synthetic oligonucleotides.
This has led to the determination of the complete base sequence of this gene, the first of a 8-HDF type PRE (35).
Although our results are on the whole in agreement with earlier publications, discrepancies were found concerning molecular weight and spectral properties of the enzyme. Saito and Werbin (5) estimated a molecular weight of 93,000 by gel filtration for Anacystk PRE, approximately twice the value we obtained. This points to a dimer molecule, in accordance with the reported tendency for aggregation (5), possibly enhanced by ammonium sulfate precipitation. We performed porous silica chromatography instead: the use of poly (ethylene glycol) for elution apparently suppresses any tendency of aggregation at this stage of purification without afflicting enzyme activity. The low level of PRE activity found between start and major protein band in nondenaturing gel electrophoresis ( Fig. 1) may indicate some residual aggregation.
The action spectrum reported by Saito and Werbin (5) is roughly similar to our results (Fig. 2), but they found a maximum at 418 nm in the absorption spectrum of native PRE compared with 437 nm in this study. Also the enzyme was reported to be fluorescent with excitation/emission maxima at 420 and 470 nm, respectively (7), while we found native PRE hardly fluorescent.
These spectral properties resemble those of released 8-OH-5-deazaflavin chromophore (see Fig.  3). In retrospect we think that the preparation used in early reports contained a substantial amount of released fluorescent 8-HDF chromophore.
We encountered a similar problem with S. griseus PRE (21) due to spontaneous release of chromophore, but we found Anacystis PRE to be far more stable.
Both spectral data and results of thin layer chromatography identified the PRE chromophores as 7,8-didemethyl-8-hydroxy-5-deazariboflavin (FO, II) and FAD, present in equimolar amounts. This classifies Anacystis PRE as a 8-HDFtype PRE. The action and absorption spectra of Anacystis (cyanobacterium) and Scenedesmus (green alga) PRE (20) are almost identical. Preliminary experiments indicate that the 8-HDF chromophore in Scenedesmus PRE has the same structure (FO) as in Anacystis PRE,' and a common origin from a pre-endosymbiontic bacterium has been suggested (20). Anacystis cells are relatively rich in PRE, compare 5.5 with 0.5 mg/kg wet cells for Scenedesmus, in accordance with the fact that cyanobacteria must relay on photoreactivation for the rapid removal of UV-induced DNA damage as efficient dark repair mechanisms seem to be absent (see Introduction). The visible absorption band of Anacystis PRE has a high molar extinction coefficient of 53,000 (437 nm) which compares well with the value of 54,300 (420 nm) reported for F420 (IV) at pH 13.5 (36). The shape of this band (Figs. 2 and 6) indicates deprotonization of the 8-hydroxy group of the 8-HDF chromophore in PRE, whereas the high molar extinction coefficient points to deprotonization of N(3), e.g. due to interaction with charged groups inside the protein. This might be crucial for the ability to act as a photosensitizer in dimer splitting.
8HDF-type PREs have the reduced FAD chromophore in common with pterin-type PREs from E. coli and S. cereuisiae and the question arises why different chromophores are present in various PREs. In Fig. 8 the absorption spectra of both types of PRE are compared. Besides the higher molar extinction coefficient of Anacystis PRE (compare 53,000 with 28,000), it is questionable whether E. coli and S. cerevisiae PREs could act with a quantum yield of 1 like S. griseus PRE (27) since they are fluorescent and will lose excitation energy. Moreover, a better overlap with the solar radiation spectrum at earth surface is obtained compared with S. cerevisiae PRE. Therefore, the presence of the 8-HDF chromophore permits a more efficient use of available sunlight compared with the reduced pt,erin or reduced FAD chromophores. This is the more important since in eutrophic waters, a normal environment for cyanobacteria, the relative contribution of the far blue and near UV region of sunlight is severely diminished. In that habitat the presence of an 8-HDF chromophore will be even more advantageous for these organisms.
Action and absorption spectra of Anacystis PRE are highly similar to each other and to the absorption spectrum of model compound 8-OH-lo-methyl-5-deazaisoalloxazine (Fig. 2). This leaves no doubt that 7,8-didemethyl-8-OH-5-deazariboflavin in the oxidized form acts as a primary photon acceptor in the photoreactivation process. Concerning the FAD chromophore, from the absence of photoreactivation at wavelengths above 500 nm (Fig. 2), it is clear that FAD semiquinone does not act as a photon acceptor in photoreactivation although it represents the major FAD species in dark equilibrated PRE. The same holds for oxidized FAD since an appreciable shoulder in the 450-500 nm region of the action spectrum is expected from superposition of absorption spectra of enzyme-bound 8-HDF and FAD (compare Fig. 6), which is not found. However, a significant difference between action spectrum and absorption spectrum of the model compound is  I  I  I  I  I  I  I  300  400  500  600  present in the 330-390 nm region (Fig. 2, shaded area) which might be attributed to FADH,. A similar discrepancy has been found for PREs from Streptomyces (27) and Scenedesmus (20). The difference spectrum has a maximum at 350 nm (as far as can be estimated from the 5-nm resolution of the action spectrum), in agreement with the absorption spectrum of FADH, bound to E. coli PRE (40), while a molar extinction coefficient of 3600 is calculated assuming a quantum yield of 1. This is comparable with the values of 3500-6200 reported for the near UV absorption band of reduced flavins in flavoproteins (37). Although other explanations cannot be completely ruled out, we assume therefore that FAD in the fully reduced form might act as a photon acceptor for photoreactivation but only in the near UV and not in the blue region. This conclusion is further corroborated by the results of experiments with Anacystis PRE lacking the 8-HDF chromophore, which is still functionally active (38) although to a lower level. A similar conclusion was reached for the pterintype PREs (39,40). From the long wavelength absorption spectrum of dark equilibrated PRE (Figs. 2 and 7), it appears that the major part of FAD is present as semiquinone. However, it is possible that PRE in living cells contains FADHz and the semiquinone is formed, despite the presence of 2-mercaptoethanol, during purification as was found for E. coli PRE (42). A rapid disappearance of the semiquinone was found on irradiation at 437 nm, which can be explained by photoreduction sensitized by the 8-HDF chromophore. This indicates a coupling of both chromophores (Fig. 9), i.e. a geometry which enables an efficient transfer of energy or electrons between the chromophores. It is conceivable that during photoreactivation the chromophores also act as a couple.
No appreciable enhancement of photoreactivating activity (as measured at 425 nm) was found after preillumination. At this wavelength only the 8-HDF and not the FADHZ chromophore will absorb light. The lack of enhancement might be explained by a mechanism in which photoreactivation, at least in the 437 nm band, mainly takes place directly through the 8-HDF chromophore, bypassing FADH* (Fig. 9). An alternative explanation is a rapid reduction of FADH' in the initial stage of photoreactivation so that in general these experiments are performed with PRE containing FADH2. The photoreduction of flavins in flavoproteins sensitized by 5-deazaflavins is well known (43).
In conclusion, the findings mentioned before led us to the following model for 8-HDF type PRE (Fig. 9). The 8-HDF (oxidized form) and possibly the FADH* chromophore can act as a primary photon acceptor in dimer splitting, although in different wavelength regions. Photoreduction of FADH ' is possible either by irradiation of the semiquinone or through sensitization by 8-HDF, indicating that these chromophores are geometrically close enough to permit energy or electron transfer from 8-HDF to FADH ' . Further experiments are necessary to refine this model, e.g. to distinguish between direct dimer splitting by 8-HDF or through a pathway with FADH2 as an intermediary.