Characterization and Mechanism of Action of DrosophilaRibosomal Protein S3 DNA Glycosylase Activity for the Removal of Oxidatively Damaged DNA Bases*

We recently demonstrated thatDrosophila ribosomal protein S3 specifically cleaved duplex oligodeoxynucleotides at sites of 7,8-dihydro-8-oxoguanine (8-oxoGua), presumably due to S3 protein possessing an N-glycosylase activity that is associated with its known apurinic/apyrimidinic (AP) lyase activity. Here we show, using DNA substrates prepared by γ-irradiation under N2O and analyzed by gas chromatography/isotope-dilution mass spectrometry, that S3 protein efficiently liberates 8-oxoGua as a free base from the damaged DNA substrate. Of the 15 additional modified bases present in the DNA substrate, the only other one acted on by S3 protein was 2,6-diamino-4-hydroxy-5-formamidopyrimidine (FapyGua). Specificity constants measured for the removal of 8-oxoGua and FapyGua indicate that S3 protein has a similar preference for both of these modified purines. Having established that S3 protein contains anN-glycosylase activity, we next examined the repair of duplex oligonucleotides containing 8-oxoGua (8-oxoGua-37-mer) positioned opposite Cyt, Gua, Thy, or Ade. Significant cleavage of the 8-oxoGua-37-mer was only detected for an opposing Cyt. Moreover, we show that an imino covalent enzyme-substrate intermediate is formed between S3 protein and 8-oxoGua-37-mer, a result similar to other DNA repair enzymes that catalyze N-glycosylase/AP lyase-type reactions at sites of DNA damage.

Cellular aerobic metabolism, ionizing radiation, and certain chemical carcinogens have in common their ability to form reactive oxygen species (1). Of these oxygen-derived species, the hydroxyl radical is highly reactive and capable of producing a large array of base modifications in DNA (2,3). These modifications, if left unrepaired, could have a role in mutagenesis, carcinogenesis, and aging (4).
Base excision repair represents a major pathway in countering the deleterious consequences of free radical attack on DNA. This pathway is initiated by N-glycosylase activity, which releases damaged DNA bases, forming an apurinic/apyrimidinic (AP) 1 site in their place. Most of the N-glycosylases that act on oxidatively damaged purines and pyrimidines also possess AP lyase activity (5). An example of this would be the Fpg protein of Escherichia coli, which contains N-glycosylase activity for the removal of oxidatively damaged purines and a second activity that mediates a concerted ␤,␦-elimination reaction for the removal of the AP site (6).
It has recently been reported that human and Drosophila ribosomal S3 protein possesses an AP lyase activity that cleaves abasic sites (7)(8)(9). The Drosophila S3 protein has also been found to act on a 5Ј end-labeled DNA duplex oligonucleotide that contained a single 7,8-dihydro-8-oxoguanine (8-ox-oGua) residue (8-oxoGua-37-mer). It therefore appeared from these results that the Drosophila S3 protein possessed an Nglycosylase activity, although this was not definitively shown by the actual release of 8-oxoGua from the 8-oxoGua-37-mer. As a means of testing for N-glycosylase activity, we have now used as a substrate calf thymus DNA that was exposed to ␥-irradiation under N 2 O. This form of damage results in numerous free radical-induced base lesions that can be identified and quantified by gas-chromatography/isotope-dilution mass spectrometry (GC/IDMS) (10,11). In this work, this technique aided us in demonstrating authentic N-glycosylase activity, as well as in identifying the substrate specificity of the Drosophila S3 protein. Our results show that 8-oxoGua and 2,6-diamino-4-hydroxy-5-formamidopyrimidine (FapyGua) were efficiently released from the irradiated DNA substrate. The inability of S3 protein to act on base mismatches opposite 8-oxoGua and the formation of an imino intermediate are also presented as a part of our characterizing the N-glycosylase activity associated with this ribosomal protein.

EXPERIMENTAL PROCEDURES
Preparation of DNA Substrates-The 37 base pair, 5Ј 32 P-end-labeled duplex DNA fragment containing a single 8-oxoGua (position 21) was described previously (9) and was synthesized at the Wells Center Oligonucleotide Synthesis Facility supported by the Riley Memorial Association, Indiana University School of Medicine. The complementary strand containing Cyt positioned opposite 8-oxoGua was also supplied by the Wells Center, whereas those containing Gua, Thy, or Ade to form a mismatch opposite 8-oxoGua were commercially synthesized by Life Technologies, Inc. Preparation of ␥-irradiated DNA substrates was done as described previously (11). DNA samples were subsequently dialyzed against 30 mM Tris-HCl buffer (pH 7.4).
Overexpression and Purification of GST-S3-Drosophila ribosomal gene S3 was overexpressed as a glutathione S-transferase (GST) fusion construct in a bacterial strain (mutM::mini-tet) that is deficient for the repair of 8-oxoGua and formamidopyrimidines such as FapyGua (12). The conditions for overexpression and purification of GST-S3 was identical to that previously reported (9).
Enzymatic Assay Using ␥-Irradiated DNA As a Substrate-Aliquots of 100 g of DNA samples were dried in a SpeedVac under vacuum. Samples were dissolved in the incubation mixture containing 35 mM Tris-HCl (pH 7.5) and 50 mM KCl. The total volume of the mixture was 110 l. Three replicates of each mixture were incubated with GST-S3 protein at 37°C in a water bath. Incubation time and enzyme amount varied depending on the experiment. As controls, DNA substrates were incubated with inactivated enzyme or without the enzyme. The inactivation of the enzyme was achieved by heating at 140°C for 15 min. For determination of excision as a function of the product concentration, 10, 22, 35, 50, and 75 g of irradiated DNA samples were supplemented with 90, 78, 65, 50, and 25 g of untreated DNA samples, respectively. Additional samples containing 100 g of irradiated DNA and unirradiated DNA as controls were also used. Three replicates of these samples were incubated with or without 6 g of GST-S3 protein at 37°C for 45 min. Following incubation, DNA samples were precipitated with 270 l of cold ethanol, kept at Ϫ20°C for 2 h, and centrifuged at 10,000 rpm for 30 min at 4°C. DNA pellets and supernatant fractions were separated. The recovery of DNA by precipitation with ethanol was near 100%, as judged by measurement of DNA amounts by UV spectroscopy.
Analysis by Gas Chromatography/Mass Spectrometry-Aliquots of stable isotope-labeled analogues of modified DNA bases were added as internal standards to pellets with known DNA amounts and to supernatant fractions (10). Analyses of pellets after hydrolysis and of supernatant fractions without hydrolysis were performed by GC/IDMS with selected ion monitoring (SIM) as described (11,13).

RESULTS
N-Glycosylase Activity of GST-S3-All DNA glycosylases catalyze the breakage of the N-C-1Ј glycosylic bond that exists between the DNA base and sugar in which the excised base can then be recovered in the supernatant fraction after ethanol precipitation of the DNA sample. To test for N-glycosylase activity, GST-S3 protein was incubated with calf thymus DNA that had been damaged by ␥-irradiation under N 2 O. Subsequent to ethanol precipitation, the supernatant fractions and hydrolyzed DNA pellets were analyzed by GC/IDMS-SIM. Of the sixteen detectable modified bases formed by irradiation under anoxic conditions (11), GST-S3 protein efficiently removed 8-oxoGua and FapyGua (Fig. 1). The amounts removed from the DNA pellet fractions by the active enzyme were accounted for by the appearance of 8-oxoGua and FapyGua in the supernatant fractions of the same samples. The results obtained with the heat-inactivated enzyme were similar to those obtained without the enzyme (not shown). Excision of 8-oxoGua and FapyGua were dependent upon the amount of protein added and the time course of the reaction (not shown).
Kinetics of 8-oxoGua and FapyGua Excision-The N-glycosylase activity of S3 protein followed Michaelis-Menten kinetics with regard to the concentration of the modified base. Kinetic constants were obtained by analysis of the data using Lineweaver-Burk plots (15) (Fig. 2) and are given in Table I. The concentration ranges of 8-oxoGua and FapyGua used for these measurements were 0.75-4.35 M and 0.27-2.9 M, respectively. The apparent K m and the turnover number (k cat ) of 8-oxoGua excision were approximately 3-fold greater than those of FapyGua. On the other hand, the specificity contants (k cat /K m ) for the removal of the two modified purines were similar, suggesting that FapyGua may also be an important physiological substrate for S3 protein.
Substrate Specificity of GST-S3 Protein-Having established that S3 protein contains an authentic N-glycosylase activity, we next examined what influence the opposing base had on the excision process. In this case, we used the 8-oxoGua-37-mer as a substrate for GST-S3 protein with the complementary strand containing Cyt, Gua, Thy, or Ade positioned opposite 8-oxoGua. A phosphoimage of products of the reaction is presented in Fig.  3, in which 100 pg of GST-S3 protein (lane 3) completely converts the 8-oxoGua/Cyt substrate to product. On the other hand, the same amount of GST-S3 protein fails to convert any of the 8-oxoGua mismatches to product (lanes 4 -6). Roughly the same amount of substrate in the control (lane 1) remains for each of the bases other than Cyt positioned opposite 8-ox-oGua, which was determined to be 95-98%.
S3 Protein Undergoes a Schiff Base Intermediate-Many DNA repair enzymes that have a combined N-glycosylase/AP lyase activity have been shown to form a Schiff base (imino) intermediate that can be trapped by NaBH 4 to generate a covalent enzyme-DNA complex (16). To examine if GST-S3 protein also undergoes a Schiff base intermediate, fresh preparations of GST-S3 protein were combined with NaBH 4 and incubated with 8-oxoGua-37-mer for various periods of time. Reactions were then terminated and applied to an SDS-polyacrylamide gel, in which GST-S3 protein alone migrates as a 56-kDa protein. If indeed GST-S3 protein undergoes a Schiff base intermediate, then the trapping of GST-S3 protein to 8-oxoGua-37-mer with NaBH 4 should increase the molecular mass to roughly 66 kDa due to the added contribution of the labeled 37-mer. This is essentially what was observed, in which samples taken at 30, 60, and 90 min show almost a complete shift from the lower molecular mass to the higher molecular mass (Fig. 4A, lanes 2-4). This shift was not observed with boiled GST-S3 protein (Fig. 4A, lane 1) or in reactions lacking NaBH 4 (lane 5). Fig. 4B shows the autoradiogram of the SDS-PAGE depicted in Fig. 4A. As can be seen, the presence of NaBH 4 shifts the radioactivity from a species migrating with the dye front (presumably uncleaved DNA products) to one that is at a molecular mass that would be anticipated if GST-S3 protein and 8-oxoGua-37-mer were covalently combined. DISCUSSION It was assumed from our previous studies using the 8-oxoGua-37-mer as a substrate that the cleavage reaction we detected was due to a combined N-glycosylase/AP lyase activity possessed by S3 protein (9). However, since Nishimura's group (17) observed in human cells a protein that had endonuclease activity directed toward 8-oxoGua, but lacked glycosylase activity, this prompted us to confirm the actual release of 8-ox-oGua from an oxidized DNA substrate. As a result, we examined the repair of calf thymus DNA that had been exposed to ␥-irradiation under N 2 O, which generates numerous different lesions that can be detected and quantified by the highly sensitive technique of GC/IDMS-SIM (10,11). Under these conditions, we found that 8-oxoGua was indeed liberated as a free base by S3 protein from the damaged DNA substrate. We also detected the removal of FapyGua. No other base modification was excised by S3 protein, including 4,6-diamino-5-formami-    dopyrimidine (FapyAde), which is an additional substrate for the Fpg protein of E. coli (11,18). It should be noted that the specificity observed for GST-S3 protein could be a particular property of the fusion protein, whereas the native protein may have a more extensive specificity.
The kinetics for the removal of the two modified purines were also determined in which turnover numbers, or k cat , of 0.32 Ϯ 0.07/min and 0.09 Ϯ 0.014/min were calculated for the repair of 8-oxoGua and FapyGua, respectively. These values are on the same order as those found for E. coli Fpg protein for excision of 8-oxoGua and FapyGua from a similar DNA substrate although k cat of FapyGua excision was greater than that of 8-ox-oGua excision (11). However, the k cat calculated here for S3 protein is considerably less than that found in another study for the removal of 8-oxoGua from a synthetic oligonucleotide (8-oxoGua-37-mer), which was determined to be 6/min (9). This difference could represent the dissimilarity of 8-oxoGua-37-mer and ␥-irradiated DNA used in these two studies. A profound effect of the nature of DNA substrate on excision rates by E. coli Fpg protein has recently been demonstrated (11). On the other hand, less than ideal reaction conditions were used in the present study to eliminate some as yet unidentified factor in our optimal reaction buffer mixture (9) that interfered with the detection, by GC/IDMS, of free bases in the supernatant fractions. This may have contributed to observed differences in excision rates between 8-oxoGua-37-mer and ␥-irradiated DNA.
A salient feature of this study is the finding that the specificity constants for the removal of 8-oxoGua and FapyGua are very similar, suggesting that FapyGua is an important physiological substrate for S3 protein. Recent studies using a similarly damaged DNA substrate as used here show that E. coli Fpg protein-catalyzed removal of 8-oxoGua and FapyGua shows no significant preference for one substrate over the other (11). Yet, it has previously been argued that the primary physiological substrate for Fpg protein is 8-oxoGua although excision of FapyGua or FapyAde was not determined (19). This conclusion is based upon the mutagenic properties of 8-oxoGua (20 -23) combined with the supposed fact that it is more widespread than FapyGua. In various cases, however, more abundant formation of FapyGua than 8-oxoGua in cells has been demonstrated in recent years (24). The abundance of 8-oxoGua and FapyGua is likely to depend upon the cellular environment. Both FapyGua and 8-oxoGua are formed as a result of hydroxyl radical attack on the C8-position of guanine, followed by reduction and oxidation of the thus formed 8-OH adduct radical of guanine, respectively (25,26). FapyGua is formed under reductive conditions, whereas the formation of 8-oxoGua is favored under oxidative conditions (26). It remains unclear whether FapyGua represents a cytotoxic lesion or a mutagenic lesion because no studies exist on the mutagenic properties of FapyGua or FapyAde (27). Recent studies using bacterial strains deficient for Fpg (mutM) gene indicate that these mutants are sensitive to H 2 O 2 (9), thus suggesting that a cytototoxic lesion is formed by this type of treatment. The mutagenicity of 8-oxoGua and the ability of S3 protein to rescue the H 2 O 2 sensitivity of mutM gene deficiency would indicate that FapyGua may be cytotoxic and a physiological substrate for repair by S3 protein.
A hallmark of N-glycosylase/AP lyase catalyzed reactions is that they form a Schiff base intermediate (16). As shown here, Drosophila S3 protein also possesses this property that has been previously demonstrated for enzymes such as E. coli Fpg and yeast Ogg1 proteins that act on oxidatively damaged DNA. It should be noted that other similarities exist among these enzymes, such as the uniform inability of each of these three enzymes to act on an 8-oxoGua/Ade mismatch. The E. coli Fpg protein, however, can work on the other 8-oxoGua mismatches, as well as when Cyt is positioned opposite 8-oxoGua. Some disagreement exists whether the yeast Ogg1 also possesses these mismatch activities (28,29). For the recently cloned human mutM homologue (30), Thy and Gua mismatches were only slightly cleaved although the amount of protein used to obtain this result was not reported. Nevertheless, it is clear that the Drosophila enzyme only acts on 8-oxoGua when Cyt is positioned opposite it.
The results presented here strengthen the notion that Drosophila protein S3 has all of the properties necessary to carry out an important role in DNA repair. It contains N-glycosylase activity for the removal of 8-oxoGua and FapyGua, AP lyase activity, and 3Ј-and 5Ј-deoxyribophosphodiesterase activity for the removal of preincised AP sites (31). The combination of these activities on 8-oxoGua or FapyGua would result in a one-nucleotide gap that appears common for ␤-polymerase-dependent base excision repair in eukaryotes (32)(33)(34)(35).