DNA Mismatch Repair Interacts with CAF-1- and ASF1A-H3-H4-dependent Histone (H3-H4)2 Tetramer Deposition*

DNA mismatch repair (MMR) is required for the maintenance of genome stability and protection of humans from several types of cancer. Human MMR occurs in the chromatin environment, but little is known about the interactions between MMR and the chromatin environment. Previous research has suggested that MMR coincides with replication-coupled assembly of the newly synthesized DNA into nucleosomes. The first step in replication-coupled nucleosome assembly is CAF-1-dependent histone (H3-H4)2 tetramer deposition, a process that involves ASF1A-H3-H4 complex. In this work we used reconstituted human systems to investigate interactions between MMR and CAF-1- and ASF1A-H3-H4-dependent histone (H3-H4)2 tetramer deposition. We have found that MutSα inhibits CAF-1- and ASF1A-H3-H4-dependent packaging of a DNA mismatch into a tetrasome. This finding supports the idea that MMR occurs before the DNA mismatch is packaged into the tetrasome. Our experiments have also revealed that CAF-1- and ASF1A-H3-H4-dependent deposition of the histone (H3-H4)2 tetramers does not interfere with MMR reactions. In addition, we have established that unnecessary degradation of the discontinuous strand that takes place in both DNA polymerase δ (Pol δ)- and DNA polymerase ϵ (Pol ϵ)-dependent MMR reactions is suppressed by CAF-1- and ASF1A-H3-H4-dependent deposition of the histone (H3-H4)2 tetramers. These data suggest that CAF-1- and ASF1A-H3-H4-dependent deposition of the histone (H3-H4)2 tetramers is compatible with MMR and protects the discontinuous daughter strand from unnecessary degradation by MMR machinery.

The DNA mismatch repair (MMR) 3 system has been conserved from bacteria to humans (1)(2)(3). Genetic stabilization provided by the MMR system suppresses both sporadic and inherited cancers (4). The MMR system has multiple functions that are involved in the genome maintenance (1)(2)(3)(5)(6)(7)(8). Among these functions MMR is the strongest contributor to the suppression of spontaneous mutation rates. Significant progress has been made in understanding MMR in Escherichia coli and eukaryotes (1,2,6,8,9). In E. coli, MMR is initiated by binding of the mismatch recognition factor MutS to a mismatch (10,11). After mismatch recognition, MutS recruits MutL in an ATP-dependent manner (12,13). The MutS-MutL complex activates the MutH endonuclease to nick the newly synthesized DNA strand at a transiently unmethylated GATC site that may be as far as 1 kb away from the mismatch (11). The MutH nick is the loading site for the excision complex that consists of UvrD (helicase II) and one of the four exonucleases (ExoI, ExoVII, ExoX, and RecJ) (14 -17). The excision complex unwinds and excises a portion of the newly synthesized strand encompassing the mismatch. The gap is filled in by the DNA polymerase III holoenzyme, and the DNA ligase seals the nick (18).
Eukaryotic MMR is more efficient on the lagging than leading strand (19). One-nucleotide deletion loops, mispaired bases, and one-nucleotide insertion loops are the most common substrates for the eukaryotic MMR reaction (1,2,6,7). In addition to these lesions, the eukaryotic reaction removes oxidized bases, one-nucleotide flaps, and single ribonucleotides that produce mispairs in DNA (20 -23). Eukaryotes have two mismatch recognition factors, MutS␣ (MSH2-MSH6 heterodimer) and MutS␤ (MSH2-MSH3 heterodimer); both are homologs of E. coli MutS (24 -26). The concentration of MutS␣ in HeLa cells is ϳ10 times higher than that of MutS␤, suggesting that the majority of MMR events in human cells involve MutS␣ (27). Consistent with this idea, MSH6, but not MSH3, is important for the suppression of carcinogenesis in humans and mice (4,28). Eukaryotic MMR is initiated by recognition of the mismatch by MutS␣ or MutS␤. Upon mismatch recognition, the MutS homolog and PCNA (29,30) loaded by RFC (31) activate the MutL homolog MutL␣ (MLH1-PMS2 heterodimer; Refs. 32 and 33-36). The activated MutL␣ incises the discontinuous daughter strand 5Ј and 3Ј to the mismatch. Incision of the discontinuous strand by MutL␣ 5Ј to the mismatch is required for the excision-dependent and excision-independent MMR pathways (33,(37)(38)(39)(40)(41)(42)(43)(44). In the excision-dependent MMR pathway, MutS␣-activated Exonuclease 1 (EXO1) removes a DNA segment containing the mismatch in a 5Ј 3 3Ј excision reaction that initiates from a 5Ј nick produced by MutL␣ endonuclease, and the generated gap is repaired by DNA polymerase ␦ (Pol ␦) holoenzyme (21, 33, 37-43, 45, 46). In the excision-independent pathway, a new 3Ј end produced by the MutL␣ incision of the discontinuous daughter strand 5Ј to the mismatch is extended by Pol ␦ holoenzyme in a DNA synthesis reaction that displaces a part of the original strand containing the mismatch (44).
Eukaryotic MMR occurs in the nucleosomal environment. The nucleosome is the major building block of chromatin. Assembly of the nucleosome is a conserved process that can be divided into two major steps. In the first step the tetrasome, a complex of the histone (H3-H4) 2 tetramer and ϳ146-bp DNA, is formed as a result of deposition of the tetramer onto DNA (47). In the second step the tetrasome is converted into the nucleosome by the addition of two histone H2A-H2B dimers. The daughter strands that emerge from the eukaryotic replication fork are rapidly packaged into nucleosomes by replicationcoupled nucleosome assembly (48,49). The histone chaperone CAF-1 orchestrates replication-coupled nucleosome assembly by depositing the histone (H3-H4) 2 tetramers onto the newly synthesized DNA (50 -53). The protein-protein interaction between CAF-1 and loaded PCNA ensures that the histone chaperone is able to act behind the replication fork (54 -56). The histone chaperones ASF1A and ASF1B (57) assist CAF-1 in replication-coupled deposition of the histone (H3-H4) 2 tetramers (58). ASF1A and ASF1B form complexes with newly synthesized histone H3-H4 dimers in cytosol and participate in their transport into the nucleus (59). In the nucleus, the heterotrimeric ASF1A-H3-H4 and ASF1B-H3-H4 complexes are thought to transfer their H3-H4 dimers onto CAF-1 molecules (60). The CAF-1-(H3-H4) 2 intermediate next forms a complex with loaded PCNA. After formation of the complex with PCNA, CAF-1 loads the histone (H3-H4) 2 tetramer onto the nascent DNA. In addition to their roles in deposition of newly synthesized histone (H3-H4) 2 tetramers, ASF1A and ASF1B have also been implicated in CAF-1-dependent loading of parental (H3-H4) 2 tetramers (61).
Recent studies have suggested that MMR coincides with CAF-1-dependent assembly of the newly synthesized DNA into nucleosomes (56,62). In this work we analyzed reconstituted human systems that support MMR and CAF-1-and ASF1A-H3-H4-dependent histone (H3-H4) 2 tetramer deposition. We found that there are several interactions between the two processes. The presence of these interactions supports the view that eukaryotic MMR occurs during replication-coupled nucleosome assembly but precedes the incorporation of the mismatch into the tetrasome.

Experimental Procedures
Proteins-Human ASF1A-H3-H4, CAF-1, histone H3-H4 complex, EXO1, MutL␣, MutS␣, MutL␣-E705K, PCNA, Pol ␦, RFC, and RPA were prepared in near homogenous forms as previously described (33,42,44,53,56). The four-subunit human Pol ␦ FLAG-tagged at the p125 subunit N terminus and His 6 -tagged at the p66 subunit N terminus was expressed in baculovirus-infected Sf9 insect cells and purified using anti-FLAG M2 affinity beads (Sigma) and Mono S and Mono Q columns (GE Healthcare). The purity of Pol ␦ obtained at the final purification step was Ͼ95%. Wild-type four-subunit human Pol ⑀ and a four-subunit Pol ⑀ variant carrying the p261 D860A and D862A substitutions were FLAG-tagged at the N terminus of the p261 subunit, expressed in baculovirus-infected insect Sf9 cells, and purified with anti-FLAG M2 affinity beads (Sigma) and Mono S and Mono Q columns (GE Healthcare). After the final purification step, the wild-type Pol ⑀ and the mutant were 90% pure. Human His 6 -GST-ASF1A was expressed in E. coli (63) and purified on Ni 2ϩ beads (Affymetrix), a Mono Q column (GE Healthcare), and a glutathione-Sepharose column (GE Healthcare). The purified His 6 -GST-ASF1A was then treated with His 6 -Tev protease, and ASF1A was separated from the His 6 -GST tag and protease by chromatography on Ni 2ϩ beads (Affymetrix) and a Mono Q column (GE Healthcare). The purified ASF1A was 85% pure.
DNA Substrates and Oligonucleotides-Preparation of 3Ј-nicked G-T DNA (3Ј G-T DNA) and 3Ј-nicked A-T DNA (3Ј A-T DNA), a relaxed covalently closed DNA (ccDNA), and single-stranded DNA primed with 12 oligonucleotides was described previously (33,42,53). 3Ј-Nicked G-T DNA (6.4 kb) carries a strand break that is 141 bp 3Ј to a G-T mispair, and 3Ј-nicked A-T DNA is the same as 3Ј-nicked G-T DNA except that it contains an A-T pair instead of the G-T mispair (42). DNA sequences of oligonucleotides that were labeled with 32 P by T4 polynucleotide kinase (New England BioLabs) and used as hybridization probes are shown in Table 1.
Histone (H3-H4) 2 Tetramer Deposition Assay-Histone tetramer (H3-H4) 2 deposition assay was based on a previously described procedure (56). Each histone tetramer (H3-H4) 2 deposition reaction was carried out in a 40-l mixture containing 20 mM HEPES-NaOH (pH 7.4), 110 mM KCl, 5 mM MgCl 2 , 2 mM ATP, 4 mM DTT, 0.2 mg/ml BSA, 1-3% (v/v) glycerol, and 0.6 nM (0.1 g) DNA (3Ј-nicked G-T DNA, 3Ј-nicked A-T DNA, or relaxed ccDNA). When indicated, human CAF-1 (15 nM), ASF1A-H3-H4 (46 nM), PCNA (21 nM), RFC (3 nM), and MutS␣ (11 nM, 22 nM, or 44 nM) were present in the reaction mixtures. After a 5-min incubation at 37°C, 35-l fraction of each reaction mixture was added to a 5-l micrococcal nuclease mixture (20 mM HEPES-NaOH (pH 7.4), 20 mM CaCl 2 , and 2.4 units/l micrococcal nuclease), and the reaction temperature was changed to 21-23°C. The DNA cleavage was carried out for 5 min and then terminated by the addition of a 4-l mixture (0.5% SDS, 70 mM EDTA, 40% glycerol, 2.5 mg/ml proteinase K, and 2.5 g/ml HindIII-cleaved plasmid DNA pAH1A (a gel loading control)). The proteins were digested at 50°C for 15-20 min, and DNA products were separated on native 1.8% agarose gels. The separated DNA products were transferred onto nylon membranes and analyzed by Southern hybridizations with the indicated 32 P-labeled probes. 32 P-Labeled DNA species were visualized with a Typhoon biomolecular imager (GE Healthcare) and quantified using the ImageQuant software (GE Healthcare). MMR Assay-The MMR assay was carried out according to a published method (44). This assay is based on the observation that nick-directed repair of the G-T mispair on 3Ј-nicked G-T DNA restores the HindIII site (43,44,64). Each MMR reaction was carried out in a 60-l mixture containing 20 mM HEPES-NaOH (pH 7.4), 110 mM KCl, 5 mM MgCl 2 , 2 mM ATP, 4 mM DTT, 0.4 mg/ml BSA, 3-5% (v/v) glycerol, 0.2 mM dGTP, 0.2 mM dATP, 0.2 mM dTTP, 0.2 mM dCTP, 0.6 nM (0.15 g) 3Ј-nicked G-T DNA, and the indicated purified human proteins. When human ASF1A-H3-H4, CAF-1, EXO1, MutL␣, MutS␣, PCNA, RFC, and RPA were present in the reaction mixtures, their concentrations were 46, 23, 3, 6, 22, 21, 3, and 52 nM, respectively. The human FLAG-tagged Pol ␦ concentration in the reaction mixtures was 0, 0.1, 0.2, 1, 5, or 10 nM, and the human Pol ⑀ concentration in the reaction mixtures was 0, 1, 2, 5, or 10 nM. MMR reactions were incubated at 37°C for 20 min, and each reaction was stopped by the addition of a 45-l solution containing 0.35% SDS, 0.4 M NaCl, 0.3 mg/ml proteinase K, 0.7 mg/ml glycogen, and 13 mM EDTA. After an incubation at 50°C for 20 min, DNA products were extracted with a phenol/ chloroform mixture and precipitated with isopropyl alcohol. The pellets were washed with 75% ethanol and dissolved in the Tris EDTA buffer. To score MMR, a fraction of the recovered DNA was digested with ClaI and HindIII, and the cleavage products separated on a native 1% agarose gel were visualized by ethidium bromide staining. Quantification of the repair products was performed using the ImageJ program. To visualize degradation of the discontinuous strand that occurs in an MMR reaction, a fraction of the recovered DNA was cleaved with ClaI, and the cleaved DNA was resolved on a denaturing agarose gel and hybridized with the indicated 32 P-labeled probe. 32 P-labeled DNAs were visualized with a Typhoon biomolecular imager (GE Healthcare), and quantification of the discontinuous strand degradation was performed using the ImageQuant program.
MutS␣ Inhibits CAF-1-, ASF1A-H3-H4-, PCNA-, and RFCdependent Formation of Mismatch-containing Tetrasomes-Having reconstituted CAF-1-, ASF1A-H3-H4-, PCNA-, and RFC-dependent deposition of histone (H3-H4) 2 tetramers onto a nicked DNA, we carried out a series of experiments to investigate whether this process is affected by the MMR system (Fig.  3). In these experiments we utilized two different DNA substrates; 3Ј-nicked A-T DNA that lacked a DNA mismatch and 3Ј-nicked G-T DNA that contained a G-T mismatch. Analysis of the reactions that occurred on 3Ј-nicked G-T DNA revealed that MutS␣, a key component of the MMR system, inhibited CAF-1-, ASF1A-H3-H4-, PCNA-, and RFC-dependent histone (H3-H4) 2 tetramer deposition both on a DNA site containing the mispair (Fig. 3A) and on a nearby site ϳ40 bp to the right of the mispair (Fig. 3B). However, MutS␣ did not affect the histone (H3-H4) 2 tetramer deposition on a site that was ϳ450 bp to the left from the mispair (Fig. 3C). Furthermore, the results of the control reactions indicated that MutS␣ did not impact the formation of tetrasomes at any of the three sites on 3Ј-nicked A-T DNA (Fig. 3, A-C). These data implied that the MMR system inhibits CAF-1-, ASF1A-H3-H4-, PCNA-, and RFCdependent formation of mismatch-containing tetrasomes.
The Effects of CAF-1-, ASF1A-H3-H4-, PCNA-, and RFC-dependent Formation of Tetrasomes on Excision-dependent and Excision-independent MMR Reactions-Current evidence suggests that mismatch-containing DNA that is packaged into nucleosomes is resistant to the action of MMR (56,65). Based on this evidence and the observations that MutS␣ suppresses two subpathways of the histone (H3-H4) 2 tetramer deposition (Ref. 56 and Fig. 3), we hypothesized that a significant fraction of the MMR events occurs during replication-coupled nucleosome assembly but before the nascent mismatch-containing DNA is packaged into nucleosomes. Because the assembly of the tetrasome is the first step in the assembly of the nucleosome, this hypothesis predicted that reconstituted MMR reactions would be able to take place during CAF-1-, ASF1A-H3-H4-, PCNA-, and RFC-dependent histone (H3-H4) 2 tetramer FIGURE 1. Human proteins used in this work. The proteins were purified as described under "Experimental Procedures." The purified proteins were separated in SDS gels, and the protein bands were visualized with Coomassie Blue R-250 staining. deposition (Fig. 3). In experiments that are described below we tested and confirmed this prediction. In addition, we analyzed the effects of histone (H3-H4) 2 tetramer deposition on reconstituted MMR reactions.
Previous research determined that EXO1, MutL␣, MutS␣, PCNA, Pol ␦, RFC, and RPA form a system that corrects mismatches on nicked DNAs (39,40,(42)(43)(44). A 3Ј or 5Ј nick-directed MMR reaction carried out by this purified system is dependent on mismatch excision by EXO1 (39,40,(42)(43)(44). Removal of EXO1 from this seven-protein system produces a system that erases mismatches on nicked DNAs via the mismatch excision-independent reaction (44). This and the other (Fig. 2) information allowed us to generate two new reconstituted systems. One of these systems that contained nine human proteins (EXO1, MutL␣, MutS␣, PCNA, Pol ␦, RFC, RPA, CAF-1, and ASF1A-H3-H4) supported the excision-dependent MMR and the four protein-dependent histone (H3-H4) 2 tetramer deposition (i.e. a histone (H3-H4) 2 tetramer deposition dependent on CAF-1, ASF1A-H3-H4, PCNA, and RFC) and the other that contained eight human proteins (MutL␣, MutS␣, PCNA, Pol ␦, RFC, RPA, CAF-1, and ASF1A-H3-H4) supported the excision-independent MMR and the four protein-dependent histone (H3-H4) 2 tetramer deposition (Figs. 2 and 4). In these two systems, Pol ␦ was the only source of DNA polymerase activity. We utilized these two systems to investigate whether the excision-dependent and excision-independent MMR were affected by the four protein-dependent histone (H3-H4) 2 tetramer deposition (Figs. 4 and 5). We first studied how the four protein-dependent histone (H3-H4) 2 tetramer deposition influenced the efficiencies of the excisiondependent and excision-independent MMR reactions (Fig. 4). The results revealed that the four protein-dependent histone (H3-H4) 2 tetramer deposition did not change the efficiency of the mismatch excision-dependent MMR but modestly increased the efficiency of the excision-independent MMR (Fig.  4B, Reactions 2,5,6,and 9). We also determined the impact of different concentrations of Pol ␦ on the excision-dependent and excision-independent MMR reactions that took place in the presence of the four protein-dependent histone (H3-H4) 2 tetramer deposition. As shown in Fig. 4C, ϳ 1 ⁄ 3 of 3Ј-nicked G-T DNA substrate was repaired in the excision-dependent reaction that was supported by 0.1 nM Pol ␦ (reaction 5), and the same level of the repair in the excision-independent reaction was achieved in the presence of 5 nM Pol ␦ (reaction 7). This observation indicated that much less Pol ␦ was required for the excision-dependent MMR reaction than for the excision-independent MMR reaction.
An earlier study has demonstrated that a strong and unnecessary degradation of the discontinuous strand occurs in the reconstituted excision-dependent MMR reaction (56). We asked whether a similar degradation of the discontinuous strand took place in the reconstituted excision-independent MMR reaction. To this end, DNA products recovered from the reaction mixtures were cleaved with ClaI, separated on denaturing agarose gels, and analyzed by Southern hybridizations with two 32 P-labeled oligonucleotide probes. One of the probes, v2505, hybridized to a discontinuous strand sequence located immediately 3Ј to the ClaI site (Fig. 5A), and the other, v2531, annealed to a discontinuous strand sequence positioned immediately 5Ј to the ClaI site (Fig. 5B). The results of the Southern hybridization analysis showed that the level of degradation of the discontinuous strand in the excision-independent MMR reaction was similar to that in the excision-dependent MMR reaction (Fig. 5, A and B, Reactions 2 and 6). We next studied whether CAF-1-, ASF1A-H3-H4-, PCNA-, and RFC-dependent histone (H3-H4) 2 tetramer deposition affected the degradation of the discontinuous strands in both the excision-dependent and excision-independent MMR reactions (Fig. 5). Inspection of the data led to the following findings. First, the degradation of the discontinuous strands was barely detectable in both the excision-dependent and excision-independent MMR reactions that took place in the presence of ASF1A-H3-H4 and CAF-1 (Fig. 5, A and B, Reactions 5 and 9). Second, the omission of either ASF1A-H3-H4 or CAF-1 increased the degradation of the discontinuous strands 3-5-fold and 2-3fold, respectively (Fig. 5, A and B, Reactions 3, 4, 7, and 8). These findings indicated that although ASF1A-H3-H4 partially suppressed the degradation of the discontinuous strands in both the excision-dependent and excision-independent MMR reactions, the most effective suppression of the degradation of the discontinuous strands was observed in the reactions that occurred in the presence of CAF-1-, ASF1A-H3-H4-, PCNA-, and RFC-dependent deposition of the histone (H3-H4) 2 tetramers. The excision-dependent MMR reaction (43) relies on the action of EXO1, MutL␣, MutS␣, PCNA, Pol ␦, RFC, and RPA and is accompanied by a strong degradation of the discontinuous strand (Ref. 56 and Figs. 4B and 5, Reaction 2). To determine whether the endonuclease activity of MutL␣ is involved in the degradation of the discontinuous strand, we performed experiments (Fig. 6) that made use of an endonuclease-deficient MutL␣-E705K mutant (33). The results of these experiments demonstrated that the endonuclease activity of MutL␣ produced the majority of discontinuous strand degradation products in the reconstituted excision-dependent MMR reaction (Fig. 6A, Reactions 2 and 4).
We next investigated whether the sizes of the discontinuous strand products visualized with the indirect labeling (Figs. 5 and 6A) correlated with those visualized with a direct labeling (Fig.  7). A comparison of the results of the indirect (Figs. 5 and 6A) and direct labeling (Fig. 7) experiments indicated that there was a significant correlation between the sizes of the products visualized with the two different methods.
MMR reactions that were analyzed above (Figs. 4 -7) depended on Pol ␦. We also studied whether the four-protein-dependent histone (H3-H4) 2 tetramer deposition affected reconstituted Pol ⑀-dependent MMR reactions (Figs. 8 and 9). We began this series of experiments by purifying recombinant human four-subunit Pol ⑀ that was expressed in baculovirusinfected insect Sf9 cells. An electrophoretic analysis (Fig. 1) and a mass spectrometry analysis (data not shown) identified that the purified Pol ⑀ contained the p261, p59, p17, and p12 subunits. The purified protein displayed DNA polymerase activity (Fig. 8B, Reaction 2). As expected from previous research (66,67), PCNA, RFC, and RPA stimulated Pol ⑀ to synthesize DNA strands that were significantly longer than those produced in the reaction that occurred in the absence of the accessory proteins (Fig. 8B, Reactions 2 and 4). This stimulation of the Pol ⑀ biosynthetic activity required the presence of all three accessory proteins (Fig. 8B, Reactions 2-4, and data not shown). We then performed the reconstitution experiments (Figs. 8, C and D, and 9A). In accord with our knowledge of the eukaryotic MMR, the reconstitution experiments demonstrated that the reactions that occurred in the presence of Pol ⑀, EXO1, MutL␣, MutS␣, PCNA, RFC, and RPA led to repair of G-T mismatches (Figs. 8C, Reactions 4 and 5, and 9A). Further analysis of the After incubation at 37°C for 20 min, the reactions were stopped. DNAs recovered from the reaction mixtures were cleaved with HindIII and ClaI, and the cleavage products separated on a native 1% agarose gel were visualized with ethidium bromide staining. A, outline of the MMR assay. B, MMR products that were formed in the indicated reconstituted reactions. Note that the DNA products that were generated in the reconstituted reactions included those that moved in the gel slower than the unrepaired DNA. These slower moving products are MMR reaction intermediates that were formed as a result of strand-displacement syntheses initiated by Pol ␦ from both the original 3Ј end and MutL␣ endonuclease-generated 3Ј ends. C, effects of the different concentrations of Pol ␦ on the efficiencies of the excision-dependent and excision-independent MMR reactions. The data shown in the graphs are the averages Ϯ 1 S.D. (n ϭ 3) and were obtained by quantification of images including those in the figure. reconstituted Pol ⑀-dependent MMR reaction revealed that it was accompanied by a strong degradation of the discontinuous strand (Fig. 9, B and C, reactions 10 -13). Supplementation of the Pol ⑀-containing system with CAF-1 and ASF1A-H3-H4 stimulated the MMR reactions by 2-fold (Figs. 8, C, Reactions 2-3, and 9A) and suppressed the degradation of the discontinuous strands (Fig. 9, B and C, . These results suggested that the four protein-dependent histone (H3-H4) 2 tetramer deposition stimulated the Pol ⑀-dependent MMR reaction and suppressed the degradation of the discontinuous strand.
To confirm that Pol ⑀ was responsible for DNA synthesis in the reconstituted MMR reaction (Fig. 8C), we carried out experiments that took advantage of a Pol ⑀-D860A-D862A mutant protein (Fig. 1) (68). In this mutant protein, the two catalytic aspartate residues are replaced with alanine residues (Fig. 8A). Consistent with a prior work (68), this Pol ⑀ mutant protein that was expressed and purified as the wild-type enzyme (Fig. 1) lacked DNA polymerase activity (Fig. 8B, Reactions 8 and 9). Replacement of Pol ⑀ with this mutant protein abolished the MMR reactions in both the nine-and seven-protein systems (Fig. 8C, Reactions 6 -9). These data indicated that Pol ⑀ re-synthesized DNA in the reconstituted MMR reactions (Fig. 8C, Reactions 2-5). We then performed the omission experiments to further characterize the reconstituted Pol ⑀-dependent MMR reaction that occurred in the nine-protein system (Figs. 8D and 9). The omission of MutS␣, MutL␣, PCNA, RFC, or Pol ⑀ completely abolished the Pol ⑀-dependent MMR reaction (Fig. 8D, Reactions 2, 6 -8, 10, and 11), but the omission of RPA only decreased the efficiency of the Pol ⑀-dependent MMR reaction (Fig. 8D, Reaction 9). Surprisingly and in contrast to the Pol ␦-dependent MMR reaction (44) (Fig. 4), the Pol ⑀-dependent reaction did not occur in the absence of EXO1 (Fig. 8D, Reaction 5). We also found that the efficiency of the Pol ⑀-dependent MMR reaction increased with increasing Pol ⑀ concentration and decreased in the absence of ASF1A-H3-H4 or CAF-1 (Fig. 9A). Taken together, the results of the above experiments (Figs. 8 and 9) provided evidence that Pol ⑀ can perform DNA synthesis in excision-dependent MMR.
Protein-Protein Interaction between MutS␣ and ASF1A-H3-H4 -Protein-protein interactions coordinate numerous processes that take place on nuclear DNA. Previous research has shown that the mismatch recognition factor MutS␣ physically interacts with the histone chaperone CAF-1 (62). To determine whether there are additional protein-protein interactions between the components of the MMR system and replication-coupled nucleosome assembly, we carried out the pulldown experiments that involved the purified MutS␣ and    Fig. 6. A fraction of each recovered DNA was cleaved with ClaI and HindIII to score MMR, and the rest was cleaved with ClaI and separated on a denaturing agarose gel. The gel was dried and exposed to a phosphorimaging screen. The image was generated with a Typhoon biomolecular imager (GE Healthcare). A, outline of the experiment. B, [ 32 P]dGMP-labeled DNA molecules formed in the indicated reactions. The level of MMR in each of the reactions is also shown. The MMR data are the averages Ϯ 1 S.D. (n Ն 2).
ASF1A-H3-H4 (Fig. 10). We found that the ASF1A-H3-H4 heterotrimers tethered to the protein A beads via the ␣-ASF1 antibodies pulled down the purified MutS␣ (Fig. 10A). The possibility of a nonspecific binding of MutS␣ to the beads was ruled out in an experiment that showed that MutS␣ molecules were not pulled down by the ␣-ASF1 antibody-containing beads that lacked ASF1A-H3-H4 (Fig. 10A). Further experiments established that the MutS␣-containing beads pulled down the purified ASF1A-H3-H4 (Fig. 10B). Based on these results, we concluded that MutS␣ physically interacted with ASF1A-H3-H4. To determine whether MutS␣ forms a complex with ASF1A, the H3-H4 dimer, or both components of the heterotrimer, we conducted pulldown experiments that included these three proteins (Fig. 10, C-E). The results revealed no evidence that MutS␣ interacted with ASF1A (Fig. 10, C and D) but showed that the histone H3-H4-containing beads pulled down the purified MutS␣ (Fig. 10E). These findings indicated that MutS␣ forms a complex with ASF1A-H3-H4, probably via the H3-H4 dimer.
Experiments summarized in Fig. 3, A and B, identified that MutS␣ inhibits CAF-1-and ASF1A-H3-H4-dependent histone (H3-H4) 2 tetramer deposition onto DNA at and near a mismatch. A different study showed that MutS␣ causes a very similar effect on CAF-1-dependent and ASF1A-H3-H4-independent histone (H3-H4) 2 tetramer deposition (56). Thus, MutS␣ interferes with two reconstituted subpathways of CAF-1-dependent histone (H3-H4) 2 tetramer deposition that can package DNA mismatches into tetrasomes. Earlier work found that DNA packaged into nucleosomes is refractory to MMR (56) and that nucleosomes inhibit mismatch recognition by MutS␣ (65). How can these findings be reconciled with the evidence that MMR coincides with CAF-1-dependent assembly of the newly replicated DNA into nucleosomes? We propose that suppression of CAF-1-dependent histone (H3-H4) 2 tetramer deposition by MutS␣ at and around the mismatch maintains the nascent DNA region free of nucleosomes, thereby allowing MMR to occur on the locally naked DNA. Once the mismatch is removed, MutS␣ no longer blocks CAF-1-dependent histone (H3-H4) 2 tetramer deposition, and the repaired DNA is packaged into nucleosomes (Fig. 11).
Unlike the MMR reaction that occurs in nuclear extracts containing CAF-1, the MMR reaction reconstituted with near homogenous EXO1, MutL␣, MutS␣, PCNA, Pol ␦, RPA, and RFC (43) causes extensive and unnecessary degradation of the discontinuous strands (56). Similar degradation of the discontinuous strands takes place in MMR reactions in a cytosolic extract that is deficient in the replicative histone chaperone CAF-1. Supplementation of the cytosolic extract with purified CAF-1 blocks the unnecessary degradation of the discontinuous strands (56). Analysis of the products formed in the cytosolic extract reactions that occurred in the presence of CAF-1 revealed that the nicked heteroduplex DNA substrate is subjected to both MMR and CAF-1-dependent packaging into nucleosomes. These data support the idea that CAF-1-dependent nucleosome assembly suppresses the unnecessary degradation of the discontinuous strand during MMR. Consistent with this idea, the unnecessary degradation of the discontinuous strand is suppressed in the reconstituted MMR reaction that takes place in the presence of CAF-1-dependent and ASF1A-H3-H4-independent histone (H3-H4) 2 tetramer deposition (56). To obtain additional insight into the mechanism of MMR in the chromatin environment, we have studied the impact of the CAF-1-and ASF1A-H3-H4-dependent histone tetramer deposition on MMR in three reconstituted systems. One of the reconstituted systems supports the excision-and Pol ␦-dependent MMR and contains EXO1, MutL␣, MutS␣, PCNA, Pol ␦, RFC, RPA, CAF-1, and ASF1A-H3-H4. Another reconstituted system is proficient in the excision-independent and Pol ␦-dependent MMR and is composed of MutL␣, MutS␣, PCNA, Pol ␦, RFC, RPA, CAF-1, and ASF1A-H3-H4. The third system supports the excision-and Pol ⑀-dependent MMR and is composed of EXO1, MutL␣, MutS␣, PCNA, Pol ⑀, RFC, RPA, CAF-1, and ASF1A-H3-H4. In agreement with previous research (56), we have determined that MMR that occurs in the presence of the CAF-1-and ASF1A-H3-H4-dependent histone deposition in the three reconstituted systems causes a very limited degradation of the discontinuous strand, but the omission of CAF-1 and ASF1A-H3-H4 increases the discontinuous strand degradation 2-4-fold (Figs. 5 and 9B). Because the increased degradation of the discontinuous strand does not increase the efficiency of MMR, such DNA degradation is unnecessary (Figs. 4, 5, 6, and 9). We note that although MutS␣ inhibits CAF-1-dependent histone (H3-H4) 2 tetramer deposition in the vicinity of the mismatch, it does not affect the histone tetrasome formation at a site that is located ϳ400 bp from the mismatch (56) (Fig. 3C). Our data indicated that histone (H3-H4) 2 tetramers deposited at such sites block the unnecessary degradation of the discontinuous strand (Ref. 56 and Figs. 5 and 9B). We have also determined that CAF-1 and ASF1A-H3-H4 stimulate the excision-independent and Pol ␦-dependent MMR reaction as well as the excision-and Pol ⑀-dependent MMR reaction (Figs. 4B and 9A). These results suggest that CAF-1-, ASF1A-H3-H4-, PCNA-, and RFC-dependent histone (H3-H4) 2 tetramer deposition creates a productive environment for MMR.
Pol ␦ and Pol ⑀ replicate the majority of nuclear DNA (70 -72). Fractionation of HeLa nuclear extracts and reconstitution studies determined that Pol ␦ re-synthesizes DNA in human excision-dependent and excision-independent MMR reactions (43,44,46,69). However, it has remained unknown whether Pol ⑀ plays a role in MMR. Our analysis has indicated that Pol ⑀ is able to support human MMR (Figs. 8 and 9). The reconstituted Pol ⑀-dependent MMR reaction requires the activities of EXO1, MutL␣, MutS␣, PCNA, and RFC (Fig. 8D, Reactions 5-8 and  10). This finding indicates that the Pol ⑀-dependent MMR reaction includes the steps of the discontinuous strand incision, mismatch excision, and DNA re-synthesis. The dependence of the Pol ⑀-dependent MMR reaction on EXO1 (Fig. 8D) suggests that unlike human Pol ␦ (44), human Pol ⑀ is not able to drive the excision-independent MMR reaction (Fig. 11). Comparison of the Pol ␦and Pol ⑀-dependent MMR reactions reveals that the specific MMR activity of Pol ␦ is ϳ20 times higher than that of Pol ⑀ (Figs. 4C, Reaction 5, and 9A). This supports the view that Pol ␦ re-synthesizes DNA in the majority of MMR events in human cells. Nevertheless, it remains possible that our biochemical analysis (Figs. 4C and 9A) underestimates the importance of Pol ⑀ for human MMR because the reconstituted system (Fig. 8D) lacks one or more factors that are necessary to increase the MMR activity of this DNA polymerase. A recent analysis has shown that the DNA replication factors Ctf4 and GINS stimulate the biosynthetic activity of Pol ⑀ (73, 74). Therefore, it may be that Ctf4 and GINS increase the MMR activity of Pol ⑀.
Our pulldown experiments have shown that MutS␣ forms a complex with ASF1A-H3-H4 (Fig. 10, A and B). Because MutS␣ is a large protein carrying both negative and positive charges on its surface and the histone H3-H4 dimer is a very basic protein molecule, it may be that the MutS␣-ASF1A-H3-H4 complex is simply an in vitro artifact of nonspecific negative-positive charge interaction(s) between MutS␣ and H3-H4. However, we regard such a conclusion unlikely because in similar experiments we detected no complex formation between ASF1A-H3-H4 and MutL␣, a large protein that, like MutS␣, contains both negative and positive charges on its surface (data not shown). If the MutS␣-ASF1A-H3-H4 interaction happens in vivo, what is the function of this interaction? We hypothesize that this interaction does not allow ASF1A-H3-H4 FIGURE 11. MMR in the presence of CAF-1-, ASF1A-H3-H4-, PCNA-, and RFC-dependent histone (H3-H4) 2 tetramer deposition. The findings described in this work are consistent with the model in which a subset of human MMR events coincides with CAF-1-, ASF1A-H3-H4-, PCNA-, and RFCdependent deposition of the histone (H3-H4) 2 tetramers onto the newly synthesized DNA. The model is in part based on the results of previous studies of human MMR (27,33,35,39,40,(42)(43)(44)56). The model also depicts that MutS␣ interacts with ASF1A-H3-H4. As described under "Discussion," this interaction may be important to facilitate MMR in the presence of CAF-1-, ASF1A-H3-H4-, PCNA-, and RFC-dependent histone (H3-H4) 2 tetramer deposition. HE, holoenzyme.
to participate in packaging of the newly synthesized mismatchcontaining DNA into the tetrasomes. Once the mismatch is removed, the local concentration of MutS␣ sharply decreases, and as a result, ASF1A-H3-H4 is no longer blocked from the involvement in the assembly of the newly replicated DNA into nucleosomes. If the interaction between MutS␣ and histone H3-H4 (Fig. 10E) occurs in vivo, it may play a role similar to that proposed above for the MutS␣-ASF1A-H3-H4 interaction.
In summary, findings described here and previously (56,62,65) support the view that the functional interactions between MMR and CAF-1-dependent nucleosome assembly ensure that the mismatch is corrected before it is packaged into the tetrasome and that unnecessary degradation of the discontinuous daughter strand does not occur during MMR.
Author Contributions-F. K. designed the experiments. E. R., L. K., and F. K. performed experiments, contributed reagents, analyzed data, and prepared the figures. F. K. and E. R. wrote the paper.