Characterization of the Action of Escherichia coli DNA Polymerase I at Incisions Produced by Repair Endodeoxyribonucleases*

The utilization for DNA synthesis by Escherichia coli DNA polymerase I of damaged circular duplex DNAs which had been incised by various endodeoxyribonu- cleases was studied. E. coli endonuclease 111 cleaves at apurinic/apyrimidinic (AP) lesions, generating 3‘-de- oxyribose and 5’-phosphornonoester termini. Whereas these 3’ termini did not efficiently prime DNA synthesis, they could be activated to support efficient DNA synthesis by incubation with E. coli DNA polymerase I in the absence of deoxyribonucleoside triphosphates. Such maximally activated termini achieved a similar rate of DNA synthesis as termini from which the 3’” site had been removed by prior exposure to human fibroblast AP endonuclease II. In all cases, the product of DNA synthesis was covalently attached to the primer and stable to alkali, indicating that during the preincubation, E. coli DNA polymerase I could remove the 3”deoxyribose termini. Indeed, deoxyribose 5-phos-phate was observed after such a reaction. Activation by the DNA polymerase also was detected with W-irradiated DNA which had been incised by T4 W en- donuclease so as to contain nicks with 3‘-deoxyribose termini and thymine:thymidylate cyclobutane dimer 5’ termini. Cleavage of depurinated DNA by E. coli endonuclease IV, human fibroblast AP endonuclease 11, or HeLa AP endonuclease generates deoxyribose 5’-phos-phate

The utilization for DNA synthesis by Escherichia coli DNA polymerase I of damaged circular duplex DNAs which had been incised by various endodeoxyribonucleases was studied. E. coli endonuclease 111 cleaves at apurinic/apyrimidinic (AP) lesions, generating 3'-deoxyribose and 5'-phosphornonoester termini. Whereas these 3' termini did not efficiently prime DNA synthesis, they could be activated to support efficient DNA synthesis by incubation with E. coli DNA polymerase I in the absence of deoxyribonucleoside triphosphates. Such maximally activated termini achieved a similar rate of DNA synthesis as termini from which the 3'" site had been removed by prior exposure to human fibroblast AP endonuclease II. In all cases, the product of DNA synthesis was covalently attached to the primer and stable to alkali, indicating that during the preincubation, E. coli DNA polymerase I could remove the 3"deoxyribose termini. Indeed, deoxyribose 5-phosphate was observed after such a reaction. Activation by the DNA polymerase also was detected with Wirradiated DNA which had been incised by T4 W endonuclease so as to contain nicks with 3'-deoxyribose termini and thymine:thymidylate cyclobutane dimer 5' termini. Cleavage of depurinated DNA by E. coli endonuclease IV, human fibroblast AP endonuclease 11, or HeLa AP endonuclease generates deoxyribose 5'-phosphate and 3'-hydroxyl nucleotide termini that efficiently supported DNA synthesis. Such synthesis occurred predominantly by strand displacement as determined by electron microscopy. However, when the deoxyribose 5-phosphate was removed from the incision site, DNA synthesis occurred primarily by nick translation. Similarly, DNA synthesis at incisions made either by Neurospora c m s a nuclease or by T4 W endonuclease plus the human fibroblast AP endonuclease I1 (upon W-irradiated DNA) occurred predominantly by nick translation. E. coli DNA polymerase I large fragment was very inefficient in utilizing duplex circles containing nicks or one-nucleotide gaps and it did not remove 3'-deoxyribose from termini in such circles.
Escherichia coli DNA polymerase I is a single polypeptide of molecular weight 109,000 (1) which may take part in excision repair of damaged DNA (2)(3)(4). In addition to its DNA polymerase activity, the enzyme has 3' + 5' and 5' 4 3' exonuclease activities (5,6). The two exonuclease functions are distinct and separable from one another by proteolytic cleavage of the intact enzyme to produce a large fragment (76,000 daltons) containing the polymerase, and 3' + 5' exonuclease activities and a small fragment (36,000 daltons) which contains the 5' + 3' exonuclease activity (7-10). The 3' + 5' exonuclease has been suggested to increase the fidelity of DNA synthesis through a "proofreading" function which removes misincorporated nucleotides (5). The 5' + 3' exonuclease hydrolyzes DNA from a 5' terminus to produce 5'-mononucleotides and oligonucleotides (1 1). Deoxyribonucleoside triphosphates stimulate this exonuclease so as to coordinate polymerization and 5' + 3' exonucleolytic degradation to result in "nick translation" (11). The excision of thymine dimers by the 5' + 3' exonuclease from UV-irradiated homopolymers (3) and from UV-irradiated DNA incised by T4 UV endonuclease (12) has been reported. Repair synthesis via nick translation might thus be initiated by a repair endonuclease cleaving one strand of a DNA duplex at o r near a DNA lesion.
One type of DNA damage, apurinic or apyrimidinic (AP)' lesions, accounts for a substantial amount of DNA damage, even under normal growth conditions (13,14). AP lesions arise either spontaneously or by virtue of the action of specific DNA N-glycosylases which initiate repair of certain types of DNA damage by base removal to initiate "base excision repair" (14). Two classes of endonucleases have been described which recognize AP sites. Class I AP endonucleases cleave on the 3"side of the AP site forming 3'-deoxyribose and 5'-phosphomonoester termini ( Fig. 1) which are not efficient primers for DNA polymerase I (15). Class I AP endonucleases include human fibroblast AP endonuclease I (15), E. coli endonuclease 111, and T4 UV endonuclease (16)(17)(18). The T4 UV endonuclease and the analogous Micrococcus luteus UV endonuclease also incise DNA at pyrimidine dimers by the combined action of a pyrimidine dimer DNA N-glycosylase and a Class I AP endonuclease (17)(18)(19) to generate 3'-deoxyribose and 5'-phosphomonoester thymine dimer termini (Fig.  1). Incisions that contain 3'-AP termini are generally poor primers for DNA polymerase I (15,16).
Class I1 AP endonucleases cleave on the 5'-side of AP lesions producing 3'-hydroxyl nucleotide and deoxyribose 5'phosphate termini (Fig. 1). Such enzymes include human fibroblast AP endonuclease I1 (15), HeLa AP endonuclease (20), and E. coli endonucleases IV and VI (16). Incisions made by Class 11 AP endonucleases are efficient primers for DNA polymerase I. In addition, Class I1 AP endonucleases can activate Class I endonuclease incisions to support efficient DNA synthesis by DNA polymerase I through the removal of deoxyribose 5-phosphate from the AP site to produce a one-nucleotide gap containing 3'-hydroxyl-and 5'-phosphomonoester termini (Fig. 1). When T4 UV endonuclease incisions are activated by a Class I1 AP endonuclease, an analogous reaction is suspected to occur (Fig. 1).
In the present report, t h e ability of DNA polymerase I t o activate Class I AP endonuclease incisions by virtue of its 3' "$5' exonuclease activity is examined. Whether DNA synthesis occurs via nick translation or strand displacement from various types of incisions made into damaged DNA is also investigated. Finally, the ability of t h e DNA polymerase I large fragment to substitute for the intact enzymes in reactions carried out on various incised circular duplex DNA primer/ templates is investigated.

Materials
Supercoiled PM2 [3H]DNA (93% Form I) was isolated from phage grown on Alteromonas espejiana thymidine auxotroph Bal 31-14 as described by Espejo and Canelo (21) and modified by Kuhnlein et al. (22). Nonradioactive PM2 DNA was isolated similarly except that A . espejiana wild type Bal 31 was used. Activated salmon sperm DNA was prepred according to the procedure of Schlabach et al. (23); poly(dA-["HIdT) was synthesized as described by Schachman et al. (24); and unlabeled poly(dA-dT) and its tri-and tetraoligomers were obtained from P-L Laboratories. Unlabeled deoxyribonucleoside triphosphates and monophosphates were purchased from Sigma.  (15) were prepared as previously described. Neurospora crassa endonuclease (Fraction IX) was isolated by the method of Linn and Lehman (27) and T4 UV endonuclease (Fraction IV) was as described by Friedberg and King (28). E. coli endonuclease IV was a gift from Dr. Tomas Lindahl (University of Goteborg, Sweden). E. coli DNA polymerase I (5050 units/mg) was either purchased from Worthington Biochemicals or a homogeneous fraction was the generous gift of Dr. Lawrence Loeb (University of Washington). Large fragment of DNA polymerase I ( S O % homogeneous), which lacked 5' + 3' exonuclease activity was obtained from Bethesda Research Laboratories and contained less than 1% intact DNA polymerase I. Homogeneous Novikoff hepatoma DNA polymerase fi was purified as described by Stalker et al. (29).
One unit of DNA polymerase activity catalyzes the incorporation of 10 nmol of total dNMP into activated DNA in 30 min at 37 "C.
Preparation of Damaged DNAs-PM2 DNA was partially depurinated at 70 "C in 10 m~ sodium citrate (pH 5.0) and 100 mM NaCl for up to 7 min so as to generate on the average less than one apurinic site per duplex DNA circle. UV-treated PM2 DNA had been irradiated at 254 nm in 10 mM Tris-HC1 (pH 7.5) and 20 mM NaCl at a fluence of 2.3 J/m'L/s for up to 10 s to produce an average of less than one pyrimidine dimer per duplex DNA circle. Depyrimidination of [32P, ~racil-~H]poly(dA-dT) was carried out using uracil DNA Nglycosylase as indicated. One unit of uracil DNA N-glycosylase activity releases 1.0 pmol of uracil/min from DNA a t 37 "C.
Incision of PM2 DNA for Polymerase Reactions-Reactions contained 50 mM Tris-HC1 (pH 7.5), 10 mM MgC12, 0.1 mM partial depurinated PM2 ["]DNA (contributing 30 m~ NaCl and 3 mM sodium citrate to the final reaction), and AP endonuclease as indicated. Incision by Neurospora crassa endonuclease was under identical conditions except that nondepurinated DNA was substrate. Incision with T4 UV endonuclease was carried out similarly in 25 mM potassium phosphate buffer (pH 7.5), 200 mM NaCl, 80 ng/ml of acetylated bovine serum albumin, 0.1 mM UV-irradiated DNA, and T4 UV endonuclease as described. Incubation was for 10 min at 37 "C followed by 3 min at 70 "C to inactivate endonuclease. Samples containing approximately 2 nmol of DNA were removed and the extent of endonuclease incision determined by the endonuclease assay. In some cases, excess AP endonuclease was then added during a subsequent 10-min incubation a t 37 "C. To reactions which did not Tris-HC1 (pH 7.5), 0.1 mM EDTA, and 0.1 mg/ml of acetylated bovine receive a second endonuclease treatment, an equal volume of 10 mM serum albumin was added and incubation carried out as above. Following incubation, the reactions were heated for 3 min at 70 "C. Endonuclease assays were also performed after the second reaction.

Methods
Endonuclease Assay-Endonuclease activity was measured using a nitrocellulose filter which preferentially retains nicked PM2 r3H] DNA after alkali treatment as described previously (22). The per cent r3H]DNA which bound to each filter was determined in a liquid scintillation spectrometer with 5 ml of Betafluor (National Diagnostics). Assuming a Poisson distribution of AP sites, the number of nicks per PM2 DNA circle was determined as described by Kuhnlein et al. (22). All assays were carried out in duplicate and reactions in which the endonuclease was omitted generally accounted for t0.2 nick/DNA circle. One unit of AP endonuclease activity will generate 1 pmol of nicks specifically into AP DNA/min at 37 "C.
DNA Synthesis Reactions-Reaction mixtures (450 pl) contained 70 mM potassium phosphate buffer (pH 7.5), 1 mM 2-mercap+,oethanol, 7 mM MgC12, 0.09 mM concentration each of dATP, dCTP, and dGTP, 15 pCi of ["'PIdTTP (4,500-15,OOO cpm/pmol), 18 nmol of PM2 ['HI DNA (6,500 cpm/nmol), and E. coli DNA polymerase I or large fragment as noted. After incubation at 37 "C for various times, 50-pl aliquots were removed and 100 p1 of 1 mg/ml of bovine serum albumin in 0.1 M sodium pyrophosphate was added on ice prior to precipitation with 500 p1 of 10% trichloroacetic acid. Precipitates were collected on Whatman GF/C filters, washed with 15 ml of 1 N HC1 in 0.1 M sodium pyrophosphate, dehydrated with 959 ethanol, and dried under a heat lamp. Acid-insoluble radioactivity was measured by liquid scintillation using double isotope counting.
Reactions using activated DNA as the substrate were performed similarly, except that 0.09 mM [32P]dTTP (33 cpm/pmol) and 56 pg of activated salmon sperm DNA were substituted. The activated DNA was subjected to a sham incision reaction to which buffer was added instead of endonuclease prior to being used for DNA synthesis.
Alkaline Sucrose Gradient Centrifugation-Reaction samples (100 p l ) were placed on ice, adjusted to 67 mM EDTA, and 100 p1 of 30 mM Tris-HC1 (pH 8.0), 250 mM NaOH, 900 mM NaCI, and 5 mM EDTA was added. After 30 s at 70 "C, samples were incubated for 30 min at 25 "C to allow alkaline hydrolysis of any residual uncleaved AP sites, Samples (200 p1) were layered onto 4.5 ml of 5-25% sucrose gradients containing 30 mM Tris-HC1 (pH 8.0), 250 mM NaOH, 900 mM NaCl, and 5 mM EDTA which had been formed over a 300-p1 cushion of saturated CsCl in 100 mM NalP04 and 1 mM EDTA. After sedimentation, 116-pl fractions were collected from the bottom of the tube, 50 p1 of 1 mg/ml of bovine serum albumin in 100 mM sodium pyrophosphate was added on ice, and acid-insoluble radioactivity was determined.
DEAE-cellulose Chromatography of Reaction Products of ("P, uracil-"Hjpoly(dA-drr)-Reactions (300 pl) contained 50 mM Tris-HC1 (pH 8.2), 4 mM EDTA, 38 nmol of r3'P, uracil-'H]poly(dA-dT), and 12 units of uracil DNA N-glycosylase. After incubation for 30 min at 37 "C, the reactions were adjusted to 10 mM MgCL and 0.04 unit of E. coli endonuclease I11 was added. Following a second incubation for 90 min at 30 "C, reactions were placed on ice, adjusted to DNA synthesis reaction conditions, and 20 units of DNA polymerase I or buffer added to a final volume of 510 pl. After 60 min at 30 " c , reactions were placed on ice, diluted to 12 ml with distilled water, and analyzed by DEAE-cellulose (DE32, Whatman) chromatography as described by Tomlinson and Tener (30) using a column (0.28 cm2 X

DNA Polymerase I Action at
Repair Incisions 577 and eluted with a 200-ml linear gradient of 0-0.2 M NaCl in 7 M urea and 2.5 mM Tris-HCI (pH 8.2) at 30 ml/h. Fractions of 1.8 ml were collected and radioactivity determined with 1-ml samples in 10 ml of Liquiscint (National Diagnostics). Electron Microscopy-Samples were adjusted to 10 m~ EDTA on ice and approximately 0.1 nmol of PM2 DNA mounted for electron microscopy as described by Davis et al. (31). DNA samples were adjusted to 0.1 M Tris and 0.01 M EDTA (pH 8.3), 50 pg/ml of cytochrome c, and 40% formamide in 50 p1 and spread over a 0.1 M Tris and 0.01 M EDTA (pH 8.3), 10% formamide hypophase. Samples were picked up on Formvar-coated grids (400 mesh), stained in uranyl acetate, rotary shadowed with tungsten, and carbon coated. Observation and photography were performed on a Phillips 301 at 6800X magnification. Molecular length measurements were made from 35mm photographic negatives projected at 28X or 54X enlargement on a Hewlett-Packard 9864A digitizer coupled with a 9810A calculator.

E. coli DNA Polymerase I Activates 3'-Deoxyribose
Termini so as to Prime DNA Synthesis Efficiently-When partially depurinated PM2 r3H]DNA is cleaved to a limit with E. coli endonuclease 111, a Class I AP endonuclease, the resulting incisions contain 3"deoxyribose and 5'-phosphornonoester termini (16) (Fig. 1). Such termini do not efficiently support DNA synthesis by DNA polymerase I (16). However, some incorporation of dNMPs (up to twice that seen for untreated PM2 DNA) was typically observed, and the rate of DNA synthesis during these reactions tended to increase as the reaction progressed (Fig. 2). Such a concave curve would be expected if 3'-deoxyribose termini were being activated in some manner during the DNA synthesis reaction so as to serve as efficient primers and if primer activation was the rate-limiting step for subsequent DNA elongation.
In order to test the possibility that DNA polymerase I could activate 3'-deoxyribose termini, partially depurinated PM2 DNA incised with an excess of endonuclease I11 was preincubated with DNA polymerase I in the absence of dNTPs and then dNTPs were added and DNA synthesis carried out. The results show an increase in the rate of DNA synthesis with increasing time of preincubation (Fig. 2  bation as was found for synthesis after a treatment with human fibroblast AP endonuclease I1 (Fig. 2). Human fibroblast AP endonuclease 11, a Class I1 AP endonuclease, removes deoxyribose 5-phosphate from the 3' termini formed by the E. coli endonuclease I11 to form a single nucleotide gap containing a 3'-hydroxyl nucleotide primer (15) (Fig. 1). Activation was dependent on DNA polymerase I or upon a second incision by a Class I1 AP endonuclease. Preincubation for 120 min in the absence of DNA polymerase I resulted in an 8.9% increased rate of DNA synthesis relative to that observed when the substrate was maximally activated with a Class I1 AP endonuclease. Thus, the majority of the activation of DNA synthesis required preincubation with DNA polymerase I and was not a result of spontaneous hydrolysis of the 3'deoxyribose termini. It should be pointed out that the incubation with AP endonuclease I1 or DNA polymerase I treatment did not cause increased nicking; thus, the stimulation of DNA synthesis observed most likely resulted from modification of the endonuclease I11 incision sites. for the times indicated at 37 "C. Reactiom from which dNTPs were omitted were preincubated at 37 "C for 60 min (A-A) or 120 min (M) with DNA polymerase I before dNTPs were added and DNA synthe. is c,arried out as above. The average number of incisions per DNA bendme after endonuclease exposure is given in parentheses. PM2 ['HIL'NA had been partially depurinated to contain about 0.7 AP site/DNA genome. The amount of DNA synthesis observed in reactions from which endonuclease 111 had been omitted was subtracted.

Activation of 3'-Deoxyribose Termini by DNA Polymerase
by guest on March 24, 2020 http://www.jbc.org/ Downloaded from I to Support DNA Synthesis Involves the Removal of the 3'-AP Site-Activation of endonuclease 111 incisions by DNA polymerase I in the absence of dNTPs could occur by one of two mechanisms. Preincubation with DNA polymerase I could lead to an increase in utilization of baseless 3'-deoxyribose termini as primers, or the polymerase could remove the 3"AP site to expose an internal 3"hydroxyl nucleotide as primer. In order to distinguish between these two alternatives, partially depurinated PM2 C3H]DNA containing 0.7 AP site/duplex circle was incised with an excess of endonuclease 111 and then the product was used as a substrate for DNA synthesis and analyzed after alkali treatment by alkaline sucrose gradient centrifugation. If DNA polymerase I was capable of utilizing a 3'-deoxyribose terminus as a primer, then the initiation point of DNA synthesis should be alkali labile. Following alkaline hydrolysis of this 3'-deoxyribose moiety, single-stranded PM2 13H]DNA should then be separable by alkaline sucrose gradient sedimentation from the short [3'PP]DNA product of DNA synthesis. However, if the 3"AP site was removed prior to DNA synthesis, then the 3H-and "P-labels would remain covalently attached after alkali treatment and sediment near full length single-stranded linear DNA. Indeed, when an average of 145 [32P]nucleotides was incorporated per nick using the DNA treated with endonuclease I11 as primer-template, both labels co-sedimented as full length, linear single-stranded PM2 DNA (Fig. 3A). Likewise, after a 60-min preincubation of the substrate with DNA polymerase I in the absence of dNTPs, a 58% stimulation of DNA synthesis resulted in the incorporation of approximately 250 nucleotides/nick, and as before, the 32P product remained covalently attached to the single-stranded, linear PM2 [3H]DNA during alkaline sucrose gradient sedimentation (Fig. 3B). In particular, no 32P was observed sedimenting at a position expected for a 250-nucleotide fragment. As a control, the substrate was totally activated by the addition of an excess of human fibroblast AP endonuclease I1 which removed the 3"AP termini. After this removal of deoxyribose 5-phosphate from the 3' termini, DNA synthesis resulted in 420 nucleotides being incorporated/nick and, as expected, the product of this reaction was also alkali stable and both the 'H-and "P-labels co-sedimented as before (Fig.  3C). Finally, when the unpolymerized, partially depurinated Form I PM2 [3H]DNA was subjected to alkaline hydrolysis followed by alkaline sucrose gradient sedimentation, similar [3H]DNA profiies were obtained as above, indicating that hydrolysis of the AP sites had occurred under these conditions (data not shown). The hydrolysis was dependent upon the alkali treatment since 0.17 nick/depurinated PM2 [3H]DNA molecule was observed prior to alkaline treatment, whereas after exposure, 0.69 nicks/molecule was detected by the nitrocellulose fiiter endonuclease assay.
The amount of singlestranded linear molecules produced by this treatment was directly proportional to the number of AP sites, assuming that the distribution of an AP site followed a Poisson distribution.
The above results argue that the 3'-deoxyribose moiety was removed by DNA polymerase I during the preincubation reaction such that the termini could efficiently prime DNA synthesis. The removal of the 3"AP site was not due to contaminating AP endonuclease in the DNA polymerase I preparation since no endonuclease (less than 6.6 X units/ unit of polymerase) was detected on partially depurinated PM2 [3H]DNA by the nitrocellulose fiiter endonuclease assay.2 In addition, preincubation of partially depurinated PM2 This value was determined as the lower limit of detection for the endonuclease assay when a 10-to 30-fold excess of DNA polymerase I was utilized. From this data, we calculate that less than 0.95 fmol of endonucleolytic incisions could have occurred during the preincubation reaction due to DNA polymerase I in the experiment of Fig. 3B. Alkaline sucrose gradient centrifugation of partially depurinated PM2 DNA that was incised by E. coli endonuclease 111 and used for DNA synthesis. PM2 ["]DNA (9 nnol) was partially depurinated to produce about 0.7 AP site/duplex DNA circle and incised as described under "Experimental Procedures" with 0.05 unit of endonuclease 111. After heat inactivation of the endonuclease, the reaction mixture was divided into three aliquots. A, one aliquot was added to a DNA synthesis reaction containing 0.24 unit of E. coli DNA polymerase I; B, a second aliquot was added to a similar reaction mixture, except that dNTPs were omitted, and after incubation with DNA polymerase I for 60 min at 37 "C, the dNTPs were added C, a third aliquot was subsequently incised with 1.8 units of human fibroblast AP endonuclease 11, heated to inactivate endonuclease, and then added to a DNA synthesis reaction as in A . After incubation with the polymerase for 45 min at 37 "C, each reaction was chilled on ice, and 50 ,u1 of 0.1 mM EDTA was added. Alkaline hydrolysis of any remaining AP sites was carried out before samples were layed onto 5-25% alkaline sucrose gradients. Centrifugation was for 4% h at 43,000 rpm and 20 "C in a Spinco SW50.1 rotor. Sedimentation is from right to left. Form I PM2 DNA sedimented to the bottom of the gradient; single strand circular DNA and full length single strand linear DNA sedimented to approximately fraction 17 and 20, respectively. The arrow indicates the theoretical position for a 250-nucleotide, singlestranded DNA fragment (40).
["HIDNA with DNA polymerase I did not activate the substrate to support additional DNA synthesis unless the substrate had previously been cleaved by endonuclease 111 (Table  I). (The incorporation observed on partially depurinated PM2 DNA which was not treated with endonuclease most likely resulted from nicks which originate spontaneously in depurinated DNA during incubation, since this material typically contains roughly 8% Form I1 DNA above the 7% background It is unlikely that such an activity could account for the 58% stimulation of DNA synthesis relative to that observed when all 110 fmol of AP sites were cleaved with human fibroblast AP endonuclease 11. However, we cannot discount the possibility that such a putative ably more efficiently than internal AP sites.  partially depurinated PM2 DNA to support DNA synthesis Four AP DNA reactions were prepared as described under "Experimental Procedures" containing 28 nmol of PM2 [3H]DNA (0.7 AP site/duplex circle) to which 0.1 unit of endonuclease I11 or buffer was added as indicated in the table. After 10 min at 37 "C and heat inactivation of the endonuclease, endonuclease assays and DNA synthesis reactions were carried out as described in the legend to Fig.  2. Preincubation (with DNA polymerase where indicated) was for 60 min at 37 "C, and DNA synthesis was for 30 min at 37 "C. Samples (4 nmol of PM2 DNA product) were removed and prepared for alkaline sucrose gradient centrifugation as described in the legend to Fig. 3 to detect the amount of 32P incorporated into full length, singlestranded PM2 DNA in the 4-nmol sample. The specific activity of the 32PrdNMP1 was 4800 comhmol. Nature of the Product Removed by DNA Polymerase I from 3'-Deoxyribose Termini-In order to elucidate the action of DNA polymerase I at 3"AP sites, the hydrolysis products were isolated. [3'P, ura~il-~H]poly(dA-dT) (an alternating dA-dT polymer containing an occasional dUMP residue in place of dTMP) was depyrimidinated with uracil DNA N-glycosylase to release 3.6 pmol of uracil or 63% of the uracil content of the polymer. After treatment with endonuclease I11 and DNA polymerase I, the hydrolysis products were analyzed by urea-DEAE-cellulose chromatography (Table 11). Two peaks of material containing only 32P were observed. One peak, containing 1.4 pmol of 32P-label, eluted coincidentally with deoxyribose 5-phosphate, while the other contained 0.34 pmol of 32P-label and eluted between the mono-and dinucleotide markers. The latter compound was not specifically identified, but migrated as though it were pAp(d-ribose). Smaller amounts of both of these products also were formed in the absence of the DNA polymerase, probably by contaminant(s) in the uracil DNA N-glycosylase preparation.
The remaining 37% of the uracil that had not been removed by the uracil DNA N-glycosylase was observed after treatment with the endonuclease and polymerase to be dUMP, as *evidenced by a 1:l ratio of 3H to 32P and co-elution with a dUMP marker. We suspect that dUMP originated from terminal dUMP residues which were not internalized by DNA polymerase p during the substrate production. Terminal dUMP residues do not seem to be a substrate for uracil DNA N-glycosylase but could be removed by the 3' + = 5"exonuclease of DNA polymerase I. In support of this interpretation, we observed that other samples of [32P, ~racil-~H]poly(dA-dT) which had been extended by DNA polymerase I were >95% sensitive to uracil DNA N-glycosylase.
E . coli DNA Polymerase I Activates T4 UV Endonuclease Incision Sites to Prime DNA Synthesis Efficiently-When UV-irradiated PM2 DNA is treated with T4 UV endonuclease, phosphodiester cleavage results in a 3'-deoxyribose juxtaposed to a thymine:thymidylate cyclobutane dimer (18) (Fig. I), and such termini do not efficiently support DNA synthesis by DNA polymerase I (16). However, upon incubation with the polymerase in the absence of dNTPs, these termini become activated to support DNA synthesis (Fig. 4). The level of units of E. coli uracil DNA N-glycosylase for 30 min at 37 "C as described under "Experimental Procedures" resulted in the removal of 3.6 pmol of uracil (63% of that present). After addition of MgClz to 10 mM, 0.04 unit of endonuclease I11 was added and incubation resumed for 90 min at 30°C. DNA synthesis reaction components were added and 20 units of DNA polymerase I or buffer was included as indicated. After 60 min at 30 "C, reactions were chilled, added to 12 ml of distilled water, and applied to a DEAE-cellulose column (0.28 cm2 X 9 cm) equilibrated in water. The column was washed and developed as described under "Experimental Procedures." Fractions (1.8 m l ) were collected. "Experimental Procedures." After heat inactivation of the endonuclease, 240 pl was removed and 0.13 unit of human fibroblast AP endonuclease I1 was added (U). The original reaction mixture received a corresponding amount of buffer, and both reactions were incubated for an additional 10 min at 37 "C. After heat inactivation, the original reaction was divided into three aliquots. DNA synthesis reaction components were added to each (A-A,

A-A,
and M) as described in the legend to Fig. 2, except that, where noted, dNTPs were omitted and preincubation was carried out at 37 "C for 60 min (A- A) or 120 min ( o " 0 ) with DNA polymerase I before the dNTPs were added. DNA synthesis was then carried out for the times indicated. The average numbers of endonuclease incisions per DNA genome are given in parentheses.
activation achieved is virtually that observed by pretreatment with human fibroblast AP endonuclease 11, a Class I1 enzyme that can remove 3'-deoxyribose termini (15) (Fig. 1). Presumably, the mechanism of activation with this substrate is similar to that described above, involving a removal of the deoxyribose 5-phosphate.
Mode of DNA Synthesis by E. coli DNA Polymerase I from Various Incision Sites-When undamaged Form I PM2 DNA is cleaved with N. crassa endonuclease, which cleaves Form I DNA to produce a single, 3'-hydroxyl-and 5'-phosphorylterminated nick (32,33) (Fig. l), the incision sites support efficient DNA synthesis by DNA polymerase I (15). After

DNA Polymerase I Action at Repair Incisions
such synthesis with an average incorporation of 1550 nucleotides/nick, the product was observed by electron microscopy.
Ninety-four per cent of the Form I1 DNA molecules observed were double-stranded DNA circles without tails, while 6% of the Form I1 molecules had tails containing an average of 300 nucleotides/tail (Table 111). Synthesis was carried out with about 1 DNA polymerase I molecule/lO incision sites, and evidently under these conditions, DNA synthesis occurs primarily by nick translation. When UV-irradiated PM2 DNA containing about one pyrimidine dimer per duplex circle was incised with a n excess of T4 UV endonuclease and then incised with human fibroblast AP endonuclease 11, the DNA substrate became an efficient primer for DNA synthesis (15). This DNA is thought to contain a one-nucleotide gap with a 3"hydroxyl nucleotide and a 5"pyrimidine dimer nucleotide terminus (15) (Fig. 1).
After DNA synthesis using this substrate, 753 nucleotides were incorporated/nick and 874 of the Form I1 DNA molecules did not contain tails (Table 111). suggesting, again, that nick translational synthesis had predominated, i.e. that the 5' + 3' exonuclease of polymerase I appears to be able to remove efficiently the pyrimidine dimers from this substrate under these conditions.
Partially depurinated PM2 DNA (0.8 AP site/duplex circle) was cleaved with the Class I1 human fibroblast AP endonuclease I1 so as to form a 3' terminus with a 3"hydroxvl nucleotide and a 5' terminus with deoxyribose 5-phosphate  After 10 min at 37 "C and subsequent heat inactivation of endonuclease, a second incision reaction was conducted with the indicated enzyme or buffer as appropriate. Incubation and heat inactivation were as above. A sample of each reaction was removed to verify that the degree of nicking roughly equalled the frequency of damage and then DNA synthesis components were added to each reaction mixture (250 pl) and incubation was carried out and samples mounted for electron microscopy (see legend to Fig. 5). The total number of molecules observed equals the Form I1 molecules described in the table plus the Form I and/or I11 molecules which were observed.  (Fig. 1). The substrate was incubated with DNA polvmerase I so as to incorporate 1 1 0 nucleotides/nick, and in contrast to the above cases, 48% of the Form I1 DNA molecules were observed to contain one or more DNA tails originating from otherwise intact relaxed circles (Table 111 and Fig. 5A). The occurrence of some Form I1 DNA molecules with multiple tails would be expected if the tails originated from AI' incision sites since their distribution among molecules is random. The vast majority of tails were almost totally duplex, and the distribution of lengths of Form I1 DNA and tail(s) was determined. Form I1 molecules had a sharp distribution with a mean length of 3.27 k 0.11 pm (Fig. 6A), which was in good agreement with the length of PM2 DNA as reported previously (34,35). Using this length as a standard for the 9500base pair PM2 molecule, the average tail was 660 nucleotides in length (Fig. 6 R ) . This value appears to be in fairly good agreement with the extent of synthesis determined by label incorporation and the number of nicks observed, particularly when considering the inability to resolve small tail structures and the fact that not all DNA synthesis necessarily produced tails (i.e. some nick translation may have occurred). Two other Class I1 A P endonucleases, HeLa A P endonuclease and E. coli endonuclease IV, were also used to incise partially depurinated PM2 DNA and then DNA synthesis was carried out to an average of 1180 and 1320 nucleotides/nick, respectively. A preponderance of tails was similarly observed (Table 111); 51% and 764 of the Form I1 molecules had one or more tails, respectively. The distribution of lengths of Form I1 DNA circles and tails produced from HeLa AP endonuclease is shown in ['HIIINA (one AP site per duplex circle) and 7.0 units of human fibroblast AP endonuclease 11. Incubation was for 10 min at 37 "C followed by heat inactivation of the endonuclease. After buffer was added, a second sham incubation was carried out for 10 min at 37 "C. R. the reaction was similar, except that 0.15 unit of endonuclease 111 was substituted in the first incubation. After incubation and heat inactivation of the endonuclease, a second reaction was conducted with 7.0 units of human fibroblast AP endonuclease 11. Incubation and heat inactivation were as before. Nicking was monitored after each reaction to assure that nicking roughly equalled the frequency of AI' sites. DNA synthesis reactions were for 6 h at 37 "C and the extent of svnthesis per nick was monitored at various times. Samples (50 pl) were adjusted to 10 mM EDTA and placed on ice to terminate that reaction. Approximately 0.1 nmol of I'M2 ["' I' , "HIDNA was mounted for electron microscopy. HeLa AP endonuclease. n is the number of molecules measured. Tail lengths were calculated from the mean circle lengths, assuming that PM2 DNA contains 9500 base pairs. of a 5"AP terminus a t a nick dramatically increases the frequency of strand displacement regardless of the Class I1 AP endonuclease used to incise the DNA. As a final substrate, partially depurinated PM2 DNA was incised to a limit with a Class I AP endonuclease, endonuclease I11 and then treated with a Class I1 AP endonuclease, either human fibroblast AP endonuclease I1 or HeLa AP endonuclease. Treatment with the second enzyme resulted in no further nicking, assuring that all AP sites were cleaved during the first incubation. The combined treatment resulted in a substrate effectively containing a gap of one nucleotide where the AP sites had been (15, 20) (Fig. 1). With this molecule as a polymerase substrate, about 95% of the Form I1 DNA product lacked tail structures after the incorporation of 930 nucleotides/nick (Table I11 and Fig. 5B). These results are similar to those observed for DNA incised by the N. crassa endonuclease and indicate a strong preference for nick translational synthesis with these small gaps. They also show that exposure of DNA to a Class I1 AP endonuclease per se does result in increased strand displacement synthesis.
Potential of Various Incision Sites to Support DNA Synthesis by E. coli DNA Polymerase I Large Fragment-Proteolytic cleavage of DNA polymerase I with subtilisin yields a larger polypeptide fragment (molecular weight 76,000) which retains both the polymerase and 3' -+ 5' exonuclease activities, but lacks the 5' + 3 exonuclease which is present on the smaller fragment (molecular weight 36,000). One might thus expect the larger fragment to be capable only of strand displacement synthesis from primer termini located at a nick. When large fragment was used for DNA synthesis on PM2 DNA cleaved by N . crassa endonuclease, approximately 10% the rate of intact DNA polymerase I was observed (Fig. 7B). The amounts of enzyme were chosen to be equal when each was used with "activated" (gapped) salmon sperm DNA primer/template (Fig. 7A). A similarly decreased rate of DNA synthesis by large fragment relative to intact DNA polymerase I was also observed when partially depurinated PM2 DNA that had been incised with human fibroblast AP endonuclease I1 was used as the substrate (Fig. 7B). These data suggest that unlike the intact DNA polymerase, the large fragment does not efficiently carry out synthesis from either of these two types of incision sites.
Since DNA polymerase I large fragment retains 3' + 5' exonuclease activity which will act on both single-stranded DNA and unpaired termini in duplex DNA (36), we tested its ability to catalyze synthesis from partially depurinated PM2 DNA cleaved with E. coli endonuclease 111. However, little if any DNA synthesis was observed under conditions wherein an equal amount of intact DNA polymerase could efficiently activate 3'-deoxyribose termini to support DNA synthesis (Fig. 8). Moreover, unlike the intact polymerase, the large fragment was also incapable of using a 3'-hydroxyl located at a one-nucleotide gap (formed by the combined action of a Class I and Class I1 AP endonuclease) to prime efficiently DNA synthesis (Fig. 8).
While the large fragment was unable to utilize endonuclease I11 incision sites for DNA synthesis, the enzyme might still be able to remove a 3"AP terminus by virtue of its 3' + 5'  were added. All reactions were then incubated as indicated at 37 "C. carried out. As expected, the DNA polymerase I large fragment did not catalyze DNA synthesis ( Fig. 9) even after preincubation. However, even when normal DNA polymerase I was added after preincubation with large fragment, the level of DNA synthesis by DNA polymerase I was equal to that found if preincubation with the large fragment had not occurred ( Fig. 9), i.e. the large fragment did not appear to remove the 3"AP terminus during the preincubation. These results suggest that the large fragment of the polymerase is not capable of utilizing efficiently nicked DNA substrates either for polymerase or exonuclease activity.

DISCUSSION
UV endonuclease activities have been reported from a variety of cell types, but the exact mechanism of incision has been described only for the M. Zuteus and T4 UV endonucleases which incise on the 5'-side of pyrimidine dimers by the joint action of a pyrimidine dimer DNA N-glycosylase and Class I AP endonuclease (17)(18)(19). Whether other UV endonucleases cleave in a similar manner remains to be determined directly, although it appears that pyrimidine dimers which are excised in vivo from UV-irradiated E. coli DNA may not be mediated by a pyrimidine dimer DNA N-glycosylase since resulting oligonucleotide products do not appear to contain a substantial number of pyrimidine dimer nucleotides (18). Nevertheless, it is desirable to understand the mechanism of action of E. coli DNA polymerase I at the coliphage T4 UV endonuclease incisions, as such action presumably applies to infected cells. It is also desirable to study the action at Class I AP endonuclease incisions since UV and y radiation results in the production of 5,6-dihydroxydihydrothymine residues (36, 37), and in E. coli these lesions are probably repaired by endonuclease I11 through the combined action of a DNA Nglycosylase and Class I AP endonuclease (18,25). Endonuclease I11 also cleaves simple AP lesions by a Class I AP endonucleolytic incision.
The removal of 3"AP termini by polymerase I was relatively slow, but was apparently a property of that enzyme, not of a contaminating AP endonuclease. AP endonuclease activity upon partially depurinated circles could not be detected in any of our polymerase preparations, and the exonucleolytic removal was equally efficient with apparently homogeneous enzyme. The 3' -+ 5' exonuclease associated with DNA polymerase I is an obvious candidate for this exonuclease activity, particularly since mismatched bases are normally removed by this activity (5). However, we were surprised to note that the polymerase appears also to release some pAp(d-ribose) as well as d-ribose 5-phosphate since the 3' + 5' exonuclease normally forms only dNMP residues. An alternative and perhaps more efficient method for activating the 3"AP sites in vivo might be the utilization of a Class I1 AP endonuclease to remove the deoxyribose 5-phosphate.
Under our reaction conditions, DNA polymerase I will catalyze predominantly coordinated polymerization and 5' + 3' exonucleolytic degradation ("nick translation") from T4 UV endonuclease incision sites as well as from nicks generated by N. crussu endonuclease. These results might seem to contradict those reported by Masamune and Richardson (35) for DNA polymerase I using PM2 DNA incised by pancreatic DNase. In the latter, an initial phase of nick translational DNA synthesis for only about 50 nucleotides was followed by strand displacement. However, in those experiments, DNA polymerase I was in vast excess over the number of3'-hydroxyl primers termini (about 30 molecules of DNA polymerase I/ terminus, whereas under conditions used here, about 1 molecule of DNA polymerase I was used/l to 10 termini). Indeed, when we carried o u t DNA synthesis using similar excesses of

DNA Polymerase I Action at Repair
Incisions 583 the polymerase, strand displacement was observed by electron microscopy from N . crassa endonuclease incisions. Thus, the level of DNA polymerase I seems to influence the mode of DNA synthesis. An excess of the polymerase tends to promote strand displacement. While many other factors probably influence the absolute level of strand displacement versus nick translational DNA synthesis in uiuo, it is still reasonable to assume that a qualitative correlation between the observations in uitro and events in uitro may exist. Whereas strand displacement predominated from nicks containing normal 5"nucleotides or 5"pyrimidine dimer nucleotides, we were surprised to observe that nicks containing 5'deoxyribose phosphate termini promoted predominantly strand displacement synthesis. The tails thus produced were double-stranded and, hence, could have come about either by template switching by the polymerase or by a branch migration reaction. Strand displacement DNA synthesis may have been preferred on this substrate (i) because of a reluctance of the 5' + 3' exonuclease to hydrolyze a deoxyribose 5"phosphate terminus; (ii) because of an increased frequency or extent of fraying of a terminus carrying a baseless site, thus allowing polymerization to supersede exonuclease hydrolysis; or (iii) by a physical blockage of the 5' + 3' exonuclease activity by an association of the AP endonuclease with the 5' termini. With respect to the latter, covalent linkage between polypeptides and 5' termini has been described for proteins involved in DNA synthesis (38, 39); however, since free deoxyribose 5-phosphate is formed by treatment of Class 11, then Class I AP endonuclease (15), this association, if occurring, is weak. In addition, incisions made by three different Class I1 AP endonuclease preparations (human fibroblast AP endonuclease 11, HeLa AP endonuclease, and E . coli endonuclease IV) each showed similar results.
As to the fist two possibilities, one might have expected that the larger pyrimidine dimer would be more refractory to hydrolysis or more frayed than the baseless sugar. Whatever the cause of the increased strand displacement with 5'-terminal baseless sites, it is not unreasonable to assume that the production of strand-displaced DNA in response to incisions made at such damage may be one of the driving forces for DNA recombination or gene duplication which have been observed to be associated with DNA repair. It may be noteworthy as well that xeroderma piginentosum fibroblasts of complementation Group D lack a Class I AP endonuclease which might normally remove deoxyribose 5-phosphate from 5' termini. The absence of such an enzyme might, by analogy, increase the frequency of strand displacement during repair in these human cells.
When DNA polymerase I large fragment was used in reactions primed by a nick, only about 10% the rate of DNA synthesis was observed compared to the intact enzyme. Since the DNA polymerase I large fragment lacks 5' + 3' hydrolytic activity, this synthesis presumably proceeds by a strand displacement mechanism. Such a product would not be expected to he substrate for DNA ligation and this expectation has been verified (35). In general, the large fragment appeared to be either reluctant or incapable of activity at any of the nicked substrates, so it might be worthwhile to measure the affinity of the fragment uersus that of intact polymerase for nicks.
In conclusion, the availability of purified, well characterized repair endonucleases now makes it possible to begin to characterize enzymatically the subsequent events of excision and repair synthesis. This study with E. coli DNA polymerase I is a beginning toward that aim.