IDENTIFICATION OF A NOVEL BINDING MOTIF IN PYROCOCCUS FURIOSUS DNA LIGASE FOR THE FUNCTIONAL INTERACTION WITH PROLIFERATING CELL NUCLEAR ANTIGEN

DNA ligase is an essential enzyme for all organisms that catalyzes a nick-joining reaction in the final step of the DNA replication, repair, and recombination processes. Herein we show the physical and functional interaction between DNA ligase and proliferating cell nuclear antigen (PCNA) from the hyperthermophilic euryarchaeon, Pyrococcus furiosus.

DNA ligase is an essential enzyme for all organisms that catalyzes a nick-joining reaction in the final step of the DNA replication, repair, and recombination processes. Herein we show the physical and functional interaction between DNA ligase and proliferating cell nuclear antigen (PCNA) from the hyperthermophilic euryarchaeon, Pyrococcus furiosus. The stimulatory effect of P. furiosus PCNA (PfuPCNA) on the enzyme activity of DNA ligase (PfuLig) was observed not at a low ionic strength, but at a high salt concentration, at which a DNA ligase alone cannot bind to a nicked DNA substrate. Based on mutational analyses, we identified the amino acid residues that are critical for the PCNA binding in a loop structure located in the N-terminal DNA binding domain (DBD) of PfuLig. We propose that the pentapeptide motif QKSFF is involved in the PCNA interacting motifs, in which Gln and the first Phe are especially important for the stable binding with PCNA.
DNA ligases catalyze nick sealing reactions via three nucleotidyl transfer steps, as described in recent review articles (1)(2)(3)(4). In the first step, DNA ligases form a covalent enzyme-AMP intermediate, by reacting with ATP or NAD + as a cofactor (step 1). In the second step, DNA ligases recognize the substrate DNA, and the AMP is subsequently transferred from the ligases to the 5'-phosphate terminus of the DNA, to form a DNA-adenylate intermediate (AppDNA) (step 2). Then, in the final step, the 5'-AppDNA is attacked by the adjacent 3'-hydroxy group of the DNA to form a phosphodiester bond (step 3). Three genes (LIG1, LIG3 and LIG4) encoding ATP-dependent DNA ligases have been identified in the human genome to date. Human DNA ligase I (Lig I) is a replicative enzyme that joins Okazaki fragments during the DNA replication process. It is well known that many eukaryotic proteins involved in DNA replication, DNA repair, and cell cycle control interact with a DNA sliding clamp, proliferating cell nuclear antigen (PCNA) (reviewed in [5][6][7]. Extensive studies of the PCNA interacting proteins revealed the existence of a consensus sequence, called the PCNA interacting protein box (PIP box) (5). The PIP box consists of the sequence "Qxxhxxaa", where "x" represents any amino acid, "h" represents hydrophobic residues (e.g. L, I or M), and "a" represents aromatic residues (e.g. F, Y or W). Furthermore, it has been proposed that other sites of the interacting protein can participate in PCNA binding.
For example, a conserved pair of Lys and Ala residues was identified as a PCNA binding motif (KA box) by using a random peptide display library (8).
However, the importance of the KA box is not obvious, because a detailed biochemical analysis of the motif has not been performed to date.
In Escherichia coli, the corresponding DNA sliding clamp is the β subunit of DNA polymerase III holoenzyme (henceforth referred to as the β clamp), which forms a toroidal dimeric structure (9).
A bioinformatics approach revealed that a pentapeptide motif (consensus QL[SD]LF) plays an important role in binding to the β clamp (10).
In higher eukaryotes, human Lig I reportedly forms a stable complex with a PCNA trimer that is topologically linked to duplex DNA via an N-terminal PIP box motif (11,12). The structures of eukaryotic DNA ligases can be divided into two major domains, an N-terminal noncatalytic domain (NCD) and a C-terminal catalytic domain (CD), which consists of an adenylation domain and an OB-fold domain (3,13). The crystal structure of human Lig I in complex with a nicked DNA and biochemical analyses of the enzyme revealed that the NCD provides most of the DNA binding affinity (14), and therefore, this domain is called the N-terminal DNA-binding domain (DBD).
Although several groups have characterized the physical interactions between human DNA ligase I and PCNA, no stimulatory or inhibitory effect on nick-joining activities has been observed in vitro (11,15). In contrast, a stimulatory effect of PCNA was also reported (16).
Thus, the detailed interaction mode between human PCNA and Lig I is somewhat unclear.
In Archaea, the third domain of life, a single homolog of the eukaryotic DNA ligase I has been identified (17)(18)(19)(20)(21). Interestingly, although most of the archaeal replicative enzymes have a eukaryotic PIP box at their C-terminus, no clear PCNA binding motif has been observed in the sequences of archaeal DNA ligases (5,22).
Recently, a physical and functional interaction between PCNA and DNA ligase from Sulfolobus solfataricus was reported (23). In the S. solfataricus DNA ligase (SsoLig), PIP box-like motifs were proposed to exist in the N-terminal region. Furthermore, a mutant SsoLig, which lacks 30 amino acids at the N-terminus, cannot interact with SsoPCNA (23, 24). However, it has not been determined whether the proposed motifs are actually important for the interaction with PCNA.
Here, we show a physical and functional interaction between PCNA and DNA ligase from the hyperthermophilic euryarchaeon, Pyrococcus furiosus. The stimulatory effect of P. furiosus PCNA (PfuPCNA) on the nick-joining reaction of DNA ligase (PfuLig) was observed under physiological conditions with an extremely high salt concentration (0.5-0.6 M). Furthermore, we show that the pentapeptide sequence QKSFF, in the DBD of PfuLig, plays an important role in binding to PfuPCNA. Interestingly, this motif is located in a loop connecting two α-helices in the DBD, but is not at the N-or C-terminus of PfuLig, based on our crystal structure (Nishida et al. Submitted). We propose a novel PCNA binding motif, which may be located inside, but not at the terminus, of the PCNA-interacting proteins.

EXPERIMENTAL PROCEDURES
Cloning the Pfu DNA ligase gene and its mutants --The DNA ligase gene (lig) was amplified by PCR directly from P. furiosus genomic DNA, using the oligonucleotides 5'-GGCCATGGGTTATCTGGAGCTTG CTCAAC-3' and 5'-GCGGATCCTTAGCTTTCCACTTT TCCTTTCATC-3' as the forward and reverse primers, respectively. The amplified gene was cloned into the pGEM-T easy vector (Promega), and its nucleotide sequence was confirmed. The cloned gene was digested by NcoI-BamHI and was inserted into the corresponding sites of pET21d (Novagene). The resultant plasmid was designated as pET-Lig. Amino acid substituted mutations and N-terminally truncated mutations were introduced into the lig gene on the pET-Lig plasmid by PCR, using the appropriate primers. Their sequences are available on request. Overproduction and purification of the Pfu DNA ligase proteins --To obtain the recombinant PfuLig, E. coli BL21 codonPlus TM (DE3)-RIL (Stratagene) carrying pET-Lig was grown in 1 liter of LB medium, containing 50 µg/ml ampicillin and 34 µg/ml chloramphenicol, at 37 . The cells were cultured to an A 600 of 0.40, and then the expression of the lig gene was induced by adding isopropyl β-D-thiogalactopyranoside to a final concentration of 1 mM and continuing the culture for 5 h at 37 . After cultivation, the cells were harvested and disrupted by sonication in buffer A, containing 50 mM Tris-HCl, pH 8.0, 0.5 mM DTT, 0.1 mM EDTA, and 10% glycerol.
The soluble cell extract, obtained by centrifugation (12,000 x g, 20 min), was heated at 80 for 20 min. The heat-resistant fraction obtained by centrifugation was treated with 0.15% polyethylenimine to remove the nucleic acids.
The soluble proteins were precipitated by 80% saturated ammonium sulfate precipitation.
The precipitate was resuspended in buffer B, containing 50 mM Tris-HCl, pH 8.0, 1 M (NH 4 ) 2 SO 4 , 0.5 mM DTT, 0.1 mM EDTA, and 10% glycerol, and was subjected to chromatography on a Hitrap Phenyl column (Amersham Biosciences). The proteins were eluted at 0 M ammonium sulfate, and the eluted proteins were dialyzed against buffer A. The dialysate was loaded onto a Hitrap Heparin column (Amersham Biosciences), and the proteins were eluted at 0.3-0.35 M sodium chloride. The eluted proteins were dialyzed against buffer A, and the dialysate was subjected to chromatography on a Hitrap Q column (Amersham Biosciences). The proteins were eluted at 0.1-0.15 M sodium chloride, pooled and stored at 4 . The mutated PfuLig proteins prepared in this study were purified by the same procedures. The purity of each proteins used in this study was evaluated by SDS-PAGE.
No extra band was detected by Comassie brilliant blue staining of the gel containing 2 µg of each purified protein.
After electrophoresis, the gels were dried and were scanned by using an image analyzer, FLA5000 (FUJIFILM), to detect the Ado-32 P-DNA Ligase adducts. DNA ligation assay --The substrate DNA used in the ligation assay was a 49 bp DNA duplex containing a single nick at the center. The 22 mer deoxynucleotide (5'-AATTCGTGCAGGCATGGTAGCT -3'), which was labeled with 32 P at the 5'-terminus, and the 27 mer deoxynucleotide (5'-AGCTATGACCATGATTACGAAT TGCTT-3') were annealed to the 49 mer deoxyoligonucleotide with a complementary sequence in TAM buffer, containing 40 mM Tris-acetate, pH 7.8, and 0.5 mM magnesium acetate. The purified PfuLig proteins (at different concentrations for each experiment, as described in the figure legends) were incubated with 5 nM of the nicked DNA substrate, prepared as described above, in 20 µl of the ligation buffer, containing 20 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 1 mM DTT, 0.01 mM ATP, 0.1% Tween 20 and 0.1 mg/ml BSA, at 60 for 15 min. Reactions were initiated by the addition of enzyme and were terminated with 5 µl of stop solution, containing 98% formamide, 10 mM EDTA, 0.1% bromophenol blue, and 0.1% xylene cyanol. Samples were heated at 100 for 5 min and chilled rapidly on ice prior to loading onto a 10% polyacrylamide gel containing 8 M urea.
After electrophoresis, the gels were dried and were scanned by using FLA5000 to detect the 32 P-labeled DNA.
Three independent experiments were carried out in succession for each ligation condition required in this study, and the standard error are shown as vertical lines on the plots in each graph. Surface plasmon resonance analysis --The BIAcore system (BIACORE) was used to study the physical interaction between PfuLig and PfuPCNA. Highly purified recombinant PfuLig or PfuPCNA (25) was fixed on a Sensor Chip CM5, research grade (BIACORE), according to the manufacturer's recommendations.
To measure the kinetic parameters, various concentrations of PfuPCNA were applied to the immobilized PfuLig. All measurements were performed at a continuous flow rate of 30 µL/min, in a buffer containing 10 mM HEPES, pH 7.4, 150 mM NaCl, and 0.005% Tween 20. At the end of each cycle, the bound protein was removed by washing with 2 M NaCl. The kinetic constants for PfuPCNA binding to PfuLig were determined from the association and dissociation curves of the sensorgrams, using the BIAevaluation program (BIACORE).

RESULTS
Biochemical properties of P. furiosus DNA ligase --PfuLig has already been characterized and is commercially available, mainly as a reagent for ligase chain reactions (Stratagene, Patent# US 5506137). In this study, we cloned and purified PfuLig independently and constructed mutant proteins to analyze the structure-function relationships of this enzyme. It was predicted from the primary amino acid sequence similarity that PfuLig is an ATP-dependent DNA ligase. Therefore, we constructed the mutant lig gene encoding PfuLig K249A, in which the lysine at the predicted adenylylation site was substituted by alanine, in parallel with the gene for the wild type PfuLig, and tested their adenylyltransferase activities in the presence of ATP. As shown in Figure  1A, wild type PfuLig can form a covalent enzyme-AMP intermediate by reacting with [α-32 P]ATP as a cofactor, but no adenylyltransferase activity was observed with the K249A mutant protein. It was reported that some thermophilic DNA ligases from archaea utilize ADP (26) or NAD + (19, 27) as a cofactor. However, we detected a distinct activity of PfuLig in the presence of ATP, but not ADP, AMP, and NAD + (Fig. 1B). A very small amount of ligation product was detected by the reaction with ADP. This result is the same as the case of DNA ligase from Pyrococcus horikoshii in the recent report (21). We think the ligation reaction may be derived from contaminant ATP (1.16 %) in our ADP reagent (Oriental yeast, Co, Osaka) according to the manufactures certificate, and therefore, we concluded that ADP is not an appropriate cofactor for PfuLig, as Shuman described for P. horikoshii DNA ligase (21).
Our crystallographic study of the wild type PfuLig revealed that the protein consists of three distinct domains, the N-terminal DNA binding domain (DBD), the middle adenylylation domain, and the C-terminal OB-fold (oligonucleotide binding fold) domain (Nishida et al., submitted). The last two domains are commonly called the catalytic core domain (CD), which is conserved in one branch of the nucleotidyltransferase superfamily, containing DNA ligase, RNA ligase, and mRNA capping enzymes (reviewed in 4). We made truncated PfuLigs, including the DBD with amino acids from 1 to 218 (N-terminal domain) and the CD from 219 to 561(the middle and C-terminal domains) to investigate the functions of each domain (Fig. 1C). A nick-joining assay was performed by using a wild type enzyme (WT) and the mutant PfuLigs, K249A, DBD, and CD (Fig. 1D). The CD protein could not complete the nick sealing reaction, even at high enzyme concentrations. A very small amount of ligation product was observed when the CD protein was added with concentration at 100 times higher than that of the wild type (Fig. 1D, lane CD).
The accumulation of AppDNA products implied that the CD protein exhibits a lower activity at the "step 3" reaction. The CD protein shares a structural similarity with the full-length Chlorella virus DNA ligase, which is the smallest ATP-dependent DNA ligase with a distinct activity in vitro (28), and therefore, it is very interesting to investigate why the CD from PfuLig lacks most of the enzyme activity. This result indicates the importance of the DBD for the overall ligation activity of PfuLig, and it can be predicted from the structural similarity that the contribution of the DBD domain to the ligation reaction is also conserved in the eukaryotic DNA ligases. Pfu DNA ligase can interact with both monomeric and trimeric PCNA proteins --To determine the physical interaction between PfuLig and PfuPCNA, we first used an immunoprecipitation (IP) method. The PfuPCNA and PfuLig proteins were incubated together and then precipitated by each antiserum.
However, significant interactions between them were not detected under several experimental conditions, probably because the protein-protein interaction is too weak to be detected by an IP method (data not shown).
We therefore performed surface plasmon resonance (SPR) experiments to analyze the weak PCNA-DNA ligase interactions. The full-length PfuLig was immobilized onto the CM5 BIAcore sensor chip, and subsequently the wild type PfuPCNA and the mutant PfuPCNA D143A/D147A, which is unable to form a stable toroidal structure in solution and thus cannot stimulate P. furiosus DNA polymerase B activity (29), were injected at different concentrations.
The physical interactions between the immobilized PfuLig and the two PCNA proteins were identified by the SPR sensorgram (Fig. 2). The calculated equilibrium constant K D values for the wild type PCNA and the D143A/D147A mutant were 1.1 x 10 -7 M and 1.4 x 10 -6 M, respectively. The K D values reported here are comparable to that of the human PCNA-p21 interaction determined by SPR analysis (K D : 3.2x10 -7 M) (30). These findings suggest that the toroidal structure is not necessarily required to form a stable PfuLig-PfuPCNA complex in vitro. PfuPCNA enhances the ligation activity of PfuLig at a physiological salt concentration --It is well known that some hyperthermophilic archaea contain strikingly high intracellular potassium ion concentrations. Based on the study of a euryarchaeon, Pyrococcus woesei, which was later proved to be a subspecies of P. furiosus (31), the potassium ion concentration in the hyperthermophilic archaeal cells was determined to range between 0.5 to 0.6 M (32). We initially examined the effect of increasing salt concentrations on the nick-joining activity of PfuLig by supplementing the reaction with KCl and K-Glu (potassium glutamate) salts. A reduction in the ligation activity was observed with each of these monovalent salts in a concentration-dependent manner, and about 90% inhibition was seen at a 200 mM salt concentration (Fig. 3A). A similar result was obtained in the enzyme assay using NaCl (data not shown). These observations are not specific for PfuLig, as the same phenomena were reported in the characterizations of other DNA and RNA ligases (18,27,33).
To determine whether PfuPCNA can stimulate the ligation activity of PfuLig at the physiological ionic strength, the proteins were assayed in a broad range of salt concentrations.
The stimulation effect of PfuPCNA on PfuLig was observed at 0.05-0.2 M KCl, but the effect was decreased over 0.2 M KCl (Fig.  3B). In the same manner, we performed the enzyme assay by adding K-Glu salt, which was reported to be an important factor contributing to the thermostability of archaeal proteins (34). As shown in Figure 3C, the stimulation effect was observed over 0.  Fig. 1). These results show that a chloride ion (Cl -) concentration over 0.2 M, but not this concentration of potassium ion (K + ) had an inhibitory effect on the enzyme activity of PfuLig. The same result was observed in the characterization of a Holliday junction resolving enzyme, Hjc, from P. furiosus, which exhibited the maximum enzyme activity at 0.2 M KCl (35). For the P. furiosus enzymes that catalyze nucleic acid modification reactions, high Clconcentrations may affect their activity. A novel PCNA binding site in the N-terminal DBD of PfuLig --To determine the region responsible for PCNA binding in PfuLig, we utilized the two truncated mutants, DBD and CD, as shown in Figure 1C. The interactions between PfuPCNA and these truncated PfuLig proteins were examined qualitatively by an SPR analysis. The wild type PfuPCNA was immobilized onto a CM5 BIAcore sensor chip, and the two truncated DNA ligase mutants were then injected. The wild type PfuLig and DBD interacted with the immobilized PfuPCNA to almost the same extent, but CD had no binding ability (Fig. 4). This SPR analysis using immobilized PfuPCNA showed very low resonance units as compared with that shown in Figure 2, in which PfuLig was immobilized. This phenomenon often happens in our experience of the SPR analyses using PfuPCNA and its binding proteins.
The difference probably depends on the direction of the proteins fixed on the sensor chip. Due to the relatively low resonance units (< 250 RU) observed in this experiment, the equilibrium constant K D was not determined. These findings suggest that the N-terminal DBD plays a critical role in the PCNA binding of PfuLig.
In Archaea, it was reported that S. solfataricus DNA ligase has a PCNA binding site in its N-terminal region (23).
To determine whether the same region of PfuLig is responsible for PCNA binding, we cloned and purified two N-terminally truncated mutants, PfuLigΔN14 (15-561) and PfuLigΔN32 (33-561), on the basis of the crystal structure of PfuLig (Fig.  5A). In addition, we carefully examined the amino acid sequence of the DBD, and found some regions that may be involved in interactions with PfuPCNA. Single amino acid substituted mutants in these regions were prepared to examine their effects on the PCNA-interaction. These regions included a candidate KA box and a PIP box-like motif found in the DBD (Fig. 5B). The Lys 67 in the candidate KA box and the two aromatic residues, Phe 106 and Phe 107 , in the PIP box-like sequence were examined by alanine substitutions. To test the stimulation activity of PCNA in the nick sealing reaction (described above) under equivalent conditions, the relative activities of these mutant proteins were determined without PfuPCNA.
The F106A/F107A mutant exhibited almost the same activity as that of the wild type PfuLig (Fig. 5C).
The decreased activity observed in the two N-terminally truncated mutants, ΔN14 and ΔN32, revealed that the integrity of the DBD is important for the overall ligation activity. In order to determine the region responsible for PCNA binding, we tested the stimulation effect of PfuPCNA on these mutants. As a result, only the ligation activity of F106A/F107A was not stimulated by PfuPCNA (Fig. 5D). Furthermore, no effect of PfuPCNA was observed with increasing concentrations of the F106A/F107A mutant PfuLig (data not shown). The PfuPCNA-dependent ligation ability of the K67A mutant was not different from that of the wild type (Fig. 5E). We concluded that at least one of the two aromatic residues, Phe 106 or Phe 107 , plays a crucial role in PCNA binding via a hydrophobic interaction. Gln 103 and Phe 106 of PfuLig are critical for the functional interaction with PfuPCNA --The PIP box-like sequence, 103 QKSFF 107 described above, is located in a loop structure in the DBD, based on the crystal structure of PfuLig (Fig. 6A). Using three single amino acid substituted mutants, Q103A, F106A, and F107A, we examined the detailed roles of each of these amino acid residues in the 103 QKSFF 107 sequence.
The specific activities of these mutant PfuLigs were confirmed to be the same ( Supplementary  Fig. 2).
The physical interactions between the mutant PfuLig proteins and the immobilized PfuPCNA were analyzed by SPR analyses. The F106A mutant was not able to bind to PCNA, but the Q103A and F107A mutants showed very weak responses (50 RU) as compared the with wild type PfuLig (240 RU) (Fig. 6B). Next, the stimulatory effect of PCNA on the ligation activity of these mutants was examined in vitro. As a result, the activities of the Q103A and F106A mutants were slightly stimulated by PfuPCNA, whereas the F107A mutant exhibited an intermediate response to PfuPCNA (Fig. 6C). These analyses indicate that Phe 106 in the 103 QKSFF 107 sequences is the most important residue in the physical and functional interactions with PfuPCNA. The Gln 103 residue may stabilize the PCNA-DNA ligase complex after connecting PfuLig to PfuPCNA with Phe 106 residue.
The F107A mutant PfuLig, which possesses the Gln 103 and Phe 106 residues, showed very weak binding activity to PfuPCNA in the SPR analysis, comparable to that of the Q103A mutant, but a distinct response to PCNA was retained in the ligation assay. Further analyses will be required to understand the detailed role of Phe 107 in the functional interaction between PfuLig and PfuPCNA.

DISCUSSION
Functional roles of the conserved DBD in eukaryotic DNA ligases --As shown in our mutational analyses, the integrity of the DBD is important for the overall ligation activity of PfuLig itself (Fig 1D). Furthermore, the other important function of the DBD is to interact with PCNA. The functional interaction between PfuLig and PfuPCNA seems to be stoichiometric. However, much excess amounts of PfuPCNA is required for stimulation of ligation reaction by PfuLig (Fig. 5D). This inconsistency could be explained by the difficulty of PCNA loading onto the DNA fragment in the assay mixture. The PCNA trimer loads by diffusion onto the double-stranded DNA fragment over the ends without clamp loader (RFC) in this case. This is probably a limited process, and efficient loading requires a large stoichiometric excess of PCNA as discussed previously for human Lig I (16).
In Eukarya and Archaea, PCNA binding proteins generally interact with PCNA via a conserved PIP box motif (e.g. archaeal DNA polymerase B, and flap endonuclease 1 have a typical PIP box motif at their C-terminus (reviewed in 7). Human Lig I has a typical PIP box at the N-terminal tail, but PfuLig lacks a long N-terminal tail.
We identified the PCNA-interaction motif of PfuLig in a loop structure, which connects two α-helices in the N-terminal DBD. Based on the information from the crystal structure of human Lig I (N-terminal truncated mutant) complexed with a nicked duplex DNA, a model structure of the DNA ligase-PCNA complex with 1:1 stoichiometry was proposed (14). This interaction is likely to involve "face to face" binding because of the proteins' similar sizes and toroidal structure of PCNA.
After binding to the PCNA-DNA complex via the PCNA binding motif in the DBD, the conformations of the CD may change freely to encircle a nicked DNA, because it has no interactive region with PCNA (Fig. 4), and subsequently the enzyme catalyzes the nick-joining reaction. PCNA is a scaffold protein for binding to DNA under physiological ionic conditions --There have been some contradictory observations about the stimulatory effect of human PCNA on DNA ligase I activity.
One group suggested that these discrepancies are due to differences in the experimental conditions (16). As shown in Figure 3B and 3C, the PfuLig activity was inhibited by PCNA at low salt concentrations In a previous report on the inhibitory effect of PCNA on human DNA ligase I, the inhibitory effect was observed at 0 and 50 mM NaCl and no effect of PCNA on ligation was observed at 100 mM NaCl (in pH 6.5 buffer) (15). Based on our findings, the stimulatory effect may be observed at 150 mM NaCl, which is near the physiological ionic strength within human cells. However, it is not easy to discuss the differences in the assay conditions, because the salt concentration of each fraction containing the purified recombinant protein is not always obvious from the presented information.
It can be predicted that, by themselves, eukaryotic DNA ligases cannot bind to DNA to catalyze the nick-joining reaction at a physiological salt concentration, but they can recognize the substrate DNA by interacting with PCNA on a nicked DNA. Most of the DNA modification enzymes can interact with substrate DNAs to express their function at low salt concentrations, but lose their activities at high salt concentrations in vitro.
The DNA binding abilities of these enzymes themselves are probably inhibited by salt in the cells. Each protein involved in DNA replication and repair has to work at a certain time in the successive processes at their appropriate sites. To control the specific timing and the position for each related protein factor to access the target DNA in vivo, the salt concentration, which prevents non-specific binding of protein factors in the cells, is especially important, and in the case of replication fork progression, for example, PCNA probably functions as a platform to control the order and the sites of interacting proteins involved in this successive reaction process. The conserved residues in the novel PCNA binding motif "QKSFF" --The well-known PIP box is generally located in the N-terminal or C-terminal region within the peptide chain of the PCNA interacting proteins.
However, the PCNA-binding motif, QKSFF, found in this study is in the middle of the PfuLig protein.
This novel PCNA binding motif resembles a putative bacterial β clamp binding motif, QL[SD]LF, which is located not only at the terminus but also in the middle of some β clamp interacting proteins. In this bacterial motif, the pair of hydrophobic residues, LF, is important for binding to the β clamp (10).
In PfuLig, the corresponding hydrophobic residues, FF, are also important for binding to PCNA, and furthermore, our work showed that the former residue, Phe 106 , is more critical than Phe 107 (Fig. 6C). Moreover, the importance of Gln 103 was revealed in our experiments. A structural comparison of the novel motif in the PfuLig crystal with that of the PIP box in the RFC large subunit in the cocrystal with PfuPCNA (36) is shown in Figure 7. Interestingly, the locations of the amino acid residues responsible for the hydrophobic and ionic interactions, respectively, clearly correspond to each other, and especially, the positions of Gln 470 and Phe 476 in the PIP box of RFCL, corresponding to Gln 470 and Phe 476 in PfuLig, are remarkably conserved among the PIP box sequences.
This new motif may represent a shorter version of the original PIP box. To determine the detailed role of each amino acid in PCNA binding, an X-ray crystallographic structure of the PfuPCNA-PfuLigase complex will be required.
Interestingly, this novel PIP box motif is widely conserved in the same region of other archaeal DNA ligases (Fig. 8A). The QKSFF sequence is completely conserved, especially in Thermococcals (Pyrococcus and Thermococcus species). In addition, the basic residues located in the region upstream of the motif are also conserved, and especially, the remarkable cluster of basic residues is conserved in the DNA ligases from Thermococcals and some Methanogens.
As we proposed previously, based on mutational analyses of the RFC large subunit (RFCL) from P. furiosus (37), these basic residues may function for the formation of the stable Lig-PCNA-DNA complex in these organisms.
We examined the sequences of the eukaryotic DNA ligases, and found that they also have the archaea-type PIP-box in the middle of the peptide chain (Fig.  8B). Eukaryotic Lig I may bind to PCNA at the site corresponding to the motif that we found in this study as discussed above, in addition to the N-terminal PIP box. These analyses show that the PCNA-DNA ligase interaction mode is also interesting from an evolutionary perspective, and we plan to investigate this possibility by introducing mutations into the conserved Glu in the human Lig I protein.  Physical interaction between PfuLig and PfuPCNA. SPR analysis was performed using a BIAcore system to detect a physical interaction between PfuLig and PfuPCNA. Purified PfuLig was immobilized on a BIAcore sensor chip, and purified PfuPCNA (3 µM) was loaded. The wild type PfuPCNA and a mutant PfuPCNA (D143A/D147A), which cannot form a stable trimeric ring structure, were used to investigate their affinities to PfuLig. The equilibrium constant K D was calculated from the obtained sensorgram.  The amino acid sequence of the DBD was examined carefully, and a new PIP box-like motif was found, in addition to the candidate KA box, as shown with a black box. The regions containing these candidate motifs were aligned with the human Lig I sequence. This PIP box-like motif is located in the loop structure connecting 6th and 7th α-helices (residues 103 to 107) of the PfuLig crystal structure. In the crystal structure of human Lig I (huLig I), a part of the corresponding region is disordered (residues 385 to 392, indicated by dashed line) (14).