Mammalian Base Excision Repair by DNA Polymerases D and E

M Stucki1, B Pascucci2, E Parlanti2, P Fortini2, SH Wilson3,UHuÈ bscher1 and E Dogliotti2

1Institut fuÈr VeterinaÈrbiochemie, UniversitaÈtZuÈrich, Winterthurerstrasse 190, 8057 ZuÈrich, Switzerland; 2Laboratory of Comparative Toxicology and Ecotoxicology, Istituto Superiore di SanitaÁ, v. le Regina Elena 299, 00161, Roma, Italy; 3Laboratory of Structural Biology, NIEHS, P.O. Box 12233, 11 TW Alexander Drive, Research Triangle Park, North Carolina, 27709, USA

Two distinct pathways for completion of base excision structure-speci®c nuclease Flap endonuclease 1 (FEN1) repair (BER) have been discovered in eukaryotes: the (Klungland and Lindahl, 1997). DNA polymerase b (Pol b )-dependent short-patch path- Several studies have attempted to address the issue way that involves the replacement of a single nucleotide of the identity of the Pols involved in the DNA and the long-patch pathway that entails the resynthesis synthesis step that ®ll in gaps created during repair. Pol of 2-6 nucleotides and requires PCNA. We have used cell d and/or Pol e have been implicated in the synthesis extracts from Pol b-deleted mouse ®broblasts to separate step of the other two major excision repair processes, subfractions containing either Pol d or Pol e. These nucleotide excision repair (Shivji et al., 1995) and fractions were then tested for their ability to perform mismatch repair (Longley et al., 1997) which both both short- and long-patch BER in an in vitro repair require PCNA (Shivji et al., 1992, Umar et al., 1996). assay, using a circular DNA template, containing a As far as BER is concerned the characteristics of single abasic site at a de®ned position. Remarkably, both mouse ®broblasts cell lines with a homozygous deletion Pol d and Pol e were able to replace a single nucleotide of Pol b (Sobol et al., 1996) give the most clear at the lesion site, but the repair reaction is delayed evidence that Pol b functions speci®cally in this compared to single nucleotide replacement by Pol b. pathway in the replacement of a single damaged Furthermore, our observations indicated, that either Pol nucleotide. In contrast, however, the identity of the d and/or Pol e participate in the long-patch BER. PCNA Pols involved in the long-patch BER is still matter and RF-C, but not RP-A are required for this process. of debate. The PCNA-dependence of this pathway Our data show for the ®rst time that Pol d and/or Pol e strongly suggested the involvement of either Pol d and/ are directly involved in the long-patch BER of abasic or e (Frosina et al., 1996) in the resynthesis step. sites and might function as back-up system for Pol b in However, the long-patch BER synthesis at a reduced one-gap ®lling reactions. AP site was strongly inhibited by antibodies against Pol b in an in vitro repair assay with cell-free extracts Keywords: DNA repair; DNA polymerases; abasic sites when a linear double-stranded DNA oligonucleotide was used as substrate (Klungland and Lindahl, 1997). In order to clarify this matter we have used whole cell extracts from Pol b-deleted mouse ®broblasts to Introduction separate subfractions containing either Pol d or Pol e. These fractions were then tested for their ability to The integrity of DNA in vivo is continuously perform both short-and long-patch BER in an in vitro challenged by a variety of exogenous and endogenous repair assay. Here, we show that either Pol d and/or agents as well as by cellular processes. Among the Pol e participate in long-patch BER and might defence mechanisms that cells have evolved to counter- function as back-up systems for Pol b in one-gap act these threats, the base excision repair (BER) ®lling reactions. Moreover, the role played by the pathway is the main strategy to correct both replication auxiliary proteins PCNA, replication spontaneous DNA damage and small DNA adducts. protein A (RP-A) and replication factor C (RF-C) in This repair system involves the removal of a base and BER is addressed. its replacement by a chain of enzymatic steps. Ecient repair of a uracil-guanine base pair present in a duplex oligonucleotide can be achieved in vitro, via replace- Results ment of a single nucleotide (short-patch BER), by the sequential action of the human proteins uracil- DNA Characterization of the Pol d and Pol e-containing glycosylase, the major apurinic/apyrimidinic (AP) fractions endonuclease HAP1, DNA polymerase b (Pol b)and either DNA ligase III (Kubota et al., 1996) or DNA Whole cell extracts from Pol b-deleted mouse ®broblasts ligase I (Prasad et al., 1996). The second BER pathway (Sobol et al., 1996) were fractionated to separate Pol d which involves gap-®lling of several nucleotides (long- and Pol e. In a ®rst fractionation step over phospho- patch BER) appears to require the same molecular cellulose, the extract was depleted from PCNA and RP- partners with the addition of the proliferating cell A. The second fractionation step over hydroxylapatite nuclear antigen (PCNA) (Frosina et al., 1996) and the led to a separation of Pol a from Pol d/e. The three fractions obtained after the third step over Mono Q, PFIIA, B and C, were analysed in more detail (Figure Correspondence: E Dogliotti 1). Poly(dA)oligo(dT) was used in combination with BP and EP contributed equally to this work Received 19 January 1998; revised 25 March 1998; accepted 26 PCNA to distinguish between Pol d- and Pol e- March 1998 containing fractions, since Pol d is inactive in this DNA polymerase d and e in base excision repair M Stucki et al 836 assay without PCNA (Weiser et al., 1991). As shown in activity of PFIIB and PFIIC was unmodi®ed in the Figure 1a, in the presence of PCNA (open circles) all presence of the Pol a-speci®c neutralizing antibody SJK three fractions were able to substain DNA synthesis, 132-20 and the absence of Pol a in these two fractions while in the absence of PCNA (closed triangles) only was also con®rmed by immunoblot analysis (data not PFIIB showed DNA synthesis activity. Fractions PFIIB shown). PFIIB and PFIIC were then named PFIIe and and PFIIA/C were therefore good candidates for Pol e- PFIId, respectively. Further immunoblot analysis and Pol d-containing fractions, respectively. Immuno- revealed that PFIIe and PFIId contained in addition blot analysis (Figure 1b) con®rmed that PFIIC DNA ligase I and replication factor C (RF-C) (Figure contained Pol d (lane 2, left side) but no Pol e (lane 2, 1c, lanes 2 and 3, respectively) but were free of PCNA right side), while PFIIB contained Pol e (lane 3, right and RP-A (data not shown). side) and was free of Pol d (lane 3, left side). Immunoblot analysis of PFIIA revealed that this Pol d and Pol e can perform short-patch base excision fraction contained another form of Pol d, with a large repair subunit migrating at approximately 130 kDa (data not shown). This fraction was not further analysed. The The experimental system used in this study (Figure 2) total level of PCNA-dependent and -independent to monitor BER involves the use of a circular duplex

Figure 1 Ion-exchange chromatography and immunological characterization of the Pol-containing fractions from Pol b-null cell extract. Chromatography was performed as described under Materials and methods. (a) Activity pro®le. Fractions were tested for activity as described under Material and methods; open circles, PCNA- dependent Pol activity, closed triangles, PCNA-independent Pol activity. PFIIB and PFIIC eluted at *260 and *430 mM potassium phosphate, respectively. (b) Immunoblot analysis of total cell extract and the MonoQ fractions PFIIB and PFIIC with antibodies against Pol d (left side) and Pol e (right side). Lane 1: 10 mg total cell extract; lane 2: 3 mg PFIIC; lane 3: 3 mg PFIIB. (c) Immunoblot analysis of total cell extract and the MonoQ fractions PFIIB and C with antibodies against DNA ligase I (left side) and RF-C (right side). Lane 1: 10 mg total cell extract; lane 2: 3 mg PFIIB (peak and side fractions); lane 3: 3 mg PFIIC (peak and side fractions). Gel electrophoresis and immunoblot analysis were performed as described in Materials and methods DNA polymerase d and e in base excision repair M Stucki et al 837 plasmid molecule containing a single abasic site (X) required for the long-patch but not for the short-patch and of dierent labeled dNTPs in the reaction mixture BER (Frosina et al., 1996; Fortini et al., 1998) the in order to distinguish between short- and long-patch dependence of the repair reaction on this auxiliary repair events. The incorporation of radiolabeled dTMP protein is an additional marker to discriminate between in the BamHI ± HindIII restriction fragment marks the two pathways. mainly the occurrence of one-gap ®lling reactions that First, the ability of these two fractions to perform are the major route for repair of AP sites, while the short-patch BER was monitored as a function of time incorporation of dCMP identi®es exclusively 2-6 (Figure 3). Repair products were analysed by nucleotides repair patches. Moreover since PCNA is autoradiography after electrophoretic separation in denaturing 15% PAGE. Three labeled DNA products were visible on the autoradiography: the internal standard (IS) which is a 5'end-labeled 60-mer added in all reaction tubes in order to correct the repair incorporation values (as measured by electronic autoradiography, Figure 3 right) for DNA recovery; the 30-mer which is a marker of fully repaired DNA and the 13-mer which is the repair intermediate arising from one-nucleotide replacement reactions. Pol d and e fractions were able to incorporate labeled dTMP at the lesion site with comparable eciency (compare lanes 1 ± 3 with lanes 4 ± 6) and, as expected, the overall repair incorporation increased as a function of time. The presence of the 13-bp repair intermediate exclusively as well as the absence of PCNA in the Figure 2 DNA substrate used in the in vitro base excision repair repair reaction (PCNA is required for long-patch BER) assay. Circular duplex plasmid DNA (pGEM-X) containing a single abasic site (X) was subjected to the repair reaction in the testi®es that, under these experimental conditions, only presence of [a-32P]dTTP or [a-32P]dCTP to monitor the one-gap ®lling reactions occur and they are due to the occurrence of short- or long-patch repair events, respectively. action of either Pol d or Pol e. Although the initial Following digestion with BamHI and HindIII the repair products ratio of 1 : 2 of ligated (30-mer) to unligated (13-mer) were analysed by denaturing polyacrylamide gel. Fully repaired fragments are expected to run with a 30-mer; a labeled 13-mer product was drastically modi®ed with time (almost 1 : 1 indicates the repair intermediate arising from one-nucleotide at 1 h incubation time) a signi®cant portion of replacement reaction. unligated molecules were still present at the last repair

Figure 3 Pol d and Pol e can perform short-patch base excision repair. Left: autoradiograph of a denaturing polyacrylamide gel. Repair replication was performed for dierent periods of time in the presence of [a-32P]dTTP. The plasmid DNA was then digested with BamHI and HindIII to release the 30-bp fragment containing originally the lesion. Lanes 1 ± 3: repair synthesis by PFIId as a function of the incubation time; lanes 4 ± 6: repair synthesis by PFIIe as a function of the incubation time; lane 7: repair synthesis performed by PFIId after 90 min incubation time; lane 8: as in lane 7 with the addition of 1 mM spermine in the last 60 min of incubation. IS: internal standard. The electrophoretic mobility of 5'end-labeled 13- and 11-bp oligonucleotides (with the same sequence of the repair products) is also shown for comparison. Right: ligated (30-mer) and unligated (13-mer) products were measured by electronic autoradiography and relative incorporation, corrected for DNA recovery, is indicated on the ordinate (net CPM). DNA polymerase d and e in base excision repair M Stucki et al 838 time tested. This is due to an inecient removal of the with both Pols (lanes 3 and 7), while an inhibitory eect 5'-deoxyribose phosphate residue from a portion of the was recorded at higher PCNA concentration (lanes 4 incised molecules. Complete ligation was in fact and 8). PFIIe gave a higher proportion of repaired observed when the b-elimination catalyst spermine molecules (lanes 5 ± 8) in comparison with PFIId (lanes was added to the repair reaction (lane 8). 1 ± 4). In both cases the repair intermediates (413-mer) We have previously shown that the short-patch BER were not visible on the autoradiography, indicating that in wild-type cell extracts is PCNA-independent the polymerization is the rate limiting step and is tightly (Frosina et al., 1996) and in Pol b-de®cient cells is coupled with the ligation step. Next, the repair kinetics slightly stimulated by PCNA (Fortini et al., 1998). The by PFIId and PFIIe were investigated in the presence of repair eciency of Pol d- and Pol e-containing PCNA. As shown in Figure 6, not only the yield of fractions was investigated in the presence of increasing repaired molecules but also the rate of the long-patch amounts of PCNA (Figure 4). The overall repair BER was higher when the repair assay was conducted synthesis was only slightly increased in the presence of with PFIIe (lanes 5 ± 8) than with PFIId (lanes 1 ± 4). In PCNA, while, on the other hand, PCNA was clearly this experiment the concentration of PFIId was doubled. stimulating the formation of the ligated product with The repair reaction reached a plateau within 1-2 hrs. both Pols (lanes 2 ± 4 and 6 ± 8). The most likely Therefore, Pol e appears to be better suited for gap ®lling explanation for this phenomenon is that in the presence repair reactions than Pol d . However, we cannot exclude of PCNA some long-patch BER takes place. Alter- that these dierences in repair eciency are due to the natively, PCNA might stimulate either the resynthesis relative content of some other critical BER components and/or the ligation process of one-nucleotide repair in the two subfractions. For instance, lower levels of patches (Jonsson and Hubscher, 1997; Levin et al., FEN1 were detected by Western blotting in PFIId than 1997; Jonsson et al., 1998). in PFIIe (data not shown).

Long-patch base excision repair by Pol d and Pol e RP-A is not required for long-patch base excision repair requires PCNA by Pol d and/or Pol e A typical feature of the long-patch BER is that it The single-stranded DNA binding protein RP-A is requires PCNA (Matsumoto et al., 1994; Frosina et al., required for DNA synthesis by Pol d and Pol e on 1996; Klungland and Lindahl, 1997). The ability of Pol primed single-stranded DNA templates (Tsurimoto and d- and Pol e-containing fractions to perform long-patch Stillman, 1989; Podust and Hubscher, 1993) and is repair synthesis was analysed in the presence of involved in nucleotide excision repair both during the increasing concentrations of PCNA (Figure 5). No damage recognition/incision step (Coverley et al., 1991; repair synthesis was observed in the absence of PCNA Shivji et al., 1992) and during the repair synthesis with either PFIId (lane 1) or PFIIe (lane 5). The optimal (Shivji et al., 1995). Consequently, the eect of RP-A PCNA concentration for long-patch synthesis was 50 ng was investigated in the long-patch BER. As shown in

Figure 4 Short-patch base excision repair by Pol d and Pol e is stimulated by PCNA. Left: autoradiograph of a denaturing polyacrylamide gel. Repair replication was performed for 1 h in the presence of [a-32P]dTTP. The plasmid DNA was then digested with BamHI and HindIII to release the 30-bp fragment containing originally the lesion. Lanes 1 ± 4: repair synthesis by PFIId as a function of the PCNA concentration; lanes 5 ± 8: repair synthesis by PFIIe as a function of the PCNA concentration; lane 9: repair synthesis by PFIIe on a control plasmid containing a normal T:A base pair. IS: internal standard. Right: ligated (30-mer) and unligated (13-mer) products were measured by electronic autoradiography and relative incorporation, corrected for DNA recovery, is indicated on the ordinate (net CPM) DNA polymerase d and e in base excision repair M Stucki et al 839 Figure 7, in the presence of PCNA, repair synthesis by PCNA and RP-A were omitted from the repair PFIIe was inhibited at increasing RP-A concentrations reaction (lane 5). These data indicated that RP-A is (lanes 2 ± 4). No repair synthesis was observed either in not required in BER neither in the short-patch (data the presence of RP-A alone (lane 6) or when both not shown) nor in the long-patch pathway.

Figure 5 The long-patch base excision repair by Pol d and Pol e is dependent on PCNA. Left: autoradiograph of a denaturing polyacrylamide gel. Repair replication was performed for 1 h in the presence of [a-32P]dCTP. The plasmid DNA was then digested with BamHI and HindIII to release the 30-bp fragment containing originally the lesion. Lanes 1 ± 4: repair synthesis by PFIId as a function of the PCNA concentration; lanes 5 ± 8: repair synthesis by PFIIe as a function of the PCNA concentration; lane 9: repair synthesis by PFIIe on a control plasmid containing a normal T:A base pair. IS: internal standard. Right: ligated (30-mer) products were measured by electronic autoradiography and relative incorporation, corrected for DNA recovery, is indicated on the ordinate (net CPM)

Figure 6 Kinetics of the long-patch base excision repair carried out by Pol d and Pol e. Left: autoradiograph of a denaturing polyacrylamide gel. Repair replication was performed for dierent periods of time in the presence of [a-32P]dCTP. The plasmid DNA was then digested with BamHI and HindIII to release the 30-bp fragment containing originally the lesion. Lanes 1 ± 4: repair synthesis by PFIId as a function of the incubation time; lanes 5 ± 8: repair synthesis by PFIIe as a function of the incubation time. IS: internal standard. Right: ligated (30-mer) products were measured by electronic autoradiography and relative incorporation, corrected for DNA recovery, is indicated on the ordinate (net CPM) DNA polymerase d and e in base excision repair M Stucki et al 840

Figure 7 The long-patch base excision repair carried out by Pol d and Pol e is not dependent on RP-A. Left: autoradiograph of a denaturing polyacrylamide gel. Repair replication was performed for 1 h in the presence of [a-32]dCTP. The plasmid DNA was then digested with BamHI and HindIII to release the 30-bp fragment containing originally the lesion. Lanes 1 ± 4: repair synthesis by PFIIe in the presence of RP-A (50, 150, 300 ng) and PCNA (50 ng); lane 5: without PCNA and RP-A; lane 6: with RP-A (150 ng) alone. IS: internal standard. Right: ligated (30-mer) products were measured by electronic autoradiography and relative incorporation, corrected for DNA recovery, is indicated on the ordinate (net CPM)

Discussion patch BER is supported by this study where we show that Pol d- or Pol e-containing fractions are well suited In light of these results and of a recent own study to ®lling of several nucleotides (2 ± 6 nucleotides) gaps (Fortini et al., 1998) we wish to reconsider the model produced as AP site repair intermediates. This process proposed for BER in mammalian cells (Klungland and is strictly dependent on the presence of PCNA. Pol e Lindahl, 1997). Whole cell extracts from Pol b-null cell appears to be better suited to this task as compared to lines (Fortini et al., 1998) as well as fractionated cell Pol d. A combination of Pol e, PCNA, RF-C, RP-A extracts containing either Pol d or Pol e (this study) and DNA ligase I was also shown to be the most were able to perform both the short- and the long- ecient repairsome for nucleotide excision repair gaps patch BER. Typically an abasic site is recognized by (Shivji et al., 1995). However, since PFIId in our an AP endonuclease that incises the phosphodiester fractions contains lower levels of FEN1 than PFIIe backbone immediately 5' to the lesion leaving a strand (data not shown) we can not exclude that this break with a 5'-abasic terminus. Pol b is able to dierence is responsible for the apparently lower catalyze both the release of this abnormal 5'-abasic repair eciency of Pol d. It is well established that, terminus as well as one-nucleotide gap ®lling reactions. on circular DNA, PCNA requires RF-C and ATP to To complete this process the nick is sealed by a DNA be loaded onto the DNA template and function as a ligase, thus restoring the genetic information. In this `sliding clamp' for Pol d and Pol e. The PCNA- work we show that, besides Pol b, Pol d and Pol e are dependence of the long-patch BER strongly suggests also able to replace a single nucleotide at the lesion site the involvement of RF-C, which is present in both in vitro, but the completion of the repair reaction is subfractions, in this BER pathway. PCNA is likely an delayed by the slow processing of the 5'-deoxyribose intracellular communicator not only for DNA replica- phosphate moieties. The addition of a b-catalyst like tion but also for important cellular events including spermine is in fact required to complete short-patch DNA repair (reviewed in Jonsson and Hubscher, BER by PFIId and PFIIe. The accumulation of repair 1997). Its role in the long-patch BER might be dual: intermediates was not observed when cell extracts from PCNA would target and anchor Pol d and/or Pol e Pol b-deleted cells were used in the short-patch BER onto the damaged template to allow the polymerization reaction (Fortini et al., 1998). It is possible that step, and it would also participate to the cleavage of molecules able to act as b-elimination catalysts were the 5'-single strand (Klungland and Lindahl, 1997) by lost in the fractionation procedure. Pol b seems to its direct interaction with FEN1 (Li et al., 1995; Wu et catalyze the excision of the 5'-terminal deoxyribose al., 1996). Interestingly all the three major repair phosphate more eciently than basic molecules processes ± nucleotide excision repair, mismatch repair (Matsumoto and Kim, 1995). This might explain why and BER ± share some molecular partners such as in vivo Pol b is the DNA polymerase of election for PCNA, FEN1, Pol d/e (reviewed in Wood, 1997; one nucleotide replacement reactions (Sobol et al., Modrich, 1997) which are also involved in replication- 1996). associated chain elongation. The overlapping among The previous observation that Pol b-de®cient cell these processes might provide the cells with alternative extracts (Fortini et al., 1998) eciently perform long- defence strategies against endogenous and exogenous DNA polymerase d and e in base excision repair M Stucki et al 841 DNA damage as well as with potential mechanism to Nucleic acid substrates coordinate DNA replication with DNA repair Poly(dA)1000 ± 1500 and oligo(dT)12±18 were purchased from (Warbrick et al., 1997). Pharmacia and Sigma, respectively. Poly(dA)oligo(dT) The removal of the 5'-abasic terminus was found to (base ratio 10 : 1) was prepared as described (Weiser et be a rate-limiting step in BER of uracil-containing al., 1991). Oligonucleotides for preparing the substrate for DNA in yeast cell-free extracts (Wang et al., 1997) and the BER assay were purchased from M-Medical. the hypothesis was formulated that this is the primary mechanism for controling the operation of the two Enzymes and proteins BER pathways. Our data strongly support this hypothesis. In fact, under conditions which allow the Restriction enzymes were from Biolabs. Human PCNA was overexpressed in E. coli strain BL21(DE3) harboring the occurrence of short-patch BER exclusively (in the expression plasmid pT7/PCNA and puri®ed as described absence of PCNA), unprocessed 5'-deoxyribose phos- by Fien and Stillman (1992). Human RP-A was over- phate moieties were detected and the addition of expressed in E. coli strain BL21(DE3) harboring the PCNA clearly increased the yield of fully repaired expression plasmid p11d-tRP-A and puri®ed according to molecules. Henricksen et al. (1994) with slight modi®cations. The kinetics of the long patch BER deserves also some comment. Here, even under conditions of low Buers repair eciency (as in the case of PFIId), no repair intermediates were detected. This ®nding suggests that Buer A: 20 mM Tris-HCl,pH7.5,NaCl(variable the strand invasion reaction by Pol d/e is tightly concentrations as indicated below), 10% glycerol (v/v), 1mM DTT, 1 mM PMSF, 1 mg/ml pepstatin, 1 mg/ml coupled with cleavage of the displaced 5'-single strand leupetin. Buer B: Potassium phosphate, pH 7.5 (variable by FEN1 and sealing of the DNA template by DNA concentrations as indicated below), 10% glycerol (v/v), ligase I. This coupling might be mediated by PCNA 1mM DTT, 1 mM PMSF, 1 mg/ml pepstatin, 1 mg/ml (Jonsson et al., 1998). leupetin. Buer C: Potassium phosphate, pH 7.5 (variable Our data show that RP-A might not be required concentrations as indicated below), 10% glycerol (v/v), for BER. The presence of the single-stranded binding 1mM DTT, 0.5 mM EDTA, 1 mM PMSF, 1 mg/ml protein is even inhibitory for the long-patch repair pepstatin and 1 mg/ml leupetin. reaction. RP-A combines with single-stranded DNA by forming a stable complex with at least 30 PCNA dependent and PCNA independent Pol assays nucleotides (Kim et al., 1992). Since RP-A is required The poly(dA)oligo(dT) assay (RF-C independent) was for nucleotide excision repair (Coverley et al., 1991) it carried out according to (Weiser et al., 1991). Brie¯y, a appears that the gap size is critical in dictating ®nal volume of 25 ml contained 50 mM BisTris-HCl, whether this protein has a functional role in DNA pH 6.5, 0.5 mg poly(dA)oligo(dT) (base ratio 10 : 1), repair. 20 mM [3H]dTTP (speci®c activity: 500 c.p.m./pmol),

The existence of two BER pathways seems to be a 10 mM MgCl2,1mM DTT, 250 mg/mlbovineserum general cellular strategy, from E. coli to humans, in albumine and enzyme fraction to be tested. When order to safeguard the genome from very common indicated, 100 ng of PCNA was added. After incubation DNA lesions. The redundancy in this repair process is at 378C for 30 min., insoluble radioactive material was in apparent contrast with the inability to generate mice determined as described in (Hubscher and Kornberg, that are defective in BER, as seen with HAP1 1979). (Xanthoudakis et al., 1996) and Pol b (Gu et al., 1994). It will be interesting to know whether the eect Extraction and fractionation of Pol b-null cells on embryonic development resides in functions other Total cell extract was prepared as described in (Frosina than DNA repair (Xanthoudakis et al., 1992; Plug et et al., 1994), with slight modi®cations. 70 ml crude al., 1997). extract (1.3 mg/ml) was dialysed against buer A, In conclusion we have shown that Pol d or Pol e can containing 100 mM NaClandloadedona20ml perform both short- and long-patch BER in Pol b- phosphocellulose column, equilibrated in the same de®cient mammalian cells. It is likely that our ®ndings buer. The column was washed with four column re¯ect two important back-up mechanisms for this volumes of equilibration buer, followed by two column essential BER pathway. A puri®ed base excision in volumes of buer A, containing 500 mM NaCl and one column volume of buer A containing 600 mM NaCl. vitro system containing a minimal set of at least seven The peak of the Pol activity, which was in the 500 mM components (uracil-DNA glycosylase, HAP1, Pol d or NaCl wash, was pooled, yielding Pol fraction I (PFI) and Pol e, RF-C, PCNA, dRPase or FEN1, DNA ligase I) dialysed against buer B, containing 20 mM potassium will give insight into the functional task(s) of the two phosphate. PFI was then loaded on a 8 ml ceramic dierent Pols in BER. hydroxylapatite column, equilibrated in buer B. The column was washed with 5 volumes of equilibration buer and proteins were eluted with a linear gradient of 20 ± 600 mM potassium phosphate in buer B. One- milliliter fractions were collected. The Pol activity, Materials and methods eluting at approximately 450 mM potassium phosphate (PFIA), was not altered in the presence of the Pol a- Chemicals speci®c neutralizing antibody SJK 132-20, thus excluding [3H]dTTP (40 Ci/mmol), [a-32P]dCTP (3000 Ci/mmol) and contamination with Pol a. This fraction was pooled [a-32P]dTTP (3000 Ci/mmol) were from Amersham and (6 ml, 1.1 mg/ml protein) and dialysed versus buer C, unlabeled dNTPs were from Boehringer. All other reagents containing 20 mM potassium phosphate. were of analytical grade and purchased from Merck and The dialysed PFIA was then loaded on a 1 ml FPLC Fluka. MonoQ column (Pharmacia), equilibrated in buer C. The DNA polymerase d and e in base excision repair M Stucki et al 842 column was washed with 5 ml of equilibration buer and priming single-stranded (+) pGEM-3Zf DNA (Promega) proteins were eluted with a 10-ml linear gradient of 20 ± with a 30-fold molar excess of uracil-containing or control 600 mM potassium phosphate in buer C. Pol activity eluted oligonucleotide and incubating with T4 DNA polymerase in three peaks (*190, *260 and *430 mM potassium holoenzyme, single-stranded DNA binding protein and T4 phosphate, respectively), yielding PFII A, B and C. The DNA ligase (Boehringer Mannheim). Closed circular DNA PFIIA and PFIIC activities were PCNA dependent and the duplex molecules were puri®ed by cesium chloride PFIIB activity was PCNA independent (Figure 1a). The equilibrium centrifugation. The oligonucleotide three ®nal fractions were dialysed against buer A, 5'-GATCCTCTAGAGUCGACCTGCA-3' was used to containing 20 mM NaCl and 0.5 mM EDTA, and dripped create plasmid molecules containing a single uracil directly into liquid nitrogen for storage in frozen bead form residue. These circular closed DNAs were then digested at 7808C. with E. coli uracil-DNA glycosylase to produce a single abasic site. The oligonucleotide 5'-GATCCTCTAGAGTC- GACCTGCA-3' was used to prepare the control plasmid. Immunoblot analysis of PFIIB and PFIIC The repair reactions were carried out essentially as Polyclonal antibodies against mouse Pol d were prepared described in (Frosina et al., 1996). Brie¯y, a standard from rabbit as described by Cullmann et al. (1993). reaction mixture contained 2 ng of puri®ed HAP1, Antibody against the 40 kDa subunit of RF-C was a gift aliquots of PFIIB or PFIIC corresponding to 0.5 U of of J Hurwitz (Chen et al., 1992). Monoclonal antibody Pol e and d respectively in reaction buer containing against human Pol e catalytic subunit was a gift of J 20 mM of each dNTP and 2 mCi of either [a-32P]dTTP or SyvaÈ oja (Uitto et al., 1995). Polyclonal antibody against [a-32P]dCTP as indicated. Puri®ed PCNA and RP-A were bovine ligase I was a gift from T Lindahl. SDS ± PAGE added when required. After dierent incubation times at was as described by Laemmli (1970). Proteins were 308C the plasmid DNA was recovered and digested with transferred to a Protran Nitrocellulose membrane (Schlei- BamHI and HindIII. The digestion products were resolved cher and Schuell) by electroblotting with a Biorad Trans- on denaturing 15% PAGE. The BER products were blot apparatus, according to the manufacturer's protocol. visualized by autoradiography and analysed by electronic The membrane was blocked with 5% (w/v) solution of milk autoradiography (Instant Imager, Packard) for quantifica- powder and then incubated in the presence of the tion. appropriate antibody. Cross-reactivity of antibody to proteins was detected by utilizing peroxidase-conjugated secondary antibodies and ECL gene detection system Acknowledgements (Amersham), following the manufacturer's protocol. The We thank R Mossi for providing puri®ed RP-A, ID membranes could be stripped of bound antibodies and Hickson for puri®ed HAP1, S Boiteux for uracil-DNA reprobed several times. glycosylase, M Lieber and T Lindahl for FEN1 and DNA ligase I antibodies, respectively. We are very grateful to F Belisario for excellent technical help and to L Gargano for In vitro base excision repair assay technical assistance. This work was partially supported by Closed circular DNA containing a single abasic site was the Swiss Cancer League (Grant KFS 164-9-1995 to MS produced as previously described (Frosina et al., 1996) by and UH) and by the Kanton of ZuÈ rich.

References

Chen M, Pan Z-Q and Hurwitz J. (1992). Proc. Natl. Acad. Kubota Y, Nash R, Klungland A, Schar P, Barnes D and Sci. USA, 89, 2516 ± 2520. Lindahl T. (1996). EMBO J., 15, 6662 ± 6670. Coverley D, Kenny MK, Munn M, Rupp WD, Lane DP and Laemmli UK. (1970). Nature, 227, 680 ± 685. Wood RD. (1991). Nature, 349, 538 ± 541. Levin DS, Bai W, Yao N, O'Donnell M and Tomkinson AE. Cullmann G, Hindges R, Berchtold MW and Hubscher U. (1997). Proc. Natl. Acad. Sci. USA, 94, 12863 ± 12868. (1993). Gene, 134, 191 ± 200. Li X, Li J, Harrington J, Lieber MR and Burgers PMJ. Fien K and Stillman B. (1992). Mol. Cell. Biol., 12, 155 ± 163. (1995). J. Biol. Chem., 270, 22109 ± 22112. FortiniP,PascucciB,ParlantiE,SobolRW,WilsonSHand Longley MJ, Pierce AJ and Modrich P. (1997). J. Biol. Dogliotti E. (1998). Biochemistry, 37, 3575 ± 3580. Chem., 272, 10917 ± 10921. Frosina G, Fortini P, Rossi O, Abbondandolo A and Matsumoto Y and Kim K. (1995). Science, 269, 699 ± 702. Dogliotti E. (1994). Biochem. J., 304, 699 ± 705. Matsumoto Y, Kim K and Bogenhagen DF. (1994). Mol. Frosina G, Fortini P, Rossi O, Carrozzino F, Raspaglio G, Cell. Biol., 14, 6187 ± 6197. Cox LS, Lane DP, Abbondandolo A and Dogliotti E. Modrich P. (1997). J. Biol. Chem., 272, 24727 ± 24730. (1996). J. Biol. Chem., 271, 9573 ± 9578. PlugAW,ClairmontCA,SapiE,AshleyTandSweasyJB. Gu H, Marth JD, Orban PC, Mossmann H and Rajewsky K. (1997). Proc. Natl. Acad. Sci. USA, 94, 1327 ± 1331. (1994). Science, 265, 103 ± 106. Podust VN and Hubscher U. (1993). Nucl. Acids Res., 21, Henricksen LA, Umbricht CB and Wold MS. (1994). J. Biol. 841 ± 846. Chem., 269, 11121 ± 11132. Prasad R, Singhal RK, Srivastava DK, Molina JT, Hubscher U and Kornberg A. (1979). Proc. Natl. Acad. Sci. Tomkinson AE and Wilson SH. (1996). J. Biol. Chem., USA, 76, 6284 ± 6288. 271, 16000 ± 16007. Jonsson ZO and Hubscher U. (1997). BioEssays, 19, 967 ± Shivji MKK, Kenny MK and Wood RD. (1992). Cell, 69, 975. 367 ± 374. Jonsson ZO, Hindges R and Hubscher U. (1998). EMBO J., Shivji MKK, Podust VN, Hubscher U and Wood RD. in press. (1995). Biochemistry, 34, 5011 ± 5017. Kim C, Snyder RO and Wold MS. (1992). Mol. Cell. Biol., Sobol RW, Horton JK, Kuhn R, Gu H, Singhal RK, Prasad 12, 3050 ± 3059. R, Rajewsky K and Wilson SH. (1996). Nature, 379, 183 ± Klungland A and Lindahl T. (1997). EMBO J., 16, 3341 ± 186. 3348. DNA polymerase d and e in base excision repair M Stucki et al 843 Tsurimoto T and Stillman B. (1989). EMBO J., 8, 3883 ± Weiser T, Gassmann M, Thommes P, Ferrari E, Hafkemeyer 3889. P and Hubscher U. (1991). J. Biol. Chem., 266, 10420 ± Uitto L, Halleen J, Hentunen T, Hoyhtya M and Syvaoja JE. 10428. (1995). Nucl. Acids Res., 23, 244 ± 247. Wood RD. (1997). J. Biol. Chem., 272, 23465 ± 23468. Umar A, Buermeyer AB, Simon JA, Thomas DC, Clark AB, Wu X, Li J, Li X, Hsieh C-L, Burgers PM and Lieber MR. Liskay M and Kunkel TA. (1996). Cell, 87, 65 ± 73. (1996). Nucl. Acids Res., 24, 2036 ± 2043. Wang Z, Wu X and Friedberg EC. (1997). J. Biol. Chem., Xanthoudakis S, Miao G, Wang F, Pan Y-CE and Curran T. 272, 24064 ± 24071. (1992). EMBO J., 11, 3323 ± 3335. Warbrick E, Lane DP, Glover DM and Cox LS. (1997). Xanthoudakis S, Smeyne RJ, Wallace JD and Curran T. Oncogene, 14, 2313 ± 2321. (1996). Proc. Natl. Acad. Sci. USA, 93, 8919 ± 8923.