Function and assembly of the bacteriophage T4 DNA replication complex: interactions of the T4 polymerase with various model DNA constructs.

Complexes formed between DNA polymerase and genomic DNA at the replication fork are key elements of the replication machinery. We used sedimentation velocity, fluorescence anisotropy, and surface plasmon resonance to measure the binding interactions between bacteriophage T4 DNA polymerase (gp43) and various model DNA constructs. These results provide quantitative insight into how this replication polymerase performs template-directed 5' --> 3' DNA synthesis and how this function is coordinated with the activities of the other proteins of the replication complex. We find that short (single- and double-stranded) DNA molecules bind a single gp43 polymerase in a nonspecific (overlap) binding mode with moderate affinity (Kd approximately 150 nm) and a binding site size of approximately 10 nucleotides for single-stranded DNA and approximately 13 bp for double-stranded DNA. In contrast, gp43 binds in a site-specific (nonoverlap) mode and significantly more tightly (Kd approximately 5 nm) to DNA constructs carrying a primer-template junction, with the polymerase covering approximately 5 nucleotides downstream and approximately 6-7 bp upstream of the 3'-primer terminus. The rate of this specific binding interaction is close to diffusion-controlled. The affinity of gp43 for the primer-template junction is modulated specifically by dNTP substrates, with the next "correct" dNTP strengthening the interaction and an incorrect dNTP weakening the observed binding. These results are discussed in terms of the individual steps of the polymerase-catalyzed single nucleotide addition cycle and the replication complex assembly process. We suggest that changes in the kinetics and thermodynamics of these steps by auxiliary replication proteins constitute a basic mechanism for protein coupling within the replication complex.


From the Institute of Molecular Biology and Department of Chemistry, University of Oregon, Eugene, Oregon 97403
Complexes formed between DNA polymerase and genomic DNA at the replication fork are key elements of the replication machinery. We used sedimentation velocity, fluorescence anisotropy, and surface plasmon resonance to measure the binding interactions between bacteriophage T4 DNA polymerase (gp43) and various model DNA constructs. These results provide quantitative insight into how this replication polymerase performs template-directed 5 3 3 DNA synthesis and how this function is coordinated with the activities of the other proteins of the replication complex. We find that short (single-and double-stranded) DNA molecules bind a single gp43 polymerase in a nonspecific (overlap) binding mode with moderate affinity (K d ϳ150 nM) and a binding site size of ϳ10 nucleotides for single-stranded DNA and ϳ13 bp for double-stranded DNA. In contrast, gp43 binds in a site-specific (nonoverlap) mode and significantly more tightly (K d ϳ5 nM) to DNA constructs carrying a primer-template junction, with the polymerase covering ϳ5 nucleotides downstream and ϳ6 -7 bp upstream of the 3-primer terminus. The rate of this specific binding interaction is close to diffusion-controlled. The affinity of gp43 for the primer-template junction is modulated specifically by dNTP substrates, with the next "correct" dNTP strengthening the interaction and an incorrect dNTP weakening the observed binding. These results are discussed in terms of the individual steps of the polymerase-catalyzed single nucleotide addition cycle and the replication complex assembly process. We suggest that changes in the kinetics and thermodynamics of these steps by auxiliary replication proteins constitute a basic mechanism for protein coupling within the replication complex.
Major cellular processes, such as DNA replication, DNA transcription, RNA translation, and protein degradation, are carried out by multiprotein assemblies (sometimes called macromolecular machines) (1), within which the constituent proteins act in a coordinated manner to perform the specific functions of the complex at issue. Furthermore, these protein machines interact within the cell and are functionally coupled to additional complexes. For example, in eukaryotes, the transcription machinery is coupled to the RNA processing complex (or spliceosome), which is itself coupled to the nuclear mRNA export system (2). Similarly, the rescue of a stalled DNA replication fork by recombination proteins illustrates the functional interactions that occur between the DNA replication, recombination, and repair assemblies (3). As a consequence, a coordination or coupling of protein activities exists, not only within a given multiprotein complex, but also between such complexes within the cell. Attaining a molecular understanding of this protein-protein coupling requires characterization of the activities and structures of the proteins as they function in isolation and also as they operate in the presence of one or more protein and/or nucleic acid interaction partners.
The DNA replication machinery is a model of a multiprotein complex that must function correctly to ensure the viability of any living organism. The DNA replication machinery of bacteriophage T4 is entirely encoded by the phage itself and requires only eight proteins to catalyze coordinated leading and lagging strand synthesis in vitro, at a rate and with a processivity and fidelity comparable with the same parameters measured in vivo (for reviews see Refs. 4 -6). Depending on the stage of phage infection, phage T4 DNA replication is initiated at either an R-loop or at a D-loop, on which the replication proteins involved in initiation assemble. The T4 (gp41) replication helicase (7,8) binds to the single-stranded part of the R-loop or D-loop. When the single-stranded DNA portion of the D-loop is covered by T4-coded single-stranded DNA-binding protein (gp32), assembly of the helicase onto the complex requires the action of gp59 (the T4-coded helicase loading protein), which functions to displace the gp32 protein from the ssDNA 1 and thus create a binding site for the helicase.
The T4-coded helicase is an A/GTPase (9) and forms a hexamer upon activation by A/GTP binding (10). This hexamer, which comprises the functional form of the gp41 T4 helicase, can, in isolation, use its single-stranded DNA-stimulated A/GTPase activity to translocate along ssDNA templates with a 5Ј 3 3Ј polarity and with moderate processivity (7)(8)(9)11). This hexameric helicase, working within the replication complex, unwinds the double-stranded DNA (dsDNA) of the genome downstream of the fork, probably (by analogy to the helicase of bacteriophage T7 (12)) by working as a ring that encircles the lagging DNA template strand and binds to the leading template strand (8). As a consequence of its strand-separating activity, the helicase provides two ssDNA templates, called the leading and the lagging DNA strands, to the replication polymerases located behind the fork.
The single-stranded sequence of the leading strand template that is exposed by the helicase is immediately "copied" by gp43 (the T4-coded DNA polymerase). The gp43 bound to the primertemplate junction at the R-or D-loop extends the nascent DNA, using the newly exposed leading DNA strand as a template. In the absence of DNA and at free protein concentrations comparable with those achieved during phage infection, the gp43 core enzyme in isolation exists as a monomer (13,14) and functions with very limited processivity to incorporate nucleotide residues (nt) into the nascent DNA at physiological salt concentrations. It gains the processivity required for genome replication by interacting with gp45 (the T4-coded sliding processivity clamp), which is loaded at the p-t junction by gp44/62 (the T4-coded processivity clamp-loading complex), resulting in a closed, ring-shaped gp45 trimer that fits loosely around the dsDNA of the genome (15).
The gp45 clamp is loaded onto the DNA at a p-t junction by the gp44/62 complex in an ATP-driven reaction, and multiple conformational changes occur in these protein complexes during the loading process (16). Whereas the ssDNA of the leading strand that is exposed by the helicase is immediately used as a template by the leading strand DNA polymerase and is thus stabilized as dsDNA, the ssDNA of the lagging strand template generated behind the helicase is covered (and protected) by gp32 before being copied by the lagging strand polymerase. gp32 binds cooperatively to ssDNA (17) and removes secondary DNA structures that might inhibit DNA synthesis, leading to a stimulation of the polymerase activity of gp43 (18). Both the leading strand DNA polymerase and the gp32 protein serve to hold the DNA of the replication fork "open" by preventing the reannealing of the two ssDNA strands unwound by the helicase and have thus been defined as molecular "reporters" of helicase function (19).
Contrary to leading strand-directed DNA synthesis, which is continuous, DNA synthesis on the lagging strand is discontinuous and requires a primase. The T4 primase (gp61) produces small RNA oligomers that bind to the lagging strand template and are then elongated into Okazaki fragments by the lagging strand polymerase. The monomeric gp61 primase interacts with the hexameric gp41 helicase at the fork and acts as an RNA polymerase to synthesize RNA pentamers of specific sequence (20 -24). Contrary to the Escherichia coli primase, which is recruited from solution each time synthesis of an Okazaki fragment is initiated (25), the gp61 primase remains associated with the T4 helicase and the DNA replication complex throughout the lagging strand synthesis cycle (14). This gp41-gp61 interaction guarantees that RNA primers are synthesized at the fork. Little is known about the regulation of gp61 priming activity, but the end result is that the lagging strand RNA primers are properly placed within the moving replication fork, as required to initiate discontinuous lagging strand DNA synthesis. gp32 protein has also been shown to be essential for confining RNA primer synthesis to sites used to start an Okazaki fragment, thus avoiding the synthesis of excess replication primers (26).
As outlined above, the assembly of the replication proteins at the fork and the movement of the assembled replication complex along the DNA while performing its task of synthesizing DNA requires highly coordinated and concerted "cross-talk" between the individual components of this multiprotein complex. For example, the DNA polymerase that carries the basic 5Ј 3 3Ј synthesis activity is functionally coupled at the fork to almost all of the other T4 DNA replication proteins, including the helicase, the sliding clamp, and the SSB proteins. This functional coupling leads to a gain in processivity and DNA synthesis rate (see, for example, Refs. 13 and 27) and also influences the fidelity of the polymerase (28 -30).
In order to define the molecular bases of the various coupling interactions in which the DNA polymerase participates during replication, we have undertaken an in depth characterization of the complexes formed between the polymerase and the DNA to which it binds during the templated single nucleotide addition (and editing) cycle and during its search for the replication origin. To this end, we have measured the binding of the replication polymerase to several model DNAs, including singlestranded, double-stranded, and primer-template junction-containing DNA molecules. These interactions have been characterized thermodynamically by measuring the dissociation constant (K d ) and the binding stoichiometry of the complex and the binding site size of the polymerase on the DNA and kinetically by measuring the rate constants for the association and dissociation of the complex. We have also analyzed the effects of the addition of dNTPs on the affinity of the T4 polymerase for specific p-t junction constructs and ssDNA molecules. These studies allow us to gain further insight into the specific role of the DNA as an assembly platform for the replication complex and also to further define the steps of the single nucleotide addition cycle that are catalyzed by the DNA polymerase. They have also helped us identify the thermodynamic and kinetic parameters of the basic DNA polymerase-DNA equilibrium that can be regulated by the various polymerasecoupled replication proteins, since these interactions underlie the function of the replication complex. We propose that the kinetic and thermodynamic modulation of the DNA polymerase-DNA equilibrium by auxiliary replication proteins constitutes a basic mechanism for protein coupling within the replication complex.

EXPERIMENTAL PROCEDURES
Reagents and Buffers-All chemicals used in these studies were of analytical reagent grade. Buffers were filtered, degassed, and temperature-equilibrated before use. Dideoxy-GTP was purchased from Invitrogen. dATP and dGTP triphosphate were from Amersham Biosciences. T4 polynucleotide kinase was from New England Biolabs.
Protein Purification and Characterization-We used a T4 polymerase mutant deficient in 3Ј 3 5Ј exonuclease activity (gp43exo Ϫ ) in these studies in order to avoid degradation of the DNA substrates during binding measurements. In this gp43 mutant, the aspartic acid residue at position 219 has been changed to alanine (Asp 219 provides one of the four essential carboxylate residues for the two-metal ion exonuclease reaction). T4 gp43exo Ϫ was purified and stored as described elsewhere (31). Stock solutions of the mutant polymerase were judged to be Ͼ95% pure on the basis of Coomassie Blue-stained SDS-polyacrylamide gels. Polymerase concentrations were determined by UV absorbance at 280 nm, using a molar extinction coefficient of 1.3 ϫ 10 5 M Ϫ1 cm Ϫ1 for monomeric gp43exo Ϫ . All polymerase concentrations are reported in units of protein monomers.
Oligonucleotides and DNA Substrates-All of the oligonucleotides utilized in this study were purchased from Oligos Etc. and were purified on a 15% polyacrylamide sequencing gel before use. The sequences of the DNA oligomers that were used in this study are shown in Table I. The fluorescein derivative used (by Oligos Etc.) to label the oligonucleotides for our fluorescence anisotropy studies was 5Ј-fluorescein phosphoramidite, which was prepared from 6-carboxyfluorescein. The various DNA p-t junction constructs were formed and "annealed" by heating the primer and template strands to 90°C for 2 min and then slow cooling to room temperature over a period of ϳ2 h. Primer and template strands were used in a 1:1 molar ratio, and the annealing buffer contained 50 mM Tris-OAc (pH 7.5), 100 mM KOAc, and 5 mM Mg(OAc) 2 . The p-t constructs were purified by electrophoresis on a 12% acrylamide native gel (30:0.8 ϭ acrylamide/bisacrylamide weight ratio). Elution from the gel was performed for 16 h at 50°C in TBE buffer (89 mM Tris, 89 mM borate, 2 mM EDTA, pH 8.0) containing 7 M urea. Sep-Pack light tC18 cartridges (Waters) were used to desalt and recover the p-t DNA constructs.
The primer was reannealed to the template by heating at 90°C for 2 min and then slow cooling in buffer A (30 mM Hepes (pH 7.6), 160 mM KOAc, 0.05 mM EDTA). A version of the 16 -20 p-t construct that contained a dideoxyguanine (ddG) residue at the 3Ј-end of the 16-mer primer was made by extending the 3Ј-end of the 15-mer of a purified 15-20-mer construct of the same sequence by one residue by incubating a 6 M concentration of the 15-20-mer construct with 500 M dideoxy-GTP and 640 nM gp43exo Ϫ for 30 min at room temperature. Under these conditions, more than 90% of the 15-mer strand was converted into a 16-mer. The DNA was then precipitated and resuspended in buffer A. The purity of the p-t constructs was determined by radiolabeling the primer strand of the final product and comparing its mobility on a native gel to that of the single-stranded primer strand as well as by monitoring (by fluorescence) the mobility of the template strand of the final product in a native gel. Gels were scanned using either the Phosphor Screen or the blue fluorescence/chemifluorescence mode of the Storm 860 PhosphorImager (Amersham Biosciences) and quantitated with the ImageQuant software, version 5.1, from Amersham Biosciences. This assay showed the p-t junction DNA constructs to be Ͼ90% pure.
Polymerase Assay-Assays monitoring DNA synthesis were performed in a buffer containing 30 mM Hepes (pH 7.5), 60 mM KOAc, 6 mM Mg(OAc) 2 , and 5 mM DTT. 400 nM DNA polymerase was preincubated for 3 min at 30°C with 400 nM 25-30 construct, 1 M gp44/62, 350 nM gp45 (trimer), and 6 mM ATP. We note that in order to maintain processivity, these assays included the gp45 processivity clamp and the gp44/62 clamp loader at the concentrations indicated, as well as ATP to drive the clamp loading process (see Refs. 16 and 32 for complete descriptions of the polymerase assay as well as the preparation and properties of the clamp and clamp loader). Single turnover elongation reactions were started by adding an equal volume of a prewarmed mixture containing the four canonical dNTPs (each at a final concentration of 250 M) and 0.02 mg/ml (final concentration) of heparin. The reaction was quenched after 30 s by adding an equal volume of a formamide/dye mixture (90% deionized formamide, 0.25% bromphenol blue, 0.25% xylene cyanol, 20 mM EDTA). The samples were then boiled, chilled, and loaded on a sequencing gel (12% acrylamide, 19:1 acrylamide/bisacrylamide). After electrophoresis, the gel was scanned using the blue fluorescence/chemifluorescence mode of the Storm 860 PhosphorImager.
The DNA elongation reaction in the presence of heparin was run to check the activity of the purified gp43exo Ϫ polymerase. Heparin has been shown to prevent "multiple-hit" extension of DNA primers by binding the polymerase that dissociates from the p-t junction after the reaction has been initiated (32). We show that a heparin concentration of 0.02 mg/ml is sufficient to trap 200 nM DNA polymerase (Fig. 1, lane  1). Furthermore, Ͼ90% of a 200 nM concentration of 25-30-mer p-t construct (see Table I) could be extended under single turnover conditions by an equivalent concentration of gp43exo Ϫ (Fig. 1, lane 2). Thus, our gp43exo Ϫ polymerase preparation is at least 90% active.
Preparation of Samples for Velocity Sedimentation and Surface Plasmon Resonance Experiments-The buffer used in the velocity sedimentation experiments contained 30 mM Hepes (pH 7.5), 100 mM KOAc, 50 M EDTA, and 1 mM DTT, whereas the buffer used in the surface plasmon resonance experiments contained 25 mM Hepes, 160 mM NaCl, 50 M EDTA, and 0.005% surfactant P20. For both sets of experiments, the proteins were dialyzed against 0.5 liters of the above buffer for 2 h at 4°C, using a Spectra/Por 16-well microdialyzer apparatus (Spectrum) containing a 10,000 molecular weight cut-off Spectra/Por CE dialysis membrane. The microdialyzer apparatus was connected to a pump with the flow rate set at 7 ml/min. Samples with volumes of 30 -60 l were loaded into each well. After dialysis, samples were centrifuged at 12,000 rpm for 15 min at 4°C, and the supernatants were saved. The protein concentrations of the supernatant solutions were measured using UV absorbance at 280 nm. From 85 to 100% of the input protein was recovered after dialysis. The recovered proteins were then diluted to the desired final concentrations with the above buffers.
Velocity Sedimentation Experiments-Analytical sedimentation ve-locity runs were performed in the Beckman Optima XL-I Analytical Ultracentrifuge. The sample and reference sectors of scrupulously cleaned 1.2-cm path length double-sector ultracentrifuge cells equipped with either aluminum or Epon aluminum-filled centerpieces and quartz windows were filled with 400-and 420-l volumes of protein (and/or DNA) and buffer, respectively. The p-t junction DNA sample (containing ϳ80% p-t junction and ϳ20% 20-mer ssDNA) used in this study was a 16-20-mer with a dG residue at the 3Ј-end of the primer. The cells were loaded into an An60Ti analytical ultracentrifuge rotor. Centrifugation was started when the preset experimental temperature (20°C) had been reached. Experiments were performed at a rotor speed of 60,000 rpm, and absorbance scans at 280 nm and/or 260 nm were collected at intervals of 2.5-8 min. (The 280-nm scans were used to follow the sedimentation of the protein component of the complexes, whereas the 260-nm scans tracked primarily the sedimentation of the DNA components.) The van Holde-Weischet method (33), the time derivative or g*(s) method (34), and direct fitting of the boundaries using numerical solutions to the Lamm equation (the Sedfit method) (35) were used for data analysis. The software for the van Holde-Weischet method was included in the UltraScan version 5.0 for Linux provided by Borries Demeler and downloaded from the World Wide Web at www.ultrascan.uthscsa.edu. The DCDT ϩ version 1.16 used to compute g*(s) was purchased from John Philo and downloaded from the World Wide Web at www.jphilo. mailway.com/. The Sedfit program, which fits the sedimentation boundaries using numerical solutions of the Lamm equation, was downloaded from the World Wide Web at www.analyticalultracentrifugation.com/. The partial specific volume (͉) for the gp43exo Ϫ polymerase (0.7362 ml/g) as well as the buffer density (1.015 g/ml) and viscosity (0.010234 poise) were calculated using the sedimentation interpretation program, Sednterp (version 1.07), which was downloaded from the World Wide Web at www.bbri.org/RASMB/rasmb.html/. A value of ͉ ϭ 0.5 ml/gm was used for all of the DNA components, and ͉ values of 0.722 ml/g and 0.715 ml/g were for the polymerase-12-mer DNA and the polymerase-16 -20-mer p-t construct, respectively. These latter values were calculated for the nucleoprotein complexes by "molecular weight averaging." Only runs analyzed using the Sedfit software are shown in the figures. The sedimentation coefficient of gp43 DNA polymerase measured by sedimentation velocity was compared with that of the highly related RB69 DNA polymerase calculated from the atomic level structure (36) using the HYDROPRO public domain computer program. HYDROPRO was downloaded from the World Wide Web at leonardo.fcu.um.es/ macromol.
Fluorescence Anisotropy Experiments-Fluorescence anisotropy data were acquired using a SLM model 8000 spectrofluorimeter (SLM Instruments, Urbana, IL). Samples were excited at a wavelength of 492 nm, and emission was monitored at 516 nm. All experiments were performed with a 440-l cuvette and at a constant temperature of 20°C in 30 mM Hepes (pH 7.5), 160 mM KOAc, 50 M EDTA, 1 mM DTT buffer. When present, the final concentration of added dNTP was 250 M.
Surface Plasmon Resonance Experiments-SPR data were collected on a Biacore X Instrument from Biosensor Amersham Biosciences. All experiments were conducted at 16°C. The p-t junction sample used in this study was the 16 -20-mer p-t DNA construct carrying a dG residue at the 3Ј-end of the primer (see Table I), and the 20-mer DNA strand was biotinylated at its 5Ј-end. The Sensor SA chip was prepared as recommended by the manufacturer, and ϳ12 response units (RUs) of the 16 -20-mer p-t construct were immobilized on the chip. To measure the association rate of the polymerase to the 16 -20-mer p-t construct, the diluted protein was injected for a period of 1 min. The dissociation reaction was monitored for 20 s. The flow rate used during both the association and the dissociation phases of the reaction was 50 l/min. The regeneration buffer contained 25 mM Hepes (pH 7.5), 1 M NaCl, 50 M EDTA, and 0.005% surfactant P20.

Binding of T4 DNA Polymerase to Nonspecific (Single-and
Double-stranded) DNA-The binding of the DNA polymerase to nonspecific DNA was measured using the steady-state fluorescence anisotropy assay. Two DNAs were used as binding ligands in these studies. The 12-mer DNA (see Table I for sequences) served as a model ssDNA, and the 16 -16-mer DNA duplex served as dsDNA. These DNA constructs were used at molar concentrations of 50 -100 nM, and changes in fluorescence anisotropy were recorded as increasing amounts of DNA polymerase were titrated into the fluorescence cell. The activity of our gp43 DNA polymerase preparation was tested using the single-turnover assay developed by Reddy et al. (32). Lane 1, the polymerase was preincubated with the 25-30-mer p-t junction DNA construct and heparin prior to starting the polymerization reaction by adding the four dNTPs. Lane 2, the polymerase was incubated with the DNA only, and heparin was added together with the four dNTPs to initiate polymerization (see "Experimental Procedures").
Complex formation between the protein and the DNA components was manifested as an increase in observed anisotropy as the titration proceeded. The polymerase should bind nonsite-specifically to these DNAs, meaning that "overlap binding" to these DNAs was expected. Therefore, a modified version (37) of the noncooperative McGhee-von Hippel (38) binding isotherm for nonspecific complexes was used to fit the data. The equation derived by Tsodikov et al. (37) is an extension of the McGhee-von Hippel equation that deals with the binding of proteins to finite oligonucleotide lattices. Fitting the data with the appropriate equation yielded apparent values of the dissociation constant (K d ) and binding site size (n). The curvature of the Scatchard plots obtained (Fig. 2, a and b) confirms that overlap binding does indeed occur when the polymerase binds to these DNA molecules. The results of these measurements are summarized in Table II and indicate that the binding of the polymerase to such nonspecific DNA lattices is only moderately strong (K d ϭ 100 -240 nM), with a calculated site size (n) of 10 nt for ssDNA binding and 13 bp for dsDNA binding.
The fact that the Scatchard plots for these nonspecific binding interactions (Fig. 2, a and b) are curved and that the curvature depends on the relationship between the site size and the lattice length provides an additional way to estimate values of n from these data. Thus, in Fig. 2a we show the best overlap binding fit to the data (solid line; n ϭ 10 nt and K d ϭ 200 nM). The dashed lines show the fits that would be expected if n were 9 or 11 nt for this reaction system, with the same value of K d . Clearly, the latter values do not fit the data, which strengthens our confidence in the value of n ϭ 10 nt for polymerase binding to ssDNA that was obtained by fitting the binding equation to the data.
Similarly, we also tested the best fit (solid line; n ϭ 13 bp and K d ϭ 95 nM) obtained from the Scatchard plot for polymerase binding to dsDNA. Here the dashed lines correspond to the fits expected for n ϭ 10 or 14 bp for the binding to dsDNA, again with the same value of K d . Clearly, this result also supports the value of n determined by direct fitting of the titration data and confirms that the size for polymerase binding to dsDNA is somewhat larger than that obtained with ssDNA (comparing bp with nt). Of course, this result is also intuitively reasonable for nonspecific binding to these two DNA forms, since on a nucleotide residue versus base pair basis, an ssDNA oligomer is more extended in space than a dsDNA duplex of the same length.
Specific Binding of T4 DNA Polymerase at p-t Junctions-Steady-state anisotropy measurements were also used to analyze the binding of the DNA polymerase to a DNA construct carrying a p-t junction as a specific DNA binding site. For this purpose, we used the 16 -20-mer p-t DNA construct (see Table   I). As Fig. 3 shows, a p-t junction does indeed constitute a specific binding site for DNA polymerase, resulting in a linear Scatchard plot as expected for single site (nonoverlap) 1:1 binding. The binding constant for the polymerase-p-t junction is also significantly (10 -20-fold) tighter than that measured for nonspecific DNAs binding in the same buffer. A best fit value of K d ϭ 14 nM was obtained for this specific complex.
Specific Polymerase Binding to a p-t Junction Is Regulated by dNTP Binding-To determine whether dNTP binding affects the strength of the interaction between the DNA polymerase and the specific p-t junction site, we constructed a 16 -20 hybrid with a dideoxynucleotide (ddG) residue at the 3Ј-end of the 16-mer. (The ddGMP residue at the 3Ј-end of the primer is, of course, complementary to its template partner residue, but the absence of the hydroxyl group on position C-3 of the deoxyribose moiety prevents the elongation of the primer during the binding measurement.) We measured the affinity of the DNA polymerase for this construct in the presence of the correct (complementary to the template) incoming dNTP (dATP) and also in the presence of an incorrect (noncomplementary) incoming dNTP (dGTP).
The results are summarized in Table II and show that the presence of the complementary dNTP (here dATP) does indeed strengthen (ϳ2-fold) the apparent affinity of the DNA polymerase for its p-t junction binding target. In contrast, the addition of a noncomplementary dNTP (here dGTP) weakens (ϳ3-fold) the interaction between the DNA polymerase and the same specific p-t target. Because the binding site on the polymerase for the next required dNTP may not be fully saturated at these dNTP concentrations (see "Discussion"), the differential standard free energy values (⌬⌬G 0 ) of ϳ0.5 kcal/mol that can be calculated from the ⌬G 0 values in Table II may represent an underestimate of the tightening effect on polymerase binding that occurs with complementary dNTPs as well as an underestimate of the weakening effect induced by noncomplementary dNTPs, that would be observed with these dNTPs at saturating concentrations.
As a control on these measurements we used the fluorescence anisotropy technique to measure the effect of both dATP and dGTP on the nonspecific affinity of DNA polymerase for the 12-mer ssDNA construct, which was intended to serve as a model of non-p-t junction construct. The result is that both of these dNTPs weaken the interaction of polymerase for the ssDNA model ϳ2-fold, with no change in the shape of the Scatchard plot (similar to Fig. 2a) and thus also no change in the apparent DNA binding site size (Table II). Since the ssDNA binding partner used included both dT and dC residues, which could (in principle) serve as complementary base pair interaction sites for the tested dNTPs, we conclude that the weakening a X corresponds to a 6-carboxyfluorescein terminal adduct.
effect of the added dNTPs in this case must reflect the fact that dNTPs bind to the active site of the polymerase and thus weaken the binding to the ssDNA (presumably by repulsive charge-charge interactions). Furthermore, since the binding is still nonspecific (as manifested by the curved Scatchard plot; data not shown), this suggests that no (presumably attractive) base pairing can occur in the nonspecific binding mode and that proper base pairing requires the presence of a primer strand to position appropriately the added dNTP in the DNA polymerase active site. DNA Polymerase Binds 5 nt of ssDNA at the p-t Junction-To determine how many residues of the template ssDNA interact at the p-t junction with T4 polymerase in the specific binding mode, we assembled a series of p-t junction constructs (Table I) that differ in the length of their single-stranded portions from 1 nt (for the 15-16-mer construct) to 8 nt (for the 15-23-mer construct). The apparent dissociation constant for polymerase binding to these constructs was measured by steady-state fluorescence anisotropy. The results (Fig. 4) show that the binding interaction between the polymerase and its specific substrate extends to the fifth nucleotide residue downstream of the p-t junction on the template strand, since the apparent K d for the interaction decreases with increasing length of the ssDNA portion of the p-t construct up to 5 nt and then plateaus at a K d value of ϳ5 nM. Combining this result with the site size (n) of 10 nt determined for the nonspecific binding of polymerase to ssDNA and the n of 13 bp measured for nonspecific binding to dsDNA (see above), suggests that the DNA polymerase probably also binds to (interact with) ϳ6 -7 bp of dsDNA upstream of the p-t junction in forming the specific polymerase-p-t junction DNA complex.
DNA Polymerase Binds to p-t Constructs and to 12-Mer ssDNA in a 1:1 Molar Ratio-Sedimentation velocity experiments were performed to determine the binding stoichiometry of DNA polymerase to a representative specific binding site on a p-t DNA construct (the 16 -20-mer was used), and also to a representative nonspecific DNA lattice (the 12-mer ssDNA molecule). Sedimentation coefficients (s 20,w ) were determined for the DNA polymerase alone, for the two DNA ligands alone and for 1:1 mixtures of DNA polymerase with each DNA ligand. Samples containing DNA only were scanned at 260 nm, whereas those containing DNA polymerase only were scanned at 280 nm. Samples containing both components were scanned at both wavelengths. The continuous c(s) distribution model was utilized with the Sedfit program to analyze the data obtained.
A typical set of scans and residuals are shown for the binding of gp43 to 12-mer ssDNA in Fig. 5a. The fits to the Lamm equation were excellent in all of our experiments, as judged by the randomness of the residuals (which ranged from Ϫ0.02 to ϩ0.02 OD) and by the value of the root mean square deviations of the residuals (which were always Ͻ0.0056 OD units). The sedimentation coefficient distribution obtained for the gp43-12-mer ssDNA complex is shown in the bottom panel of Fig. 5a and demonstrates that this complex sediments as a single homogeneous peak with s 20,w ϭ 5.91 S. The quality of all the other sedimentation experiments was comparable, but only the sedimentation coefficient distribution profiles are shown for the other runs. The distributions for the 12-mer ssDNA, the polymerase alone, and the nonspecific complex of polymerase with 12-mer ssDNA, are shown in Fig. 5b, and the distributions for the 16 -20-mer p-t construct and the specific complex of polymerase with the 16 -20-mer p-t DNA construct are shown in Fig. 5c.
The DNA polymerase, the 16 -20-mer p-t construct, and the 12-mer DNA yielded s 20,w values of 5.55, 2.28, and 1.26 S, respectively (see Fig. 5, a-c). The sedimentation coefficient of the T4 DNA polymerase reported here is in good agreement with an s 20,w of 5.59 S calculated from the atomic structure of the RB69 DNA polymerase (36), using the bead-modeling methodology developed by the José Garcia De La Torre's group (39). We attribute the species with s 20,w ϭ 2.01 S in the 16 -20-mer DNA sample to excess 20-mer present in this DNA preparation. The 2.01 S peak represents 15% of the total DNA, in good agreement with the sample purity deduced by radiolabeling the DNA.
Hydrodynamic The sample that contained polymerase and 16 -20-mer DNA construct in equimolar concentrations showed a single bound-ary. Using Sedfit, we were able to resolve two independent nucleoprotein complex species from the data collected at both 260 and 280 nm (Fig. 5c). We propose that the peak at s 20,w ϭ 6.89 S corresponds to DNA polymerase complexed with the 16 -20-mer p-t junction DNA construct, whereas the peak with s 20,w ϭ 6.13 S corresponds to polymerase complexed with the excess 20-mer ssDNA strand. The value of f/f o ϭ 1.16 obtained for the gp43-16-20-mer complex suggests that the complex may be more spherical than the two constituent reaction components.
The sample containing an equimolar concentrations of DNA polymerase and 12-mer ssDNA showed a single boundary as well and was characterized by an s 20,w ϭ 5.91 S when the sedimenting complex was monitored at either 260 or 280 nm (Fig. 5, a and b). The f/f o ratio for this nonspecific complex was 1.33, which is significantly greater than the f/f o measured for the specific complex between gp43 and the 16 -20-mer p-t junction DNA. This result suggests that the nonspecific complex formed between polymerase and 12-mer ssDNA and the specific complex formed between polymerase and the specific 16 -20-mer p-t DNA construct have different shapes, and we propose that they represent two different binding modes of the T4 polymerase to DNA (see "Discussion"). We note that the values of s 20,w measured for both complexes are consistent with a 1:1 molar ratio of polymerase to DNA for these DNA samples, as expected from the polymerase binding site sizes measured above. We note also that there is no evidence of gp43 polymerase dimerization, either when the polymerase is sedimenting alone (see also Ref. 13) or when it sediments as a specific or a nonspecific complex with DNA. This point is considered further under "Discussion." The Rate of Association of DNA Polymerase to the p-t Junction Is Close to Diffusion-limited-We have also used surface plasmon resonance (SPR) methods to examine the equilibrium and the kinetics of the binding interaction between the DNA polymerase and a specific p-t junction DNA target site. A value of K d and apparent association (k on ) and dissociation (k off ) rate constants were measured by SPR using a 16 -20-mer p-t junction construct carrying a biotin adduct on the 5Ј-end of the 20-mer strand. The p-t construct was immobilized onto a streptavidin sensor chip through the biotin adduct. DNA polymerase was injected into the apparatus at different concentrations to monitor the level of binding of the protein to the immobilized DNA construct as well as to determine the apparent association and dissociation rate constants for the reaction.
The relative equilibrium plateau values for the binding of polymerase to the tethered p-t junction DNA constructs are shown in Fig. 6a. In Fig. 6b, we plot these relative heights as a function of injected protein concentration and use the resulting titration curve to determine a dissociation binding constant value of K d ϭ ϳ10 nM for the polymerase-p-t junction DNA construct binding interaction. We note that this SPR measurement of K d depends only on the relative levels of the apparent  plateaus in the sensorgrams and does not require knowledge of the constituent rate constants. This binding constant is in good accord with the K d values measured by steady-state fluorescence anisotropy (e.g. see Fig. 3 and Table II). Fig. 6a also shows that the rate of binding of the polymerase to its specific p-t junction DNA target site is very fast, as indicated by the short time required for the apparent binding to reach the level of the equilibrium plateau. Dissociation of the DNA polymerase from the DNA, as measured by the time required to reach base line from the equilibrium plateau after stopping polymerase injection, is rapid as well (Fig. 6a).
The observed association and dissociation processes were too fast to be measured reliably with our SPR instrument. We therefore used simulation techniques, together with the measured SPR value of K d , to estimate lower limits for both k on and k off . The results indicate that k off is Ͼ2 s Ϫ1 , whereas k on for the binding of DNA polymerase to its specific p-t substrate is Ͼ2 ϫ 10 8 M Ϫ1 s Ϫ1 , which is close to the calculated diffusion limit for reactions of this type. The method used to perform these simulations is described under "Appendix."

DISCUSSION
In this study, we have used a combination of biophysical tools (velocity sedimentation, steady-state fluorescence anisotropy, and SPR methodologies) to characterize the interactions of the replication polymerase (gp43) of the bacteriophage T4 DNA replication system with a variety of DNA molecules. The DNA components were chosen to mimic the DNA states that serve as cofactors for the polymerase, not only during the various steps of the single nucleotide addition cycle, but also during replication complex assembly.
Interaction Properties of T4 DNA Polymerase Bound Specifically and Nonspecifically to DNA-The critical facts that have emerged from our studies of the interactions of gp43 with model DNA constructs are the following. (i) A single T4 DNA polymerase binds site-specifically to p-t DNA junction constructs with high affinity (K d ϳ5 nM) at physiological salt concentration. (ii) Polymerase bound specifically at a p-t junction binding site interacts with 5 nt of ssDNA upstream of the 3Ј-end of the primer strand and with 6 -7 bp of dsDNA down-stream of the p-t junction. (iii) The specific binding affinity for polymerase at the p-t junction increases in the presence of the next required (complementary to the template) dNTP substrate and decreases in the presence of a noncomplementary dNTP. (iv) The T4 polymerase also binds in a nonspecific "overlap" mode to ssDNA and dsDNA. The K d for this form of binding is 20 -40-fold weaker than is site-specific binding to a p-t junction site at the same salt concentration, and this affinity is further weakened by binding dNTPs to the active site of the DNA polymerase. (v) The binding site size (n) of polymerase in this nonspecific binding mode is 10 nt for ssDNA and 13 bp for dsDNA. (vi) The specific (p-t junction DNA bound) and nonspecific (ssDNA-or dsDNA-bound) states of T4 polymerase have different conformations, as manifested in part by the different frictional ratios observed for the two types of complexes in sedimentation velocity experiments, and we have no evidence for another binding mode of the DNA polymerase to DNA. (vii) The rate of binding of polymerase to a specific p-t junction is very fast (effectively diffusion-controlled). (viii) T4 DNA polymerase shows no tendency to form oligomers, either when free in solution or when specifically or nonspecifically bound to DNA.
As will be discussed below, these facts all find interpretation in the various steps of the single nucleotide addition cycle catalyzed by DNA polymerase and also in the sequential process of DNA polymerase assembly at the fork, and they lead us to propose a general protein-protein coupling mechanism for polymerase within the full DNA replication complex.

Regulation of DNA Synthesis by Control of the Relative Rates and Equilibria of the Individual Steps of the Polymerase-catalyzed Single Nucleotide Addition Cycle-
The functions of DNAdependent DNA polymerases, like those of their DNA-dependent RNA polymerase analogues (40,41) can be regulated either through control of the relative rates of the single nucleotide addition/excision reaction on the leading and lagging strand sides of the replication fork or by control of the stability of the leading and/or lagging strand polymerases within the replication complex. In this view, the polymerase-catalyzed DNA synthesis pathway can be considered to be in kinetic competition with other potential reaction pathways. These may involve competition of reaction pathways within the replication complex (e.g. misincorporation and editing (basic to fidelity control) or polymerase dissociation (basic to processivity control)). Alternatively, the competition may be between pathways involved in the function of different DNA manipulation complexes, such as DNA recombination, DNA repair, and RNA transcription. Fig. 7a presents a schematic summary of some of the important steps that are thought to occur within the single nucleotide addition (and excision) cycle catalyzed by a replication polymerase. Some of these ideas have developed from thermodynamic insights (e.g. see Refs. 41 and 42), some from structural considerations (e.g. see Refs. [43][44][45][46][47], and some from detailed kinetic studies (e.g. see Refs. 48 -51).

The Equilibrium between State 2 and State 4 Involves a Specific and Tight Binding Interaction of the Polymerase with the p-t Junction-
The polymerase-DNA interaction represented by state 2 in Fig. 7a can be modeled thermodynamically, using our measurements of K d ϭ ϳ8 nM for the specific site binding of gp43 to the 16 -20-mer (ddG) p-t junction DNA construct in the absence of dNTP (see Table II). In this state, the 3Ј terminus of the elongating primer strand is located within what might be called (using nomenclature developed in the context of RNA polymerase mechanisms; see Ref. 41) the product binding subsite of the synthesis active site of the polymerase. In this state, the product subsite is appropriately aligned to interact with an incoming (template-matched) dNTP that can bind in the so-called substrate binding subsite of the polymerase. This specific polymerase-DNA interaction is further tightened (to a K d of Ͻ4 nM; see Table II) when the polymerase moves to state 4, in which the correct dNTP is bound in the substrate binding subsite of the synthesis active site of the polymerase. One possible explanation of this tighter specific binding of the DNA polymerase to a p-t junction in the presence of the correct dNTP might be that such dNTP binding to the substrate subsite (and the complementary base on the template) increases the affinity of the polymerase for the p-t junction DNA by the addition of a component of favorable free energy contributed by the base-pairing of the (correct) dNTP in the substrate binding of the polymerase with its complementary template base. Furthermore, since the nonspecific binding of DNA polymerase on the ssDNA 12-mer is weakened by the addition of either dATP or dGTP (see Table II), our data show that the primer strand itself on the p-t junction is important for tightening this specific interaction of the DNA polymerase with the p-t junction DNA in the presence of the correct dNTP.

State 1 Represents a Nonspecific and Loosely Bound Interaction of the Polymerase with the p-t Junction DNA That Allows
Easy One-dimensional Sliding-When the correct dNTP is present in the substrate binding subsite of the polymerase, phosphodiester bond formation will occur in state 6 (the catalytically competent complex), resulting in the formation of state 1, in which the 3Ј-end of the DNA primer strand is located in the substrate binding subsite of the polymerase. State 1 (perhaps concomitantly with the release of PP i ) then undergoes a conformational change, resulting in the release of the 3Ј-end of the primer strand from the substrate binding subsite of the polymerase to form a nonspecifically (and electrostatically) bound "sliding state" of the polymerase on the DNA. We propose that this nonspecifically bound state of the DNA polymerase at the p-t DNA junction is thermodynamically similar to the binding state of the DNA polymerase on nonspecific DNAs (single-stranded or double-stranded), which (at physiological salt concentrations; see Ref. 42) represents a significantly less stable polymerase-DNA complex (K d ϭ 100 -240 nM; see top entries in Table II; see "Results").
Eventually, random one-dimensional sliding of the polymerase on the DNA in this nonspecific (electrostatically bound) state results in the translocation of the free 3Ј-end of the extended primer strand (at an effectively diffusion-controlled rate; see kinetic results derived from our SPR measurements) back to the product binding subsite of the polymerase, which then binds again in the specific p-t binding mode (state 2). In this state, the complex is ready to bind the next correct dNTP and again move into state 4, thus launching another cycle of DNA chain extension. Such a "loose" binding mode of the DNA polymerase at the p-t junction might be important, not only for the polymerase translocation from state 1 to state 2, but also in at least two other situations.
One such situation might arise when the DNA polymerase encounters a lesion in the template strand that blocks further synthesis. In this case, the polymerase may either slide back or transiently move off the template without complete dissociation (due to the flexible tether between the DNA polymerase and its processivity factor; see Refs. 52 and 53) to permit access to the family of Y DNA polymerases that specialize in translesion DNA synthesis. Another situation in which such a sliding mode of the polymerase might play an important role could occur when the lagging-strand DNA polymerase reaches the the rate of formation of the catalytically competent complex (reaction iii), or by increasing the rate of catalysis itself (reaction iv).

FIG. 7.
Reaction steps in the single nucleotide synthesis (and excision) cycle of wild type T4 DNA polymerase. The exonuclease active site and the substrate and product binding subsites within the active site for synthesis of the DNA polymerase are depicted as a triangle, a square, and a circle, respectively (see key). a, six different states of the complex between DNA polymerase and p-t junction DNA that are shown differ in the location of the 3Ј-OH primer terminal residue (filled circle) and the presence or absence of a dNTP (designated by a star) in the substrate binding subsite of the DNA polymerase. In state 1, the 3Ј-OH primer terminus is located in the substrate binding subsite of the synthesis active site of the polymerase. The polymerase then slides backward along the DNA to position the 3Ј-OH primer terminus in the product binding subsite of the polymerase active site (state 2). If the terminal base pair is not complementary to the template, the 3Ј-OH primer terminus may detach from the synthesis active site of the polymerase and bind to the exonuclease active site (state 3). When in either state 2 or state 3, polymerase with the 3Ј-terminal end of the primer strand bound to the product binding subsite can bind a dNTP to the substrate binding subsite, leading to state 4 or state 5, respectively. State 6 corresponds to the "closed" and catalytically competent form of the DNA polymerase bound to a p-t junction with the correct incoming dNTP. This state has been described crystallographically. The single nucleotide incorporation (or excision) cycle ends with phosphodiester bond formation and PP i release (not shown), which converts state 6 back to state 1 and allows the cycle to begin again. Not indicated in the figure is the possibility that polymerase at each state can dissociate (locally) from the p-t junction DNA, although remaining tethered within the replication complex (see "Discussion"). b, equilibria between states 2 and 4 (as well as the probability of dissociation) are shifted by the binding of the correct (or the incorrect) dNTP (see "Discussion"). We have indicated in parentheses the measured K d values for the various equilibria. c, shifts within inter-state equilibria that could increase the fidelity of DNA synthesis. State 4 of the polymerase-DNA complex is bound to an incorrect dNTP. Replication proteins coupled to the polymerase that increase the fidelity of DNA synthesis may work by increasing the rate of dNTP release (reaction i), by increasing the rate of polymerase dissociation (reaction ii), by slowing the rate of formation of the catalytically competent complex (reaction iii), or by slowing the rate of catalysis itself (reaction iv). d, shifts within inter-state equilibria that could decrease the fidelity of DNA synthesis. State 4 of the polymerase-DNA complex is bound again to an incorrect dNTP. Replication proteins coupled to the polymerase that decrease the fidelity of DNA synthesis may work by slowing the rate of dNTP release (reaction i), by slowing the rate of polymerase dissociation (reaction ii), by increasing end of an Okazaki fragment (basically a dsDNA sequence with a nick) and must detach from the DNA to reinitiate synthesis at a new primed site on the lagging strand within the replication fork.
Our Data Confirm That the Binding of the Correct Incoming dNTP to the Polymerase (and to the Template) Is Not, by Itself, Sufficient to Achieve Full Selectivity and Fidelity-Using the equation linking standard free energy change and dissociation constant, we have calculated the standard free energy change for the reaction involving the DNA polymerase, its specific substrate, and the correct or incorrect incoming dNTP (Table  II). It is important to realize that one weakness of the above calculation is that we have assumed that the added dNTP (correct or incorrect) is indeed bound by the polymerase-p-t DNA complex. The K d values for the equilibrium between the gp43-p-t complex and dNTP are not known. However, Turner et al. (54) recently reported that 20 M of the correct incoming dNTP (dATP) was sufficient to drive the formation of a ternary (Klenow fragment-p-t-dATP) complex. Furthermore, the weakening effect of dNTP addition on the interaction between the gp43 polymerase and our ssDNA 12-mer suggests that dNTPs do bind in the active site of the DNA polymerase (although full dNTP binding may not be reached under our experimental conditions). Gillin and Nossal (55) and Topal et al. (56) reported that the apparent Michaelis constant (K m ) of the T4 DNA polymerase for a complementary dATP is lower than that for a noncomplementary dGTP (17-38 M and 190 -605 M for dATP and dGTP, respectively).
The K m values for the noncomplementary incoming dNTP suggest that, in the steady state and under our experimental conditions (250 M dGTP), between 30 and 57% of the polymerase-p-t DNA complex should be complexed with dGTP. It is interesting to note that the difference between the standard free energy change for the reaction involving the DNA polymerase, its specific substrate, and the correct or incorrect incoming dNTP (ϳ1 kcal/mol) corresponds rather well to the difference in standard free energy between a correct A-T base pair and a mismatched T-G base pair (ϳ2 kcal/mol), for such base pairs located within duplex oligonucleotides free in solution (57). However, this 1-2 kcal/mol difference is less than what is required to account for the selectivity and fidelity of the DNA polymerase enzyme.
Consequently, if our assumption that the DNA polymerasep-t junction DNA complex binds dNTP under our experimental conditions is correct, our data confirm that the binding of the incoming correct nucleotide by the DNA polymerase is itself not sufficient to achieve full dNTP binding selectivity and synthesis fidelity. Checking for proper base pairing by the polymerase takes place after dNTP binding and may include a preferential exclusion of water from the newly formed (correct) base pair, since it has been proposed that the free energy difference between a matched and a mismatched base pair can be amplified by water exclusion (58) as well as checking for the proper active site configuration (e.g. absence of steric clashes, proper stacking with the base at the 3Ј-end of the primer, etc.). These additional checks, although condensed within a single reaction step (state 4 3 state 6 in Fig. 7a), are certainly complex and could be driven by a series of protein conformational changes that lead to the closed form and catalytically competent state of the DNA polymerase. Mutagenesis studies indicate that amino acid residues located both within and outside of the polymerase active site can participate in these checking events (see Refs. 59 and 60).
The Weakening Effect of Binding the Incorrect dNTP by the DNA Polymerase-p-t Junction Nucleoprotein Complex-Our experiments show that the presence of the noncomplementary incoming dNTP weakens the interaction between the DNA polymerase and its specific substrate (K d Ͼ 24 nM). Similar results have been reported for the Klenow fragment of polymerase I (61)(62)(63). It is interesting to note that results from Thompson et al. (64), who performed exonuclease site binding titrations using a p-t DNA with four consecutive mismatched terminal base pairs, suggest that a binary polymerase-p-t junction complex with the 3Ј-end of the primer strand in the exonuclease site of the polymerase (state 3 of Fig. 7) is less stable than a binary complex with the 3Ј-end of the primer strand in the product binding site of the enzyme (state 2). (They estimated that the dissociation constant of a binary nucleoprotein complex in which the 3Ј-end of the primer is bound to the 3Ј 3 5Ј exonuclease site of the Klenow fragment was 33 nM, as opposed to a 0.5-0.03 nM dissociation constant when the DNA is bound in the polymerization active site of the enzyme (62,64).) Although we do not know the position of the 3Ј-end of primer strand within the enzyme (i.e. whether it is in the product binding subsite of the synthesis active site or in the exonuclease active site) when the incorrect dNTP is in the substrate binding site of the enzyme, it is possible that the weakening effect of the binding the incorrect dNTP may be due, at least in part, to the displacement of the primer strand from the product binding subsite toward the exonuclease active site. In either case, the weaker interaction between the DNA polymerase and the p-t DNA when the substrate binding subsite is filled with an incorrect dNTP may be viewed as a "thermodynamic signal" for the DNA polymerase, favoring either the dissociation of the polymerase from the template or (more likely) the release of the incorrect dNTP, resulting in the latter case in reversion of the system to state 2. These alternative possibilities and their consequences are presented schematically in Fig. 7b. Auxiliary replication proteins that affect fidelity, such as the processivity factor or the SSB protein, may also modulate the relative rates at which these competing reactions occur.

Auxiliary Replication Factors as Modulators of the Equilibria within the Single Nucleotide Incorporation Cycle: A Possible
Protein-Protein Coupling Mechanism?-Auxiliary replication factors, such as the processivity clamp (28 -30, 65) and the SSB protein (66), have been reported to influence the fidelity of the DNA polymerase, resulting in either a decrease or an increase in replication fidelity depending on the sequence context and the resulting mutation type (e.g. a substitution or a frameshift). As a possible molecular basis for this effect, we propose that DNA polymerase-coupled proteins may modify the fidelity of synthesis by modulating the kinetics and/or the thermodynamics of the various equilibria involved in the single nucleotide incorporation cycle. Fig. 7, c and d, shows examples of relative reaction rate changes (heavy versus light arrows) within the single nucleotide incorporation cycle that could lead to an increase or a decrease in fidelity. More specifically and starting with a DNA polymerase-p-t complex filled with an incorrect dNTP, a DNA polymerase-coupled protein can increase the DNA polymerase fidelity by accelerating dNTP release (state 4 3 state 2) or polymerase dissociation from the DNA. (We note that this dissociation may leave the polymerase bound, perhaps by the processivity clamp, within the replication complex and thus be potentially reversible.) Similarly, a regulatory protein coupled to the polymerase can decrease fidelity by stabilizing the ternary complex complexed with an incorrect dNTP (state 4 in Fig.  7d) or by increasing the rate at which base pair checking (state Higher Level Assembly of the Functional T4 DNA Replication Complex-Binding information of the sort we have gathered here is also needed to attain a molecular (thermodynamic and kinetic) understanding of the additional steps that are involved in the incorporation of gp43 into the functional multicomponent T4 DNA replication complex, since these assembly steps (from a thermodynamic and kinetic perspective) also represent potential targets for regulation of the replication complex. Our goal has thus been not only to gather thermodynamic and kinetic information about the interactions of polymerase with DNA in the various steps of the synthesis cycle but also to show how such in depth thermodynamic and kinetic characterization of these interactions can provide fundamental insight into the molecular origins of the assembly pathways for the entire DNA replication complex.
The Search for a p-t Junction-We report here that the binding of the DNA polymerase to nonspecific ssDNA or dsDNA is only moderately strong (240 and 100 nM for singlestranded and double-stranded DNA, respectively) (Table II and Fig. 2) but is probably still significant in the cell, given the concentrations of these nonspecific substrates relative to the concentration of the specific DNA binding target (the p-t junction). These nonspecific DNAs could comprise "landing sites" for the DNA polymerase, suggesting that a one-dimensional sliding process (67) can be used in the final stages of p-t junction target location by gp43. While bound to a nonspecific site, the T4 DNA polymerase adopts what we have previously called a nonspecific (mostly electrostatic and salt-sensitive) binding mode (42) and covers 10 nt or 13 bp on ssDNA or dsDNA, respectively. This picture is in good agreement with previous biochemical (e.g. see Refs. 68 and 69) or structural (see Refs. 70 and 71) studies.
The "Footprint" of the T4 Polymerase at a p-t Junction-Our experiments performed with p-t junctions carrying singlestranded portions of varying length indicate that the DNA polymerase interacts mostly with the first 5 nt of the (singlestranded) template downstream of the primer terminus. The path taken by the template strand may lie between the aminoterminal domain and the exonuclease domain of the DNA polymerase, as suggested by the crystal structure of the RB69 DNA polymerase with p-t DNA and dTTP (72). Interestingly, the Klenow fragment contacts at least the first 4 nt in the singlestranded region beyond the primer terminus, with a fingers subdomain (from Tyr 766 to Phe 771 ) being the major interaction surface (54). We therefore propose that the polymerase binds at a p-t junction by interacting with the last 6 -7 bp of the dsDNA upstream of the 3Ј-primer terminus and with the first 5 nt of the ssDNA template downstream of the terminus. This result is in a good agreement with findings from footprinting (73), crosslinking (74), and structural (52,72) studies, thus validating our use of our steady-state fluorescence anisotropy measurements to map the detailed dimensions of the specific p-t binding site of the polymerase.
How Might Two gp43 Polymerases Interact within the Replication Fork?-The sedimentation velocity experiments we report here strongly suggest that under our experimental conditions, with a DNA binding partner carrying a single specific polymerase binding site, the polymerase-DNA construct complex exists in a 1:1 molar ratio. Ishmael et al. (75) have recently reported being able to cross-link two monomers of T4 DNA polymerase in the presence of DNA. It is important to realize that in this cross-linking study, the DNA construct used carries multiple nonspecific binding sites for the DNA polymerase. As a consequence, this nonspecific binding by itself could have brought two DNA polymerases into close proximity, with the proteins eventually forming cross-links not because of high interprotein affinity or DNA-induced conformational changes but instead because of the high local protein concentration enforced by such adjacent binding sites. It would therefore have been interesting to test the specificity of this polymerasepolymerase interaction.
Our data clearly demonstrate that in the absence of a second polymerase binding site on the same DNA, a polymerase bound to a DNA substrate cannot recruit another polymerase. Two DNA polymerases are required at the replication fork. It has been shown that the DNA polymerase on the lagging strand is not recruited from solution at each Okazaki fragment but instead remains bound to the entire replication machinery (14). This suggests that the functional T4 DNA replication polymerase dimer that operates within the full replication complex when leading and lagging strand DNAs are both being synthesized must be held together (coupled) by additional subunits of the full replication complex (such as the sliding clamp, the helicase (13), or the helicase loader (75), if this last protein complex is still present in the elongating replication complex) rather than by a significant intrinsic affinity between the polymerase monomers.
A Ͼ1800-Fold Decrease in the Dissociation Rate Constant (k off ) of the DNA Polymerase from the p-t Junction Results in High Polymerase Processivity-Our SPR measurements indicate that the binding of the DNA polymerase to a specific p-t DNA junction proceeds at a rate that is close to the diffusioncontrolled limit. Therefore, the tight binding of the DNA polymerase at a p-t junction in the presence of the processivity factor is likely to be achieved largely by decreasing the rate at which the DNA polymerase dissociates from the DNA. Assuming that the DNA polymerase still binds at a rate of 2 ϫ 10 8 M Ϫ1 s Ϫ1 in the presence of the processivity factor and knowing that the affinity of the DNA polymerase-sliding clamp couple for a p-t junction is strong (K d ϭ 30 pM) (76), we can estimate that the dissociation rate constant of the DNA polymerase-sliding clamp couple is ϳ0.006 s Ϫ1 (i.e. (30 ϫ 10 Ϫ12 M) ϫ (2 ϫ 10 8 M Ϫ1 s Ϫ1 ), where the first number is the estimated dissociation constant of the equilibrium between the DNA polymerase-sliding clamp couple and a p-t junction, and the second number represents the lower limit of the association rate constant that we have estimated).
As a consequence, the coupling between the DNA polymerase and its processivity factor depends on a physical interaction between the proteins that manifests itself kinetically as a 300fold decrease in a dissociation rate constant of the polymerase from the replication complex. It has been estimated that the half-time of binding of the T4 gp41 helicase within the T4 replication complex is ϳ11 min (27). If we assume that the half-time for the disassembly of the entire replication machinery is also ϳ11 min and that it dissociates as a first order reaction, we can estimate that the dissociation rate constant of the DNA polymerase, when coupled to the other replication proteins (helicase, helicase loader, processivity factor, and SSB protein), is 0.001 s Ϫ1 (i.e. ln 2/(11 ϫ 60)), resulting in another 6-fold decrease in the dissociation rate constant of the polymerase from the replication complex.

CONCLUSIONS
The data presented in this paper have helped us refine the parameters of the single nucleotide addition cycle performed by the T4 DNA polymerase. We propose that the polymerase can exist in two different binding modes at the p-t junction, depending on the location of the 3Ј-end primer terminus within the enzyme and that these two modes differ thermodynamically. When the 3Ј-end primer terminus is in the product binding subsite, the polymerase is in a specific and tight binding state that is poised to bind and incorporate the correct dNTP into the nascent strand. In contrast, after the catalysis reaction has occurred, the polymerase is in a nonspecific and loose (electrostatic) binding state and is poised to translocate along the DNA at the replication fork in order to vacate the substrate binding subsite and reposition the extended 3Ј-end primer terminus in the product binding subsite.
In addition, we have proposed that the nonspecific and loose binding state of the DNA polymerase is important not only for translocation along the DNA but also for polymerase dissociation at the end of an Okazaki fragment or to provide access to DNA polymerases charged with lesion bypass on the DNA template. Our data show that the binding of dNTP modulates the affinity of the DNA polymerase for its p-t junction, the binding being weaker with the incorrect incoming dNTP compared with the correct dNTP. We propose that the weakening effect of binding the incorrect dNTP opens a window of opportunity for dNTP or (local) polymerase dissociation from the p-t junction.
In addition, we have suggested that nonspecific DNA sequences serve as "landing sites" for the DNA polymerase, from which the enzyme bound in its nonspecific and electrostatic mode searches for its specific DNA binding target by random one-dimensional diffusion. We envision that the structure of the DNA fork (containing two p-t junctions) and the presence of the other replication proteins are likely to be responsible for assembling and "tethering" the leading and lagging strand DNA polymerases at the fork. We estimate that the processivity clamp, together with the helicase, helicase loader, and SSB proteins, stabilize the DNA polymerase at the p-t junction by decreasing the rate at which the DNA polymerase can dissociate from its specific substrate by a factor of at least 1800.
Finally, this analysis illustrates how knowing thermodynamic and kinetic parameters can be central to understanding the mechanisms of the assembly and function of multiprotein complexes (macromolecular machines). We suggest that auxiliary replication factors, such as the sliding clamp, the helicase, or the SSB protein, couple their activity with the DNA polymerase by modulating the intrinsic polymerase-DNA interaction. Such schemes for modulating the thermodynamics and/or kinetics of protein-protein or protein-nucleic acid interactions can thus provide plausible (and general) mechanisms for coupling reactions within macromolecular machines. APPENDIX As Fig. 6a shows, binding of the polymerase to the DNA is very fast, as indicated by the short time required to reach the equilibrium binding plateau (less than 2 s). Dissociation of the polymerase from the DNA, which was measured by determining the time needed to reach base line from the equilibrium plateau after stopping the injection of polymerase, was rapid as well. These apparent association and the dissociation rates are too fast to measure directly on the Biacore X instrument. Therefore, simulation techniques were used to estimate lower limit values of k on and k off from our SPR data. The simulations were performed using the "basic kinetics module" provided with the BIAsimulation version 1.2 software from Biacore. We simulated the various sensorgrams by entering pairs of rate constants that were constrained so that their k off /k on ratio was equal to the measured (by SPR) dissociation (binding) constant (K d ϭ 10 nM; see Fig. 6, a and b, and "Results").
The other parameters used for the simulations (see reactant properties and experimental parameters) (Table III) were as follows: the reactant concentration was 50 nM, the preinjection delay was 2 s, the association and dissociation times were 10 s, and R max was set at 8 RUs. As the simulated sensorgrams of Fig. 8 show, the times required to reach the equilibrium plateau during the association phase or to reach baseline during dissociation depend on the values of k on and k off that are entered.
The k on and k off parameters that permit the reaction to reach equilibrium in less than 2 s during association and to reach base line in less than 2 s during dissociation provide lower limits for these rate constants. The results of such simulations showed that k on for the DNA polymerase to its specific p-t substrate must be Ͼ2 ϫ 10 8 M Ϫ1 s Ϫ1 (close to the calculated diffusion limit for processes of this kind), whereas k off must be Ͼ2 s Ϫ1 .  Table III. Filled diamonds, k on ϭ 0.15 ϫ 10 9 M Ϫ1 s Ϫ1 , k off ϭ 1.5 s Ϫ1 ; filled squares, k on ϭ 0.075 ϫ 10 9 M Ϫ1 s Ϫ1 , k off ϭ 0.75 s Ϫ1 ; filled triangles, k on ϭ 0.03 ϫ 10 9 M Ϫ1 s Ϫ1 , k off ϭ 0.3 s Ϫ1 ; filled circles, k on ϭ 0.01 ϫ 10 9 M Ϫ1 s Ϫ1 , k off ϭ 0.1s Ϫ1 ; stars, k on ϭ 5 ϫ 10 6 M Ϫ1 s Ϫ1 , k off ϭ 0.05 s Ϫ1 ).