Regulatory roles of p21 and apurinic/apyrimidinic endonuclease 1 in base excision repair.

Many types of DNA damage induce a cellular response that inhibits replication but allows repair by up-regulating the p53 pathway and inducing p21(Cip1, Waf1, Sdi1). The p21 regulatory protein can bind proliferating cell nuclear antigen (PCNA) and prohibit DNA replication. We show here that p21 also inhibits PCNA stimulation of long patch base excision repair (BER) in vitro. p21 disrupts PCNA-directed stimulation of flap endonuclease 1 (FEN1), DNA ligase I, and DNA polymerase delta. The dilemma is to understand how p21 prevents DNA replication but allows BER in vivo. Differential regulation by p21 is likely to relate to the utilization of DNA polymerase beta, which is not sensitive to p21, in the repair pathway. We have also found that apurinic/apyrimidinic endonuclease 1 (APE1) stimulates long patch BER. Furthermore, neither APE1 activity nor its ability to stimulate long patch BER is significantly affected by p21 in vitro. We propose that APE1 serves as an assembly and coordination factor for long patch BER proteins. APE1 initially cleaves the DNA and then facilitates the sequential binding and catalysis by DNA polymerase beta, DNA polymerase delta, FEN1, and DNA ligase I. This model implies that BER can be regulated differentially, based upon the assembly of relevant proteins around APE1 in the presence or absence of PCNA.

DNA base excision repair (BER) 1 is the process responsible for the targeting and removal of damage incurred to an individual base within the cellular DNA template (1). Damage to DNA bases can be caused by methylating and oxidizing agents as well as by other genotoxicants. Bases can also be modified by deamination. During repair, a DNA glycosylase removes the altered base, generating an abasic site. Additionally, abasic sites can arise as a result of spontaneous hydrolysis of the N-glycosylic bond (2). Apurinic/apyrimidinic (AP) sites are noncoding lesions that can lead to misincorporation during replication and transcription. Therefore, these damaged sites need to be repaired promptly. Repair is initiated by an apurinic/ apyrimidinic endonuclease (APE) that cleaves 5Ј to the abasic site (3,4). In mammalian cells, the predominant AP endonuclease is APE1 (HAP1, REF1) (5,6). Cleavage is followed by removal of the 5Ј-deoxyribose phosphate moiety, resynthesis, and ligation to generate repaired double-stranded DNA (7). Upon APE1 cleavage, the downstream events can occur via two separate pathways, dependent upon the oxidation state of the abasic site (8,9). In short patch repair, only the damaged nucleotide is replaced (10); however, in long patch repair, the damaged nucleotide is replaced along with several downstream residues that are removed in the form of a flap (9,(11)(12)(13).
During short patch repair, DNA polymerase ␤ (pol ␤) removes the 5Ј-sugar phosphate residue by catalyzing a ␤-elimination reaction (10). Although the short patch pathway is the predominant route in mammalian cells (9), oxidation of the deoxyribose renders the abasic site resistant to ␤-elimination, necessitating removal via the long patch repair pathway. Significantly, many of the proteins involved in long patch BER are also involved in chromosomal DNA replication. These include DNA polymerase ␦ (pol ␦), flap endonuclease 1 (FEN1), DNA ligase I, and replication protein A (RPA) (11,13). The sliding clamp replication protein, proliferating cell nuclear antigen (PCNA), is thought to act as a factor that promotes the assembly of these proteins at an incised abasic site. PCNA greatly stimulates several of the aforementioned proteins by tethering them to the DNA (14 -16).
The p21 Cip1, Waf1, Sdi1 regulatory protein is involved in DNA replication and DNA repair (17). p21 is a potent and universal inhibitor of cyclin-dependent kinases (18). The p21 protein was initially identified as a component of a quaternary complex that includes PCNA, cyclin D, and a cyclin-dependent kinase (19). After DNA damage is incurred, p53 is induced and p53 up-regulates p21 (18). In vivo and in vitro studies have shown that p21 inhibits PCNA-dependent DNA replication (20 -23). Inhibition of PCNA by a p21-derived PCNA-binding peptide also results in inhibition of DNA synthesis in vivo and in vitro (24,25). Because PCNA is common to both DNA replication and long patch BER, this protein might play regulatory roles in several essential biological processes, including cell cycle progression, DNA replication, and DNA repair.
During mammalian BER, AP sites are the central intermediate, and these sites must be processed by APE1. This pivotal enzyme recognizes and cleaves the DNA phosphodiester backbone 5Ј to the AP site to generate a free 3Ј-OH end for polymerase repair synthesis (26). In recent years, new evidence has suggested that APE1 also coordinates the DNA repair steps (27). There is an orderly transfer of the DNA damage site from DNA glycosylases to APE1 and from APE1 to pol ␤ that is mediated by APE1. The extensive APE1 interaction surface allows APE1 to displace a bound glycosylase from the AP site. Subsequently, the interaction of the APE1⅐DNA complex with pol ␤ then recruits the DNA repair synthesis enzymes to areas of damaged DNA (27,28).
APE1 is involved in the repair of oxidative damage in vivo (29). In addition, the p21 regulatory protein is up-regulated during oxidative stress (30). Consequently, both of these proteins have proposed regulatory roles during long patch BER. In this report, we initiate an investigation of the consequences of the interactions between p21 and the enzymes responsible for BER. We also examine whether APE1 can coordinate DNA repair in the absence of PCNA interactions with the BER proteins.
Recombinant human APE1 was expressed in Escherichia coli BL21(DE3) from the His-tagged APE1 expression plasmid that was generated by cloning the human APE1 cDNA into a Novagen pET-28b expression plasmid. BL21(DE3) pET28b-APE1 was grown at 37°C with shaking (200 rpm) to mid-log phase (A 600 ϭ 0.5). Expression was induced by the addition of isopropyl-1-thio-␤-D-galactopyranoside at a final concentration of 400 M, and the culture was incubated for an additional 3 h at 37°C with shaking. The bacteria were then harvested by centrifugation, and the cell pellet was resuspended with cold phosphate-buffered saline and stored at Ϫ80°C until purification. Human APE1 was purified by a two-chromatography step procedure. The bacterial pellet was resuspended in lysis buffer (50 mM HEPES-KOH, pH 7.5, 500 mM KCl, 1 mM imidazole, 0.1 mM DTT, 0.1 mM EDTA, 10% glycerol, 1 mM phenylmethylsulfonyl fluoride, 2 g/ml aprotinin, 2 g/ml leupeptin, and 1 g/ml pepstatin A). After lysing the cells by passing them twice through a French press, the lysate was clarified by centrifugation for 30 min at 30,000 ϫ g. The supernatant was loaded onto a 10-ml precharged Qiagen nickel resin fast-protein liquid chromatography column at 1 ml/min. The column was washed with 10 column volumes of nickel resin buffer (50 mM HEPES-KOH, pH 7.5, 500 mM KCl, 1 mM imidazole, 0.1 mM DTT, 0.1 mM EDTA, and 10% glycerol), and the protein was eluted with a 20 -100 mM imidazole gradient. Peak fractions were pooled and dialyzed into Mono-S buffer (50 mM HEPES-KOH, pH 7.5, 100 mM KCl, 0.1 mM DTT, 1 mM EDTA, and 10% glycerol). The sample was then loaded onto a 1-ml Mono-S column (obtained from Amersham Biosciences, Inc.) at 0.1 ml/min. The column was washed with 10 column volumes of Mono-S buffer, and the protein was eluted with a 100 -250 mM KCl gradient. Purified protein was dialyzed into storage buffer (50 mM HEPES-KOH, pH 7.5, 100 mM KCl, 0.1 mM DTT, 1 mM EDTA, and 10% glycerol) and stored at Ϫ80°C.
Oligonucleotide Substrates-Oligomer sequences are listed in Table  I, and the primer-template substrates were constructed as described in the figure legends. In all substrates, the 3Ј-end regions of the downstream primers share homology with the 5Ј-ends of their respective templates. For the flap substrate, the downstream primer creates a substrate with an unannealed 5Ј-flap. The respective upstream primer was annealed to create a complementary one-nucleotide 3Ј-tail. For the nick substrate, the respective upstream primer was annealed to the proper template to create a nick between the 3Ј-end of the upstream primer and the 5Ј-end of the downstream primer. Prior to annealing, the 5Ј-radiolabeled primers were generated utilizing [␥-32 P]ATP and T4 polynucleotide kinase according to the manufacturer's instructions. Downstream primer D 1 was annealed to T 1 and radiolabeled at the 3Ј-end using [␣-32 P]dCTP and Sequenase (version 2.0). Unincorporated radionucleotides were removed with Micro Bio-Spin 30 chromatography columns. All radiolabeled primers were purified by gel isolation from a 15% polyacrylamide, 7 M urea denaturing gel prior to annealing.
Substrates were annealed by mixing 2 pmol of the respective downstream primer with 5 pmol of the corresponding template in annealing buffer (10 mM Tris base, 50 mM KCl, and 1 mM EDTA, pH 8.0) to a final volume of 30 l. The mixtures were heated to 95°C for 5 min and allowed to cool to room temperature. A corresponding upstream primer (10 pmol) was subsequently added and annealed by incubating at 37°C for 1 h. The polymerization substrate was generated by annealing the upstream primer (U 2 ), the corresponding template (T 2 ), and the downstream primer (D 2 ) at a molar ratio of 1:2.5:5, respectively. A mixture of the upstream primer and the template was heated to 95°C for 5 min and subsequently cooled to room temperature. The downstream primer was annealed by incubating at 37°C for 1 h. The BER substrate was generated by annealing a 73-mer oligonucleotide (T 3 ) with an internal deoxyuridine to the corresponding template (T 4 ) in the same manner as described previously. This substrate contains two residues that overhang at both 3Ј-ends to prevent degradation of the substrate by exonucleases and to distinguish BER products from DNA synthesis run-off products during analysis. The uracil base was removed using uracil DNA glycosylase at 37°C for 90 min. The substrate (2 pmol) was then incubated at 37°C with 4 pmol of human APE1 for 15 min to specifically cleave on the 5Ј side of the internal deoxyribose 5-phosphate moiety. After the incubation period, human APE1 was removed by sedimenting the substrate through a Micropure EZ column to give the starting substrate for the BER assays. Fig. 1 illustrates the representative substrates.
Enzyme Assay-The reactions containing the indicated amounts of substrate and enzymes were performed in reaction buffer (30 mM HEPES, pH 7.6, 40 mM KCl, 0.01% Nonidet P-40, 0.1 mg/ml bovine serum albumin, 8 mM MgCl 2 , and 0.1 mM ATP). The reactions were incubated at 37°C, terminated with 20 l of formamide dye (90% formamide (v/v) with bromphenol blue and xylene cyanole), and heated to 95°C for 5 min. After separation on a 15% polyacrylamide, 7 M urea denaturing gel, products were detected by PhosphorImager (Molecular Dynamics) analysis. For the reconstituted base excision repair assays, each reaction contained 50 fmol of DNA substrate, 0.825 pmol of [␣-32 P]dATP, 0.125 nmol of each dNTP, and various combinations of enzymes in reaction buffer. The applicable proteins were added simultaneously to the reaction mix. Reactions were incubated at 37°C for 15 min and terminated by adding an equal volume of formamide dye.
The underlined nucleotide is a deoxyuridine.

Regulatory Roles of p21 and APE1 in BER
Reaction products were resolved on a 15% polyacrylamide, 7 M urea denaturing gel and detected by PhosphorImager (Molecular Dynamics) analysis. All assays were performed at least in triplicate.

RESULTS
p21 Inhibits PCNA-directed Stimulation-Other groups have shown that the C-terminal region of the p21 regulatory protein is responsible for binding to PCNA (24,25). For our studies, we have designed a synthetic peptide resembling this region of p21. The regions of FEN1 and DNA ligase I that interact with PCNA have been identified, and these regions share homology with the PCNA-binding motif of p21 (16). Because several other DNA replication proteins also contain a similar PCNA-binding motif, most PCNA-binding proteins appear to bind the same or at least overlapping regions of PCNA (15,16). Consistent with this notion, p21, or a PCNA-binding peptide of p21, inhibits the binding of FEN1 (36), DNA ligase I (37), pol ␦ (23, 38), and several other proteins to PCNA. Considering there are three binding sites on each assembled PCNA, it can bind more than one FEN1 or p21. However, complexes of all three molecules are not observed, demonstrating that FEN1 and p21 binding is exclusive (36). Because perturbing the binding of FEN1 or DNA ligase I to PCNA should result in the loss of the PCNA stimulatory effect, exclusive binding implies that p21 is designed to regulate the stimulation process. Fig. 2A shows an examination of the effects of the p21 peptide on PCNA-directed stimulation of FEN1 cleavage activity. Lanes 5-9 illustrate that the p21 peptide does not alter FEN1 activity in the absence of PCNA. The addition of PCNA to the reaction leads to an approximate 26-fold stimulation of cleavage product formation (lane 10); however, titration of the p21 peptide leads to a notable reduction in the amount of observable stimulation (lanes [11][12][13][14][15]. This reduction is presumably a result of the destabilization of PCNA complexes with FEN1. Therefore, the addition of p21 does lead to an overall reduction in nuclease activity.
Likewise, the p21 peptide attenuates PCNA stimulation of DNA ligase I activity (Fig. 2B). The presence of PCNA leads to an approximate 5-fold enhancement of ligation activity (lane 10), but titration of the p21 peptide into the reactions results in a marked decrease of ligated product (lanes [11][12][13][14][15]. Therefore, the p21 peptide can reduce PCNA stimulation of both FEN1 and DNA ligase I by disrupting the physical interactions required to effect stimulation. However, both FEN1 and DNA ligase I will retain a basal level of activity that is not altered by the p21 peptide. We considered the possibility that the activity of APE1 is stimulated or regulated by PCNA. Because PCNA is apparently involved in coordinating the actions of proteins during long patch BER (12), we anticipated that PCNA might target APE1 to abasic sites. To analyze this possibility, PCNA was added to reactions with APE1 to determine whether any change in activity would be observed. In Fig. 2C, PCNA was added to a reaction with APE1 (lane 10). There is no noticeable change in product formation. Therefore, PCNA does not appear to interact with APE1 in the manner in which it interacts with FEN1 and DNA ligase I. In addition, titration of the p21 peptide into reactions without PCNA (lanes 5-9) and with PCNA (lanes 11-15) does not lead to any inhibition of APE1 activity.
During both DNA replication and long patch BER, flap intermediates generated by strand displacement synthesis by a polymerase need to be processed to yield intact doublestranded DNA. Fig. 3 depicts an experiment whereby a flap substrate was analyzed, and the processing of this substrate by FEN1, DNA ligase I, and PCNA was monitored. The substrate was radiolabeled at the 3Ј-end of the downstream primer so that intermediates generated during the processing reactions could be observed. Additionally, the upstream primer contains a complementary one-nucleotide 3Ј-tail (as illustrated in Fig.  1A). Biochemical evidence suggests that the physiologically relevant substrate for FEN1 consists of a one-nucleotide 3Ј-tail upstream of the nick. In this way, cleavage of the 5Ј-flap one nucleotide into the annealed region leads to the generation of a ligatable substrate. The 18-nt product represents FEN1 cleavage of the 5Ј-flap one nucleotide into the annealed region. The 44-nt product is a result of ligation by DNA ligase I of the upstream primer to the resultant 18-nt downstream primer produced by FEN1. Lane 6 illustrates the amount of ligated product generated by FEN1 and DNA ligase I in the absence of PCNA. The addition of PCNA results in the enhancement of ligation product formation (lane 12). Although the p21 peptide does not alter the amount of ligated product in the absence of PCNA (lanes 7-11), addition of the peptide to reactions with PCNA leads to a reduction in ligation activity (lanes 13-17). These results are in agreement with the observations noted above for Fig. 2. Therefore, the p21 peptide should affect the processing of flap intermediates during both DNA replication and long patch BER.
pol ␤ Activity Is Unaffected by PCNA or p21-pol ␤ plays an integral role in both the removal of the 5Ј-sugar phosphate residue by catalyzing a ␤-elimination reaction as well as the polymerization steps during short patch BER (10, 39). Re-  ). B, nick substrate, containing a nick between the 3Ј-end of the upstream primer and the 5Ј-end of the downstream primer. C, polymerization substrate, containing a one-nucleotide gap between the 3Ј-end of the upstream primer and the 5Ј-end of the downstream primer. D, BER substrate. pS represents a 5Ј-deoxyribose phosphate moiety. Uracil DNA glycosylase was utilized to remove the uracil base. Subsequently, APE1 was used to specifically cleave on the 5Ј side of the internal deoxyribose phosphate residue.
cently, Podlutsky et al. (40) have proposed that pol ␤ is responsible for initiating synthesis during long patch BER. Because pol ␤ is presumably involved in the long patch repair of AP sites (40 -42), an experiment was performed to determine whether this polymerase has a functional interaction with PCNA (Fig.  4). The polymerization substrate contains a one-nucleotide gap (as illustrated in Fig. 1C), and this substrate is representative of an intermediate in long patch BER. The addition of PCNA does not result in increased extension of the upstream primer, which on this substrate requires displacement of the downstream primer (lane 10). Also, titration of the p21 peptide into reactions devoid of PCNA and into reactions containing PCNA did not lead to any change in polymerase activity. Therefore, pol ␤ does not appear to be affected by either PCNA or the p21 peptide. Consequently, pol ␤ can still effectively perform its role in repair during the induction of p21 upon DNA damage.
p21 Disrupts PCNA Enhancement of pol ␦ Activity-pol ␦ requires PCNA for processive synthesis (43). In Fig. 4, pol ␦ alone does not yield high levels of synthesis (lane 16). Because pol ␦ catalyzes inefficient and distributive synthesis in the ysis of DNA ligase I activity using 1 fmol of DNA ligase I per reaction (lanes 4 -15). The substrate consists of D 3 :U 2 :T 2 (Fig. 1B) with a ␥-32 P radiolabel at the 5Ј-end of the downstream primer. The reactions including PCNA contained 500 fmol of PCNA (lanes 10 -15). Addition of PCNA leads to an approximate 5-fold stimulation of ligation activity (lane 10). Lanes 2 and 3 represent control lanes containing PCNA only and the p21 peptide only, respectively. C, analysis of APE1 activity using 0.5 fmol of APE1 per reaction (lanes 4 -15). The substrate is comprised of T 3 :T 4 with the uracil base removed by uracil DNA glycosylase (intermediate substrate illustrated in Fig. 1D). The T 3 oligonucleotide was radiolabeled at the 5Ј-end with ␥-32 P. The reactions containing PCNA included 500 fmol of PCNA (lanes 10 -15). Lanes 2 and 3 contain PCNA only and the p21 peptide only, respectively. LIG. I, DNA ligase I; nt, nucleotide.

FIG. 2. PCNA-directed stimulation of FEN1 and DNA ligase I is inhibited by the p21 peptide, but APE1 activity is not affected.
Reactions of 20 l containing 5 fmol of DNA substrate were performed as described under "Experimental Procedures." The following amounts of the p21 peptide were added (as indicated by the triangles): 50, 100, 150, 200, and 250 ng. The reactions were incubated at 37°C for 10 min. Substrate and product sizes are as indicated. The conversion of substrate to product was determined by quantitating the substrate and product utilizing PhosphorImager (Molecular Dynamics) analysis. A, analysis of FEN1 cleavage activity using 1 fmol of FEN1 per reaction (lanes 4 -15). The substrate is comprised of D 1 :U 1 :T 1 (Fig. 1A)  absence of PCNA, the small amount of full-length product observed might be the result of synthesis by a small quantity of a contaminating polymerase. As expected, the addition of PCNA greatly increases the amount of full-length product (lane 22), which is likely formed by the highly processive complex of PCNA and pol ␦. Titration of the p21 peptide into reactions with pol ␦ alone does not significantly affect strand displacement synthesis activity (lanes [17][18][19][20][21]. Polymerization is altered by the p21 peptide in the reactions with PCNA (lanes 23-27). Increasing the amount of the p21 peptide results in a marked decrease in the generation of full-length product. Because pol ␦ is involved in both DNA replication (44) and several DNA repair processes (45)(46)(47), inhibition of polymerization activity by p21 should affect both DNA replication and DNA repair.
Long Patch BER Is Affected by p21-Because PCNA interacts with several of the proteins involved in long patch BER (12), inhibiting these interactions would presumably lead to a decrease in the level of repair. The p21 peptide binds to PCNA and disrupts other protein interactions with PCNA (20,22,36,37). Therefore, p21 may affect long patch BER by disrupting repair complexes. To investigate this possibility, long patch BER was reconstituted with purified proteins to determine whether the p21 peptide could inhibit the formation of repair product (Fig. 5A). The 33-nt band corresponds to incorporation of a radioactive nucleotide at the position immediately successive to the abasic site (the substrate is illustrated in Fig. 1D). Lane 4 indicates the amount of repair product generated with pol ␤, pol ␦, FEN1, DNA ligase I, and RPA in the absence of PCNA. The p21 peptide does not appear to alter the modest level of repair product synthesis (lanes 5-9). Because PCNA facilitates long patch BER (12), we anticipated and found that addition of PCNA leads to the generation of a higher level of repair product (lane 10). Titration of the p21 peptide into the PCNA-containing reactions results in the reduction of repair product formation (lanes 11-15). Therefore, p21 is capable of inhibiting long patch BER in vitro by a mechanism presumed to involve p21 binding to PCNA. Analysis of the percent inhibition plot (Fig. 5B) reveals that there is a 50% inhibition of the repair reaction at the approximate p21 peptide concentration of 1.6 M.
APE1 Stimulates Long Patch BER-The long patch BER pathway involves the coordinated actions of many proteins and accessory factors. These include a DNA glycosylase, APE1, FEN1, DNA ligase I, RPA, PCNA, and a polymerase (11,13). Two polymerases have been proposed to be involved in BER. In short patch BER, pol ␤ is responsible for excision of the abasic site and polymerization (10,39). During long patch repair, the abasic site is not removed by pol ␤, and FEN1 becomes necessary for removal of the lesion (11,13). The long patch BER pathway involves pol ␦ (8, 48); however, recent studies have shown that pol ␤ is potentially responsible for initiating synthesis in long patch BER (40,42). The model proposed by Podlutsky et al. (40) suggests that, after pol ␤ adds the first nucleotide during repair synthesis, this polymerase dissociates from the AP site if the site is resistant to ␤-elimination. Subsequent synthesis and strand displacement could then be performed by pol ␦. Of the two polymerases, pol ␤ is not affected by the inhibitory action of p21 due to its inability to form a complex with PCNA. Because long patch BER can be performed utilizing pol ␤ as the sole polymerase, the synthesis steps of this repair pathway should not be significantly affected by p21.
Long patch BER was reconstituted with all of the proteins listed above and pol ␤ (Fig. 6A). Lane 21 shows that the addition of the p21 peptide to reactions lacking APE1 leads to effective inhibition of the formation of the fully repaired product (as compared with lane 15). Therefore, this process is also affected by p21. Presumably, inhibition occurs because p21 disrupts PCNA interactions with FEN1 and DNA ligase I. Titration of APE1 into the repair reactions (lanes 10 -14, 16 -20, and 22-26) leads to an increase in repair. One possible explanation for this observation is that APE1 may serve as a coordination factor for repair. An equivalent titration of either bovine serum albumin or Escherichia coli single-stranded DNA-binding protein into the reconstitution reactions did not result in any stimulation of repair activity (data not shown). In addition, APE1 stimulates the formation of a significant amount of repair product even in the presence of sufficient p21 to prevent PCNA-directed stimulation (lanes 22-26). Evaluation of the different amounts of APE1directed stimulation in the absence of PCNA, in the presence of PCNA, and in the presence of both PCNA and the p21 peptide reveals that the -fold stimulation by APE1 is greater in the absence versus the presence of PCNA (Fig. 6B).
The repair process was subsequently reconstituted with all of the relevant components, including both pol ␤ and pol ␦ (Fig.  7A). Titration of APE1 into these reactions (lanes 12-16, 18 -22,  and 24 -28) gave results similar to those seen in Fig. 6A. There-  (Fig. 1C) with a ␥-32 P radiolabel at the 5Ј-end of the upstream primer. Reactions containing PCNA included 500 fmol of PCNA (lanes 10 -15, 22-27). The following amounts of the p21 peptide were added (as indicated by the triangles): 50, 100, 150, 200, and 250 ng. The reactions were incubated at 37°C for 10 min. Substrate and product sizes are as indicated. The 61-nt product represents synthesis to the end of the template. Lanes 2 and 3 are control lanes with PCNA only and the p21 peptide only, respectively. nt, nucleotide. fore, APE1 does not appear to specifically affect pol ␦. Again, stimulation was not as prevalent in the presence of PCNA, and inhibition of PCNA interactions by the p21 peptide led to the recovery of some stimulation (Fig. 7B). These results implicate APE1 as an assembly and coordination factor for long patch BER proteins and suggest that APE1 can act to partially compensate for p21-directed inhibition of PCNA. In this way, the cell could utilize APE1, p21, and PCNA as part of a mechanism to differentially regulate BER. DISCUSSION A dilemma arises in explaining the differential regulation of DNA replication and long patch BER because both processes employ a number of the same proteins (11,13,49). We examined the properties of BER reactions reconstituted from puri-fied proteins in vitro to explore this issue. Our results indicate that p21 can inhibit long patch BER, and these results suggest that variances in inhibition of replication and repair observed in vivo relate primarily to the synthesis steps. Furthermore, APE1 has been proposed to be a coordinating factor for the BER process (27,28). Our results suggest that the functional interaction between BER proteins and APE1 partially compensates for the inhibitory effects of p21 on PCNA.
In this study, we analyzed the interactions among the Cterminal PCNA-interacting peptide of p21 and various combinations of BER proteins in reconstitution reactions. Judging from the effects on the concentrations of reaction intermediates and products, the p21 peptide reduced PCNA-directed stimulation of the BER steps catalyzed by FEN1 and DNA ligase I. Because FEN1 and DNA ligase I activities are enhanced by PCNA through a physical interaction (14,15), the mechanism of inhibition evidently involves disruption of the stimulatory interactions by competitive binding. APE1 is a unique component of BER as compared with DNA replication, and the activities of this protein are unaffected by PCNA or p21. Therefore, the repair steps performed by APE1 and the subsequent coordinating roles that APE1 may play will not be attenuated by p21.
Another important step in BER is synthesis by pol ␤ and pol ␦. pol ␤ is the major DNA polymerase involved in the short patch repair pathway (10). Although several studies have indicated that pol ␦ is the major polymerase participating in long patch BER (11)(12)(13), recent studies have suggested a role for pol ␤ in the long patch pathway (40 -42). Podlutsky et al. (40) have proposed that pol ␤ is responsible for incorporating the first nucleotide during repair of reduced AP sites followed by subsequent synthesis and strand displacement by either pol ␤ or pol ␦/⑀. Their model suggests that pol ␤ would dissociate from the repair complex, if AP sites were resistant to ␤-elimination, and pol ␦ would subsequently bind. Because both pol ␤ and pol ␦ are participants in the current model of long patch BER, the effect of the p21 peptide on these polymerases was investigated. Results show that pol ␤ is not influenced by either PCNA or the p21 peptide. Our observations showing PCNA stimulation of pol ␦ activity and p21 inhibition of the PCNA-dependent stimulation are in agreement with those seen previously (23,38). These properties of the two polymerases suggest that p21 induced by DNA damage inhibits PCNA-dependent pol ␦ activity, but pol ␤ is spared. If pol ␤ substitutes and partially compensates for the activity of pol ␦ in vivo, the synthesis steps of BER can continue at p21 levels that stop DNA replication by preventing essential pol ␦-directed DNA primer elongation. Although the long patch mode of repair does not appear to be predominant in mammalian systems, this BER pathway is essential for the removal of AP sites containing an altered sugar (9).
Control of DNA replication and DNA repair by p21 is likely   FIG. 6. APE1 stimulates long patch BER with pol ␤. A, reactions of 20 l containing 50 fmol of DNA substrate were performed as described under "Experimental Procedures." The substrate consists of T 3 :T 4 that was treated with uracil DNA glycosylase and APE1 (Fig. 1D) to be dependent on the differential basis of regulation lying in the primer elongation step. Long tracts of DNA synthesis are required during replication, whereas only short tracts are employed for BER. The need for longer periods of uninterrupted interaction between pol ␦ and PCNA during replication, as compared with repair, should make the replication synthetic reactions relatively more sensitive to transient displacement of PCNA from the polymerase by p21. This would allow actively dividing cells that undergo up-regulation of p21 upon DNA damage to utilize a p21 concentration-dependent threshold to inhibit replication.
Why then has DNA repair evolved to be sensitive to p21? Presumably, when a cell is heavily damaged, inhibition of both DNA replication and DNA repair to promote cell death would be preferred. Such a response would aid natural selection in lower organisms and could inhibit carcinogenesis in higher organisms. In this way, p21 would be required to simultaneously disrupt both replication and repair.
The C-terminal region of p21 appears to be designed to regulate PCNA binding interactions with other proteins (24,25). Because PCNA is responsible for targeting multiple proteins to their substrates (16,50), PCNA stimulates a variety of enzymes (14 -16). The p21 peptide will bind to PCNA, disrupt protein interactions, and decrease stimulatory effects. The inhibition of long patch BER by p21 is the direct result of inhibiting PCNA-directed stimulation of FEN1, DNA ligase I, and pol ␦. Based upon our reconstitution results and the reports of others (27,28), it appears that APE1 is involved in the coordination of BER reactions. Our results further suggest that APE1 is capable of partially compensating for the loss of PCNAdirected stimulation upon binding to p21. APE1 bound to DNA has been reported to interact with pol ␤ to recruit the DNA repair synthesis enzymes to regions of DNA damage (27,28). Our observation that APE1 enhances the efficiency of long patch BER demonstrates a functional consequence of this interaction. Although APE1 also stimulates BER in the presence of PCNA, the -fold stimulation is less than that observed in reactions lacking PCNA. When the p21 peptide is subsequently added, repair efficiency is somewhat reduced, but there is an increase in the -fold stimulation by APE1. This effect with APE1 was observed in reconstituted reactions with pol ␤ as the only polymerase and in reactions containing both pol ␤ and pol ␦. In vivo BER is primarily conducted in the presence of APE1; however, BER may or may not have the benefit of fully functional PCNA. The observation that the stimulatory effects of APE1 and PCNA are not additive is consistent with the need for APE1 to maintain a desirable level of repair efficiency in the absence of functional PCNA. Possibly, there is some intentional overlap in the mechanism of stimulation such that additive stimulation is not attainable when both proteins are active. In fact, additive stimulation may not be needed for the most desirable level of BER. Overall, our results are consistent with a role for APE1 in the maintenance of BER activity when p21 is induced.
In summary, p21 is involved in the regulation of cell cycle progression, DNA replication, and DNA repair. p21 can inhibit BER in vitro; however, the precise biological significance of this inhibition requires further study. The impact of p21 on BER may depend on the intracellular concentration of this regulatory protein at the onset of repair relative to the PCNA level. Additionally, pol ␤ can potentially substitute for pol ␦ after the induction of p21 to alleviate inhibition of PCNA-dependent synthesis during BER. APE1 coordination and stimulation of BER is likely to further compensate for the inhibitory effects of p21 on BER. Therefore, the cell could employ at least two mechanisms to differentially regulate DNA replication and BER.