Differential requirement for proliferating cell nuclear antigen in 5' and 3' nick-directed excision in human mismatch repair.

Proliferating cell nuclear antigen (PCNA) is involved in mammalian mismatch repair at a step prior to or at mismatch excision, but the molecular mechanism of this process is not fully understood. To examine the role of PCNA in mismatch-provoked and nick-directed excision, orientation-specific mismatch removal of heteroduplexes with a pre-existing nick was monitored in human nuclear extracts supplemented with the PCNA inhibitor protein p21. We show here that, whereas 3' nick-directed mismatch excision was completely inhibited by low concentrations of p21 or a p21 C-terminal fusion protein, 5' nick-directed excision was only partially blocked under the same conditions. No further reduction of the 5' excision was detected when a much higher concentration of p21 C-terminal protein was used. These results suggest the following. (i) There is a differential requirement for PCNA in 3' and 5' nick-directed excision; and (ii) 5' nick-directed excision is conducted by a manner either dependent on or independent of PCNA. Our in vitro reconstitution experiments indeed identified a 5' nick-directed excision pathway that is dependent on PCNA, hMutSalpha, and a partially purified fraction from a HeLa nuclear extract.

The DNA mismatch repair (MMR) 1 system plays an important role in maintaining genomic stability in all living organisms, from bacteria to humans. The importance of the MMR system in humans is underscored by the fact that defects in this system are the pathological basis of hereditary non-polyposis colorectal cancer (HNPCC) and other sporadic cancers (1)(2)(3). At the present time, seven proteins (MutS␣, MutS␤, MutL␣, exonuclease 1, polymerase ␦, PCNA, and replication protein A) are known to be required for human MMR, and completion of the repair involves the concerted action of these and other unidentified proteins. Initiation of MMR is thought to occur through the binding of either MutS␣ (the MSH2-MSH6 heterodimer) or MutS␤ (the MSH2-MSH3 heterodimer) to a mismatch or a small insertion/deletion mispair, followed by the recruitment of MutL␣ to form an initiation complex. Removal of the mispaired base(s) occurs by strand excision. In this poorly understood process, a strand break located either 5Ј or 3Ј to the mismatch can serve as a starting point for the unwinding of the DNA duplex to allow exonuclease digestion of the nicked strand. Recent studies have implicated exonuclease I (Exo1), a 5Ј 3 3Ј exonuclease, in both 5Ј and 3Ј nick-directed excision (4,5), but how this enzyme conducts excision for both orientations is not understood. Once the strand is degraded beyond the mismatch, DNA resynthesis occurs through the catalysis of the polymerase ␦ (6) in the presence of PCNA (7) and replication protein A (8,9). The remaining nick is then sealed by an unidentified ligase, completing the repair process (for reviews, see Refs. [1][2][3]. Increasing evidence suggests that PCNA plays an important role in the MMR process. As a cofactor that greatly enhances the processivity of DNA polymerase ␦, PCNA is required for DNA resynthesis in MMR (7). Work by Umar et al. (10) suggested a role for PCNA in MMR at a step prior to DNA resynthesis. Subsequently, PCNA was found to physically interact with MSH6 and MSH3 (11)(12)(13)(14). More recent evidence indicates that PCNA may transfer MutS␣ to the mismatched site (15). Additionally, other studies suggest that PCNA also binds to many different kinds of DNA replication and repair proteins (for a review, see Ref. 16), including the flap endonuclease 1 (FEN-1). However, the involvement of PCNA in MMR initiation/excision is still not fully understood.
In this report, we show a differential requirement for PCNA in 3Ј versus 5Ј nick-directed excision in human MMR. Our data show that, whereas 3Ј nick-directed excision is completely abolished by the PCNA-inhibitor p21, 5Ј nick-directed excision is only partially sensitive to p21. These results suggest that, while 3Ј nick-directed excision absolutely requires PCNA, 5Ј nick-directed excision occurs in both PCNA-dependent and -independent manners. Therefore, we hypothesize that, as seen with Escherichia coli MMR, multiple exonucleases are involved in the human 5Ј nick-directed pathway.

EXPERIMENTAL PROCEDURES
Nuclear Extract Preparation and Protein Purification-HeLa S 3 cells were either purchased from the National Cell Culture Center (Minneapolis, MN) or cultured in our own laboratory, and nuclear extract was prepared as described previously (17). Constructs for the bacterial overexpression of human N-terminal hexahistidine-tagged PCNA and human N-terminal hexahistidine-tagged p21 Cip1/WAF (referred throughout as p21) were generously provided by Jerard Hurwitz (Memorial Sloan-Kettering Cancer Center, New York, NY) and Yue Xiong (University of North Carolina, Chapel Hill, NC), respectively. These proteins were overexpressed in E. coli BL21 DE3 (pLysS) and BL21 DE3 cells, respectively, and purified to homogeneity as described previously (18,19). Constructs coding for the overexpression of a glutathione S-transferase (GST) fusion protein containing the C-or N-terminal domain of p21 (referred to as p21C or p21N, respectively) were generously provided by Anindya Dutta (University of Virginia, Charlottesville, VA), and these proteins were overexpressed in E. coli BL21 DE3 (pLysS) cells and purified to homogeneity using anion exchange and glutathione-Sepharose chromatography essentially as described (20). As originally constructed (21), the amino acid compositions for the p21N and p21C fusion proteins were GST (1-218) -SDP-p21  and GST (1-218) -SDPM-p21 (87-164) , respectively. hMutS␣ was purified from Sf9 insect cells that were co-infected with baculoviruses containing the hMSH2 and hMSH6 cDNAs (a gift from Joe Jiricny, University of Zü rich, Switzerland) as described (22). Before use, all proteins were centrifuged at 16,000 ϫ g for 15 min at 4°C to remove aggregated material, and the concentrations of the protein in solution were determined from UV absorbance measurements of native samples using extinction coefficients calculated from the amino acid sequences of the protein. All proteins were judged to be Ͼ95% pure.
Heteroduplex Preparation-Heteroduplexes used in this study contained a single G-T mismatch and a strand break at either 125 bp 5Ј to the mismatch (5Ј G-T substrate) or 172 bp 3Ј to the mismatch (3Ј G-T substrate, see Fig. 1). The 5Ј G-T substrate was constructed by hybridizing Sau96I-digested f1MR3 dsDNA with f1MR1 single-stranded DNA and purified essentially as described previously (23). The 3Ј G-T substrate was constructed by hybridizing PstI-digested M13mp18-UKY1 dsDNA with M13mp18-UKY2 single-stranded DNA. First, doublestranded M13mp18 DNA was double-digested with BsmBI and BtsI, and the larger fragment was purified and ligated with the oligonucleotide duplex 5Ј-ctggTCGACCATTAGCATTAGCGGCATCTTACACTGC-Aag-3Ј, where the nucleotides in small case indicate overhangs and the nucleotides in capital case represent the duplex. The resulting M13mp18 derivative was used to construct-UKY1 and UKY2 phages by replacing the BamH I-SpHI sequence of the derivative with oligonucleotide duplex I (UKY1; 5Ј-GATCATGCACTCGAGACATG-3Ј and 3Ј-TACGTGAGCTCT-5Ј) and oligonucleotide duplex II (UKY2; 5Ј-GATC-ATGCATTCGAGACATG-3Ј and 3Ј-TACGTAAGCTCT-5Ј), respectively, where the bold typed base pairs are the only difference between these two duplexes. The resulting phages were confirmed by DNA sequencing.
Nick-directed Excision Assay-Nick-directed excision was assayed essentially as described previously (17) in a 15-l reaction mixture containing 100 ng of heteroduplex DNA, 50 g of nuclear extract, 10 mM Tris-HCl (pH 7.6), 5 mM MgCl 2 , 110 mM KCl, and 1.5 mM ATP. No exogenous dNTPs were added to the mixture to minimize DNA resynthesis. After incubation at 37°C for various times, reactions were terminated by proteinase K digestion, and DNA samples were recovered. To score the amount of excision beyond the mismatch, the 3Ј substrate was digested with the restriction enzymes BseRI and HindIII (the scoring enzyme; see Fig. 1), and the 5Ј substrate with BspDI and NheI (the scoring enzyme; see Fig. 1). The products were electrophoresed on 1% agarose gels, and DNA bands were visualized by UV illumination in the presence of ethidium bromide. For the 5Ј substrate, the band located at 6.4 kb corresponds to excision products (loss of NheI site). For the 3Ј substrate, the band located at 7.1 kb corresponds to excision products (loss of HindIII site). When p21 or the p21C was added, it was pre-incubated with extract for 15 min on ice before assembling the reactions.
Southern Blot Analyses to Detect DNA Excision-DNA excision intermediates were visualized using the Southern blotting technique performed essentially as described (24). Briefly, MMR reactions were carried out as described above, and DNA samples were recovered by phenol extraction and ethanol precipitation. For the 5Ј substrate, DNA was digested with SspI, separated on 6% denaturing polyacrylamide gels, and electrotransferred onto a nylon membrane. Membranes were exposed to a 32 P end-labeled oligonucleotide probe (5Ј-ATTGTTCTG-GATATTACC-3Ј) that is complementary to the sequence on the nicked strand near the SspI restriction site (as indicated on Fig. 1). For the 3Ј substrate, DNA was digested with SspI and DraIII, and excision intermediates were probed with 32 P-labeled oligonucleotide 5Ј-AATTTA-ACGCGAATTTTAAC-3Ј, which is complementary to the sequence on the nicked strand near the SspI restriction site. Reaction products were visualized by autoradiography.

RESULTS
p21 Inhibits 3Ј but Not 5Ј Nick-directed Mismatch Excision-To more closely examine the role of PCNA in MMR prior to the DNA resynthesis step, we monitored the time course of mismatch-provoked, strand-specific excision in HeLa nuclear extract in the presence or absence of p21 Cip1/WAF1 (p21), a protein that tightly binds to PCNA and blocks MMR (7, 10) and other DNA metabolic pathways that require PCNA (21,25). A characteristic of MMR-associated excision is that the process starts at the strand break, proceeds to the mismatch along the shorter path between the two sites, and stops at a point ϳ150 nucleotides (nt) beyond the mismatch, thereby generating a single stranded gap on the substrate (26). Taking advantage of this property, we constructed nicked circular heteroduplex substrates that contain a restriction endonuclease site immediately 3Ј or 5Ј (depending on the excision orientations) to the mismatch (see Fig. 1). Thus, under conditions of limited DNA synthesis (i.e. in the absence of exogenous dNTPs), orientationspecific excision by MMR can be scored by conversion of the restriction sequence (NheI for the 5Ј substrate and HindIII for the 3Ј substrate) from double-stranded DNA to single-stranded DNA, rendering the DNA resistant to the enzyme (4, 27). Fig. 2 shows a time course of mismatch-provoked excision by this analysis. At time zero, almost all of the DNA substrate (both 5Ј and 3Ј substrates) could be digested into two smaller fragments, indicative of no generation of a single-stranded DNA gap (i.e. no excision). As incubation time increased, reactions without p21 accumulated a larger molecular weight DNA species for both the 5Ј and 3Ј substrates, suggesting that a single strand gap extended (from the pre-existing strand break) beyond the HindIII (3Ј substrate) or NheI (5Ј substrate) site, because the excision products were resistant to digestion by the corresponding scoring enzymes (Fig. 2, A and B). At 25 min, the accumulation of the larger species of DNA reached a plateau, with 40 and 32% for 3Ј and 5Ј substrates, respectively. These numbers are essentially equivalent to the percentage of heteroduplexes repaired by HeLa nuclear extracts (data not shown; Ref. 17). The great majority of the excision apparently occurred in a manner dependent on a functional MMR system and on the presence of a mismatch, because nuclear extracts derived from hMLH1-deficient HCT116 (H6) cells (28,29) could generate no more than 10% of HindIII-or NheI-resistant products ( Fig. 2E and data not shown) after 25 min of incubation. Additionally, much less excision was also observed in reactions containing a homoduplex (Fig. 2E).
Interestingly, when p21 was added to the reaction to inhibit The 5Ј G-T substrate (6.4 kb in size) was constructed from the f1MR series, and the 3Ј substrate (7.1 kb in size) was from the M13mp18-UKY series, as described under "Experimental Procedures." In addition to a G-T mismatch, the substrates also contain a strand break located either 125 bp 5Ј (5Ј G-T substrate) or 172 bp 3Ј (3Ј G-T substrate) to the mismatch. Several restriction enzymes used for scoring excision or repair are shown. The black bars indicate where the probes used in Southern blot assay bind. endogenous PCNA, there was a large difference in nick-directed, mismatch-provoked excision between reactions containing 5Ј and 3Ј substrates. For the 3Ј substrate, p21 at a concentration of 1.5 M reduced gapped molecules by ϳ 75% (compare A and C, Fig. 2,). However, little reduction in gapped molecules was detected in reactions containing the 5Ј substrate under the same conditions (compare B and D, Fig. 2). These results strongly suggest a differential requirement for PCNA in 5Ј and 3Ј nick-directed, mismatch-provoked excision. To determine whether a higher concentration of p21 could completely block the 5Ј nick-directed excision, titration experiments were performed. We found that Ͼ95% of 3Ј nick-directed excision was inhibited by p21 at a concentration of 3 M, but p21 had no significant effect on 5Ј-directed excision at concentrations approaching 10 M (data not shown). Because of limitations in the solubility of recombinant p21, the highest useable concentration of this protein in the in vitro assay was at 10 M. To confirm the results obtained with the full-length p21 protein, a fusion protein containing the C-terminal portion of p21 (p21C), a domain that is known to bind to PCNA with very high affinity (20), was used in this analysis. Fig. 3 shows the dependence of the level of MMR-dependent excision for both the 5Ј and 3Ј substrates on the concentration of p21C protein used over a two-order of magnitude range. Whereas p21C at a concentration of 0.3 M completely blocked 3Ј nick-directed excision, the majority (63%) of 5Ј nick-directed excision was insensitive to p21C at much higher concentrations (from 0.3-3 M). For a negative control, we also carried out similar experiments using a fusion protein containing the N-terminal domain of p21, which does not bind to PCNA (20). No obvious inhibition in excision was detected for either substrate at any of the concentrations of this protein used (data not shown). The results obtained using p21C largely confirm those that were obtained using the recombinant full-length p21 protein described above, i.e. there is a different utilization of PCNA in 3Ј and 5Ј nickdirected excision in human MMR. It is also noteworthy that, although the majority of the 5Ј nick-directed excision was tolerant to high concentrations of p21C, a significant fraction (37%) of the excision was blocked by the fusion protein at lower concentrations (Fig. 3). As demonstrated below in our in vitro reconstitution experiments, the p21C-inhibited 5Ј nick-directed mismatch excision is dependent on both a mismatch and hMutS␣. Therefore, these experiments suggest that the 5Ј nickdirected excision may contain both the PCNA-dependent and -independent components.
To confirm the observation made using the restriction enzyme assay, Southern blot analysis was performed to directly visualize excision intermediates under conditions of limited DNA synthesis (24). DNA products were digested with SspI (in the case of 5Ј substrate) or SspI-DraIII (in the case of 3Ј substrate), and excision intermediates were mapped using a probe that hybridizes at a site on the nicked strand well beyond the mismatch (see the position of the probes in Fig. 1). As shown in Fig. 4, nicked strands that undergo mismatch-provoked excision are indicated by the appearance of shortened fragments on the gel as compared with the nick strand of the un-reacted substrate (Fig. 4, lane 1 for 3Ј G-T and lane 5 for 5Ј G-T). As

FIG. 2. p21 differentially blocks orientation-specific excision in MMR.
A-D, 100 ng of 3Ј or 5Ј G-T substrate were incubated with 50 g of HeLa nuclear extract in the absence of exogenous dNTPs at 37°C as described under "Experimental Procedures," and samples were removed at various time points for analysis. DNA was purified and digested with two restriction enzymes (HindIII and BseRI for the 3Ј G-T substrate, and NheI and BspD 1 for the 5Ј substrate) and electrophoresed on 1% agarose gels to analyze the extent of strand excision. Formation of the top band (arrow) indicates loss of a restriction site (HindIII for 3Ј G-T substrate, and NheI for 5Ј G-T substrate) due to excision of the nicked strand beyond this site. A, reactions with 3Ј G-T substrate in the absence of p21. B, reactions with 5Ј G-T substrate in the absence of p21. C, reactions with 3Ј G-T substrate in the presence of p21. D, reactions with 5Ј G-T substrate in the presence of p21; when present, exogenous p21 concentration added was 1.5 M. E, excision of 5Ј G-T and 5Ј A-T by MMR-proficient (HeLa) and MMR-deficient (H6) nuclear extracts in the presence or absence of p21 under limited DNA synthesis conditions. 100 ng of DNA substrates were incubated with 50 g of nuclear extracts at 37°C for 20 min, and excision was scored by the resistance of DNA substrates to NheI as described above. expected, shortened fragments (Ͻ669 nt for the 3Ј substrate and Ͻ544 nt for the 5Ј substrate) were generated in the nicked strand when either the 3Ј G-T (lane 2) or the 5Ј G-T (lane 6) was incubated with HeLa nuclear extracts under conditions of limited DNA synthesis, indicative of occurrence of excision. For the 3Ј G-T substrate, this fragment pattern was found to largely disappear upon the addition of p21C to the reaction (Fig. 4, lane  3), suggesting that excision of the 3Ј substrate was inhibited by p21C in these conditions. The p21C inhibition on 3Ј substrate excision was apparently reversed when excess exogenous PCNA was added to the reaction mixture, as evidenced by the fact that excision intermediates reappeared in the reaction (Fig. 4, lane 4). In contrast to the results seen for the 3Ј substrate, the 5Ј substrate was found to undergo excision in the presence of p21C at levels similar to that seen in the absence of the fusion protein (compare lanes 6 and 7, Fig. 4). These observations further confirm that there is a differential requirement for PCNA in 3Ј and 5Ј nick-directed excision in human MMR.

p21 Inhibits MMR of PCNA-independent 5Ј Nick-directed Excision at the DNA Resynthesis
Step-Previous studies have demonstrated that p21 completely inhibits MMR for both the 5Ј and 3Ј substrates (10). Our data show here that the inhibition for 3Ј substrates is at a step prior to or at excision, but this is largely not the case for the 5Ј substrate. To determine whether p21 blocks 5Ј nick-directed MMR at the DNA resynthesis step, repair assays were carried out in the presence of exogenous dNTPs, and repair products were assayed for the formation of a homoduplex at the mismatch site by virtue of their sensitivity to NsiI (for 3Ј G-T), or HindIII (for 5Ј G-T) (see Fig. 1), followed by Southern blot analysis. As expected, a band (497 nt in length for 3Ј G-T and 414 nt for 5Ј G-T; see Fig. 5, lanes 2 and 6,  respectively) representing the repaired products was detected under normal conditions. When p21C was added to these reactions, the band representing the repair products was not detected for either substrate (Fig. 5, lanes 3 and 7 for 3Ј and 5Ј G-T, respectively). For the 3Ј substrate, two major NsiI-resistant bands were observed; the top band (964 nt in size) was the directly ligated substrate, and the lower band was the unreacted, nicked substrate. In the case of the 5Ј substrate, many bands smaller than 544 nt (Fig. 5, lane 7) were evident, indicative of the formation of excision intermediates. This result suggests that the excision in the reaction occurred, but the resynthesis was inhibited. However, the defect in DNA resynthesis was restored by the addition of excess amount of exogenous PCNA to the reaction (lane 8) as judged by the disappearance of the excision intermediates and the concomitant appearance of the repair product, i.e. the 414-nt band. These observations indicate that p21 inhibits 5Ј nick-directed MMR largely at the step of DNA resynthesis.
Partial Reconstitution of the PCNA-dependent 5Ј Excision-Although the 5Ј nick-directed mismatch excision is much less sensitive to p21 or p21C as compared with the 3Ј nick-directed excision, the addition of p21C at a relatively low concentration (0.6 M) reduced the 5Ј nick-directed excision by 37%, suggesting an involvement of PCNA in the reaction. To explore this possibility, we utilized a partially purified phosphocellulose fraction (PF) that is required for mismatch-provoked, nickdirected excision in in vitro MMR. 2 We incubated this fraction along with a 5Ј G-T substrate in the presence or absence of exogenous hMutS␣ or PCNA. As shown in Fig. 6A, either PF alone or PF with either exogenous hMutS␣ or PCNA catalyzed minimal amounts of excision activity on the 5Ј G-T substrate (lanes 2-4). However, when all three components were mixed together (lane 5), 5Ј nick-directed excision was significantly stimulated compared with the levels found for the two component mixtures (PF-PCNA or PF-hMutS␣). Notably, the level of excision using this partially purified extract and purified proteins reached a level that was even higher than that found for the whole cell nuclear extract. The stimulation of excision in  4. Visualization of PCNA-dependent and -independent excision tracts by Southern analysis. 200 ng of 3Ј or 5Ј substrate were incubated with 100 g of HeLa nuclear extract, as indicated, in the absence of exogenous dNTPs at 37°C for 20 min. Repair DNA products in all reactions were digested with SspI (5Ј substrate) or SspI and DraIII (3Ј substrate) and fractionated on denaturing 6% polyacrylamide gel electrophoresis, followed by electrotransfer onto nylon membranes. The membranes were blotted with 32 P end-labeled probes as described under "Experimental Procedures." Corresponding restriction fragments for each substrate are diagrammed on the left side for the 3Ј substrate and the right side for the 5Ј substrate. Black bars indicate where the probe binds. The top band seen in lanes 2-4 and lanes 6 -8 represents the substrate that is ligated before being processed by the MMR system (26).
FIG. 5. p21 blocks 5 nick-directed MMR at the step of resynthesis. 200 ng of 3Ј or 5Ј substrate were incubated with 100 g of HeLa nuclear extract, as indicated, in the presence of exogenous dNTPs at 37°C for 20 min. Repair DNA products from the 5Ј substrate were digested with HindIII (to score the conversion of heteroduplex to homoduplex) and SspI (to fragment the substrate of interest for Southern blot analysis), and DNA products derived from the 3Ј substrate were cut with NsiI (to score for repair) and SspI-DraIII (to isolate the fragment of interest). The digested DNA was fractionated using denaturing 6% polyacrylamide gel electrophoresis and then electrotransferred onto nylon membranes. The membranes were blotted with 32 P-end-labeled probes as described under "Experimental Procedures." Corresponding restriction fragments for each substrate are diagrammed on the left side for the 3Ј substrate and the right side for the 5Ј substrate. Black bars indicate where the probe binds. The top band seen in lanes 2-4, and lanes 6 -8 represents a substrate that is ligated before being processed by the MMR system (26). this reaction, however, was completely inhibited by p21C (Fig.  6A, lanes 7 and 8); hence, our results suggest that the PFcatalyzed excision activity requires both hMutS␣ and PCNA.
Similar experiments were performed using a 5Ј A-T homoduplex as a substrate in otherwise identical conditions. Excision of this substrate was minimal with PF alone or PF incubated with either hMutS␣ or PCNA. However, unlike that seen for the heteroduplex, excision was not stimulated by the addition of all components beyond the levels seen for the two component mixtures (PF-PCNA or PF-hMutS␣; Fig. 6A, lanes 10 -12), and the excision was insensitive to p21C (lanes [13][14][15]. These results indicate that the homoduplex-directed excision is a nonspecific excision and that the 5Ј nick-directed hMutS␣and PCNA-dependent excision must occur in a manner dependent on a mismatch. The PF fraction was unable to catalyze any 3Ј nick-directed excision (see Fig. 6B), and this result likely occurs because the fraction does not contain any 3Ј-directed exonucleases. From these results (Fig. 6) and the fact there is a considerable amount of 5Ј-directed and mismatch-provoked excision that is not blocked by p21C (Fig. 3), we conclude that there are at least two types of 5Ј nick-directed mismatch excision in human cells, one that is dependent on PCNA and another that is not. DISCUSSION In this report, we demonstrate a differential requirement for PCNA for 3Ј versus 5Ј nick-directed mismatch removal, suggesting that distinct mechanisms are used for mismatch excision at different orientations. This is demonstrated in Figs. 2 and 3, where p21 completely blocks 3Ј but not 5Ј nick-mediated excision. Our work supports previous observations that PCNA is required for MMR at a step prior to or at mismatch-provoked excision (10), in addition to participating in the step of DNA resynthesis (7). We also find, however, using a partially purified nuclear extract, that 5Ј nick-directed excision is likely to be carried out using both PCNA-dependent and -independent manners.
It is known that mismatch excision requires an assembly of MMR initiation factors at the mismatch site, followed by recruitment of exonuclease(s) to execute excision. PCNA has been shown to interact with human MutS and MutL homologs (7, 10 -14) and may function to recruit these proteins to the mismatch site (15), indicating an involvement of PCNA in the assembly of the MMR initiation complex. On the other hand, PCNA may help to recruit or activate exonuclease(s). This assumption is based on the fact that PCNA is known to interact with and greatly enhance the activity of FEN-1 (reviewed in Ref. 16), a nuclease that may play a role in MMR (30), and that PCNA may be capable of interacting with other nucleases. However, it is not known whether the PCNA differentiation for mismatch excision in different orientations is made at the repair initiation level or at the excision level, or both.
Using the p21 C-terminal fusion protein, we show that even though the majority (63%) of 5Ј nick-directed mismatch excision seemed to be insensitive to p21C (or independent of PCNA), more than one-third of excision was inhibited by the fusion protein (Fig. 3). This phenomenon suggests that 5Ј nickdirected excision can occur in both PCNA-dependent and -independent manners. Available evidence appears to support this hypothesis. In our in vitro reconstitution experiments, we show that PCNA-hMutS␣ greatly stimulated 5Ј nick-directed excision, suggesting the presence of a PCNA-dependent excision pathway (Fig. 6). Although Exo1 (a 5Ј 3 3Ј exonuclease) has been shown to be involved in 5Ј nick-directed MMR (4, 5), the PCNA-dependent excision we observed may not be conducted by Exo1. During the preparation of this manuscript, a study was published that demonstrated that human Exo1 catalyzed 5Ј nick-directed excision in a manner independent of PCNA (31). Based on the fact that PF-catalyzed 5Ј excision relies on PCNA and that HeLa extract-catalyzed 5Ј-excision is largely insensitive to p21C, we hypothesize that there are at least two types of nucleases that are involved in 5Ј nick-directed excision in human MMR, i.e. one whose activities are not enhanced by PCNA (i.e. Exo1), and another whose activates are enhanced by PCNA (and whose identities remain to be identi- FIG. 6. Involvement of PCNA in 5 nick-directed excision. DNA substrates were incubated with reaction mixtures containing different combinations of 1.5 g of a PF, 26 nM hMutS␣, and 67 nM of PCNA, as indicated, for 20 min at 37°C. DNA products were digested with NheI and BspDI (5Ј substrate) or HindIII and BseRI (3Ј substrate) to score for the generation of gapped molecules as described under "Experimental Procedures." When present, p21C concentration was 0.06, 1.6, or 4.8 M, and bovine serum albumin (BSA) was 100 ng. PF was obtained by fractionating HeLa nuclear extracts in a phosphocellulose column. Briefly, HeLa nuclear extracts were loaded onto a phosphocellulose column equilibrated with buffer A (20 mM Hepes, pH 7.6, 2 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 1 mM EDTA, and 1 g/ml leupeptin) containing 0.1 M KCl. After extensive washing with buffer A containing 0.3 M KCl, the column was eluted with the same buffer containing 0.35 M KCl. The protein peak fractions were pooled, concentrated, and stored in small aliquots at Ϫ80°C before use. A, comparison of 5Ј nick-directed excision between a heteroduplex and a homoduplex in the partially reconstituted reaction. B, 3Ј nick-directed excision in the partially reconstituted system. fied). The involvement of multiple exonucleases for the 5Ј 3 3Ј orientation excision in the human cells would be homologous to the E. coli MMR system, as at least four exonucleases (two for each orientation) are implicated in mismatch excision in E. coli (32). Evidence from recent Exo1 knockout studies also supports this hypothesis (5). First, even though Exo1 Ϫ/Ϫ mice display increased Hprt mutability, the mutation rate is 5-fold lower than that in Msh2 Ϫ/Ϫ mice. Second, Exo1 Ϫ/Ϫ cells still possess residual activity (ϳ20% compared with wild type) to repair a 1-nt insertion/deletion mispair, a heteroduplex that is believed to be processed only by the MMR pathway. Finally, whereas all Msh2 Ϫ/Ϫ animals were dead of cancer by 12 months of age, only 50% of Exo1 Ϫ/Ϫ mice died at 17 months (5). The weaker mutator phenotype and reduced tumorigenicity in Exo1 Ϫ/Ϫ mice suggest that MMR activity is not completely blocked in Exo1 Ϫ/Ϫ mice, which is consistent with a notion that Exo1 is not the only nuclease involved in mammalian MMR. Additionally, evidence from yeast studies has suggested that, in addition to Exo1, additional nucleases, such as Rad27 (also called Rth1, a FEN-1 homolog) and the exonuclease activities associated with DNA polymerases ␦ and ⑀, may also function at the excision step of MMR (30,33,34).