in vitro repair of synthetic ionizing radiation-induced multiply damaged dna sites

Article No. jmbi.1999.2892 available online at http://www.idealibrary.com on J. Mol. Biol. (1999) 290, 667±684

In Vitro Repair of Synthetic Ionizing Radiation-inducedMultiply Damaged DNA Sites

Lynn Harrison, Zafer Hatahet and Susan S. Wallace*

Department of Microbiologyand Molecular Genetics, TheMarkey Center for MolecularGenetics, The University ofVermont, Stafford HallBurlington, VT 05405-0068, USA

Present addresses: L. Harrison, DLA 71130, USA; Z. Hatahet, Depart75708, USA.

Abbreviations used: MDS, multipformamidopyrimidine DNA glycosyexcision repair; TCR, transcription-c

E-mail address of the correspond

0022-2836/99/280667±18 $30.00/0

When ionizing radiation traverses a DNA molecule, a combination oftwo or more base damages, sites of base loss or single strand breaks canbe produced within 1-4 nm on opposite DNA strands, forming a multi-ply damaged site (MDS). In this study, we reconstituted the base excisionrepair system to examine the processing of a simple MDS containing thebase damage, 8-oxoguanine (8-oxoG), or an abasic (AP) site, situated inclose opposition to a single strand break, and asked if a double strandbreak could be formed. The single strand break, a nucleotide gap contain-ing 30 and 50 phosphate groups, was positioned one, three or six nucleo-tides 50 or 30 to the damage in the complementary DNA strand.Escherichia coli formamidopyrimidine DNA glycosylase (Fpg), whichrecognizes both 8-oxoG and AP sites, was able to cleave the 8-oxoG orAP site-containing strand when the strand break was positioned three orsix nucleotides away 50 or 30 on the opposing strand. When the strandbreak was positioned one nucleotide away, the target lesion was a poorsubstrate for Fpg. Binding studies using a reduced AP (rAP) site in thestrand opposite the gap, indicated that Fpg binding was greatly inhibitedwhen the gap was one nucleotide 50 or 30 to the rAP site.

To complete the repair of the MDS containing 8-oxoG opposite a singlestrand break, endonuclease IV DNA polymerase I and Escherichia coliDNA ligase are required to remove 30 phosphate termini, insert the``missing'' nucleotide, and ligate the nicks, respectively. In the absence ofFpg, repair of the single strand break by endonuclease IV, DNA polymer-ase I and DNA ligase occurred and was not greatly affected by the 8-oxoG on the opposite strand. However, the DNA strand containing thesingle strand break was not ligated if Fpg was present and removed theopposing 8-oxoG. Examination of the complete repair reaction productsfrom this reaction following electrophoresis through a non-denaturinggel, indicated that a double strand break was produced. Repair of thesingle strand break did occur in the presence of Fpg if the gap was onenucleotide away. Hence, in the in vitro reconstituted system, repair of theMDS did not occur prior to cleavage of the 8-oxoG by Fpg if the oppos-ing single strand break was situated three or six nucleotides away,converting these otherwise repairable lesions into a potentially lethaldouble strand break.

# 1999 Academic Press

Keywords: multiply damaged sites; in vitro DNA repair; ionizing radiationDNA damage; base excision repair; 8-oxoguanine
*Corresponding author
epartment of Physiology, LSU Medical Center, 1501 Kings Highway, Shreveport,ment of Biochemistry, University of Texas Health Center at Tyler, Tyler, TX

ly damaged sites; 8-oxoG, 8-oxoguanine; AP, abasic site; Fpg,lase; rAP, reduced AP; SSB, single strand breaks; endo, endonuclease; BER, baseoupled repair UDG, uracil DNA-glycosylase.ing author: [email protected]

# 1999 Academic Press

mailto:[email protected]

668 Repair of Multiply Damaged Sites in DNA

Introduction

Fewer DNA lesions are required to produce alethal event in a mammalian cell when they areintroduced by ionizing radiation or drugs such asbleomycin A2 and neocarzinostatin compared to adamaging agent such as hydrogen peroxide (Wardet al., 1987). Ionizing radiation and hydrogen per-oxide introduce similar types of base damage, sitesof base loss (AP sites) and single strand breaks(SSB) by the generation of hydroxyl radicals inclose proximity to the DNA. The difference in thelethality produced by ionizing radiation is believedto be due to the spatial distribution and repairabil-ity of the damage in the DNA molecule. Ionizingradiation track structure simulations and calcu-lations of damage induction have indicated thatmultiple lesions can be produced from low energyX and g-rays within a single helical turn of theDNA molecule (Brenner & Ward, 1992; Goodhead,1994). It has been proposed that clustering oflesions in the DNA, called multiply damaged sites(MDS), may inhibit the base excision repair process(for reviews, see Demple & Harrison, 1994;Wallace, 1997) or enhance the number of double-strand breaks formed by base excision repair pro-cessing leading to lethal consequences. In supportof the latter hypothesis, when irradiated cells areallowed time to repair the damage, an increase inthe level of double strand breaks is found(Ahnstrom & Bryant, 1982; Bonura et al., 1975;Dugle et al., 1976). Chaudhry & Weinfeld (1995a)demonstrated that when an MDS contains twoidentical closely opposed base damages, only onedamage is removed and a single strand break isgenerated. The two base damages had to be atleast three nucleotides apart before the glycosy-lase/AP lyase that recognized them could generatea double strand break. This is in agreement withour previous work (Harrison et al., 1998) whichshowed that the Escherichia coli DNA glycosylases,endonuclease III (endo III) and endonuclease VIII(endo VIII), are unable to release the damagedbase if the damage site is situated opposite andone nucleotide away from a strand break. Whenthe strand break is three or more nucleotides fromthe base damage, cleavage by endo VIII takesplace and appears to correlate with its ability tobind to the lesion. Disruption of glycosylase bind-ing or cleavage is also dependent on the orien-tation of the strand break with respect to the basedamage and is different for the two enzymes (endoIII and endo VIII) that have very similar substratespeci®cities (Harrison et al., 1998).

We and others have postulated that attempts torepair MDS can convert non-lethal lesions (e.g.dihydrothymine) or mutagenic lesions (e.g. 8-oxoG) into lethal double strand breaks. Dianovet al. (1991) demonstrated that when a plasmidcontaining two uracil bases, positioned in oppositestrands and separated by 12 bp, was treated withcell-free extracts or transformed into bacteria, adouble strand break or deletion was formed,

respectively. So far studies concerning the repair ofoxidative DNA damage in MDS (Chaudhry &Weinfeld, 1995a,b, 1997; Harrison et al., 1998) haveonly addressed the question of double-strandbreak formation with regard to whether the DNAglycosylase, the ®rst step in the base excision repair(BER) process, can cleave the DNA at a MDS. Inthe presence of all the BER enzymes, it is possiblethat the lesion in one strand would be repairedbefore the second lesion in the MDS is convertedto a break. Here, we have tried to address thisquestion by reconstituting the base excision repairprocess: glycosylase action, endonuclease IVremoval of 30 blocking termini, insertion of anucleotide by E. coli DNA polymerase I and lig-ation by E. coli DNA ligase. Reconstitution of thebase excision repair pathway with prokaryotic(Dianov & Lindahl, 1994) or eukaryotic proteins(Kubota et al., 1996) has previously been achievedfor the removal of uracil from DNA. For a model,biologically relevant MDS produced by ionizingradiation, we used an 8-oxoguanine (8-oxoG) clo-sely opposed to a single strand break. MDS of thistype can be produced by the multiple radicals gen-erated by an ionizing radiation track, and arehighly probable given the preponderance of singlestrand breaks and base damages formed. SuchMDS are also formed, as we have previouslyshown, by the cleavage of a DNA glycosylase/APlyase at a pyrimidine damage closely opposed toan 8-oxoG (Harrison et al., 1998). This is the ®rstreconstitution of the BER process that examines therepair of multiple oxidative lesions in DNA. Here,we demonstrate that a double strand break can beformed in vitro even when formamidopyrimidineDNA glycosylase (Fpg), endo IV, DNA polymeraseI and E. coli DNA ligase are present, and thatdouble-strand break formation depends on thepositioning of the lesions in the MDS.

Results

Reconstitution of BER with an MDS substrate

Single strand breaks formed by ionizing radi-ation have 50 phosphate and 30 phosphate or 30phosphoglycolate termini (Henner et al., 1983) dueto radical attack on the deoxyribose moiety, whichresults in fragmentation of the deoxyribose (Ward,1988). Thus a representative radiation-inducedsingle-strand break is a nucleotide gap with both 30and 50 phosphate termini. A similar strand break isgenerated by certain DNA-glycosylases thatcontain an associated AP lyase activity. Ionizingradiation also introduces many types of DNA basedamages, including 8-oxoG and sites of base loss(AP sites). Statistically, a MDS containing a nucleo-tide gap with 30 and 50 phosphate termini closelyopposed to an 8-oxoG or an AP site is likely to beformed by ionizing radiation, and is therefore abiologically relevant lesion. A similar MDS canalso be produced by the partial repair of twoopposing base damages, AP sites or a base damage

Table 1. Oligonucleotides

Strand Damage Sequence

Position ofdamage relativeto X in strand A

A - 50ATTCCAGACTGTCAATAACACGGGGGACCAGTCGATCCTGGGCTGCAGGAATTC 30 -

AX�8oxoGAP, rAP, U 50ATTCCAGACTGTCAATAACACGGXGGACCAGTCGATCCTGGGCTGCAGGAATTC 30 -

B - 30TAAGGTCTGACAGTTATTGTGCCCCCTGGTCAGCTAGGACCCGACGTCCTTAAG 50 -B 1nt gap 30TAAGGTCTGACAGTTAT GTGCCCCCTGGTCAGCTAGGACCCGACGTCCTTAAG 50 6 nt 50B 1nt gap 30TAAGGTCTGACAGTTATTGT CCCCCTGGTCAGCTAGGACCCGACGTCCTTAAG 50 3 nt 50B 1nt gap 30TAAGGTCTGACAGTTATTGTGC CCCTGGTCAGCTAGGACCCGACGTCCTTAAG 50 1 nt 50B 1nt gap 30TAAGGTCTGACAGTTATTGTGCCC CTGGTCAGCTAGGACCCGACGTCCTTAAG 50 1 nt 30B 1nt gap 30TAAGGTCTGACAGTTATTGTGCCCCC GGTCAGCTAGGACCCGACGTCCTTAAG 50 3 nt 30B 1nt gap 30TAAGGTCTGACAGTTATTGTGCCCCCTGG CAGCTAGGACCCGACGTCCTTAAG 50 6 nt 30

The double-stranded substrates were produced by annealing strand A with a strand B as described in Materials and Methods. Forsubstrates with a nucleotide gap, two oligonucleotides were used to form strand B. The oligonucleotides forming the gap had 50 or30 phosphate termini (see Figure 1). nt, nucleotide.

Figure 1. Diagram of a MDS containing 8-oxoG, blackpurine, and a closely opposed single strand break con-taining 30 and 50 phosphate groups positioned one, threeor six nucleotides 30 to the 8-oxoG.

Repair of Multiply Damaged Sites in DNA 669

and an AP site (Chaudhry & Weinfeld, 1995a,1997; Harrison et al., 1998).

Complete repair of a MDS containing a basedamage opposite a SSB requires a DNA glycosy-lase to remove the damaged base, a 30 repairdiesterase or 30 phosphatase to generate 30hydroxyl termini, a DNA polymerase to replacethe ``missing'' nucleotide and a DNA ligase to sealthe ``nicks''. Since the E. coli base excision repairenzymes have been puri®ed and well character-ized, we chose Fpg, endonuclease IV, E. coli DNApolymerase I and E. coli DNA ligase to reconstitutethe complete repair of a MDS containing 8-oxoGopposite a SSB. Previous studies have demon-strated that Fpg is a DNA glycosylase with anassociated AP lyase activity that cleaves the N-glycosylic bond of 8-oxoG in DNA (Castaing et al.,1993; Tchou et al., 1991), producing an AP site. In aconcerted reaction, Fpg then catalyzes a b,d-elimin-ation reaction, converting the AP site to a nucleo-tide gap, the termini of which have attached 30 and50 phosphate moieties (Bailly et al., 1989; Bhagwat& Gerlt, 1996; O'Connor & Laval, 1989). Endo-nuclease IV is a 30 repair diesterase and a class IIAP endonuclease. It is able to remove a phosphatemoiety from a 30 terminus to generate a 30hydroxyl terminus, which can then be extended byE. coli DNA polymerase I using the complementarystrand sequence to insert any missing nucleotides.Because of the nick translation activity of E. coliDNA polymerase I (Klett et al., 1968), it wasnecessary to limit the types of the deoxynucleotidetriphosphates in the repair reaction to thoserequired to replace 8-oxoG (dGTP) and to repairthe nucleotide gaps (dTTP or dCTP, see Materialsand Methods). When all four dNTPs were present,a complete repair product (the size of a 54-mer)was detected even when E. coli DNA ligase wasnot present (data not shown). When reactions wererun at 37 �C some breakdown of the substrate wasobserved in the presence of DNA polymerase I, soall reactions were performed at 20 �C. In fact, allthe enzyme reactions examined, even those notcontaining E. coli DNA polymerase I, were per-

formed at 20 �C and in the buffer (see Materialsand Methods) that was optimal for all the enzymesinvolved in the complete repair reaction. As shownbelow, each stage of the repair process was initiallystudied to optimize the amount of each enzymerequired to achieve maximum activity underthe conditions required for the complete repairreaction.

Fpg-catalyzed removal of 8-oxoG closelyopposed to a single nucleotide gap

Strand A (54-mer) containing an 8-oxoG at pos-ition 24 (Table 1) was 50 labeled with 32P andannealed to strand B (as described above) to formdouble-stranded substrates with 8-oxoG closelyopposed to a nucleotide gap one, three or sixnucleotides 30 to the 8-oxoG lesion (Figure 1).Following incubation of the substrates with Fpg,

Figure 2 (Legend shown opposite)


Figure 2. Fpg cleavage at an 8-oxoguanine (8-oxoG) or an AP site situated in close opposition to a nucleotide gap.(a) and (b) Substrate containing 20 nM 8-oxoG was incubated for 15 minutes at 20 �C with 0, 10, 40 or 80 nM Fpg inbuffer A. Samples were subjected to electrophoresis through a 12 % polyacrylamide, 7 M urea gel. (a) Products of thereactions were visualized after autoradiography of the gel. The extent of cleavage was quanti®ed using a BIO-RADMolecular Imager. Reactions were performed in triplicate and the average and standard error is shown in (b). Thecontrol substrate (m) consists of a 54-mer with an 8-oxoG at position 24 (strand A, see Table 1), annealed to an intactDNA strand. The MDS substrates have a nucleotide missing in strand B at position one (&), three (&) or six (F), 50or 30 to the 8-oxoG in strand A (see Table 1 and Figure 1). Bars to the left of the control correspond to substrateswith the nucleotide gap 50 to the damage in strand A and bars to the right of the control correspond to substrateswith the nucleotide gap 30 to the damage. (c) and (d) Substrate containing a 20 nM AP site was incubated for 15 min-utes at 20 �C with 0, 2, 5 or 10 nM Fpg in buffer A. Samples were analyzed as above. An example of the cleavagereaction is shown in (a). The average of the percentage of cleaved substrate and the standard error quanti®ed fromtriplicate reactions is represented graphically in (b). The control substrate (m) consists of a 54-mer with an AP site atposition 24 (strand A, see Table 1), annealed to an intact DNA strand. The MDS substrates have a nucleotide gap instrand B at position one (&), three (&) or six (F), 50 or 30 to the AP site in strand A (see Table 1 and Figure 1). Barsto the left of the control correspond to substrates with the nucleotide gap 50 to the damage in strand A and bars tothe right of the control correspond to substrates with the nucleotide gap 30 to the damage.


strand A was cleaved at the 8-oxoG (Figure 2(a))producing a 23-mer with a 30 phosphate (b,d-elim-ination product) or a 23-mer with an attached 30unsaturated aldehyde (b-elimination product). It isevident from Figure 2(a) that Fpg was able toremove 8-oxoG and catalyze a b,d-elimination reac-tion when incubated with the control substrate(strand A annealed to an intact strand B) and thesubstrates containing a nucleotide gap at positionsix. However, as the distance between the nucleo-tide gap and the damage in strand A wasdecreased, Fpg activity decreased. This is particu-larly apparent when the gap was at position one,since a drastic reduction in the level of cleavage(Figure 2(a)) was observed. Inhibition was alsodetected when the gap was at position three 50 tothe damage site in strand A, but the inhibition wasmuch less than when the gap was at position one.The d-elimination activity of Fpg was clearly inhib-ited by the gap at position one to three 30 to the 8-oxoG (Figure 2(a)), since a b-elimination product as

well as the b,d elimination product was detected.However, quanti®cation of the products revealedthat the percentage of total cleavage (b and b,dproducts) for the substrate with a gap at positionthree 30 was equivalent to the amount of b,dcleavage product observed with the controlsubstrate (Figure 2(b)).

The time-course of the Fpg reaction using thecontrol and MDS with a nucleotide gap 30 to thedamage in strand A showed that the rate of clea-vage of the substrate with a gap at position sixwas very similar to that of the control substrate,and the slowest reaction rate was detected with thesubstrate that contained a gap at position one; Fpgactivity on 8-oxoG in strand A was only slightlyreduced when the gap was at position three (datanot shown).

An 80 nM concentration of Fpg was used forsubsequent repair experiments that includedadditional enzymes. At this concentration of Fpg,with the gaps 30 to the 8-oxoG, cleavage equal to


the control was observed with gaps at the threeand six positions, while less than 40 % cleavagewas found with the gap at position one.

To measure Fpg activity on an AP site closelyopposed to a single nucleotide gap, strand A (54-mer) containing a uracil at position 24 (Table 1)was labeled with 32P and annealed to strand B (asdescribed above); the uracil was then converted toan AP site by uracil DNA glycosylase as describedin Materials and Methods. The resulting double-stranded substrates contained an AP site closelyopposed to a single nucleotide gap in the samepositions as described above for 8-oxoG. Theresults with AP sites (Figure 2(c)) were qualitat-ively similar to those obtained with 8-oxoG,although much less Fpg was required to achievethe same level of cleavage, since an AP site is thebest substrate for the enzyme (Z.H. et al., unpub-lished results). When the gap was at position six,cleavage of the AP site was as ef®cient as the con-trol while much less cleavage was observed whenthe gap was at position one (Figure 2(d)). Cleavageat the AP site, however, was more extensive thanat 8-oxoG in a similar MDS. When the gap was atposition three 30, cleavage by Fpg at the AP sitewas inhibited about 50 % even though Fpg activitywas not affected by the gap at position three 50 tothe AP site. The opposite was true when the MDScontained 8-oxoG (Figure 2(b)).

Does a decrease in the extent and rate ofreaction correspond to the ability of Fpg tobind to the MDS substrate?

Fpg has previously been shown to bindspeci®cally to a reduced AP site (Castaing et al.,1992) and to a tetrahydrofuran residue (Tchouet al., 1993) which are not substrates for theenzyme due to the absence of the aldehydegroup required for b-elimination. To examine theFpg binding to a MDS, a reduced AP (rAP) sitewas introduced into strand A 50 labeled with32P. The binding of Fpg to the rAP site closelyopposed to a nucleotide gap at position one,three or six 50 or 30 to this damaged site wascompared to a control binding substrate (strandA contained a rAP site and strand B was intact)by gel shift analysis. As can be seen inFigure 3(a), binding to the AP site by Fpg wasdrastically reduced compared to the control ifthe nucleotide gap was at position one, butbinding was not affected by the gap if it was atposition three or six 50 or 30 to the rAP site(Figure 3(b)).

Removal of 30 phosphate termini in MDS byendonuclease IV

As mentioned above, Fpg produces a nucleotidegap with 30 and 50 phosphate termini following theremoval of 8-oxoG from DNA. Therefore, therepair of a MDS containing an 8-oxoG in onestrand and a single strand break in the other

requires the removal of the 30 phosphate followingFpg cleavage and the 30 phosphate from the singlestrand break prior to extension of both 30-OHgroups by DNA polymerase I. E. coli endonucleaseIV can carry out this function. To determinewhether endonuclease IV could remove the 30phosphate at a strand break in close proximity toan 8-oxoG and whether it could complete its func-tion in a MDS in the presence of Fpg, both strandswere examined individually. This was achieved bygenerating MDS substrates with either strand A(Figure 4(a)) or strand B (Figure 4(b)) labeled foreach type of MDS substrate. Substrates containing8-oxoG on strand A and a nucleotide gap onstrand B at position one, three or six 30 to the 8-oxoG were studied.

Substrate labeled on strand A (8-oxoG)

When 80 nM Fpg and increasing amounts ofendonuclease IV were incubated with the 50labeled strand A, the 30 blocked termini (b-elimin-ation product or 30 phosphate) were converted to30 hydroxyl termini (Figure 4(a)). The conversion ofthe 30 phosphate to a 30 hydroxyl in strand A of thecontrol substrate required a greater concentrationof endonuclease IV than for the comparable reac-tion when the substrates contained a nucleotidegap at position three or six. With the gap at pos-ition one, little cleavage by Fpg was observed butwhen increasing concentrations of endonuclease IVwas added, a decrease in intact strand A and anincrease in fragmented strand A with a 30 hydroxylwas observed, albeit small.

Substrate labeled on strand B (singlenucleotide gap)

Reaction products for substrates containing an 8-oxoG or G at position 24 on strand A and a nucleo-tide gap on strand B at position one 30 to the 8-oxoG or G are shown in Figure 4(b). Here, the 50terminus of strand B was labeled with 32P andupon loss of the duplex substrate structure in thedenaturing gel, the 50 fragment of B electrophor-esed as a 29-mer with an attached 30 phosphategroup. Incubation of the control (strand A contain-ing a G residue at position 24) and MDS substratewith endonuclease IV alone resulted in removal ofthe 30 phosphate to produce a slower mobility pro-duct, a 29-mer with a 30 hydroxyl terminus. Incu-bation of the MDS substrate with Fpg alone didnot alter strand B and the removal of the 30 phos-phate by endonuclease IV occurred in the presenceor absence of Fpg. However, it took a lower con-centration of endonuclease IV to remove the 30phosphate group from strand B when Fpg waspresent in the reaction. Similar results wereobserved for the removal of the 30 phosphategroups from MDS substrates labeled on strand Bwith opposing gaps at positions three and six (datanot shown).

Figure 3. Fpg binding to areduced AP site in control andMDS substrates. Substrate (1 nM)was incubated with 0, 2, 5 or10 nM Fpg in 10 mM Tris-HCl(pH 7.5), 1 mM EDTA, 3.2% gly-cerol at 20 �C for ®ve minutes.Samples were subjected to electro-phoresis through a 15 % polyacryl-amide gel at �300 V at 4 �C. (a) Anexample of an autoradiographshowing 5 nM Fpg binding to thesubstrates in (a). (b) Binding reac-tions were performed in triplicateand the average and standarderrors are shown. The control sub-strate (m) consists of a 54-mer witha reduced AP site at position 24(strand A, see Table 1), annealed toan intact DNA strand. The MDSsubstrates have a nucleotide gap instrand B at position one (&), three(&) or six (F), 50 or 30 to the 8-oxoG in strand A (see Table 1).Bars to the left of the control corre-spond to substrates with thenucleotide gap 50 to the damage instrand A and bars to the right ofthe control correspond to substrateswith the nucleotide gap 30 to thedamage.


For subsequent experiments, a concentration of5 nM endonuclease IV was chosen. This concen-tration gave virtually complete conversion of the30-P to 30-OH in the control, the 8-oxoG, and gap-containing strands.

Insertion of ``missing'' residues in MDSsubstrates by DNA polymerase I

Following the action of Fpg and endonucleaseIV, the ``missing'' nucleotide must be inserted atthe 30 hydroxyl terminus before DNA ligase canseal the nicks and complete repair. The same sub-strates as above were examined. The action ofDNA polymerase I on strand A and strand B wasstudied separately by labeling either strand A orstrand B.


Substrates were incubated with 80 nM Fpg,5 nM endonuclease IV and increasing amounts ofDNA polymerase I. Due to the DNA sequence 30 tothe 8-oxoG on strand A (see Table 1) and becauseonly dGTP and dTTP (or dCTP) were added to thereaction as described earlier, strand A, which wasfragmented to a 23-mer with a 30 hydroxyl group

after cleavage by Fpg and endonuclease IV, wasexpected to be extended by only three dGTP resi-dues resulting in a 26-mer after denaturation. Ascan be seen in Figure 5(a), DNA polymerase I wasable to extend the 23-mer to a 26-mer in the reac-tions containing the control substrate and the MDSsubstrates that contained a gap at position three orsix 30 to the 8-oxoG. With the MDS substrate con-taining a gap at position 1, little cleavage at 8-oxoG by Fpg was observed. However extension ofall the cleaved substrate resulted in a 24-mer.Increasing the concentration of DNA polymerase Iin the presence of Fpg and endo IV also appearedto increase the ef®ciency of Fpg cleavage on thesubstrate with a gap at position 1. This was seen(Figure 5(a)) by a decrease in the intensity of the54-mer (strand A) and an increase in the intensityof the 23-mer with a 30 phosphate group.

Substrate labeled on strand B (singlenucleotide gap)

Figure 5(b) shows the products of reacting aMDS substrate containing a gap at position onewith Fpg, endonuclease IV and DNA polymerase Iwhen the 50 terminus of the 29-mer forming part ofstrand B was labeled with 32P. In the control sub-

Figure 4. Removal of 30 phosphate termini by endonuclease IV. Substrate (®nal concentration 20 nM) was added toan enzyme mix on ice, to obtain ®nal concentrations of 0, 1, 2, 5 or 7.5 nM endonuclease IV (E IV) and 0 or 80 nMFpg. Reactions were performed in buffer A or for the substrate with a nucleotide gap at position one on strand B,buffer B. After 15 minutes at 20 �C, reactions were stopped by addition of 5 ml of formamide, 0.03 % bromophenolblue, 0.03 % xylene cyanol and analyzed as described (see Materials and Methods). (a) Strand A or (b) strand B waslabeled with 32P (*). The substrates examined are shown above the lanes of the gel. Y corresponds to an 8-oxoG resi-due at position 24 on strand A and #1, 3 or 6 corresponds to the position of the nucleotide gap on strand B 30 to the8-oxoG. Substrates examined in (b) contain a nucleotide gap at position one and the 50 fragment of strand B (29-mer)was labeled with 32P. A deoxyguanosine residue was at position 24 when a Y is not indicated on the substrate.


strate, G is at position 24 on strand A; in the MDSsubstrate, 8-oxoG is at position 24 on strand A.Here, the reaction buffer contained dGTP anddCTP, which allowed DNA polymerase I to extend

a maximum of ®ve nucleotides following endonu-clease IV removal of the 30 phosphate from thelabeled 29-mer. A 34-mer was the principal pro-duct following incubation of the enzymes with the

Figure 5. Insertion of ``missing'' residues by DNA polymerase 1 after treatment with Fpg and endonuclease IV.(a) Substrate (®nal concentration 20 nM) was added to an enzyme mix on ice, which contained 0, 0.004, 0.01, 0.02 or0.027 unit of DNA polymerase I and ®nal concentrations of 0 or 80 nM Fpg, and 0 or 5 nM endonuclease IV (E IV).Strand A was labeled with 32P (*). (b) Substrate containing a nucleotide gap at position one was added (®nal concen-tration 20 nM) to an enzyme mix on ice, which contained 0, 0.004, 0.02, 0.04 or 0.06 unit of DNA polymerase I and®nal concentrations of 0 or 80 nM Fpg, and 0 or 5 nM endonuclease IV (E IV). The 50 fragment of strand B (29-mer)was labelled with 32P (*). The reaction containing 0.04 unit of DNA polymerase I for the control substrate (i.e. sub-strate containing a gap at position one on strand B, but a deoxyguanosine residue at position 24 on strand A) is notshown. Reactions were performed in buffer A or buffer B, described in the legend to Figure 4, for 15 minutes at 20 �Cand analyzed as described (see Materials and Methods). The substrates examined are shown above the lanes of thegel. Y corresponds to an 8-oxoG residue at position 24 on strand A and #1, 3 or 6 corresponds to the position of thenucleotide gap on strand B 30 to the 8-oxoG.


control substrate. In the absence of Fpg, and at thelower concentrations of DNA polymerase I (<0.02unit in the reaction), 30-mer and 31-mer productswere predominantly detected in the reactions con-taining the MDS substrate, although 34-mers wereobserved. Even at the highest level of polymerase(0.06 unit/reaction) and in the absence of Fpg, the31-mers were still observed although now the 34-

mers were more abundant. These results indicatethat the 8-oxoG was inhibitory to 30 extension ofstrand B after insertion of the nucleotide opposite8-oxoG. A similar transient inhibition of Klenowfragment extension of a C residue opposite 8-oxoGwas previously reported (Shibutani et al., 1991). Inthe presence of Fpg, and at the highest concen-tration of polymerase the major products were par-


tially extended 30, 31 and 32-mers, although 34-mers were present. The 30-mer product (generatedby the insertion of one nucleotide) most likelyre¯ects the proportion of the 8-oxoG substratecleaved by the high concentration of Fpg used(80 nM) and the subsequent loss of the full-lengthtemplate. This is supported by the prominence ofthe 30-mer, indicating insertion of only one nucleo-tide. A concentration of 0.02 unit/reaction of DNApolymerase was chosen for the reconstituted sys-tem which gave almost complete extension of thecontrol 30 hydroxyl terminus.

Similar experiments using MDS with a gap atposition three or six 30 and in buffer containingdGTP and dTTP resulted in products that did notextend to the base opposite the 8-oxoG due to therequirement for dCTP. Insertion of one or fournucleotides was expected and detected in MDSsubstrates with gaps at position three and six,respectively, and was not affected by the presenceof the 8-oxoG in strand A (see Figure 6(c) and (d)).

Complete repair of MDS substrates

For complete repair to occur in the MDS sub-strates on strand A, 8-oxoG and the resulting 30phosphate has to be removed, a G inserted at pos-ition 24, and the ``nick'' sealed by DNA ligase; onstrand B, the 30 phosphate must be removed, themissing nucleotide inserted and the ``nick'' sealed.In order to prevent the formation of a doublestrand break, strand B has to be repaired beforeFpg cleaves strand A. The MDS substrates wereadded to a mix of 80 nM Fpg, 5 nM endonucleaseIV, 0.02 unit of DNA polymerase I and 0.5 unit ofE. coli DNA ligase. As previously stated, the ®nalconcentrations of the enzymes were optimized inprior experiments and resulted in a maximal con-version of the substrate into product for eachenzyme. The results of the reactions were analyzedby electrophoresis through a denaturing gel(Figure 6) or through a native gel (Figure 7), whichallowed examination of not only the size of theproducts but also the integrity of the 54-merduplex molecule.


To examine the repair of the 8-oxoG, substratewas labeled on strand A and the reaction productselectrophoresed through a denaturing gel(Figure 6(a)). Using a phosphorimager, the amountof substrate the size of a 54-mer (complete repairproduct or uncleaved substrate) or the sizes of 24to 26-mers (the unligated DNA polymerase I exten-sion products) were quanti®ed and the percentageof the 54-mer determined. Without DNA ligase,�19 % of strand A of the control substrate was stilla 54-mer; this increased to �63 % when DNAligase was added to the reaction, demonstratingthat complete repair of a proportion of the controlsubstrate was achieved. Addition of DNA ligase toreactions containing the MDS substrates did not

increase the percentage of strand A at the size of a54-mer, indicating that complete repair did nottake place. Addition of DNA ligase to the MDSsubstrate with a nucleotide gap at position onedecreased the amount of intact strand A (54-mer)and increased the 24-mer strand A.

Substrate labeled on strand B (one nucleotide gap)

In order to examine the repair of the nucleotidegap in the MDS, strand B was labeled andannealed to form the MDS or control (G at position24 on strand A) substrates. The percentage ofstrand B at the size of a 54-mer (complete repairproduct) or at the size of the unligated DNA poly-merase I extension products for each substrate wasdetermined as described above for strand A. In theabsence of Fpg, 79 % of strand B was repaired inthe MDS substrate with a gap at position one,which was similar to the repair of the gap if G wasat position 24 in strand A (Figure 6(b)). Similarresults were found in the absence of Fpg for therepair of strand B in MDS substrates with a gap atposition three (Figure 6(c)) or six (Figure 6(d)). Inthe presence of Fpg, however, DNA ligase wasonly able to signi®cantly complete repair of strandB when the gap was one nucleotide away andcleavage of strand A by Fpg was inhibited. Withgaps at positions three or six, very few moleculeswere completely repaired to a 54-mer. Also, thelevel of complete repair of strand B in the presenceof Fpg (see, for example, Figure 6(c) for gaps atposition three opposite 8-oxoG) was approximatelyequal to the amount of uncleaved substrate instrand A, about 20 % in this case. Therefore theability of the enzymes to repair strand B wasdependent on the inability of Fpg to cleave the 8-oxoG on strand A.

Formation of double strand breaks

To determine whether incomplete repair resultedin fragmentation of the duplex substrate, the reac-tion products of strand A substrates were subjectedto electrophoresis through a native gel (Figure 7).Incubation of the control substrate with Fpg(Figure 7(a)) resulted in the detection of twobands. One band corresponded to the Fpg-boundsubstrate and the other exhibited a slightly slowermobility than the intact substrate, which webelieve to be Fpg-cleaved substrate with a nucleo-tide gap. Reaction of the MDS substrates with Fpg(Figure 7(a) and (b)) resulted in an extra productwhich had faster mobility than the substrate, but aslower mobility than a single-stranded 29-mer witha 30 phosphate terminus, the expected products ofa double strand break. Fpg alone therefore resultedin the generation of a double strand break.

Addition of Fpg, endonuclease IV, and DNApolymerase I decreased the amount of control sub-strate in the shifted position as compared to Fpgalone; the products had a similar mobility to adouble-stranded 54-mer (Figure 7(a)). However,

Figure 6 (a) and (b)


Figure 6 (Legend shown on opposite page)



similar reaction with MDS substrates did not resultin an increase of duplex 54-mer product. In thecase of substrates with a gap at position three orsix, the products were all at the position of doublestrand break products, and treatment of substratewith a gap at position one also resulted in anincrease of the fragmented product. Analysisunder denaturing conditions of a similar reactionalso showed an increase in fragmentation of strandA (Figure 5).

The addition of ligase in all cases resulted inmultiple bands with slower mobilities than theduplex 54-mer (Figure 7(a) and (b)). It is possiblethat these bands represent the formation ofmultiple DNA-protein complexes or a low level ofligated oligomeric substrates. However, a pro-portion of the MDS substrate can still be seen atthe size of fragmented substrate.

Discussion

Previous studies have demonstrated that ioniz-ing radiation can introduce multiply damaged sitesin DNA (Goodhead, 1994; Goodhead et al., 1993),and that the increased lethality of ionizing radi-ation compared to other oxidizing agents may beexplained by the decreased ability of the MDS tobe repaired (Ward et al., 1987). It has beensuggested that the initiation of the repair of MDSby the base excision repair pathway introduces asingle strand break at AP sites or base damages,and results in the formation of a double strandbreak at the MDS. Double strand breaks areformed in plasmid DNA containing closelyopposed uracil residues incubated in vitro withE. coli extracts (Dianov et al., 1991). Transformationof E. coli with plasmids containing a single uracilbase or two closely opposed uracil bases resultedin a deletion of sequences when the uracil baseswere ¯anked by direct repeats. The deletion fre-quency was tenfold higher for plasmids containingopposing uracil residues and greater in ung�

E. coli (wild-type genotype for uracil DNA glycosy-lase) compared to ungÿ E. coli, suggesting thatinitiation of base excision repair of opposed uracilbases in vivo by uracil DNA glycosylase results ina double strand break (Dianov et al., 1991). Calcu-lations of the production of base damages by ioniz-ing radiation have estimated that �2.5 � 103 basedamages are produced per cell per gray, and thatthis number is �20 times greater than the level ofsingle strand breaks (Ward, 1995). A MDS consist-

Figure 6. Reconstitution of the base excision repair pathwbuffer A or buffer B as described in the legend to Figure 4. Wby �) the ®nal concentrations were 80 nM Fpg, 5 nM endonof E. coli DNA ligase. The strand labeled with 32P is indicatand then incubated for 15 minutes at 20 �C. (a) The produ(b), (c) and (d) The products of the reaction for substrates wively. The substrates examined are shown above the lanes o24 on strand A. A deoxyguanosine residue was at position 2

ing of two opposing base damages or a basedamage closely opposed to a single strand break istherefore a signi®cant and biologically relevantMDS lesion produced by ionizing radiation. Pre-viously we demonstrated that two closely opposedbase damages could be converted to a basedamage opposite a single strand break (Harrisonet al., 1998). We therefore chose an 8-oxoG opposedto a single strand break as a substrate for the studyof complete repair of a MDS lesion.

The repair of a MDS in vivo requires a DNA gly-cosylase/AP lyase, a 30 repair diesterase, a DNApolymerase and a DNA ligase. In the case of aMDS such as a SSB opposed to a base damage, it ispossible that the SSB could be repaired before theintroduction of a strand break at the site of thebase damage by the glycosylase/AP lyase. Byreconstituting the base excision repair pathway, wehave extended our initial studies on the processingof MDS by the ®rst enzyme in the reaction, a DNAglycosylase/AP lyase, to an examination of howeach of the stages of repair of a MDS is affected bythe complex lesion and whether complete repaircan occur in vitro.

We demonstrated that Fpg is able to initiaterepair of an 8-oxoG or AP site if a nucleotide gapis positioned in the opposite strand 5three nucleo-tides away. It was evident, however, that Fpg cata-lysis of the d-elimination reaction was inhibited ifthe nucleotide gap was situated within threenucleotides and 30 to the 8-oxoG or AP site andwithin one nucleotide 50 to the AP site. Endo VIIIactivity has also previously been shown to beaffected not only by the distance, but also theorientation of the nucleotide gap with respect tothe target lesion (Harrison et al., 1998). Thisappeared to be due to disruption of endo VIIIbinding, as suggested by the gel shift analysis ofendo VIII with a reduced AP site in the MDS(Harrison et al., 1998), and the footprint whichshowed endo VIII binding predominantly to the 30side of a reduced AP site (Jiang et al., 1997). Similargel shift analysis for Fpg binding to a reduced APsite also indicated (Figure 3) that Fpg demon-strated diminished binding if the nucleotide gapwas at position one 50 or 30 to the reduced AP site,and may explain why Fpg cleavage at 8-oxoG oran AP site in a similar MDS was drasticallyreduced compared to the control substrate. Bind-ing, however, was not perturbed by the gapsituated at position three. It is unlikely, therefore,that reduced binding of Fpg to the initial target

ay. Reactions were performed using 20 nM substrate inhen the enzymes were present in the reaction (indicateduclease IV, 0.02 unit of DNA polymerase I and 0.5 united by *. Substrate was added to the enzyme mix on icects of the reaction when strand A is labeled with 32P.ith nucleotide gaps at position one, three or six, respect-f the gel. Y corresponds to an 8-oxoG residue at position4 when a Y is not indicated on the substrate.

Figure 7. Examination of the double-stranded structure of the complete repair products. Reactions were performedas described in Figure 6. The ®nal enzyme concentrations were 80 nM Fpg, 5 nM endonuclease IV, 0.02 unit of DNApolymerase I and 0.5 unit of E. coli DNA ligase. The strand labeled with 32P is indicated by *. Reaction products werevisualized by autoradiography following electrophoresis through a 15 % (w/v) polyacrylamide gel. (a) The productsof the reactions for the control substrate (8-oxoG on strand A annealed to an intact strand B) and the MDS with anucleotide gap in strand B at position one 30 to 8-oxoG. Lanes labeled 29 (30P) and 54 contained labeled single-stranded 29 and 54-mers respectively. (b) The products of the reactions for MDS substrates with a nucleotide gap instrand B at position three or six 30 to 8-oxoG. The lane labeled 54 contains a labeled single-stranded 54-mer.


Figure 8. Cartoon depicting base excision repair pro-cessing of MDS containing 8-oxoG closely opposed to asingle strand break.


lesion in the MDS accounted for the inhibition ofthe d-elimination step of Fpg activity.

Following Fpg cleavage, a 30 phosphate terminusis generated (strand A) similar to the 30 terminuson the opposing strand (B) at the single strandbreak. Both of these phosphate groups have to beremoved before polymerase can ®ll the gap. In thissystem, endonuclease IV, a 30 repair diesterase,was examined. Removal of the 30 phosphate fromstrand B by endonuclease IV was not inhibited bythe 8-oxoG in strand A (Figure 4(b)) or the posi-tioning of the nucleotide gap (one, three or sixnucleotides 30 to the 8-oxoG). However, in the pre-sence of Fpg, less endonuclease IV was required toremove the 30 phosphate group from strand B.There are at least two explanations to account forthese observations. Either the double strand breakformed following Fpg cleavage of strand Aallowed endonuclease IV greater access to the 30phosphate termini, or endonuclease IV bound tothe MDS substrate at the 30 phosphate group onstrand A, could more readily remove the 30 phos-phate group on strand B as the enzyme was inclose proximity to this second 30 phosphate group.However, it cannot be ruled out that Fpg andendonuclease IV together act more ef®ciently thanwhen examined individually. Fpg activity didappear to cleave strand A to a greater extent at theMDS with the nucleotide gap at position one asthe endonuclease IV concentration increased(Figure 4(a)). In fact, an increase in Fpg activity(albeit small) was also seen upon addition of DNApolymerase I and DNA ligase (Figure 6(b)).

E. coli DNA polymerase I was added to the reac-tion to insert the ``missing'' nucleotide in strand Aand strand B. Extension greater than one nucleo-tide was observed due to the nick translationactivity of this enzyme when the available dNTPand sequence context allowed, except for the sub-strate with a gap at position one 30 to the 8-oxoG.For this substrate, the predominant product evenat the highest DNA polymerase I concentrationwas insertion of one nucleotide to strand A and B,indicating that a double strand break had formedin that proportion of the substrate cleaved by Fpg.In the absence of Fpg, the 8-oxoG caused a slightinhibition to extension of strand B 30 to the C resi-due inserted opposite the 8-oxoG (Figure 5(b)).A similar transient inhibition has been previouslyreported for Klenow fragment (Shibutani et al.,1991).

To complete repair, E. coli DNA ligase was usedto seal the nicks remaining in strands A and B.Figure 6 shows that a high proportion of the con-trol substrates (for repair of strand A, an 8-oxoG instrand A annealed to an intact strand B and forrepair of strand B, a nucleotide gap on strand Bwith a G residue at position 24 on strand A) wascompletely repaired; however, MDS substrates thatwere ef®ciently cleaved by Fpg were not. Thestrand containing a gap at position one in the MDSwas repaired, and the amount of repair observedwas equivalent to the proportion of substrate that

was not cleaved by Fpg. Examination of the reac-tions on a native gel revealed that when Fpg wasable to cleave at the 8-oxoG, a double strand breakwas formed. In general, when the closely opposedgap was positioned three or more nucleotidesaway from 8-oxoG, Fpg cleaved the 8-oxoG-con-taining strand and a double strand break wasformed (Figure 8). In contrast, when the gap wasonly one nucleotide away from the 8-oxoG, the 8-oxoG was a poor substrate for Fpg and cleavageonly occurred at high concentrations of the enzyme(Figure 2). If cleavage did not occur, the gap-con-taining strand could be repaired and double strandbreaks were not formed as depicted in Figure 8.

This in vitro system does not take into accountthe chromatin structure in mammalian cells and itis possible that the proteins bound to DNA mayhold the cohesive breaks together for the timerequired for DNA repair. However, cellular studiesdo exist to support the hypothesis that attemptedrepair of MDS can lead to double strand breaks.Double strand breaks increase in irradiated cellsafter they are allowed time to repair, and over-


expression of E. coli endonuclease III in Chinesehamster ovary double strand break repair de®cientcells, xrs-7 (Harrison et al., 1992) enhances thesensitivity of the cells to bleomycin sulfate, alsoknown to introduce MDS.

It has been demonstrated that ionizing radiationDNA damage is preferentially repaired in the tran-scribed strand of an active human gene (Leadon &Cooper, 1993) and that the base excision repair ofthymine glycol is coupled to transcription (Cooperet al., 1997), but other oxidative lesions have notbeen studied. In a cell, transcription-coupled repair(TCR) of oxidative DNA damage could alter theproduct of the repair of a MDS. The biologicaleffect of TRC on MDS repair will depend onwhether it is a phenomenon common to all oxi-dative DNA damage.

When considering the lethality of a DNA lesionit is necessary not just to consider what type ofDNA lesion exists, i.e. a base damage or an APsite, but also how the lesions are distributed in theDNA. A simple base damage such as 8-oxoG maybe mutagenic on its own, but in a MDS it can beconverted into a lethal lesion. This new class oflesion called multiple damaged sites has now beenadded to the list of potentially lethal DNA lesions.

Materials and Methods

Oligonucleotides

Oligonucleotides (see Table 1) containing uracil, 8-oxoG or 50 and 30 phosphate termini were synthesized inthe Department of Microbiology and Molecular Genetics,University of Vermont, or purchased from Operon Tech-nologies. The oligonucleotides were puri®ed by electro-phoresis in a 12 % polyacrylamide, 7 M urea gel,electroeluted from the excised gel fragment into 8 Mammonium acetate, and desalted using a NAP 52

column (Pharmacia).

Enzymes

Formamidopyrimidine DNA-glycosylase and endonu-clease IV were puri®ed using procedures described(Hatahet et al., 1994a,b; Melamede et al., 1994). UracilDNA-glycosylase (UDG) and E. coli DNA polymerase Iwere purchased from USB and phage T4 polynucleotidekinase (lacking 30 phosphatase activity) and E. coli DNAligase were purchased from Boehringer Mannheim.

Preparation of duplex substrates

Oligonucleotides (3-5 pmol) were 50 labeled with 32Pusing one unit of phage T4 polynucleotide kinasethat was de®cient in 30 phosphatase activity and50 mCi of [g-32P]ATP (6000 Ci/mmol; 10 mCi/ml; NENDupont) in 25 ml of the supplied reaction buffer at 37 �Cfor 30 minutes. Unincorporated [g-32P]ATP was removedfollowing puri®cation of the oligonucleotide using aNENSORB2 20 cartridge (NEN Dupont). The labeled oli-gonucleotide was eluted in 50 % (v/v) ethanol, dried andresuspended in double-distilled water at 100 fmol/ml. Inorder to obtain substrate with a speci®c activity of �100-200 dpm/fmol, the labeled strand A (or B), was mixedwith unlabeled strand A (or B) in a ration of 1:10 or 1:20.

This was then annealed to 1.2-fold molar excess of theunlabeled complementary strand (Table 1) to produceduplex molecules at concentrations of 100-200 fmol/ml in50 mM Tris-HCl (pH 7.5), 50 mM NaCl, 0.1 mM EDTA,5 mM MgCl2, 1 mM DTT, 20 mM dGTP and either20 mM dTTP (buffer A) or 20 mM dCTP (buffer B) (seethe individual experiments for buffer content of dTTP ordCTP). To generate duplex molecules with a singlenucleotide gap, two oligonucleotides were used to formstrand B; both oligonucleotides were at 1.2-fold molarexcess of strand A if strand A was labeled with 32P. Thehybridization reaction was heated to 75 �C for approxi-mately two minutes and slowly cooled to room tempera-ture. This procedure was carried out over approximatelytwo hours.

To obtain a labeled AP site containing double-stranded substrate, strand A containing uracil waslabeled, puri®ed and annealed to strand B as describedabove at 200 fmol/ml of duplex DNA. UDG (0.5 unit)was incubated with 1 pmol of duplex DNA in 5.5 ml ofbuffer A for 30 minutes at 37 �C. To determine if the ura-cil had been removed, a small aliquot of the resultingsubstrate was boiled for 30 minutes and the productsvisualized after electrophoresis through a 12 % polyacryl-amide, 7 M urea gel. Approximately 90 % of the sub-strate was fragmented after boiling.

Fpg cleavage reactions

Duplex substrate (20 nM ®nal concentration) wasmixed with Fpg (2-80 nM) in 5 ml of buffer A on ice andthen incubated at 20 �C for 15 minutes. Reactions werestopped on ice by the addition of 5 ml of formamide,0.03 % (w/v) bromophenol blue, 0.03 % (w/v) xylenecyanol. To examine the rate of the cleavage reaction,duplex substrate (100 fmol) in the above reaction bufferwas pre-warmed to 20 �C and the reaction started by theaddition of Fpg to a ®nal concentration 0, 5 or 10 nM.Reactions were performed at 20 �C for 0.5-15 minutesand stopped by the addition of 5 ml of formamide,0.03 % bromophenol blue, 0.03 % xylene cyanol. Sampleswere subjected to electrophoresis through a 12 % poly-acrylamide, 7 M urea gel at 1500 V for �two hours in1 � TBE (89 mM Tris, 89 mM boric acid, 2 mM EDTA(pH 8)) and then dried. The reaction products werevisualized by autoradiography and quanti®ed using aBIO-RAD Molecular Imager.

Preparation of duplex binding substrates

Duplex binding substrates containing a reduced APsite at position 24 on strand A and either an intact strandB or a strand B containing a nucleotide gap at positionone, three or six 50 or 30 to the reduced AP site were pre-pared as described (Harrison et al., 1998).

Gel electrophoretic mobility shift analysis

Reduced AP site-containing duplex substrates (1 nM®nal concentration) were mixed on ice with Fpg(2-10 nM ®nal concentration) in 5 ml of 10 mM Tris-HCl(pH 7.5), 1 mM EDTA, 50 mM NaCl, 3.2 % (v/v) glycer-ol and incubated at 20 �C for ®ve minutes. After theaddition of 1 ml of 20 % (v/v) glycerol, 125 mM Tris-HCl(pH 7.5), the samples were subjected to electrophoresisthrough a 15 % polyacrylamide gel containing 2.5 % (v/v) glycerol and 0.5 � TBE; 0.5 � TBE was used for the


buffer and electrophoresis was performed at 4 �C and200-300 V for �three hours.

Repair reactions

The control substrate and the substrates with a nucleo-tide gap on strand B at position one, three and six 30 tothe 8-oxoG in strand A were incubated in separate reac-tions with 1-7.5 nM endonuclease IV, 5 nM endonucleaseIV plus 0.004-0.06 unit of E. coli DNA polymerase I, and5 nM endonuclease IV, plus 0.02 unit of E. coli DNApolymerase I and 0.5 unit of E. coli DNA ligase in thepresence and/or absence of 80 nM Fpg. The repairenzymes were mixed on ice in 4 ml of buffer A, exceptfor the substrate with the nucleotide gap at position one30 to 8-oxoG where buffer B was used. For reactionsincluding E. coli DNA ligase, NAD was added to a ®nalconcentration of 1 mM. The reaction was started by theaddition of 100 fmol duplex substrate and the sampleswere incubated at 20 �C. After 15 minutes, the reactionswere placed on ice. Samples were either stopped by theaddition of 5 ml of formamide, 0.03 % bromophenol blue,0.03 % xylene cyanol and analyzed as described for theFpg cleavage reactions or, to examine the duplex struc-ture of the reaction products, 1 ml of 20 % (v/v) glycerol,125 mM Tris-HCl (pH 7.5) was added before electro-phoresis through a 15 % polyacrylamide gel containing2.5 % glycerol and 0.5 � TBE; 0.5 � TBE was used for thebuffer and electrophoresis was performed at 4 �C at 200-300 V for �three hours. The gel was dried and visualizedby autoradiography.

Acknowledgments

This work was supported by National Institutes ofHealth grant R37 CA22657 awarded by the NationalCancer Institute. L.H. was the recipient of a LakeChamplain Cancer Research Organization PostdoctoralFellowship awarded by the Vermont Cancer Center. Theauthors are grateful to Dr Michael Weinfeld for helpfuldiscussions.

References

Ahnstrom, G. & Bryant, P. E. (1982). DNA double-strand breaks generated by the repair of X-raydamage in Chines hamster cells. Int. J. Radiat. Biol.41, 671-676.

Bailly, V., Verly, W. G., O'Connor, T. R. & Laval, J.(1989). Mechanism of DNA strand nicking at apuri-nic/apyrimidinic sites by Escherichia coli [formami-dopyrimidine] DNA glycoslyase. Biochem. J. 262,581-589.

Bhagwat, M. & Gerlt, J. A. (1996). 30- and 50-strand clea-vage reactions catalyzed by the Fpg protein fromEscherichia coli occur via successive beta- and delta-elimination mechanisms, respectively. Biochemistry,35, 659-665.

Bonura, T., Smith, K. C. & Kaplan, H. S. (1975). Enzy-matic induction of DNA double-strand breaks ingamma-irradiated E. coli K-12. Proc. Natl Acad. Sci.USA, 72, 4265-4269.

Brenner, D. J. & Ward, J. F. (1992). Constraints onenergy deposition and target size of multiplydamaged sites associated with DNA double-strandbreaks. Int. J. Radiat. Biol. 61, 737-748.

Castaing, B., Boiteux, S. & Zelwer, C. (1992). DNA con-taining a chemically reduced apurinic site is a highaf®nity ligand for the E. coli formamidopyrimidine-DNA glycosylase. Nucl. Acids Res. 20, 389-394.

Castaing, B., Geiger, A., Seliger, H., Nehls, P., Laval, J.,Zelwer, C. & Boiteux, S. (1993). Cleavage and bind-ing of a DNA fragment containing a single 8-oxo-guanine by wild type and mutant FPG proteins.Nucl. Acids Res. 21, 2899-2905.

Chaudhry, M. A. & Weinfeld, M. (1995a). The action ofEscherichia coli endonuclease III on multiplydamaged sites in DNA. J. Mol. Biol. 249, 914-922.

Chaudhry, M. A. & Weinfeld, M. (1995b). Induction ofdouble-strand breaks by S1 nuclease, mung beannuclease and nuclease P1 in DNA containing abasicsites and nicks. Nucl. Acids Res. 23, 3805-3809.

Chaudhry, M. A. & Weinfeld, M. (1997). Reactivity ofhuman apurinic/apyrimidinic endonuclease andEscherichia coli exonuclease III with bistrandedabasic sites in DNA. J. Biol. Chem. 272, 15650-15655.

Cooper, P. K., Nouspikel, T., Clarkson, S. G. & Leadon,S. A. (1997). Defective transcription-coupled repairof oxidative base damage in Cockayne syndromepatients from XP group G. Science, 275, 990-993.

Demple, B. & Harrison, L. (1994). Repair of oxidativedamage to DNA: enzymology and biology. Annu.Rev. Biochem. 63, 915-948.

Dianov, G. & Lindahl, T. (1994). Reconstitution of theDNA base excision-repair pathway. Curr. Biol. 4,1069-1076.

Dianov, G. L., Timchenko, T. V., Sinitsina, O. I.,Kuzminov, A. V., Medvedev, O. A. & Salganik, R. I.(1991). Repair of uracil residues closely spaced onthe opposite strands of plasmid DNA results indouble-strand break and deletion formation. Mol.Gen. Genet. 225, 448-452.

Dugle, D. L., Gillespie, C. J. & Chapman, J. D. (1976).DNA strand breaks, repair, and survival in X-irra-diated mammalian cells. Proc. Natl Acad. Sci. USA,73, 809-812.

Goodhead, D. T. (1994). Initial events in the cellulareffects of ionizing radiations: clustered damage inDNA. Int. J. Radiat. Biol. 61, 7-17.

Goodhead, D. T., Thacker, J. & Cox, R. (1993). Effects ofradiations of different qualities on cells: molecularmechanisms of damage and repair. Int. J. Radiat.Biol. 63, 543-556.

Harrison, L., Skorvaga, M., Cunningham, R. P., Hendry,J. H. & Margison, G. P. (1992). Transfection of theEscherichia coli nth into radiosensitive Chinesehamster cells: effects on sensitivity to radiation,hydrogen peroxide, and bleomycin sulfate. Radiat.Res. 132, 30-39.

Harrison, L., Hatahet, Z., Purmal, A. A. & Wallace, S. S.(1998). Multiply damaged sites in DNA: interactionswith Escherichia coli endonucleases III and VIII.Nucl. Acids Res. 26, 932-941.

Hatahet, Z., Kow, Y. W., Purmal, A. A., Cunningham,R. P. & Wallace, S. S. (1994a). New substratesfor old enzymes. 5-Hydroxy-20-deoxycytidine and5-hydroxy-20-deoxyuridine are substrates forEscherichia coli endonuclease III and formamidopyri-midine DNA N-glycosylase, while 5-hydroxy-20-deoxyuridine is a substrate for uracil DNA N-glyco-sylase. J. Biol. Chem. 269, 18814-18820.

Hatahet, Z., Purmal, A. A. & Wallace, S. S. (1994b).DNA Damage Effects on DNA Structure and ProteinRecognition (Wallace, S. S., Van Houten, B. & Kow,Y. W., eds), vol. 726, pp. 346-348.


Henner, W. D., Rodriguez, L. O., Hecht, S. M. &Haseltine, W. A. (1983). g-Ray induced deoxyribo-nucleic acid strand breaks. J. Biol. Chem. 258, 711-713.

Jiang, D., Hatahet, Z., Melamede, R. J., Kow, Y. W. &Wallace, S. S. (1997). Characterization of Escerichiacoli endonuclease VIII. J. Biol. Chem. 272, 32230-32239.

Klett, R. P., Cerami, A. & Reich, E. (1968). ExonucleaseVI, a new nuclease activity associated with E. coliDNA polymerase. Proc. Natl Acad. Sci. USA, 60,943-950.

Kubota, Y., Nash, R. A., Klungland, A., Schar, P.,Barnes, D. E. & Lindahl, T. (1996). Reconstitution ofDNA base excision-repair with puri®ed human pro-teins: interaction between DNA polymerase betaand the XRCC1 protein. EMBO J. 15, 6662-6670.

Leadon, S. A. & Cooper, P. K. (1993). Preferential repairof ionizing radiation-induced damage in the tran-scribed strand of an active human gene is defectivein Cockayne syndrome. Proc. Natl Acad. Sci. USA,90, 10499-10503.

Melamede, R. J., Hatahet, Z., Kow, Y. W., Ide, H. &Wallace, S. S. (1994). Isolation and characterizationof endonuclease VIII from Escherichia coli. Biochemis-try, 33, 1255-1264.

O'Connor, T. R. & Laval, J. (1989). Physical associationof the 2,6-diamino-4-hydroxy-5-(N-methyl) forma-midopyrimidine-DNA glycosylase of Escherichia coli

and an activity nicking DNA at apurinic/apyrimidi-nic sites. Proc. Natl Acad. Sci. USA, 86, 5222-5226.

Shibutani, S., Takeshita, M. & Grollman, A. P. (1991).Insertion of speci®c bases during DNA synthesispast the oxidation-damaged base 8-oxodG. Nature,349, 431-434.

Tchou, J., Kasia, H., Shibutani, S., Chung, M. H., Laval,J., Grollman, A. P. & Nishimura, S. (1991). 8-oxo-guanine (8-hydroxyguanine) DNA glycosylase andits substrate speci®city. Proc. Natl Acad. Sci. USA,88, 4690-4694.

Tchou, J., Michaels, M. L., Miller, J. H. & Grollman, A. P.(1993). Function of the zinc ®nger in Escherichia coliFpg protein. J. Biol. Chem. 268, 26738-26744.

Wallace, S. S. (1997). Oxidative damage to DNA and itsrepair. In Oxidative Stress and the Molecular Biology ofAntioxidant Defenses (Scandalios, J., ed.), pp. 49-90,Cold Spring Harbor Laboratory Press, Cold SpringHarbor, NY.

Ward, J. F. (1988). DNA damage produced by ionizingradiation in mammalian cells: identities, mechan-isms of formation, and reparability. Prog. Nucl. AcidRes. Mol. Biol. 35, 95-125.

Ward, J. F. (1995). Radiation mutagenesis: the initialDNA lesions responsible. Radiat. Res. 142, 362-368.

Ward, J. F., Evans, J. W., Limoli, C. L. & Calabro-Jones,P. M. (1987). Radiation and hydrogen peroxideinduced free radical damage to DNA. Cancer, 55,105-112.

Edited by J. H. Miller

(Received 9 November 1998; received in revised form 4 May 1999; accepted 6 May 1999)

in vitro repair of synthetic ionizing radiation-induced multiply damaged dna sites

Documents