Proc. Natl. Acad. Sci. USA Vol. 87, pp. 5842-5845, August 1990 Biochemistry Mismatch-specific thymine DNA glycosylase and DNA polymerase P8 mediate the correction of GOT mispairs in nuclear extracts from human cells (DNA repair enzymes/cell-free extracts/cytosine deamlnation/5-methylcytosine) KARIN WIEBAUER* AND JOSEF JIRICNY* Friedrich Miescher Institute, P.O. Box 2543, CH-4002 Basel, Switzerland Communicated by Walter J. Gehring, May 21, 1990 ABSTRACT To avoid the mutagenic effect of spontaneous enzyme responsible for its removal from G-T mispairs would hydrolytic deamination of 5-methylcytosine, G-T inspirs, have to be, unlike the uracil and hypoxanthine glycosylases, arising in DNA as a result of this process, should always be fully inactive on single-stranded and matched double- corrected to G-C pairs. We describe here the identification of stranded substrates. It therefore seemed likely that a thymine a DNA glycosylase activity present in nuclear extracts from DNA glycosylase would have to act in conjunction with HeLa cells, which removes the mispaired thymine to generate another enzyme. We postulated two possible scenarios. In an apyrimidinic (AP) site opposite the guanine. We further the first, the glycosylase would be guided to the mispair by show, using a specific antibody and inhibitors, that the single a G-T mismatch binding protein, analogous to the MutS nucleotide gap, created upon processing of the AP site, is filled protein of E. coli (11). In the absence of this protein on the in by DNA polymerase (3. This rmding substantiates the DNA, the glycosylase would be inactive. In the second, a GOT proposed role of this enzyme in short-patch DNA repair. mismatch modifying enzyme would convert the mispaired T to a thymine derivative such as thymine glycol or hydroxy- Efficient correction of base-base mismatches, arising as er- methyluracil, which would then be removed by their respec- rors of DNA polymerase or through recombination, requires tive DNA glycosylase (12, 13). To provide experimental that the cellular machinery be able to direct the repair to the evidence supporting either of these two pathways, we de- strand carrying the original information. For this purpose, cided to analyze the in vitro G-T repair reaction in greater secondary signals such as adenine methylation or strand detail. nicks are used (for a comprehensive review, see ref. 1). The Our previous results have also shown that the processing GOT mispair is the only one that can arise also by an ofthe apyrimidinic (AP) site, arising in the DNA following the alternative pathway in "resting" (i.e., nonreplicating, non- removal ofthe mispaired thymine, leads to the appearance of recombining) DNA-through the spontaneous hydrolytic a single nucleotide gap opposite the guanine (10). We were deamination of 5-methylcytosine. In the correction of this interested in finding out which DNA polymerase is involved latter type of G-T mismatch, no strand discrimination is in the gap-filling reaction, as this information could provide required, as it should always be corrected to a G-C. Indeed, us with further insights into the mechanism of repair of this repair pathways thought to be dedicated to the correction of unique type of deamination damage, the G-T mismatch. the deamination-associated GOT mispairs have been identified in Escherichia coli (2-4) as well as in mammalian cells (5, 6). In all organisms studied to date, correction ofdeamination MATERIALS AND METHODS damage is mediated by base-specific glycosylases, which Synthesis of the Tritiated G-T Duplex. The 90-mer oligonu- remove the deaminated base from the sugar-phosphate back- cleotide A (15 pmol) and the 5'-phosphorylated 49-mer primer bone by cleaving the glycosylic bond. Uracdil DNA glycosyl- (10 pmol) (see Fig. 2) in 100 1.L of Sequenase buffer (40 mM ase and hypoxanthine DNA glycosylase remove the products Tris-HCI, pH 7.5/10 mM MgCl2/50 mM NaCl) were heated of cytosine and adenine deamination, uracil (U) and hypo- for 7 min at 800C and then allowed to stand at room temper- xanthine (H), respectively (for review, see refs. 7 and 8). ature for 10 min. To 50 jl ofthe annealed mixture was added Both enzymes are highly substrate specific and can remove 5 1ul of0.1 M dithiothreitol, 30 pmol of [methyl-3H]thymidine their respective substrates from matched (A-U and C-H) as 5'-triphosphate (43 Ci/mmol, 1 mCi/ml; 1 Ci = 37 GBq; well as mismatched (G-U and T.H) duplexes. As neither U Amersham) and the total volume was adjusted to 74 1Ld with nor H is a natural DNA base, their excision can be mediated water. One microliter of Sequenase (13 units/,ul; United by the base-specific enzymes from both double- and single- States Biochemical) was added and the mixture was left for stranded DNA. 5 min at room temperature; 7.5 Al of dNTPs (2.5 mM each) In our previous studies, we reported that G-T mispairs, was then added and the reaction was allowed to proceed for incorporated in the simian virus 40 genome and transfected a further 5 min at 370C. The mixture was extracted with an into monkey CV-1 (5, 6) or human (9) cells, were corrected equal volume of phenol/chloroform and the oligonucleotide with high efficiency and mostly to G-C pairs. Our in vitro was recovered by ethanol precipitation. The dried pellet was experiments (10) demonstrated that in nuclear extracts from suspended in 4.5 jul of formamide loading dye, heated for 3 human (HeLa) cells, the mispaired thymidine was excised min at 950C, and applied on an 8% denaturing polyacrylamide from DNA to generate a single nucleotide gap. Preliminary gel. A duplicate experiment was carried out with a 5' 32p- evidence also suggested that the initial step of this repair labeled 49-mer (see Fig. 2), and this product mixture was process was mediated by a DNA glycosylase. This was an loaded in an adjacent lane as a radioactive marker. The band unexpected finding. Thymine is a natural DNA base and any migrating with the same Rf as the 32P-labeled control was cut The publication costs of this article were defrayed in part by page charge Abbreviation: AP, apyrimidinic. payment. This article must therefore be hereby marked "advertisement" *Present address: IRBM, via Pontina km 30,600, 00040 Pomezia, in accordance with 18 U.S.C. §1734 solely to indicate this fact. Rome, Italy. 5842- Downloaded by guest on September 24, 2021 Biochemistry: Wiebauer and Jiricny Proc. NatL. Acad. Sci. USA 87 (1990) 5843 Sail Accl EcoRI Hinc Hind III 5. VW a.. ACGT TGTAAAACGACGGCCAGTGAAT TCCCGGGGATCCGTC R ACCTGCAGCCAAGCT TGGCGTAATCATGGTCATAGCTGT T TCCTGTGT TGCAACAT TT TGCTGCCGGTCACT TAAGGGCCCCTAGGCAG YTGGACGTCGGTTCGAACCGCAT TAGTACCAGTATCGACAAAGGACACA S. A" S. FIG. 1. Sequence of the synthetic 90-mer substrate. R, A or G; Y, C or T. Cleavage of the duplex G C, labeled at the 5' end of the C strand, with Sal I (v), Acc I (v), and HincII (v) produced the indicated marker bands. out and the tritiated 90-mer T was eluted into 300/A of0.5 M or whether it first modifies it to, for example, thymine glycol ammonium acetate/5 mM EDTA in the presence of 15 fig of or hydroxymethyl thymine, both ofwhich would be substrates tRNA by shaking at room temperature overnight. The oligo- for already characterized glycosylases (12, 13). For this pur- nucleotide was recovered by ethanol precipitation and sus- pose we synthesized a G-T 90-mer duplex (see Materials and pended in sterile water to a concentration of 0.05 pmol/4l. Methods), in which the mispaired thymine residue was labeled Tritiated 90-mer T (1 pmol) was annealed with 12 pmol of to a high specific activity with 3H (Fig. 2). This duplex was then either 90-mer A or 90-mer G in 30 Al of 10 mM Tris HCl/10 incubated with the HeLa nuclear extract. After a suitable mM MgCl2 by immersion in boiling water for 5 min and slow period oftime, an aliquot ofthe reaction mixture was analyzed cooling to room temperature to give duplexes APT and GOT, by TLC. As shown in Fig. 3a, the markers thymine (T), respectively, where the mispaired thymine is tritiated to high 2'-deoxythymidine (dT), and the oligonucleotides are well specific activity. EDTA was added to 10 mM to chelate the resolved in this eluent system. Furthermore, the most prob- free magnesium. able metabolites, thymine glycol and hydroxymethyl thymine, Glycosylase Activity Assay. The A-T or G-T duplexes (0.25 migrate with Rf values lower than that of thymine (data not pmol) were incubated with the HeLa nuclear extract (2 Al; 8 shown). Fig. 3b shows that incubation ofthe perfectly matched mg/ml) in binding buffer (25 mM Hepes-KOH, pH 7.9/0.5 duplex APT with the nuclear cell extracts failed to liberate any mM EDTA/0.01 mM ZnCl2/0.5 mM dithiothreitol) in a total free radioactivity from this oligonucleotide substrate. In con- volume of 20 /l for 15 hr at 30'C. Five microliters each of a trast, of the total saturated aqueous solution of thymine and thymidine was z17% radioactivity present in the starting added to the reaction mixture, and 5 Al was spotted on a silica G'T duplex was found to comigrate with the thymine marker gel TLC plate (4 x 8 cm; Polygram Sil G/UV254, Macherey (see also Fig. 3c). We thus conclude that the first step of the & Nagel) and eluted with chloroform/ethanol (8.5:1.5, vol/ repair reaction is mediated by an as yet uncharacterized DNA vol). The dried plate was photographed under UV (254 nm), thymine glycosylase.