t Binds and Organizes Escherichia coli Replication Proteins through Distinct Domains

The t subunit dimerizes Escherichia coli DNA polymerase III core through interactions with the a subunit. In addition to playing critical roles in the structural organization of the holoenzyme, t mediates intersubunit communications required for efficient replication fork function. We identified potential structural domains of this multifunctional subunit by limited proteolysis of C-terminal biotin-tagged t proteins. The cleavage sites of each of eight different proteases were found to be clustered within four regions of the t subunit. The second susceptible region corresponds to the hinge between domain II and III of the highly homologous d* subunit, and the third region is near the C-terminal end of the t-d* alignment (Guenther, B., Onrust, R., Sali, A., O’Donnell, M., and Kuriyan, J. (1997) Cell 91, 335–345). We propose a five-domain structure for the t protein. Domains I and II are based on the crystallographic structure of d* by Guenther and colleagues. Domains III–V are based on our protease cleavage results. Using this information, we expressed biotin-tagged t proteins lacking specific protease-resistant domains and analyzed their binding to the a subunit by surface plasmon resonance. Results from these studies indicated that the a binding site of t lies within its C-terminal 147 residues (domain V).

The DNA polymerase III holoenzyme is responsible for the replication of the Escherichia coli chromosome. Like other replicases from eukaryotes and prokaryotes, the holoenzyme contains three functional subassemblies (for reviews see Refs. [1][2][3]: the DNA polymerase III (␣⑀) core, the ␤ sliding clamp processivity factor, and the DnaX complex, a clamp assembly apparatus. The DNA polymerase III core contains the ␣, ⑀, and subunits and provides the polymerase function. The DnaX complex ( 2 ␥␦␦Ј,) is a multiprotein ATPase that recognizes the primer terminus and loads the ␤ processivity factor onto DNA.
The and ␥ subunits are different translation products of the dnaX gene (4 -7). The subunit plays central roles in the structure and function of the holoenzyme. It interacts with the core polymerase to coordinate leading and lagging strand synthesis (8,9). also interacts with DnaB helicase to couple the replicase with the primosome and mediate rapid replication fork movement (10,11). These two important functions of reside in C-, a proteolytic fragment consisting of its unique C-terminal 213 amino acid residues. binds tightly to the ␣ subunit; the shorter translation product ␥ does not. C-is a monomer and binds ␣ with a 1:1 stoichiometry as determined by sedimentation equilibrium analyses (12). Results from a recent study indicated that C-binds DnaB, can partially replace full-length in reconstituted rolling circle replication reactions, and effectively couples the leading strand polymerase with DnaB helicase at the replication fork (12). DnaB helicase is composed of six identical subunits and is a stable hexamer over a wide range of concentration in the presence of magnesium ions (13,14).
In the preceding manuscript, we reported that comprises five potential structural domains (15). Domains I, II, and III are common to both ␥ and . Domain IV includes 66 amino acid residues of the C-sequence and the C-terminal 17 residues of ␥. Domain V corresponds to the 147 C-terminal residues of the subunit. Based on these assignments, biotin-hexahistidinetagged proteins lacking specific domains were produced. Results from binding studies employing these truncated fusion proteins indicated that the binding site of for ␣ subunit lies within its C-terminal 147 amino acid residues (domain V).
The objective of this study was to determine the domain(s) of the subunit involved in binding DnaB. Biotin-hexahistidinetagged proteins lacking specific domains were expressed and purified. Analysis of DnaB binding to these truncated proteins by surface plasmon resonance permitted the assignment of the DnaB-binding domain of .

EXPERIMENTAL PROCEDURES
Strains-E. coli DH5␣ and HB101 were used for initial molecular cloning procedures and plasmid propagation. E. coli BL21( DE3) was used for protein expression.
Buffers-Buffer L, Buffer W and HKGM Buffer were prepared as previously described (15).
Proteins-Three biotin-tagged proteins C(0), C-⌬147, and * This work was supported by Grant GM36255 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
N-⌬413 as well as holoenzyme subunits were prepared as previously described (15).
Construction of the Fusion Plasmids-Plasmids P A1 -C-⌬213 and pET11-N-⌬430 were constructed to express the fusion proteins C-⌬213 and N-⌬430, respectively. Fusion protein C-⌬213 corresponds to ␥, the shorter of the two potential dnaX products. In this construct, the 213 C-terminal residues found exclusively in the subunit are replaced by a peptide tag, which includes hexahistidine and a 13-amino acid residue biotinylation sequence. N-⌬430 corresponds to C-; the N-terminal 430 amino acids found in both and ␥ are replaced by the hexahistidine/biotinylation tag. P A1 is a semi-synthetic E. coli RNA polymerase-dependent promoter containing two lac operators (16). The pET11 vector is under the control of the T7 promoter.
The starting material for construction of plasmid P A1 -C-⌬213 was P A1 -C(0), which encodes the C-terminal tagged full-length protein (15). PCR primer C-213P1 (Table I) is complementary to a 110-nucleotide stretch upstream of the RsrII site within dnaX. Primer C-213P2 (Table I) corresponded to dnaX codons 423-430 preceded by a noncomplementary SpeI restriction site. P A1 -C(0) was digested with RsrII and SpeI. The PCR product generated by use of primers C-213P1/C-213P2 was cleaved with RsrII and SpeI and then ligated into the linearized vector to generate plasmid P A1 -C-⌬213.
Primers N-430p1 and N-430p2 (Table I) and plasmid P A1 -N-⌬1, which encodes an N-terminal tagged protein (15), were used to generate a PCR product for the construction of pET11-N-⌬430. The resultant PCR product consisted of a PstI restriction site within the noncomplementary 5Ј region followed by dnaX codons 431-436 and a KpnI site near the 3Ј end. The KpnI site located downstream of the natural dnaX stop codon. After digestion with PstI and KpnI, the resultant 929-base pair fragment was used to replace the dnaE gene of vector pET11-N0 (16) to produce plasmid pET11-N-⌬430.
Protein Purification-The procedures for purification of C-⌬213 and N-⌬430 were similar to those described for other truncated fusion proteins (15). Briefly, induced cells (14 g for C-⌬213 or 22 g for N-⌬430) were lysed in the presence of lysozyme (2.5 mg/ml), EDTA (5 mM), benzamidine (5 mM), and phenylmethylsulfonyl fluoride (1 mM) for 2 h at 4°C and 6 min at 37°C. For purification of C-⌬213, 0.226 g of ammonium sulfate was added to each milliliter of the resulting supernatant, and the precipitate was collected by centrifugation at 23,300 ϫ g at 4°C for 1 h. The pellets were then resuspended in Buffer L. Each suspension was mixed with 1 ml of Ni2 ϩ -NTA resin and pre-equilibrated with Buffer L, and the slurries were then packed into 1-ml columns. Columns were washed with ϳ30 column volumes of buffer W containing 23 mM imidazole. Bound proteins were then eluted with buffer W containing 150 mM imidazole in a single step. 13 mg of C-⌬213 were obtained in the preparation used for these studies. The purification of N-⌬430 was as that for C-⌬213 except that: 1) the supernatant proteins were precipitated with 65% ammonium sulfate, 2) 3 ml of pre-equilibrated Ni 2ϩ -NTA resin were used, 3) the columns were washed with buffer W containing 10 mM imidazole, and 4) elutions were effected by a 10 -100 mM imidazole gradient in buffer W. 61 mg of purified N-⌬430 were obtained in the preparation used for these studies.
Surface Plasmon Resonance-A BIAcore TM instrument was used for protein-protein binding studies. Research grade CM5 sensor chips were used in all experiments. Streptavidin was captured onto sensor chips by N-hydroxysuccinimide/1-ethyl-3-[ (3-dimethylamino)propyl)] carbodiimide coupling as previously described (15). The biotinylated proteins were then injected over the immobilized streptavidin sensor chip. Binding analyses of to DnaB (0.025-1 M) were performed in HKGM buffer at 20°C. Kinetic parameters were determined using the BIAevalua-tion™ 2.1 software.
Other Procedures-DNA polymerization assays, protein determinations, and SDS-polyacrylamide gel electrophoresis were performed as described in the preceding paper (15).

Expression and Purification of the Truncated Fusion Proteins-
The subunit binds to DnaB helicase and is the only subunit within the holoenzyme shown to interact with DnaB (10). The unique C terminus of (C-) bound DnaB in a coupled immunoblotting method (12). To confirm this observation and more precisely map the DnaB binding region of , a series of truncated proteins lacking specific domains were produced, and their interactions with DnaB helicase were quantified using BIAcore methodology. The fusion proteins employed in this study included C(0) (domains I-V), C-⌬147 (domains I-IV), C-⌬213, which was equivalent to ␥ (domains I-III plus 17 amino acids of domain IV), N-⌬413 (domains IV and V), and N-⌬430, which was equivalent to C-(the C-terminal 66 residues of domain IV plus domain V in its entirety). The truncated terminus of each fusion protein was tagged with a peptide containing both a hexahistidine sequence to aid in purification as well as a short biotinylation sequence. The biotinylation sequence enabled oriented immobilization of the fusion proteins onto BIAcore sensor chips via biotin-streptavidin binding. C(0), C-⌬147, and N-⌬413 were expressed and purified as previously described (15). C-⌬213 and N-⌬430 were expressed in the BL21 ( DE3) strain by induction with isopropyl-␤-D-thio-galactoside and reached similar expression levels (2-5% of total cell proteins). Both C-⌬213 and N-⌬430 were purified by Ni 2ϩ -NTA affinity chromatography. After Ni 2ϩ -NTA purification, C-⌬213 was obtained at 80% purity, and N-⌬430 at 90% purity as determined by SDS-polyacrylamide gel electrophoresis analysis (Fig. 1). The activities of the fusion proteins were ascertained by their ability to replace ␥ or in DNA polymerase III reconstitution assays (15). The specific activity of C-⌬213 was 5.5 ϫ 10 6 units/mg, similar to that of full-length C(0) (5.7 ϫ 10 6 units/mg). As expected, no holoenzyme reconstitution activity was detected for N-⌬430, which lacks the ␥ sequence required for assembly of the ␤ processivity factor on DNA.
DnaB Binding to Proteins Containing Domain IV-The interaction between DnaB and C(0) was first characterized via use of BIAcore technology. C(0) (2025 RU) was immobilized onto a streptavidin sensor chip. DnaB solutions of varying concentrations were passed over the immobilized C(0), and binding activity was monitored ( Fig. 2A). Attempts to fit the dissociation phase to a single first-order dissociation equation were unsuccessful, suggesting that a more complex mechanism was operative. To simplify the kinetic analysis, a limited interval (35-125 s following the starting point of dissociation) was analyzed from each binding curve and fit to a model in which two simultaneous independent dissociation processes occur. The two apparent dissociation rate constants k off major and k off minor (Table II) corresponded to 70 -80% and 20 -30% of the dissociating species, respectively. k off major was used to calculate the apparent association rate (k on ). The apparent K d was calculated from k on and k off . The interaction between DnaB and C(0) had an apparent K d of 4 nM (Table II). Under the conditions employed in these studies, DnaB is known to exist as a
Next, DnaB samples (0.05-0.5 M) were injected over immobilized C-⌬147 (2860 RU) (Fig. 2C). An apparent K d of 5 nM was obtained, which is similar to that of the C(0)-DnaB interaction (Table II). This suggests that C-⌬147 contains elements sufficient for binding to DnaB at the same level observed for the intact subunit. C-⌬147 (domains I-IV) bound DnaB, but C-⌬213 (domains I-III) did not, localizing the region required for DnaB binding to somewhere within domain IV.
DnaB Recognizes a 66-Amino Acid Sequence within Domain IV-To confirm that domain IV was the DnaB-binding domain, N-⌬430 (1200 RU) was captured onto a BIAcore sensor chip, and its interaction with DnaB was assessed (Fig. 3A). The dissociation phase did not fit to a single first-order dissociation equation, so the binding data were fit to the model that assumes two parallel dissociation processes. The apparent K d was about 8 nM, which was similar to that of the interaction between DnaB and C(0) ( Table II). The sum of these results indicates that the DnaB binding site is located within the unique C-terminal 66 residues of the subunit.
The C-terminal 17 amino acid residues of domain IV are lacking in N-⌬430. To investigate whether these 17 residues provide additional binding energy for the -DnaB interaction, DnaB binding studies using fusion protein N-⌬413 were per-

FIG. 2. DnaB interacts with C(0) and C-⌬147 but not C-⌬213.
Streptavidin was chemically immobilized onto sensor chip as described under "Experimental Procedures." fusion proteins were captured onto the streptavidin sensor chip via biotin-streptavidin interaction. DnaB diluted in HKGM to the indicated concentrations was injected over immobilized for 6 min at 5 l/min. Following injection, buffer was passed over the sensor chips for 30 min to permit dissociation of the bound DnaB protein from immobilized derivatives. A, DnaB interacts with C(0). 2025 RU of C(0) were captured on the sensor chip, and varying concentrations of DnaB were passed over it. Control injections over a streptavidin sensor chip were performed and subtracted from the data shown. B, DnaB does not interact with C-⌬213. 3390 RU of C-⌬213 were captured on the sensor chip, and DnaB at 1 M was injected over the immobilized C-⌬213. The control injection over a streptavidin sensor chip was also shown. C, DnaB interacts with C-⌬147. 2860 RU of C-⌬147 were captured onto the streptavidin sensor chip varying concentrations of DnaB were passed over it. Control injections over a streptavidin sensor chip were performed and subtracted from the data shown. formed. However, the DnaB/N-⌬413 interaction was characterized by an apparent K d of 5 nM, which is similar to that observed for the interaction of DnaB with N-⌬430 ( Fig. 3B and Table II). Thus, it is unlikely that the C-terminal 17 residues of domain IV contribute significantly to DnaB binding interactions.
More than One Protomer Binds a DnaB Hexamer-In the preceding experiment, the binding ratio of DnaB 6 to the monomeric N-⌬430 was less than 0.1. This value is significantly different from 0.72, the observed ratio for the interaction between DnaB 6 and C(0) 4 . One potential underlying cause of the low binding ratio for the former interaction is the binding of DnaB 6 to more than one immobilized N-⌬430 molecule. To test this hypothesis, we examined the interactions of DnaB (1 M) with sensor chips bearing differing amounts of N-⌬430 (146 RU-2700 RU). The corresponding densities of the six different levels of N-⌬430 tested are shown in Table III. Binding of DnaB to immobilized N-⌬430 at 146 RU was not observed. Increased binding ratios of DnaB to N-⌬430 were observed for surfaces bearing greater densities of N-⌬430 (Fig. 4). If we assume that each DnaB 6 binds two (N-⌬430) 1 molecules, then the binding ratio of DnaB to N-⌬430 is increased from 0.04 to 0.24 within the range of the amount of immobilized N-⌬430 tested (Table III). The same apparent dissociation and association rate constants for the DnaB and N-⌬430 interaction were obtained at different N-⌬430 density as reported in Table  II. These results are consistent with the multivalent binding of DnaB and N-⌬430. The observed K d is the product of the individual K d values for single site binding interactions. No binding was observed at low N-⌬430 density, suggesting that the monomeric -DnaB 6 interaction is too weak to be observed with the BIAcore methodology. The apparent K d values of the DnaB-N-⌬430 interaction and DnaB-C(0) interaction were the same, suggesting that the interaction between DnaB and C(0) is also multivalent; more than one C(0) monomer binds each DnaB 6 . The binding ratio between DnaB and N-⌬413 was also N-⌬413 density-dependent and increased with increased immobilized N-⌬413 density (data not shown).
To ensure that the observed binding ratio of the hexameric DnaB to the tetrameric C(0) was not density-dependent, the binding ratio of DnaB 6 to C(0) was examined at an increased density (4262 RU) of C(0) on a sensor chip. In a previous experiment, 2025 RU of C(0) was used ( Fig. 2A), and a binding ratio of 0.72 DnaB 6 to C(0) 4 was observed. The C(0) concentrations in these two different experiments corresponded to 568 FIG. 3. N-⌬430 and N-⌬413 bind DnaB. Streptavidin was chemically coupled onto the BIAcore sensor chips as described under "Experimental Procedures." 1200 RU of N-⌬430 (A) and 1293 RU of N-⌬413 (B) were captured onto streptavidin-derivatized sensor chips, and DnaB diluted in HKGM buffer at the indicated concentrations was injected over the chips bearing immobilized N-⌬430 and N-⌬413, respectively. Control injections over a streptavidin derivatized sensor chip were performed and subtracted from the data shown.   and 270 M of C(0) as monomer, within the density range of N-⌬430 used in the density dependence experiment (Table  III). The observed binding ratio of DnaB 6 to C(0) 4 was 0.69, which was not significantly different from the ratio obtained when using with 2025 RU of C(0) ( Table III).

DISCUSSION
In the preceding paper, we detailed our use of limited proteolysis studies to identify five putative structural domains of the protein (15). Domains I-III are common to both and ␥. Domain IV is composed of 17 amino acid residues from the C-terminal end of ␥ plus 66 amino acids from the unique C terminus of . Domain V is located at the C-terminal end of . One function of is to bind DnaB, coupling the holoenzyme with the primosome at the replication fork. C-, the unique C terminus of , bound DnaB in a coupled immunoblotting method (12). In reconstituted rolling circle replication reactions, C-can partially replace full-length in coupling the leading strand polymerase with the DnaB helicase at the replication fork (12).
This study further defined the DnaB binding domain of by analyzing the interactions of DnaB with several truncated proteins. N-⌬413, N-⌬430, C-⌬147, and C(0) bound DnaB with similar apparent K d values. Because complicated binding kinetics were operative, the apparent K d values we obtained in this study were not the true constants. However, the resulting apparent K d values presumably contain the same systematic errors and therefore permit a quantitative comparison of relative affinities. The relative binding affinities of the different fusion proteins for DnaB indicate that amino acid residues 431-496 are sufficient for DnaB binding. This 66-residue stretch corresponds to the C-terminal portion of domain IV.
Although similar apparent K d values were obtained for the interactions of DnaB 6 -C(0) 4 and DnaB 6 -(N-⌬430) 1 , the binding ratios for the DnaB 6 -(N-⌬430) 1 was density-dependent. We conclude that more than one N-⌬430 monomer is required to bind DnaB 6 and that the interactions between DnaB and involved multivalent binding. Thus, the true microscopic K d for binding of DnaB 6 to a single N-⌬430 was too weak to observe using a BIAcore. At higher N-⌬430 densities, binding was observed between DnaB 6 to two or more N-⌬430 molecules; the observed macroscopic K d is roughly equal to the product of each of the constituent microscopic K d values. 2 Consistent with this interpretation, the number of DnaB 6 binding to a BIAcore chip surface increases with the density of immobilized N-⌬430. Increases in N-⌬430 density would result in increased numbers of N-⌬430 molecules becoming located within each DnaB 6 binding sphere. The DnaB 6 binding sphere is a function of the diameter of the distance between two binding sites within each DnaB 6 molecule. Within each binding sphere, a certain number (n) of N-⌬430 molecules can be accommodated; n is equal to the maximum potential binding stoichiometry of N-⌬430 to DnaB 6 .
Recently, a model for quantifying the principal aspects of multivalent binding was developed (19). We used this model to estimate the probability of more than one N-⌬430 molecules binding DnaB simultaneously. The proportion of spheres containing a given number of N-⌬430 molecules was calculated assuming a binomial distribution. The DnaB-binding sphere was defined as the volume within which binding of the DnaB by two N-⌬430 molecules can occur, and it was calculated using the following equation: V S ϭ 4/3**D 3 , where D is the distance between two binding sites on DnaB. The DnaB hexamer is a cyclic structure and contains six chemically identical subunits (14,20,21). Based on hydrodynamic and electron microscopic studies, the cyclic structure of the DnaB hexamer has an outside diameter of ϳ140 Å and an inner channel of ϳ40 Å. The N-⌬430 would be in the 900 M range provided that there was no cooperativity involved ((8 nM) 1/2 ϭ 900 M). This low (900 M) affinity range is consistent with the lack of detected interaction. A streptavidin-derivatized sensor chip lacking the bound N-⌬430 provided a blank control, and the subtracted data are shown. DnaB at 1 M diluted in HKGM buffer was injected over the six N-⌬430 immobilized sensor chips for 4 min at 5 l/min. The schematic at right indicates that if the N-⌬430 density on a sensor chip is so low that only one N-⌬413 molecule binds each DnaB 6 molecule, the resultant interaction is too weak to be observed using this methodology. However, at higher densities of immobilized N-⌬430, multiple N-⌬413 molecules bind each DnaB 6 molecule, and the resultant interaction is strong enough to be detected. expected DnaB binding sphere would be in the range from 4/3**(40 Å) 3 to 4/3**(140 Å) 3 (268 -11480 nm 3 , respectively). If we assume that the interaction between and DnaB involves two N-⌬430 molecules, the calculated DnaB binding sphere is 2500 nm 3 , within the possible range for an interaction between two protomers and DnaB 6 .
The notion that two protomors bind each DnaB hexamer is consistent with the presence of a dimer at the replication fork (Fig. 5). 2 functions to dimerize the DNA polymerase III core to enable simultaneous synthesis of leading and lagging strands. We already know that the leading strand polymerase is tethered to DnaB (12). The findings presented in this report indicate that the same DnaB molecule couples both of the leading and lagging strand polymerases. Thus, a double tethers exists between the leading and lagging strand polymerase, one is through thelink and the additional one through the -DnaB link. This second tether might help keep the lagging strand associated with the replication fork and may serve to help retarget the dissociated lagging strand polymerase to the next primer synthesized at the replication fork (Fig. 5). Our mapping results demonstrate that the DnaB helicase binds domain IV and that the polymerase ␣ subunit binds domain V (15). These findings indicate important roles that the C terminus plays in DNA synthesis.