Aromatic Residues Located Close to the Active Center Are Essential for the Catalytic Reaction of Flap Endonuclease-1 from Hyperthermophilic Archaeon Pyrococcus horikoshii *

Flap endonuclease-1 (FEN-1) possessing 5 (cid:1) -flap endonuclease and 5 (cid:1) 3 3 (cid:1) exonuclease activity plays important roles in DNA replication and repair. In this study, the kinetic parameters of mutants at highly conserved aromatic residues, Tyr 33 , Phe 35 , Phe 79 , and Phe 278 Phe 279 , in the vicinity of the catalytic centers of FEN-1 were examined. The substitution of these aromatic residues with alanine led to a large reduction in k cat values, although these mutants retained K m values similar to that of the wild-type enzyme. Notably, the k cat of Y33A and F79A decreased 333-fold and 71-fold, respectively, compared with that of the wild-type enzyme. The aromatic residues Tyr 33 and Phe 79 , and the aromatic cluster Phe 278 -Phe 279 mainly contributed to the recognition of the substrates without the 3 (cid:1) projection of the upstream strand (the nick, 5 (cid:1) -recess-end, single-flap, and pseudo-Y substrates) for the both exo- and endo-activities, but played minor roles in recognizing the substrates with the 3 (cid:1) projection (the double flap substrate and the nick substrate with the 3 (cid:1) projection). The replacement of Tyr 33 , Phe 79 , and Phe 278 -Phe 279 , with non-charged aromatic residues, but not with

Flap endonuclease-1 (FEN-1) 1 has important roles in DNA replication, repair, and recombination. It belongs to a family of structure-specific nucleases, which are conserved from archaea to eukaryotes and share homology with the 5Ј-nuclease domain associated with DNA polymerase I in prokaryotes and 5Ј33Ј exonuclease of bacteriophages (1)(2)(3)(4)(5).
In DNA replication, FEN-1 removes the RNA primers during the maturation of the Okazaki fragment (6), in conjunction with Dna2 having an endonuclease activity and a helicase activity (7). For DNA repair, FEN-1 removes damaged nucleotides in long patch base excision repair (8) and is required for non-homologous end joining of double-stranded DNA breaks (9).
FEN-1 possesses 5Ј-flap endonuclease and 5Ј33Ј exonuclease activities (10,11). The flap endonuclease activity has been shown to require the upstream primer, which fills up the junction portion, and expanding the 3Ј of the upstream primer known as a double flap structure elevates the activity (12)(13)(14). The 5Ј33Ј exonuclease activity digested the double-stranded DNA containing the nick, gap, and 5Ј-recess-end, and the activity was elevated by expanding the 3Ј-end of the upstream primer in the nick portion (1,14).
Three crystal structures of FEN-1 have been reported exclusively in thermophilic archaea (15)(16)(17). The molecular structure of the members of the FEN-1 family, T5 exonuclease, T4 RNase H, and the exonuclease domain of Taq polymerase were also reported (18 -20). All these structures have a conserved helical arch located above the globular domain that contains the active site, which is thought to recognize the 5Ј-end of the flap strand, tracking the length of the tail and cleaving near the junction between double-stranded and single-stranded DNA (21).
Recently, DNA binding sites of FEN-1 in human, archaea, and T5 bacteriophage were identified using site-direct mutagenesis (12, 13, 17, and 22-27), and two DNA binding models were postulated based on the identified binding sites and the molecular structure (22,25). Several residues identified as DNA binding sites were located on the large loop (12, 17, 25, and 26), one of them was in contact with the bottom of the flap strand at the junction between the single-and double-stranded DNA (25). Two binding models for Pyrococcus furiosus and T5 bacteriophage FEN-1 showed that the helix-hairpin-helix region interacted with the downstream duplex DNA, and the flap strand interacted with the helical arch of FEN-1. However, the DNA binding mechanism is still unclear, because the number of DNA binding sites identified was too small to understand the substrate recognition.
We previously investigated the substrate specificity of phFEN-1 in detail using 35 different substrates (1). The substrate specificity was similar to that of eukaryote FEN-1, except that phFEN-1 could cleave a flap strand having a double strand, suggesting that phFEN-1 might recognize strongly the junction portion of the flap strand rather than the 5Ј-end. Furthermore, the crystal structure of phFEN-1 was solved, and the large loop and small loop 2 ( Fig. 2) were suggested to be very flexible by superimposing three kinds of structures from archaea FEN-1s (17), indicating that the structures of two loops might change remarkably for binding to the substrate DNA. The four DNA binding sites on the large loop and small loops were identified by site-directed mutagenesis (17). These consist of basic residues presumably bound to phosphate moieties of the DNA strand. We successfully elucidated the functions of the four DNA binding sites that form productive ES complexes for exo-and endo-activities, probably by bending the substrates.
In this study, we investigated the functions of aromatic residues close to the active center and report here the significant roles for these residues and the substrate recognition mechanism of FEN-1 in the formation of the productive intermediates probably through a partial unwinding of the substrates via stacking interactions with the central aromatic residues.

EXPERIMENTAL PROCEDURES
Chemicals-The vector pET-11b and Escherichia coli strain BL21(DE3) were purchased from Novagen (Madison, WI). The restriction enzymes were purchased from Promega (Madison, WI). Isopropyl-␤-D-thiogalactopyranoside was obtained from Takara Shuzo (Japan). Oligonucleotides labeled with the fluorescent dye fluorescein (FAM) were purchased from Sawady Technology, Ltd. (Tokyo, Japan). Oligonucleotides substituted with 2Ј-O-methyl at the 2Ј position of the deoxyribose were purchased from Espec Oligo Service Corp. (Japan).
Expression and Purification of Recombinant phFEN-1s-Each mutant gene was constructed on the basis of the vector pGEMEX/FEN-1 using site-directed mutagenesis and the PCR method. The wild type and mutated genes were inserted into the vector pET-11a using NdeI and XhoI sites, resulting in the expression vectors pET-11a/phFEN-1s. The presence of desired mutations and the absence of unwanted mutations were verified by sequencing.
The E. coli strain BL21(DE3) was transformed with the expression vectors. The transformant was propagated in 200 ml of 2ϫ YT medium containing ampicillin at 37°C. The recombinant protein was induced at A 600 ϭ 1.0 -1.5 with 1 mM isopropyl-␤-D-thiogalactopyranoside for 12 h. The cells were collected by centrifugation and stored at Ϫ20°C. The frozen cells were then thawed at room temperature and mixed with 1 ml of 50 mM Tris-HCl buffer (pH 8.0). The sample was heated at 83°C for 30 min and centrifuged at 2300 ϫ g for 30 min, then the supernatant was again heated at 83°C for 15 min to remove endogenous proteases completely before being centrifuged. The supernatant was loaded on a HiTrap SP column (Amersham Biosciences) and eluted with a linear gradient of 1 M NaCl in a 50 mM Tris-HCl (pH 8.0) buffer. The recombinant phFEN-1s were eluted with buffer containing 200 -300 M NaCl.
Measurement of Kinetic Parameters-The kinetic parameters were determined using the fluorescence substrates. The reaction mixture was comprised of 50 mM Tris-HCl (pH 7.4) containing the labeled substrate, 1.5 mM MgCl 2 , and 1.5 g of bovine serum albumin. The enzyme and substrate concentrations ranged from 0.025 to 200 ng and from 0.1 to 75 pmol, respectively. A 10-fold volume (150 l) of the reaction mixture was preheated at 60°C for 5 min, and 14 l was added to 10 l of stop solution as a control for a zero time reaction. Nine volumes of enzyme were added, and 15 l of the reaction mixture was added to 10 l of the stop solution every 1 min for 7 min. The samples were denatured at 95°C for 10 min. Five microliters of sample was loaded onto a 12% polyacrylamide gel (10 ϫ 10 cm) with 7 M urea and 1ϫ TBE, and electrophoresed with the same buffer for 1.5 h at 200 V. The reaction products were visualized and quantified using a Fluor-Imager 585 (Molecular Dynamics, Inc.). The initial velocities were obtained directly from the initial slopes of the time course plots. The K m and k cat values were calculated using the Michaelis-Menten equation and the least squares method (28).

Measurement of Cleavage Position and Recognition
Sites of Substrates-To determine of cleavage positions in each substrate, a 12% polyacrylamide gel (35 cm ϫ 42.5 cm) containing 7 M urea and 1ϫTBE was used, and the electrophoresis was performed for 2 h at 3000 V. The size markers of oligonucleotides were labeled at the 5Ј terminus with FAM and designed to have the same sequence as the products. The enzyme amounts of WT and alanine mutants were from 0.1 to 100 ng, and the substrate amounts were from 3 to 5 pmol. The reaction time was from 5 to 10 min. The other reaction conditions were the same as for the kinetics analysis.
The measurement of the recognition sites was performed using substrates substituted with 2Ј-O-methyl at the 2Ј position in the deoxyribose. The oligonucleotides having deoxyribose substituted with 2Ј-Omethyl at each position and labeled at the 5Ј terminus with FAM was annealed to each oligonucleotide to make the substrates as described above. The relative activity of WT enzyme was determined using nick and 5Ј-recess-end substrates. The enzyme and substrate amounts were 10 ng and 3 pmol, respectively. The reaction time was 5 min. The other conditions were the same as for the kinetics analysis.

RESULTS
Purification of Mutant Protein-The residues corresponding to Tyr 33 , Phe 35 , Phe 79 , and Phe 279 of phFEN-1are conserved among FEN-1s from eukaryotes and archaea, although the residues corresponding to Phe 278 are variable within hydrophobic residues as shown below in Fig. 2A. They were located close to the active center as shown in Fig. 2 (B and C). To analyze the function of these aromatic residues, they were replaced with alanine, a hydrophobic residue, and other aromatic residues. Twenty mutant genes were constructed by site-directed mutagenesis. The expression level of each mutant enzyme in E. coli cells was almost the same as that of the wild type. These mutants were purified using the same procedure as was used for the wild type. The electrophoretic mobility of each mutant protein on SDS-gel was the same as that of the wild type protein.
Tyr 33 , Phe 35 , Phe 79 , and Phe 278 -Phe 279 Recognize Nick and 5Ј Recess-end Substrates for Exo-activity-The nick and 5Ј-recessend substrates were used for a kinetic analysis of exo-activity as shown below in Fig. 3. The highly conserved residue Tyr 33 is located in the N-terminal region of FEN-1s from eukaryotes and archaea as shown in Fig. 2A. The residue is present close to magnesium ion 1 bound to the catalytic residues (Asp 27 , Asp 80 , Glu 152 , and Glu 154 ) and faces the active center as shown in Fig. 2 (B and C). For the exo-activity against the 5Ј-recessend substrate, the k cat values of Y33A and Y33L decreased 333-fold and 1180-fold, respectively, compared with that of the wild type enzyme (WT), although the k cat values of Y33F, Y33W, and Y33H were recovered to 20 -29% of WT. The K m values of the mutants altered at 33rd position changed slightly compared with that of WT. For the exo-activity against the nick substrate, the k cat values of Y33A and Y33L decreased 53-fold and 134-fold, respectively, compared with that of WT, indicating that the k cat values of Y33A and Y33L for the nick substrate are ϳ6 and 9 times higher than those for the 5Ј-recess-end substrate, respectively, although the only structural difference between the nick and 5Ј-recess-end substrates is the presence of a upstream strand as shown in Fig. 1. The k cat values of Y33F, Y33W, and Y33H for the nick substrate were recovered to 29 -200% of WT. The K m values of the mutants altered at the 33rd position varied moderately, but not significantly, compared with that of WT. These results clearly indicated that Tyr 33 could be substituted only with aromatic amino acids to maintain high k cat values for both substrates.
The residue Phe 35 is conserved among FEN-1s from eukaryotes and archaea and is located also to magnesium ion 1 like Tyr 33 as described above (Fig. 2). The k cat and K m of F35A and F35L showed no significant decrease compared with those of WT, however, the k cat value of F35Y decreased about 17-fold and 24-fold with the nick and 5Ј-recess-end substrates, respectively. The K m values of F35Y were about 4-and 3-fold higher than those of WT with the nick and 5Ј-recess-end substrates, respectively. These results suggested that the 35th position preferred hydrophobic residues to polar residues, because the hydroxyl group of tyrosine showed inhibitory effects on both binding and catalysis for these substrates.
The conserved Phe 79 residue is adjacent to a catalytic residue, Asp 80 , as shown in Fig. 2 (A and B). For the 5Ј-recess-end substrate, the k cat value of F79A decreased 71-fold compared with that of WT, and the k cat value of F79Lwas restored to 20% of the WT value as shown in Fig. 3. For the nick substrate, the k cat value of F79A decreased 25-fold compared with that of WT, and the k cat value of F79Lwas restored to 20% of that of WT. The k cat values of F79Y and F79W for both substrates were restored to almost the same level as those of WT. The k cat value of F79H for the 5Ј-recess-end substrate was decreased to ϳ50% of the WT value, and the k cat value of F79H for the nick substrate was seven times lower than that of WT. The K m value of F79H for the 5Ј-recess-end substrate was 13-fold higher than that of WT, whereas the K m value of F79H for the nick substrate was ϳ2 times higher than that of WT. The charged imidazole group of histidine inhibited the catalytic activity. These results suggested that the function of Phe 79 could be satisfied only by non-charged aromatic residues.
The aromatic cluster Phe 278 -Phe 279 is located in the vicinity of magnesium ion 2 according to the molecular structure of phFEN-1 (Fig. 2C). Phe 279 is conserved among FEN-1s from eukaryotes and archaea, although Phe 278 is variable within hydrophobic residues as shown in Fig. 2A. The k cat value of F278Y/F279Y was restored to around 70% of that of WT with both substrates, whereas the k cat values of F278A/F279A, F278L/F279L, and F278W/F279W were decreased ϳ10 -20% with both substrates as shown in Fig. 3. Notably, the k cat value of F278H/F279H decreased dramatically, 83-fold for the 5Јrecess-end substrate and 150-fold for the nick substrate compared with those of WT. For both substrates, the K m values of F278L/F279L, F278Y/F279Y, and F278W/F279W were lower than those of WT, although the K m value of F278A/F279A was severalfold higher than that of WT. The K m value of F278H/ F279H for the 5Ј-recess-end substrate was elevated 5 times compared with that of WT, whereas for the nick substrate the K m value of F278H/F279H was 60% that of WT. These results indicated that both Phe 278 and Phe 279 could be substituted by tyrosine but not by histidine.  Fig. 1 was used to determine the kinetic parameters for endo-activity. As summarized in Fig. 3, the K m value of Y33L, F79A, and F278H/F279H increased 4-, 7-, and 4-fold, respectively, and their k cat /K m values decreased 5-, 6-, and 14-fold, respectively, compared with that of WT. As mentioned above, in the hydrolysis of both the 5Ј-recess-end and nick substrates, Y33L, F79A, and F278H/F279H represent the mutants with the most severely affected k cat /K m values, decreased 2353-, 58-, and 454-fold for the 5Ј-recess-end substrate, and 270-, 75-, and 80-fold for the nick substrate, respectively, compared with that of WT. Hence, the small reduction rates of the mutants, Y33L, F79A, and F278H/F279H for the k cat /K m values against the double flap substrate indicated that the residues Tyr 33 , Phe 35 , and Phe 79 , and the aromatic cluster Phe 278 -Phe 279 made minor contributions to recognizing the double-flap substrate.
Furthermore, to investigate the function of Tyr 33 , Phe 79 , and Phe 278 -Phe 279 in the endo-type hydrolysis, the kinetic parameters of mutants were determined using the single flap and pseudo-Y substrates without the 3Ј projection of the upstream strand (Fig. 1). The results are summarized in Fig. 4. The k cat of Y33A and Y33L for the single flap substrate decreased 30and 433-fold, respectively, and the k cat of Y33A and Y33L for the pseudo-Y substrate decreased 485-and 3233-fold, respectively, compared with that of WT, although the k cat values of Y33F for the single flap and pseudo-Y substrates were restored to 38 and 20% of the value of WT, respectively. However, the K m values of Y33A, Y33L, and Y33F for both the single flap and pseudo-Y substrates were 18 -200% of the WT value. The K m of the mutants altered at the 33rd position varied moderately, but not significantly, compared with that of WT. These high reduction rates for the k cat values of endo-activity against both the single flap and pseudo-Y substrates were similar to those for the recess-end and nick substrates for exo-activity. These results clearly indicated that Tyr 33 could be substituted with an aromatic residue to maintain high k cat values in both exo-and endo-type reactions.
For the single flap and pseudo-Y substrates, the k cat of F79A decreased 31-and 37-fold, respectively, compared with that of WT, and the k cat of F79L decreased to 17 and 7% as shown in Fig. 4. The k cat values of F79Y for both substrates were restored to almost the same level as those of WT. The K m of F79L and F79Y for the two substrates was varied moderately, but not significantly, compared with that of WT, whereas the K m value of F79A for the substrates was ϳ3-4 times higher than that of WT. These results suggested that the function of Phe 79 could be satisfied by an aromatic residue, with high k cat values retained for both exo-and endo-type reactions.
For both substrates, the k cat /K m of F278A/F279A decreased ϳ20-fold, compared with that of WT, whereas the k cat /K m of F278L/F279L recovered to around 15% of the value for WT. On the other hand, the k cat /K m values of F278Y/F279Y were similar to those of WT, suggesting that the cluster Phe 278 -Phe 279 could be replaced with tyrosine to maintain the high k cat values in both the exo-and endo-type reactions.
Tyr 33  WT. Except for the K m value of F79A, which increased 4-fold compared with that of WT, the K m , k cat , and k cat /K m of these alanine mutants did not show a significant difference from those of WT (data not shown). The kinetic profiles of the alanine mutants for the nick substrate with the 3Ј protrusion of the upstream strand were similar to those for the double flap substrate with the 3Ј protrusion of the upstream strand as shown in Fig. 3. Taking into account all of the kinetics parameters described above, it was concluded that Tyr 33 , Phe 79 , and Phe 278 -Phe 279 play significant roles in the catalysis of the substrates without the 3Ј projection of the upstream strand (the nick, 5Ј-recess-end, single flap, and pseudo-Y substrates), whereas they are minor determinants in the catalytic reaction of the double flap and nick substrates with the 3Ј protrusion of the upstream strand.
Tyr 33 where ⌺K j corresponds to the sum of the binding constants for all binding forms of the substrate. K j equals the sum of K p (the binding constant of ES complex leading to the product) and K q (the binding constant of ES complex failing in production). k int is the intrinsic rate constant of the hydrolysis. The k int value was postulated as invariant among mutant and WT enzymes, because, except for Y33L, F79A, F278H/F279H, the K m , k cat , and k cat /K m values of mutants for the double flap substrate were between 25 and 300% of those for WT as shown in Fig. 3. With these equations and the kinetic parameters shown in Figs. 3 and 4, K j (the binding constant) and k int K p (identical to the k cat /K m value) for all the substrates were evaluated, and compared among alanine mutant and WT enzymes as shown in Fig. 5. For the substrates without the 3Ј projection of the upstream strand (the nick, 5Ј-recess-end, single flap, and pseudo-Y substrates), K p values were reduced remarkably by mutation at the three substrate binding sites (Tyr 33 , Phe 79 , and Phe 278 -Phe 279 ), because the K int was postulated as invariant, whereas, for the double flap and nick substrates with the 3Ј protrusion of the upstream strand, the K p values were not markedly changed by the mutations. Furthermore, the K j values for all the substrates were not significantly changed by the mutations. These results strongly suggest that, for the substrates without the 3Ј projection of the upstream strand, the residues Tyr 33 , Phe 79 , and Phe 278 -Phe 279 might contribute to transforming the ES complexes into the productive transient states rather than to forming merely the initial ES complexes, because the K p values for the substrates without the 3Ј projection were drastically changed, whereas the K j values for the substrates were not significantly reduced by the mutations. On the other hand, the three sites probably play minor roles to form the productive intermediates for the double flap and nick substrates with the 3Ј protrusion, because the K p values for the substrates with the 3Ј projection were moderately maintained even with the mutations.  Fig. 6A. Oligonucleotide markers with the same base sequence as the products were used to negate the difference in electrophoretical mobility due to each specific sequence. For the nick substrate, a major 1-mer product (95%) and a minor 3-mer (5%) product were produced as shown in Fig.  6A, lane 1, indicating that the cleavage positions were 1 base and 3 bases inside from the 5Ј-end of the downstream strand. For the 5Ј-recess-end substrate, a major 3-mer product and a trace 1-mer product were detected as shown in Fig. 6A, lane 2, indicating that the cleavage position was 3 bases inside from the 5Ј-end of the downstream strand. Thus, cleavage positions differed between the nick and 5Ј-recess-end substrates. For the pseudo-Y substrate having a 4-mer projection of the flap strand, only a 7-mer product was detected as shown in Fig. 6A, lane 6, indicating that the pseudo-Y substrate was cleaved 3-mer inside of the duplex end, and that the cleavage position was the same as that of the 5Ј-recess-end substrate. For the single flap substrate having a 4-mer projection of the flap strand, 4-mer (66%), 5-mer (23%), and 7-mer (11%) products were detected as shown in Fig. 6A, lane 7 Fig. 6A, lane 14. Furthermore, the influence of alanine mutations on the cleavage patterns of the substrates was analyzed, however, no significant difference was observed for any of the substrates examined (data not shown).
Because it was reported that the substitution with 2Ј-Omethyl at the 2Ј position of deoxyribose moieties had an inhibitory effect on the FEN-1 activity due to the steric-hindrance induced at the sites contacted with the enzyme (25), the sites on the substrates recognized by phFEN-1 were identified using the same method, with the nick and 5Ј-recess-end substrates substituted with 2Ј-O-methyl. The methyl groups were introduced into the deoxyribose moieties from the 2nd to 5th positions of the 5Ј-end of strand C, and from the 24th to 28th positions of the 5Ј-end of strand A of the nick and 5Ј-recess-end substrates as shown in Fig. 6B. A prominent decrease in activity was demonstrated by the 2Ј-O-methyl substitution at the 3rd and 4th positions of strand C, and at the 27th and 28th positions of strand A in the nick substrate as shown in Fig. 6B  (part I). For the 5Ј-recess-end substrate, a significant decrease was observed following substitutions at the 3rd and 4th positions of strand C and at the 26th and 27th positions of strand A as shown in Fig. 6B (part II). Because phFEN-1 cleaved the nick substrate 1-mer and 3-mer inside of the 5Ј-end of strand C, and cleaved the 5Ј-recess-end substrate 3-mer inside of the 5Ј-end of strand C as mentioned in the former section, the results of the substitution experiments indicated that phFEN-1 recognized both strands covering a few nucleotides adjacent to the cleavage positions of the nick and 5Ј-recess-end substrates.

DISCUSSION
Aromatic Residues Located Close to the Active Center Are Essential for the Catalytic Reaction of FEN-1 Probably through Stacking Interactions with the Substrates-The activities of phFEN-1s were measured using substrates with a highly fluorescent dye, fluorescein (FAM), at the 5Ј-end, because the catalytic activities of phFEN-1 against fluorescent substrates are the same as those against radiolabeled substrates as reported previously (17). In this report, we compared the catalytic activities of mutants altered at three sites (Tyr 33 , Phe 79 , and Phe 278 -Phe 279 ) for both fluorescent and radiolabeled substrates; however, no detectable difference between the substrates was found for any mutant (data not shown). Furthermore, the cleavage positions on the nick, single flap, and pseudo-Y substrates were compared between the fluorescent and radiolabeled substrates. The results revealed that the digestion points on the fluorescent substrates were the same as those of the radiolabeled ones, although the digestion point on the radiolabeled recess-end substrate was shifted 1-mer toward the 5Ј-end, compared with that of the fluorescent one (data not shown). Thus, the fluorescent substrates were used to measure the kinetic parameters of mutant enzymes due to their ease of handling.
The molecular structure of FEN-1 from archaea has been reported (15)(16)(17), and recently several DNA binding sites were found in the molecules (12-13, 17, and 22-27). The aromatic amino acids were reported to form a stacking interaction with single-stranded DNA in a helicase and a DNA-binding protein, replication protein A (34 -36). Interactions with bases of double-stranded DNA were found in the hydrolysis reaction of resolvases and Mut S proteins belonging to nucleases (37)(38)(39). The residues Tyr 33 and Phe 79 in the vicinity of the active center of the phFEN-1 molecule could be substituted with aromatic amino acids to maintain high k cat values in both exo-and endo-type reactions as shown in Figs. 3 and 4. However, the replacement of aromatic residues with alanine (disappearance of aromatic groups) at the 33rd and 79th positions of phFEN-1 resulted in a decrease in the k cat value of 333-and 71-fold for the 5Ј-recess-end substrate, and 53-and 25-fold for the nick substrate, respectively, compared with WT, whereas the K m of the mutants varied moderately, not significantly, compared with that of WT. Furthermore, the change to an aliphatic hydrophobic residue, leucine, at the 33rd and 79th positions induced a remarkable decrease in k cat for the substrates without the 3Ј protrusion of the upstream strand, compared with that of WT, whereas the K m values of the leucine mutants did not change significantly. The results indicated significant roles for the aromatic groups of Tyr 33 and Phe 79 probably through stacking interactions with the substrates in the catalytic reaction. The aromatic cluster Phe 278 -Phe 279 showed a more limited specificity for the aromatic residues than did the former two sites (Tyr 33 and Phe 79 ). The cluster preferred phenylalanine and tyrosine residues, but not histidine presumably due to a repulsion between the two charged imidazole rings of the serial histidine residues.
As shown in Fig. 5, for the substrates without the 3Ј projection of the upstream strand (the nick, 5Ј-recess-end, single flap, FIG. 5. Comparison of the K j and K int K p values among WT and alanine mutants using the substrates with and without the 3 projection of the upstream strand for endo-and exo-activities. The K j and K int K p values were evaluated from the K m and k cat values according to the equations described under "Results." and pseudo-Y substrates), the K p values were reduced remarkably by mutation at the three substrate binding sites (Tyr 33 , Phe 79 , and Phe 278 -Phe 279 ), because the K int was postulated as invariant, whereas, for the double flap and nick substrates with the 3Ј protrusion of the upstream strand, the K p values were not drastically changed by the mutations. Furthermore, the K j values for all substrates were not significantly changed by the mutations. These results suggest that Tyr 33 , Phe 79 , and Phe 278 -Phe 279 might play essential roles in forming the transient productive ES complex via multiple stacking interactions with the substrates without the 3Ј projection of the upstream strand using these central aromatic residues located close to the active site, because the K p values for the substrates without the 3Ј projection were drastically changed, whereas the K j values for the substrates were not significantly reduced by the mutations. On the other hand, the three sites probably play minor roles in forming the productive intermediates for the double flap and nick substrates with the 3Ј protrusion, because the K p values for the substrates with the 3Ј projection were moderately maintained even with the mutations as shown in Fig. 5.
As shown in Fig. 2A, the aromatic groups at the 33rd and 79th positions of phFEN-1 are conserved completely not only among FEN-1s but also among the family enzymes containing T5 and T7 exonuclease, and the exonuclease domains of E. coli polymerase 1 and Taq polymerase 1. In the superimposition of the molecular structures shown in Fig. 2B, these two aromatic residues are kept at the same positions and in a similar direction. Interestingly, as shown in Fig. 2B, the aromatic ring of the Phe 112 residue for the T5 exonuclease occupies the same position as that of the Phe 79 residue of phFEN-1, although the two residues Phe 112 and Phe 79 are derived from distinct portions of the superimposed structures. These results suggested that the cooperative stacking interactions with the substrates with the conserved aromatic residues close to the active center might work commonly as a key determinant in the hydrolysis reactions of the family enzymes.
The Stacking Interactions of the Residues Tyr 33 and Phe 79 Might Play Important Roles in Fixing the Template Strand and the Downstream Strand, Respectively-According to the molecular structure (17), the kinetic parameters and the substrate digestion patterns as summarized in Figs. 3-6, a possible model of the binding of phFEN-1 with the nick substrate is proposed in Fig. 7. The large loop (81-128) colored in pink (Fig.  2C) was removed from the structure to reduce steric hindrance between the loop and substrate, and to keep the active center accessible to the substrate, because the large loop was demonstrated to be very flexible and to hide the active center (17). For the orientation of the nick substrate to the active cleft, the reported DNA binding models of FEN-1s was referred to (22,25). The downstream duplex DNA was arranged to bind to the HhH motif (25), and the upstream duplex was designed to bind FIG. 6. The cleavage positions and recognition sites of the substrates. A, measurement of the product size. The assay conditions were described under "Experimental Procedures" using the WT enzyme. The amount of WT was 1-10 ng, and the amount of substrates was 3-5 pmol. The structure of each substrate is shown above the corresponding lanes. Lanes 1, 2, 6, and 7 correspond to the products from the nick substrate, the 5Ј-recess-end substrate, the pseudo-Y substrate with a 4-mer flap strand, and the single flap substrate with a 4-mer flap strand, respectively. Lane 12 corresponds to the products from the double flap substrate with a 1-mer 3Ј-end projection of the upstream strand and with a 19-mer-flap strand. Lane 14 corresponds to the products from the nick structure having a 1-mer 3Ј-end projection of the upstream strand. Lanes 3-5, 8 -11, 13, and 15-17 contain 1-mer, 2-mer, 3-mer, 4-mer, 5-mer, 6-mer, 7-mer, 20-mer, 1-mer, 2-mer, and 3-mer oligonucleotide markers labeled with FAM, respectively. These oligonucleotides had the same sequence as the cleaved products. The positions of FAM are shown by an asterisk. B, the recognition sites in the nick and 5Ј-recess-end substrates for the phFEN-1 molecule. The assay conditions were described under "Experimental Procedures." In B: part I, nick substrates with FAM labeling at the 5Ј-end of strand C and with the 2Ј-O-methyl substitution of the deoxyribose moiety were used. The abbreviations, C2-C5, correspond to the substituted deoxyriboses located at the 2nd through 5th positions from the 5Ј-end of strand C. The abbreviations, A24 -A28, correspond to the substituted deoxyriboses located at the 24th through 28th positions from the 5Ј-end of strand A. The relative activities toward the substituted substrates were calculated with the activity for the substrate without the 2Ј-O-methyl substitution as 100%. The  Fig. 2C, which were identified as the major DNA binding sites for the nick substrate (17). In the molecular structure shown in Figs. 2C and 7, Tyr 33 , Phe 79 , and Phe 278 -Phe 279 are located close to active magnesium ions 1 and 2. The residue Phe 79 and the cluster Phe 278 -Phe 279 were in alignment with magnesium ions 1 and 2, whereas Tyr 33 was located at right angles to the Phe 79 aligned with the two magnesium ions. Given these structural constraints, the Tyr 33 residue was arranged close to the template strand, and the Phe 79 residue and the cluster Phe 278 -Phe 279 were situated in the vicinity of the downstream strand in the modeled complex as shown in Fig. 7. The k cat values of Y33A and Y33L for the nick substrate are ϳ6 and 9 times higher than those for the 5Ј-recess-end substrate, respectively, although the only structural difference between substrates lies in the presence of the upstream strand as shown in Fig. 1. The substrate binding, in which the residue Tyr 33 was accessible to the nick portion of the substrate as shown in Fig. 7, was well supported by the experimental results mentioned above, because the nick position should be influenced by the presence or absence of the upstream strand. The binding was also consistent with the finding that the replacement of Tyr 33 with leucine more markedly affected the k cat values for the 5Ј-recess-end and nick substrates (1180-and 134-fold decreases, respectively) than the replacement with alanine (333-and 53-fold decreases, respectively), suggesting the preference for a methyl group rather than bulky aliphatic group at the 33rd position, probably due to the tight space surrounding the aromatic group of Tyr 33 caused by the distorted nick portion owing to the bent conformation of the template strand.
According to the complex structure modeled with the nick substrate (Fig. 7), the residue Tyr 33 exists close to the nick portion of the template strand and is able to undergo a stacking interaction with the base moiety of the distorted nick portion. The Tyr 33 residue might play essential roles in fixing and distorting the template strand through the stacking interaction. The location of Phe 79 is suitable for the stacking interaction with the base moiety at the 5Ј-end of the downstream strand. The cluster Phe 278 -Phe 279 is located in the vicinity of the third base moiety from the 5Ј-end of the downstream strand and is able to undergo the stacking interaction with the base flipping out. In the modeled structure (Fig. 7), the second phosphate moiety from the 5Ј-end of the downstream strand is suitable for forming a hydrogen bond with the active magnesium ion 1, suggesting that the phosphoester linkage should be cleaved. The manner in which the substrate was recognized in the modeled structure might be elucidated well by the following experimental results: 1) the nick substrate was cleaved at the 1st nucleotide from the 5Ј-end of the downstream strand (Fig.  6A); 2) the 3rd and 4th nucleotides from the 5Ј-end of the downstream strand were specifically recognized by the phFEN-1 molecule (Fig. 6B (part I)); 3) the 26th, 27th, and 28th bases from the 5Ј-end of the template strand were strongly recognized by the phFEN-1 molecule (Fig. 6B (part I)).
For the other substrates (the 5Ј-recess-end, pseudo-Y, and single flap substrates), the same mechanism of binding as that of the nick substrate might occur, although the cleavage positions on the substrates might be adjusted depending on the specific binding of each strand to these central aromatic residues. For the pseudo-Y and single flap substrates, the flap strands might pass through the large loop via interaction with the Phe 79 residue.
In the preceding report (17) we identified four DNA binding sites (Arg 40 -Arg 42 , Arg 94 , Arg 118 -Lys 119 , and Lys 193 -Arg 194 -Lys 195 ) located on a flexible loop at the surface of the molecular structure. Using these DNA binding sites far apart from the active center, phFEN-1 might bend the substrates to make accessible the downstream strands to the active center located interior of the molecule. Simultaneously, the stacking interactions of Tyr 33 and Phe 79 adjacent to the active center might play important roles in fixing the hinge portion of the template strand and the 5Ј-end of the downstream strand as shown in the modeled complex. The stacking interactions might cause the partial unwinding of the 5Ј-end of the downstream strand from the template strand to form the productive transient state leading to the hydrolysis.