Potential of Escherichia coli GTP Cyclohydrolase II for Hydrolyzing 8-Oxo-dGTP, a Mutagenic Substrate for DNA Synthesis*

MutT protein of Escherichia coli prevents the occurrence of A:T → C:G transversion by hydrolyzing an oxidized form of dGTP, 8-oxo-7,8-dihydro-2′-deoxyguanosine 5′-triphosphate (8-oxo-dGTP), which is produced by active oxygen species. In a search formutT-related genes, we found that the ribAgene, encoding GTP cyclohydrolase II, is able to reduce the increased level of mutation frequency of the mutT strain to almost the normal level, provided that the gene product is overproduced. Purified preparations of Escherichia coli GTP cyclohydrolase II protein as well as the histidine hexamer-tagged recombinant GTP cyclohydrolase II protein efficiently hydrolyze 8-oxo-dGTP and 8-oxo-GTP, producing 8-oxo-dGMP and 8-oxo-GMP, respectively. dGTP was not hydrolyzed by these preparations. GTP cyclohydrolase II catalyzes conversion of GTP to 2,5-diamino-6-hydroxy-4-(5-phosphoribosylamino)-pyrimidine, which constitutes the first step for riboflavin synthesis. TheK m values for the three types of guanine nucleotides, GTP, 8-oxo-GTP, and 8-oxo-dGTP, were almost the same. In the mutT − background,ribA − cells showed higher spontaneous mutation frequencies as compared with that of ribA +cells. Thus, GTP cyclohydrolase II, the ribA gene product, has a potential to protect genetic material from the untoward effects of endogenous oxygen radicals.

Organisms are always exposed to attacks of active oxygen species that are generated not only by exogenous environmental factors such as ionizing radiation and redox-cycling chemicals but also through endogenous oxygen metabolism (1,2). Active oxygen species are able to produce modifications in proteins, lipids, carbohydrates, and nucleotides (3). Among the various types of DNA damage caused by active oxygen species, 8-oxo-7,8-dihydroguanine (8-oxo-G) 1 may be responsible for a significant portion of spontaneous mutations, which would lead to induction of cancer as well as other age-related disorders (4). The 8-oxo-G residue present in DNA efficiently induces G:C 3 T:A transversion both in vitro and in vivo (5,6), and two genes of Escherichia coli, mutM and mutY, are involved in the repair process of this lesion (7)(8)(9)(10). The protein encoded by the mutM gene (MutM or Fpg protein) possesses a DNA glycosylase activity that specifically removes 8-oxo-G from DNA, whereas the MutY protein has an enzymatic activity that removes an adenine base from an A:8-oxo-G pair as efficiently as from an A:G pair in DNA.
Oxidation of guanine also proceeds in the form of free nucleotides, and an oxidized form of dGTP, 8-oxo-dGTP, is a potent mutagenic substrate for DNA synthesis (11). In contrast to the consequence of 8-oxo-G arising in DNA, 8-oxo-dGTP can induce A:T 3 C:G as well as G:C 3 T:A transversion (9). In E. coli, 8-oxo-dGTP can be eliminated from the nucleotide pool by MutT protein, the product of the mutT gene, which hydrolyzes the mutagenic nucleotide to 8-oxo-dGMP (11). 8-Oxo-dGMP is further hydrolyzed by other enzymes and finally eliminated from the cellular nucleotide pools (12). Lack of the mutT gene increases the occurrence of A:T 3 C:G transversion but not G:C 3 T:A (13). In the mutT mutant, 8-oxo-G misincorporated opposite the C residue of the template may be removed by the MutM protein before the next round of DNA replication. mutT homologues have been found in human, mouse, and rat (14 -16). Thus, 8-oxo-dGTP hydrolysis by MutT and related proteins may play a crucial role in reducing spontaneous mutation frequency in a wide range of organisms.
In view of the importance of preserving genetic material, it is not surprising that organisms possess multiple pathways for removing oxygen-induced lesions. For example, oxidative pyrimidine residues, such as thymine glycol, in DNA can be removed by both endonuclease III and endonuclease VIII glycosylases in E. coli (17,18). In mammalian cells, N-methylpurine-DNA glycosylase functions to repair 8-oxo-G in DNA, in addition to 8-oxoguanine DNA glycosylase, a mammalian MutM homologue glycosylase (19,20). Then, we have pursued a possibility that E. coli may possess a backup system for MutT, which plays a crucial role in high fidelity of DNA replication. By phenotypic rescue, we were able to clone a second E. coli gene that suppresses increased mutation frequency of mutT mutants. The gene was subsequently shown to be the ribA, which encodes GTP cyclohydrolase II (GCHII) that catalyzes the first step of riboflavin biosynthesis (21)(22)(23). We have demonstrated that GCHII can indeed hydrolyze 8-oxo-dGTP and its ribonucleotide counterpart, at a rate higher than that for GTP, the substrate for riboflavin biosynthesis.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Plasmids-The bacterial strains used in this study were all derivatives of E. coli K12. Strain SY11 is a mutT::Cm derivative of SY5, which is an F factor-negative derivative of JM107 (24). Strain T-198 has the mutT1 allele (25). T-198R is a ribA::Km derivative of T-198. The mutT::Cm and ribA::Km alleles were made by the gene replacement method using the JC7623 (recBC sbcB) strain (26,27). An F episome from strain CC101, derived from P90C (28), was transferred to SY11, yielding strain SY11F. The lacZ gene of the episome carries G:C 3 T:A transversion mutation at position 461 of the gene, and thus, only A:T to C:G transversion would revert the strain to the wild-type phenotype (28). XL1-Blue MRFЈ (Stratagene) was used as the host of the M13 phage. Plasmid pMT11, carrying a mutT::Cm allele, is a derivative of pSK6 (29). Plasmid pBST was constructed by transferring a DNA fragment carrying the mutT gene from pSK6 into pBluescript II. Plasmid pGRA, carrying the E. coli ribA gene, is a derivative of pBluescript II. Plasmid pBS29 possesses the BglI-EcoRI fragment carrying the ribA gene. A His-tag fusion vector, pQE9, was purchased from Qiagen. The 0.9-kbp DNA fragment encoding the full-length GCHII was amplified by polymerase chain reaction (PCR) using pBS29 as a template, using primers A (5Ј-CATGCAGCTTAAACGTGTGGCA-3Ј, corresponding to sequence Ϫ1 to ϩ21 of the ribA gene) and B (5Ј-TGTAAAACGACGGCCAGT-3Ј, which is an M13 universal primer). The PCR fragment was treated with Klenow polymerase, then SalI, and ligated to the Klenow polymerase-treated BamHI site and SalI site of vector pQE9, yielding pHT29. Vector plasmids, pJKKmf(Ϫ), pTZ18R, pTZ19R, and pBluescript II, were from our laboratory stocks.
Reagents and Media-The Luria-Bertani (LB) broth, LB plates, M9 medium, and phosphate buffer used were described (30). The minimal glucose (or lactose) agar medium is composed of M56 salt plus 0.35% glucose (or lactose), 1.5% agar and supplemented with 1 g of thiamine per ml (27). Ampicillin (Ap), chloramphenicol (Cm) and kanamycin (Km) were included, if necessary, in the medium at concentrations of 50, 30, and 50 g/ml, respectively. Riboflavin, purchased from Wako Pure Chemical Industries, Ltd. (Osaka, Japan), was included in the medium at a concentration of 50 g/ml. Enzymes and reagents used for DNA manipulation and DNA sequencing were purchased from Takara Shuzo Co., Ltd. (Kyoto, Japan) and Applied Biosystems Inc. (Foster City, CA).
Library Construction and Screening-DNA was isolated from E. coli strain SY11 and digested with EcoRI. DNA fragments were inserted into the EcoRI site of plasmid pJKKmf(Ϫ), and the resulting recombinant DNA mixture was used to transform SY11F cells. The transformation mixture was plated on MacConkey agar containing 50 g of Km per ml at a density of 200 -500 colonies per plate. After incubation for 2 days at 37°C, colonies with a few papillae were selected and individually grown overnight at 37°C in LB broth containing Km. Plasmids were prepared from each of the cultures and examined for generation of LacZ ϩ reversion.
DNA Manipulation-Restriction enzyme fragments were subcloned into pTZ18R, pTZ19R, and pBluescript II vectors, and their sequences were determined. Both strands were sequenced by the dideoxy chain termination method, using an ABI automated DNA sequencer model 373A. A homology search was achieved with the aid of the BLAST program. 2 Purification of GCHII-LB broth (2 liters) containing 50 g of Ap per ml was inoculated with 50 ml of an overnight culture of E. coli SY11 (mutT::Cm) carrying plasmid pBS29 with the ribA gene. The cells were grown at 37°C for 16 h and harvested. The cells (7.5 g) were suspended in 20 ml of 200 mM Tris-HCl, pH 8.0, supplemented with 2 mM phenylmethylsulfonyl fluoride. The suspension was sonicated and then centrifuged at 35,000 ϫ g for 30 min. The supernatant was applied to a column (2.5 ϫ 21 cm) of DEAE-Sephacel (Amersham Pharmacia Biotech), equilibrated with buffer A (25 mM Tris-HCl, pH 8.0, 0.2 mM EDTA, 0.5 mM MgCl 2 ) containing 150 mM of NaCl. After the column was washed with 400 ml of the same buffer, elution was performed with a linear gradient of 150 -500 mM NaCl in buffer A. Fractions containing a significant amount of GCHII were pooled and concentrated by ultrafiltration with a YM10 membrane (Amicon). After dialysis, the sample was loaded onto an HPLC column of DEAE-5PW (Tosoh), and the column was extensively washed with buffer A containing 125 mM of NaCl. The column was developed with a linear gradient of 125-150 mM NaCl in buffer A at a flow rate of 0.5 ml/min, and the enzyme was eluted at around 144 mM NaCl. The fractions with high levels of enzyme activity were pooled, dialyzed against buffer A, and stored at Ϫ20°C.
For the preparation of His-tag fusion GCHII, E. coli XL1-BlueMRFЈ carrying pHT29 was grown at 37°C for 16 h. The cells (5.5 g) were suspended in 15 ml of a buffer (50 mM sodium phosphate, pH 8.0, 300 mM NaCl) and disrupted by sonication. The supernatant was loaded onto a Ni-NTA-agarose (Qiagen) column. After the column was washed with a buffer (50 mM sodium phosphate, pH 6.0, 300 mM NaCl, 10% glycerol), the protein was eluted with a gradient of 0 to 0.5 M imidazole in 150 ml of the buffer. Samples were used for analysis by SDSpolyacrylamide gel electrophoresis and for enzyme assay.
Assay of Enzyme Activities-To measure GTP cyclohydrolase II activity, the reaction was carried out with 5 mM GTP in a reaction mixture containing 200 mM Tris-phosphate, pH 8.7, 10 mM MgCl 2 , and 20 mM dithiothreitol. After incubation for 30 min at 37°C, an equal volume of 11.4 mM diacetyl was added, and the mixture was heated at 95°C for 30 min to convert 2,5-diamino-6-hydroxy-4-(5-phosphoribosylamino)-pyrimidine, the product of the enzyme reaction, to 6,7-dimethylpterin (23). After centrifugation, 15 l of each sample was subjected to HPLC with an Oligo R3 (Perspective Biosystems) column equilibrated with 0.1 M ammonium formate and 40% methanol, and the product was detected by fluorometric monitoring (excitation, 365 nm; emission, 435 nm). One unit of GCHII activity was defined as the amount of enzyme that produced 1 nmol of 2,5-diamino-6-hydroxy-4-(5-phosphoribosylamino)pyrimidine/h at 37°C. [␣-32 P]GTP and [␣-32 P]dGTP were purchased from Amersham Pharmacia Biotech. [␣-32 P]8-Oxo-GTP and [␣-32 P]8oxo-dGTP were prepared as described previously (12). Reactions were carried out with an appropriate amount of purified GCHII or His-tag GCHII in 10 l of a reaction mixture containing 20 mM Tris-HCl, pH 7.5, 0.8 g of bovine serum albumin, 8 mM MgCl 2 , 5 mM dithiothreitol, 4% glycerol, and 20 M either dGTP, GTP, 8-oxo-dGTP, or 8-oxo-GTP, each labeled at the ␣-32 P position (20,000 cpm), as reported previously (31). The mixture was incubated at 30°C for 30 min, and the reaction was stopped by addition of an equal volume of 50 mM EDTA. A portion of the reaction mixture was spotted on a PEI-cellulose F thin layer chromatography plate (Merck) and developed with 1 M LiCl for 30 min. The product was quantitated by autoradiographic analysis with a molecular imager GS-250 (Bio-Rad). For product analysis, ␣-32 P-labeled 8-oxo-dGTP or 8-oxo-GTP was incubated with GCHII at 30°C for 30 min. A portion of the reaction mixture was applied to a Mono Q column and eluted with a linear gradient of 0.01 to 1 M triethylammonium hydrogen carbonate at a flow rate of 1 ml/min. The radioactivity of the fractions was measured.
Mutation Assays-Cells derived from an independent clone were inoculated into M9 medium containing, if necessary, 50 g of riboflavin, 30 g of Cm, 50 g of Km, and/or 50 g of Ap per ml, and the culture was incubated overnight with shaking at 37°C. After the cells were washed by centrifugation, serial dilutions were made. Appropriate dilutions were plated on minimal lactose agar for a Lac ϩ reversion assay and minimal glucose agar for counting the number of viable cells. The viable cells were counted after 2 days of incubation at 37°C, and Lac ϩ revertants were counted after 3 days. Aliquots of the appropriate dilutions were also plated on streptomycin (50 g/ml)or rifampicin (100 g/ml)-containing LB plates, and numbers of resistant colonies were counted after overnight incubation at 37°C.

Isolation of an E. coli Gene That Suppresses the MutT Mutator-
The mutator phenotype of strain SY11F, defective in the mutT gene, is evident when colonies are produced on MacConkey agar. Numerous red papillae appeared, reflecting the frequent occurrence of A:T to C:G transversion in the lacZ gene, and this was the basis for screening of a second gene that suppresses the mutT mutator. DNA prepared from SY11 cells was digested with EcoRI, and the resulting DNA fragments were inserted into a multi-copy plasmid vector, pJKKmf(Ϫ). The genomic library was introduced into SY11F cells, and bacterial colonies that exhibit a reduced number of papillae were isolated. Among several thousand transformants examined, one clone which showed distinctly lower mutation frequencies when measured with other characters was isolated. The plasmid carried a 4.5-kbp insert, and further analysis revealed that a 1.4-kbp BglI-EcoRI fragment possessed an ability to suppress the mutator phenotype. The DNA fragment was inserted into pBluescript II, yielding plasmid pBS29.
To ascertain the extent of suppression of the mutator effect more precisely, mutation frequencies of three distinct characters were determined. As shown in Table I, the mutation frequencies of SY11F cells with pBS29 were considerably lower than those of SY11F cells with the vector alone. The extent of suppression achieved with pBS29 was, however, low compared with the effect of the authentic mutT ϩ gene, located on pBST.
Identification of the Gene-The nucleotide sequence of the 1.4-kbp insert of pBS29 was determined using the dideoxy method. The sequence contained a 588-bp open reading frame, which would code for a protein with 196 amino acid residues. Computer-aided analysis revealed that it is identical to the ribA gene, encoding GCHII of E. coli (23). To see whether the gene of pBS29 can complement the ribA-defective character, strain JC7623ribA was transformed with plasmid pBS29. The transformant was able to grow in a medium without riboflavin, indicating that the plasmid carries the ribA gene.
GCHII is one of the enzymes involved in biosynthesis of riboflavin. It catalyzes conversion of GTP to 2,5-diamino-6hydroxy-4-(5-phosphoribosylamino)-pyrimidine, releasing formate and pyrophosphate (21). This GTP pyrophosphorolysis partly mimics the reaction catalyzed by MutT protein, although there is no ring opening in the latter reaction (see Fig. 4). We have then asked whether GCHII acts on oxidized forms of guanine nucleotide.
Purification of the Gene Product-GCHII protein was overproduced in SY11F cells carrying pBS29 and purified to appar-ent physical homogeneity. His-tag GCHII, which carries additional amino acids residues (MRGSH 6 GC) at the N-terminal, was also purified from an extract of XL1-Blue MRFЈ cells harboring pHT29, by using a different purification procedure. On SDS-polyacrylamide gel electrophoresis, the 21.8-kDa GCHII protein moved slightly faster than did the 25.8-kDa His-tag GCHII protein (Fig. 1A). As shown in Fig. 1B, GTP cyclohydrolase activity was evidently present in the preparation of GCHII. A similar result was obtained with His-tag protein (data not shown).
Hydrolysis of Oxidized Forms of dGTP and GTP by GCHII-We examined abilities of purified preparations of GCHII and His-tag GCHII proteins to hydrolyze 8-oxo-dGTP and 8-oxo-GTP. Fig. 2 shows that both preparations can hydrolyze 8-oxo-dGTP and 8-oxo-GTP, yielding unique fast-moving products. The R F values of the fast-moving products coincided with those of 8-oxo-dGMP and 8-oxo-GMP, respectively. GTP was also hydrolyzed by the enzyme preparations under the same conditions; however, the autoradiographic profile of the fast-moving radioactive materials was smeary, probably due to the instability of 2,5-diamino-6-hydroxy-4-(5-phosphoribosylamino)-pyrimidine (21), a product of the reaction of GTP with GCHII. dGTP was not totally hydrolyzed by GCHII or His-tag GCHII. Fig. 3 shows the results of reactions of GCHII with various nucleotides at 20 M. 8-Oxo-dGTP and 8-oxo-GTP were hydrolyzed efficiently at almost the same rates. GTP was cleaved at a lower rate, but dGTP was not hydrolyzed. Based on these results, we calculated the kinetic values for 8-oxo-dGTP, 8-oxo-GTP, and GTP (Table II). The apparent K m values for hydrolysis of 8-oxo-dGTP and 8-oxo-GTP were 30 and 21 M, respectively. K m of GTP is 29 M, consistent with the value calculated previously by Foot and Brown (21).
We next analyzed the reaction products by HPLC. The retention volume on HPLC of the product of hydrolysis of 8-oxo-dGTP by GCHII was 11 ml, which corresponds to that of an authentic 8-oxo-dGMP sample (data not shown). The retention volume of the product of 8-oxo-GTP hydrolysis by GCHII was also the same as that for 8-oxo-GMP (data not shown). It is evident, therefore, that GCHII possesses a pyrophosphatase activity for 8-oxo-dGTP and 8-oxo-GTP, in addition to the previously demonstrated cyclohydrolase/pyrophosphatase activity for GTP.
Characteristics of ribA Mutants-Spontaneous mutation fre- Purified GCHII (1 g) was incubated with 5 mM GTP for 30 min at 37°C. The reaction was terminated by the addition of diacetyl, and the mixture was heated at 95°C for 30 min to convert the product, 2,5-diamino-6-hydroxy-4-(5-phosphoribosylamino)-pyrimidine, to 6,7-dimethylpterin. Samples (15 l) were applied to an Oligo R3 column equilibrated with 0.1 M ammonium formate, 40% methanol, and examined by fluorescence monitoring. 1, the reaction mixture in the absence of GCHII; 2, the reaction mixture in the presence of GCHII; 3, synthesized 6,7-dimethylpterin (0.25 nmol) as a reference. a Among 10 cultures, 9 cultures had no Lac ϩ revertants. Mutation frequency was then calculated by assuming that all the Lac ϩ mutations occur randomly following Poisson distribution. The fraction of the cultures containing no Lac ϩ mutant, P(0), is given by equation, P(0) ϭ c ϪmN , where m is mutation frequency, and N is the mean final cell count per culture (37). Now, P(0) is 9/10, and N is 1.41 ϫ 10 9 . Thus, mutation frequency, m, is calculated to be 0.0075 ϫ 10 Ϫ6 . quency of ribA Ϫ strain (JC7623ribA Ϫ ) is as low as that of wild-type strain, JC7623 (Table III). Thus, under the mutT ϩ background, RibA exerts no significant effect on mutation frequency. However, when bacteria lack MutT protein, an effect of ribA deficiency was observed. Spontaneous mutation frequencies of ribA Ϫ mutT Ϫ strain were 2.5 to 4 times higher than those of mutT Ϫ strain, with two different characters. This result may be taken as evidence that the GCHII protein can substitute the MutT function in vivo, although its efficiency is relatively low. DISCUSSION 8-OxoG is produced in DNA by reactive oxygen species generated in cell as by-products of aerobic metabolism, and this production is enforced by oxidative stress (32,33). This modified base can pair with cytosine and adenine with an almost equal efficiency and thus has the potential to induce G:C 3 T:A transversion (5,6). Oxidation of guanine also occurs in the cellular nucleotide pool, and 8-oxo-dGTP and 8-oxo-GTP thus formed can be incorporated in DNA and RNA, respectively. MutT protein of E. coli degrades 8-oxo-dGTP and 8-oxo-GTP to the corresponding nucleoside monophosphates (11,34), thereby preventing occurrence of transversion mutations as well as of transcription errors.
Considering the significant roles of MutT protein in DNA replication as well as transcription, we think that a cell might be equipped with a mechanism that can substitute the MutT function. With this in mind, we initiated screening of an E. coli genomic DNA library, prepared from mutT-deficient strain to clone a gene with an ability to suppress the mutT mutation. The gene isolated turned out to be the ribA gene, encoding GTP cyclohydrolase II, which functions at the first step of riboflavin biosynthesis. The RibA protein indeed has an ability to hydrolyze 8-oxo-dGTP and 8-oxo-GTP to the corresponding nucleoside monophosphates, providing the biochemical basis for this unexpected finding. It should be stressed that the RibA protein does not hydrolyze dGTP under the conditions tested (Figs. 2  and 3). The RibA protein, therefore, can specifically degrade the potent mutagenic substrate, 8-oxo-dGTP.
The K m for 8-oxo-dGTP of GCHII is 30 M, a value considerably higher as compared with that of the MutT protein, being 0.48 M (11). Since the concentration of 8-oxo-dGTP in the cellular nucleotide pool is extremely low, it seems that the role of RibA protein in control of mutation frequency in E. coli cells may be rather limited. Indeed, the ribA mutation afforded no increase in spontaneous mutation frequency of the mutT ϩ cells. However, in the mutT Ϫ background the ribA deficiency causes increased frequencies in spontaneous mutation. The antimutagenic effect of RibA was further evident when the gene product was overproduced in mutT Ϫ cells. Under ordinary conditions, MutT protein exerts almost exclusively its function to degrade   (36) suggested that the mutT gene may be one of the exogenous genes, since its codon usage is different from other E. coli genes. The ribA gene, on the other hand, exhibits the ordinary codon usage and may be authentic. It can be inferred that the ancestral RibA protein might function to eliminate oxidized forms of guanine nucleotides during the early evolutionary phase when oxygen tension was relatively low. Once E. coli acquires a strong anti-mutator mutT to adapt to the aerobic state, the biological role of the ribA gene might be restricted, e.g. biosynthesis of riboflavin.
The amino acid sequence for the GCHII shows no similarity with that for the MutT nor has specific motif such as MutT box. Thus, the mechanism for hydrolysis of 8-oxo-dGTP and 8-oxo-GTP by GCHII is different from that of MutT protein.
In the biosynthesis of riboflavin, GTP is first converted to an imidazole ring-opened dephosphorylated compound, 2,5-diamino-6-hydroxy-4-(5-phosphoribosylamino)-pyrimidine. This reaction, accompanying the release of formate and pyrophosphate, is catalyzed by GCHII itself (21). How is this reaction related to hydrolysis of 8-oxo-dGTP and 8-oxo-GTP? The GCHII protein may catalyze two types of reactions, one hydrolyzing the ␣-␤ phosphoryl bond and the other cleaving the C-8 -N-9 bond of the imidazole ring, the latter being associated with release of formate (Fig. 4). In the case of 8-oxo-dGTP and 8-oxo-GTP, in which the C-8 position is oxidized, the ring opening reaction would not occur, leaving only the dephosphorylation reaction. In hydrolysis of GTP, 2-amino-6-hydroxy-4-(5-triphosphoribosylamino)-5-formamidopyrimidine may be produced as an intermediate of the imidazole ring opening reaction (compound 3 in Fig. 4). A more complicated mode of the reaction with GTP, as compared with those of the oxidized counterparts, 8-oxo-GTP and 8-oxo-dGTP, may explain the re-quirement of larger number of enzyme molecules for the former (see Fig. 3).
It is highly probable that 8-oxoguanine-related mutagenesis is an important part of spontaneous mutagenesis in higher organisms and that similar systems address the threat of oxidation of guanine residues. An 8-oxo-dGTPase similar to MutT protein is present in mammalian cells, and the gene responsible was named MTH1 for mutT homologue (12, 14 -16). In MTH1-deficient mouse cells, there is yet a low level of enzyme activity to degrade 8-oxo-dGTP. It is of interest to see whether the protein responsible has some sequence homology with GCHII even though mammalian cells are devoid of an ability to synthesize riboflavin.