Plasmid replication initiator interactions with origin 13-mers and polymerase subunits contribute to strand-specific replisome assembly

Significance

Research on DNA replication initiation has not revealed the exact mechanism for replication complex de novo assembly at the origin or how the directionality of replication is determined. To date, no evidence for direct involvement of a replication initiation protein (Rep) in the process of polymerase recruitment has been reported. This work demonstrates that a plasmid Rep, in addition to its already described functions in origin opening and helicase recruitment, can serve as a DNA polymerase anchoring factor. Through its interaction with 13-mer sequences on one strand of initially unwound DNA and interactions with the subunits of DNA polymerase, the initiation protein facilitates strand-specific replisome assembly at the replication origin. This step determines the direction of DNA replication.

Abstract

Although the molecular basis for replisome activity has been extensively investigated, it is not clear what the exact mechanism for de novo assembly of the replication complex at the replication origin is, or how the directionality of replication is determined. Here, using the plasmid RK2 replicon, we analyze the protein interactions required for Escherichia coli polymerase III (Pol III) holoenzyme association at the replication origin. Our investigations revealed that in E. coli, replisome formation at the plasmid origin involves interactions of the RK2 plasmid replication initiation protein (TrfA) with both the polymerase β- and α-subunits. In the presence of other replication proteins, including DnaA, helicase, primase and the clamp loader, TrfA interaction with the β-clamp contributes to the formation of the β-clamp nucleoprotein complex on origin DNA. By reconstituting in vitro the replication reaction on ssDNA templates, we demonstrate that TrfA interaction with the β-clamp and sequence-specific TrfA interaction with one strand of the plasmid origin DNA unwinding element (DUE) contribute to strand-specific replisome assembly. Wild-type TrfA, but not the TrfA QLSLF mutant (which does not interact with the β-clamp), in the presence of primase, helicase, Pol III core, clamp loader, and β-clamp initiates DNA synthesis on ssDNA template containing 13-mers of the bottom strand, but not the top strand, of DUE. Results presented in this work uncovered requirements for anchoring polymerase at the plasmid replication origin and bring insights of how the directionality of DNA replication is determined.

DNA synthesis of prokaryotic and eukaryotic replicons requires the coordinated action of several enzymes (reviewed in detail in 1, 2). These enzymes cooperate to form specific nucleoprotein complexes during the course of DNA replication. The formation of the initial complex is a result of a replication initiation protein (Rep) or origin recognition complex binding to dsDNA within the origin of DNA replication initiation (ori). This interaction of the replication initiators with DNA results in origin opening [i.e., destabilization of the DNA unwinding element (DUE)]. Origin opening provides ssDNA for helicase (3, 4), primase (5), and polymerase.

It has been demonstrated that during the opening of the bacterial chromosomal origin (oriC), the chromosomal replication initiator, DnaA, binds to specific sequences (DnaA boxes) (6) and forms a filament on ssDNA (7, 8). Specific interaction between DnaA and the DnaB helicase (9, 10) recruits the helicase and contributes to its loading by a helicase loader, the DnaC protein (11). Interactions between DnaB and the τ-subunit of polymerase (12), as well as DnaB and primase (13), contribute to replisome assembly at Escherichia coli oriC. The primase requires contact with single-stranded DNA-binding protein (SSB) (14) to remain bound to the RNA primer. Disruption of this interaction mediated by the polymerase clamp loader leads to primase displacement (14); β-clamp loading on primed DNA (15, 16); and, finally, interaction of the polymerase core subunits with the β-clamp–loaded template (14, 17). β-clamp loading is a complex reaction involving clamp opening and then positioning around the DNA with the use of the clamp loader (reviewed in ref. 18). The scenario for the polymerase assembly at the replication origin is mainly assumed, based on investigations of the mechanism of leading and lagging DNA strand synthesis, conducted with in vitro assays on primed circular DNA but not on supercoiled templates. It is not clear how the replisome is assembled on supercoiled dsDNA after origin opening and whether the helicase interactions with primase and τ-subunit are the only factors contributing to de novo replisome assembly at the replication origin.

In case of bacterial plasmids, involvement of both the plasmid-encoded Rep and the host-encoded replication initiator DnaA was reported as essential for origin opening and helicase complex recruitment (19–21). DNA replication of the broad-host-range plasmid RK2 (reviewed in ref. 19) is initiated by the RK2 plasmid encoded Rep protein (TrfA), which binds to direct repeats (iterons) localized at the plasmid’s replication origin (oriV) (22) (Fig. 1A). In contrast to DnaA (23), TrfA, as well as other plasmid Reps, does not contain a DNA binding domain (DBD). Instead, the plasmid Reps are similar to eukaryotic replication initiators and contain a winged helix (WH) domain for DNA interaction (24, reviewed in 25). TrfA interaction with the iterons leads to origin opening assisted by host HU and DnaA proteins (26). TrfA plays a crucial role in DnaB helicase recruitment and positioning at the AT-rich region of the oriV (27, 28). In contrast to DnaA protein, no data for plasmid Rep filament formation on ssDNA have been provided to date. Recently, it was shown that TrfA interacts with ssDNA containing 13-mer sequences of one strand of the plasmid origin DUE (29). Interestingly, the specific motif (QL[S/D]LF) determining interaction with the β-clamp has been identified in plasmid Reps (30); however, the relevance of the interaction between the β-clamp and Reps has not been established. The QAMSLF motif (related to QLSLF) was identified in the δ-subunit of the clamp loader, and it has been shown to be required for δ-interaction with the β-clamp and clamp loading (16). In this work, we investigate the significance of the β-clamp interaction with the TrfA replication initiator in the process of RK2 plasmid DNA replication initiation. We also examine the requirements for the assembly of the replication complex at the plasmid origin.

Fig. 1.

TrfA ∆LF is not active in RK2 DNA replication either in vitro or in vivo. (A) Plasmid RK2 minimal origin of replication. The scheme presents the RK2 origin (oriV) region comprising the cluster of 17-bp direct repeats (iterons), four DnaA boxes, and the DUE region with four 13-mers. (B) In vitro replication with a crude extract (FII) prepared from E. coli C600. (C) In vitro DNA replication reaction reconstituted with purified proteins (Recon. system). (B and C) Both in vitro replication experiments were established with increasing amounts of wt TrfA or TrfA mutants as noted (0, 30, 60, 90, 120, 150, 210, and 300 nM) (results are presented from n = 3 replicates, with the SD shown). (D) TrfA in vivo activity test. Plasmid pSV16 containing oriV was used to transform E. coli DH5α cells with plasmids carrying variants of trfA as indicated. The transformation frequency is reported as the number of colony-forming units per 1 μg of supercoiled plasmid pSV16 DNA averaged from three independent experiments, with the SD shown.

Results

To address the question concerning the significance of TrfA interaction with the β-clamp, we constructed TrfA variants with mutations in the QLSLF motif that were expected to disrupt this interaction. With the use of an ELISA test, it was previously shown that LF deletion within the QLSLF motif results in TrfA defective in β-clamp binding (30). Using PCR-based site-directed mutagenesis, we constructed plasmids carrying genes for TrfA ΔLF and TrfA F138A (where Phe in the QLSLF motif was changed into Ala). Wild-type TrfA (wt TrfA) and TrfA variants with alterations within the QLSLF motif were purified by affinity chromatography (Materials and Methods). To assess the quality of the purified proteins, we performed an analysis of the isothermal CD spectra (Fig. S1) and calculated the content of the respective secondary structures for each TrfA variant. The results did not reveal any substantial differences in the secondary structure’s content of the analyzed TrfA variants in comparison to the wt TrfA. We then determined how the TrfA variants interacted with the β-clamp using surface plasmon resonance (SPR) (SI Materials and Methods and Fig. S2). The wt TrfA immobilized on a CM5 sensor chip interacted with the β-clamp, whereas TrfA ΔLF–β-clamp complex formation was severely impaired. Under the same experimental conditions, TrfA F138A interacted with the β-clamp, although slightly less efficiently than was observed for the wt TrfA.

Fig. S1.

CD spectra analysis of TrfA variants and Western blot analysis of anti-α and anti-β antibodies. (A) Isothermal CD spectra measurements of wt TrfA, TrfA F138A, and TrfA ∆LF, measured from 200–260 nm, were performed as described by Pierechod et al. (43). (B) Content of respective secondary structures for each protein was calculated from the spectra using the SELCON3 method with CDPro software (62) (lamar.colostate.edu/∼sreeram/CDPro). [θ]MRW, the mean residue molar ellipticity. (C and D) Analysis of the specificity of the anti-α and anti-β antibodies was performed using the Western blot technique. (C) Complete Western blot of FII and purified α with the α-antiserum is shown. (D) Complete Western blot of purified β, TrfA variants, and α is shown.

Fig. S2.

WT TrfA interacts with β-clamp, and TrfA ∆LF is defective in β-clamp binding. Formation of the complex involving β-clamp and TrfA variants in real time was studied using SPR on a CM5 sensor chip with wt TrfA (A), TrfA F138A (B), or TrfA ∆LF (C) immobilized on its surface. The β-clamp was injected onto the sensor chip at the indicated concentrations. (D and E) Steady-state response (Req) obtained from the sensograms was plotted against increasing β-clamp concentration (black circles) and fitted to a steady-state affinity model (solid line) for determination of the equilibrium dissociation constant (KD) values for wt TrfA and TrfA F138A with the use of BiaEvaluation software. All results are presented from n = 5 replicates, with the SD shown.

To determine whether the mutations in the QLSLF motif that altered TrfA’s interaction with the β-clamp influenced RK2 DNA synthesis, we tested the replication activity of the purified TrfA variants in an in vitro replication assay. The test was based on E. coli cell crude extract (FII) that allows replication of supercoiled dsDNA in the presence of the plasmid replication initiator, TrfA. The soluble FII extract contains all proteins necessary for plasmid DNA synthesis, including polymerase III (Pol III) holoenzyme and chaperones for TrfA activation. DNA synthesis in this assay, measured as the total amount of incorporated nucleotides, reached a maximum when 90 nM wt TrfA was added to the assay mix and was inhibited by larger amounts of this protein (Fig. 1B). Based on the amount of nucleotides incorporated into the template, we calculated that 25% of DNA templates were typically copied, similar to results presented by others during experiments with an oriC in vitro system (31, 32). TrfA ΔLF was defective in DNA synthesis; replication reactions carried out in the presence of varying amounts of this mutant protein remained at background levels. Replication activity of the TrfA F138A mutant was reduced in comparison to wt TrfA but showed a similar activity profile with a peak at 90 nM protein and an inhibition of DNA synthesis at larger amounts of protein. Similar results (Fig. 1C) were obtained when we performed in vitro replication assays using purified enzymes and a supercoiled dsDNA oriV template according to the method of Konieczny and Helinski (28), with modifications (SI Materials and Methods).

The in vitro replication results were consistent with the replication results obtained in transformation efficiency tests. Plasmid pSV16 was used to transform E. coli cells containing pAT plasmid encoding one of the trfA gene variants: wt trfA, trfAΔLF, or trfAF138A. Plasmid pSV16 contains the RK2 origin of replication but not the gene for TrfA; thus, transformant colony growth indicates that the pAT encodes a replicatively active form of TrfA. When E. coli carrying pATwt trfA was transformed with pSV16 plasmid, colonies were obtained with a transformation frequency of 3.8 × 10⁴ (Fig. 1D). No colonies were observed when E. coli cells containing plasmid pATtrfAΔLF were transformed with pSV16. When E. coli containing pATtrfAF138A was transformed with pSV16, the obtained transformation frequency was slightly lower compared with the transformation frequency observed with pATwt trfA encoding wt TrfA. In a control experiment, when pBBR1MCS2 plasmid (Table S1), replicating independent of the TrfA, was used instead of pSV16, we observed colony growth of E. coli strains with all three pAT variants tested (Fig. 1D).

Table S1.

Strains and plasmids used in the study

Strain or plasmid	Genetic characteristic	Source
E. coli strains
BL21(DE3)	F− ompT gal dcm lon hsdSB(rB− mB−) λ(DE3 [lacI lacUV5-T7 gene 1 ind1 sam7 nin5]); used for protein overproduction	(63)
C600	[thr leu thi lacY supE44 tonA]; used for in vitro assays	(64)
DH5α	fhuA2Δ(argF-lacZ) U169 phoA glnV44 Φ80Δ(lacZ)M15 gyrA96 recA1 relA1 endA1 thi-1 hsdR17; used for plasmid transformation	(65)
XL1-Blue	recA1 endA1 gyrA96 thi-1 hsdR17 supE44 relA1 lac [F′ proAB lacIqZΔM15 Tn10 (Tetr)]; used for ssDNA template construction and rescue	(66)
Plasmids
pATtrfAF138A	trfAF138A gene cloned in pTrcHisA with an N-terminal His6 tag, expression controlled by a trc promoter, Amp resistance	This study (33)
pATwt trfA	wt trfA gene cloned in pTrcHisA with an N-terminal His6 tag, expression controlled by a trc promoter, Amp resistance	(33)
pATtrfAΔLF	trfAΔLF gene cloned in pTrcHisA with an N-terminal His6 tag, expression controlled by a trc promoter, Amp resistance	(30, 33)
pBluescript KS(−)	Phagemid vector, Amp resistance	Stratagene
pBluescript oriVbottom	pBluescript KS(−) phagemid vector derivative containing HindIII-XhoI RK2 oriV DNA fragment	This work
pBluescript oriVtop	pBluescript KS(−) phagemid vector derivative containing AleI-HindIII RK2 oriV DNA fragment	This work
pBAD24ClpX	E. coli clpX gene cloned in pBAD24, Amp resistance	(43)
pBBR1MCS2	Broad-host-range plasmid, Kan resistance	(67)
pBSoriC	Plasmid containing oriC	(68)
pBS59	E. coli gyrAB genes, gyrase expression controlled by a tandem lambda promoter, Amp resistance	Nicholas Dixon (University of Wollongong)
pET3b	E. coli dnaN gene, β-subunit of DNA Pol III expression controlled by a T7 promoter, Amp resistance	Deepak Bastia (Medical University of South Carolina)
pET-δ	E. coli holA gene, δ-subunit of DNA Pol III expression controlled by a T7 promoter, Amp resistance	Nicholas Dixon (University of Wollongong) (52)
pET-δ′	E. coli holB gene, δ′-subunit of DNA Pol III expression controlled by a T7 promoter, Amp resistance	Nicholas Dixon (University of Wollongong) (52)
pET-θ	E. coli holE gene, θ-subunit of DNA Pol III expression controlled by a T7 promoter, Amp resistance	Nicholas Dixon (University of Wollongong) (69)
pET-χ	E. coli holC gene, χ-subunit of DNA Pol III expression controlled by a T7 promoter, Amp resistance	Nicholas Dixon (University of Wollongong) (58)
pET-ψ	E. coli holD gene, ψ-subunit of DNA Pol III expression controlled by a T7 promoter, Amp resistance	Nicholas Dixon (University of Wollongong) (58)
pGK1	E. coli dnaA gene cloned in pBAD24 with a C-terminal His6 tag, Amp resistance	(48)
pJC491	E. coli dnaX gene, τ-subunit of DNA Pol III expression	Nicholas Dixon (University of Wollongong) (70)
pJC490	E. coli dnaX gene, γ-subunit of DNA Pol III expression	Nicholas Dixon (University of Wollongong) (56)
pKD19L1	Plasmid carrying the minimal RK2 origin region, Tet resistance	(71)
pND261	E. coli dnaN gene, β-subunit of DNA Pol III expression	Nicholas Dixon (University of Wollongong) (72)
pND517	E. coli dnaE gene, α-subunit of DNA Pol III expression	Nicholas Dixon (University of Wollongong) (52)
pPS562	E. coli dnaC gene, Amp resistance	(49)
pRC50	E. coli dnaB gene cloned in pBAD24 with a C-terminal His6 tag, Amp resistance	(47)
pRLM1078	E. coli HU gene, expression controlled by a lac promoter, Amp resistance	Roger McMacken (Johns Hopkins University)
pRLM96	E. coli dnaG genes, primase expression controlled by a λ-promoter, Amp resistance	Roger McMacken (Johns Hopkins University)
pSH1017	E. coli dnaQ gene, ε-subunit of DNA Pol III expression controlled by a T7 promoter, Amp resistance	Nicholas Dixon (University of Wollongong) (69)
pSV16	Plasmid carrying oriV, Kan resistance	(73)
VCSM13	Helper phage, a derivative of M13KO7	Stratagene

Our experiments revealed that besides the defects in TrfA interaction with β-clamp, the analyzed mutations within the QLSLF motif result in a deficiency of plasmid DNA synthesis both in vitro and in vivo. We then asked at which stage of replication initiation the ΔLF and F138A mutations caused the observed deficiencies of the in vitro and in vivo replication. To answer this question, we examined the individual steps of the initiation process in vitro. First, we tested if mutations within the QLSLF motif might affect TrfA interaction with DNA. TrfA binds to oriV iterons as a monomer (33), and chaperone proteins are required for TrfA activation by converting TrfA dimers to monomers (34, 35). Thus, before the DNA interaction tests, we used the E. coli ClpX chaperone to activate the purified TrfA variants (Materials and Methods). After incubation with the chaperone, we used a gel mobility shift assay (GMSA) (SI Materials and Methods) to analyze binding of the ClpX-activated TrfA variants to fluorescently labeled dsDNA fragments containing a five-iteron sequence of RK2 oriV. Our experiments revealed no substantial differences among the tested proteins in their ability to form nucleoprotein complexes; TrfA ΔLF and TrfA F138A bind to the DNA fragment and produce retarded bands as efficiently as wt TrfA (Fig. S3 A–C). It must be pointed out that, as has been previously demonstrated, the ClpX does not interact with DNA (36), which we also verified during our experiments (Fig. S3D).

Fig. S3.

Affinity of TrfAF138A and TrfA ∆LF for DNA is similar to the affinity of wt TrfA. Binding of (A) wt TrfA and (B and C) its variants to DNA was studied by GMSA with a fluorescently labeled, linear DNA fragment containing five iterons. Increasing molar concentrations of ClpX-activated TrfA variants (0.2, 0.4, 0.56, 0.75, and 0.9 μM) were incubated with DNA (50 nM). (D) ClpX was tested for its binding to iterons. Arrows indicate the position of free DNA probes.

Because TrfA is involved in DnaB helicase complex formation at oriV (27, 28), we decided to test if the TrfA ΔLF and TrfA F138A, which were defective in replication, were able to recruit the helicase. Using gel filtration on a Sepharose CL-4B column (GE Healthcare), we tested helicase complex formation on a supercoiled DNA template containing oriV (pKD19L1) in the presence of ClpX-activated TrfA variants (Fig. 2A). In addition to supercoiled DNA, TrfA variants, and DnaB, the reaction mixture contained DnaA and DnaC proteins, which are required for helicase complex formation at oriV (28). Under these experimental conditions, plasmid DNA was found in the column’s void volume fractions (Fig. 2A, Top). Other panels of Fig. 2 illustrate DnaB detected by a protein-specific antiserum in consecutive fractions collected from the column. In a control experiment, when wt TrfA was not added to the reaction mixture, we did not detect DnaB in the void volume, which indicated a lack of helicase at oriV in the absence of TrfA. In contrast, DnaB protein was observed in void volume fractions when the reaction mixture contained any of the tested variants of TrfA, showing that the TrfA ΔLF supports helicase complex formation on RK2 DNA as well as the wt TrfA (Fig. 2A). TrfA F138A supports helicase complex formation less efficiently. In control experiments, we did not detect DnaB in void volume when pUC18 supercoiled DNA or no DNA was present in the reaction mixture (Fig. 2A).

Fig. 2.

In the presence of TrfA ΔLF or TrfA F138A, the DnaB helicase complex is loaded and activated on RK2 DNA. (A) Formation of nucleoprotein complexes was studied using gel filtration as described in SI Materials and Methods. Reaction mixtures containing each of the ClpX-activated TrfA variants (120 nM) if present, DnaB (200 nM), DnaA (7.5 nM), DnaC (1.1 μM), and pKD19L1 DNA (2.3 nM) were used. A 20-min incubation was followed by gel filtration. Fractions (80 μL) collected from the column were analyzed by SDS/PAGE and Western blotting for the presence of DnaB and by agarose electrophoresis for the presence of DNA. (Top) Plasmid DNA was found in void volume fractions. (B) E. coli DnaB helicase activity on RK2 DNA was determined by the FI* assay. The position of the FI* DNA band, indicating helicase activity, is marked with arrows. Reaction mixtures were as described in SI Materials and Methods with increasing TrfA concentrations (0, 30, 60, 90, 120, and 150 nM in lanes 1–6, respectively).

It has been demonstrated that in addition to DnaB loading, TrfA is involved in helicase activation at oriV (28). Thus, formation of the DnaB helicase complex on DNA might not necessarily result in helicase activity. We therefore tested the TrfA variants as helicase-activating factors. E. coli DnaB helicase activity on supercoiled RK2 plasmid DNA template was identified electrophoretically by the formation of an extensively unwound form of the supercoiled plasmid DNA, designated FI* (28). The formation of the FI* form is a result of the combined activities of plasmid and bacterial Reps, with the presence of an active form of TrfA being essential. ClpX activation of TrfA variants was performed as for the previous in vitro tests, and the other reaction components were then added. Experiments were carried out with increasing concentrations of the TrfA variants (Fig. 2B, lanes 2–6; lane 1 is the reaction mixture without added TrfA). In agreement with previous results (34), in the presence of ClpX-activated wt TrfA, we observed the FI* DNA band that indicates helicase unwinding activity. The FI* form of DNA was also observed in the presence of both TrfA mutants tested, TrfA ΔLF and TrfA F138A; however, TrfA F138A was slightly less efficient in helicase activation (Fig. 2B).

To acquire helicase unwinding activity on plasmid DNA, a number of steps, including the formation of an initial complex, origin opening, and helicase complex assembly and its activation, have to be accomplished in the presence of an active form of TrfA. Our results indicate that TrfA ΔLF, as well as TrfA F138A, promotes helicase activation but not plasmid DNA synthesis. These results, together with the confirmed TrfA–β-clamp interaction, strongly suggest a role for TrfA during the assembly of the replisome complex at the plasmid replication origin. To further examine TrfA's role in replisome assembly, we studied the formation of nucleoprotein complexes on supercoiled RK2 plasmid DNA using gel filtration with a fluorescently labeled β-clamp (Fig. 3). As in previous experiments, wt TrfA, TrfA F138A, or TrfA ΔLF was initially activated by incubation with ClpX. Reaction mixtures were prepared as for in vitro replication assays using purified enzymes (SI Materials and Methods). After incubation, the reaction mixtures were applied on a Sepharose CL-4B column and the collected fractions were analyzed for the presence of DNA and fluorescently labeled β-clamp. A fluorescent signal in the void volume fractions containing plasmid DNA would indicate that the β-clamp was found in a nucleoprotein complex. In the presence of wt TrfA, the fluorescent signal was detected in the void volume fractions (Fig. 3A), whereas when the TrfA ΔLF variant was used, fluorescence remained at background levels in the void volume. The observed fluorescence was significantly reduced when TrfA F138A was tested under the same conditions (Fig. 3A). We estimated that when wt TrfA was present in the reactions, ∼15% of the total amount of the β-clamp used for the experiment was detected in DNA-containing fractions. It should be pointed out that a similar efficiency was obtained with the DnaA-dependent reaction on oriC-containing templates (Fig. S4). We also performed control experiments where reaction mixtures contained only fluorescently labeled β-clamp and RK2 plasmid DNA or all components except TrfA. In both cases, the fluorescence signal did not appear in void fractions (Fig. S5 E and F). These results indicated that TrfA, with an intact QLSLF motif, is specifically required for β-clamp complex formation at the RK2 plasmid origin. We then determined what other proteins in the reaction mixture were indispensable for the β-clamp complex formation with supercoiled RK2 plasmid DNA. A fluorescent signal was not detected in void fractions in the absence of all of the other Pol III subunits (Fig. 3B), γδδ′χψ (Fig. 3D), DnaA (Fig. 3E), DnaB (Fig. 3F), or primase (Fig. 3G). The presence of Pol III core subunits (Fig. 3C) or gyrase (Fig. 3H), however, was not essential for β-clamp nucleoprotein complex formation on RK2 plasmid DNA, because we still detected fluorescent signal in the void volume in the absence of either of these factors. When GTP, UTP, and CTP and/or dNTPs were omitted from the reaction mixture, we still observed a fluorescent signal in the void fractions (Fig. S5 B–D). Because β-clamp nucleoprotein complex formation was dependent on TrfA and γδδ′χψ in our gel filtration experiments, we tested how the composition of clamp loader affects the RK2 DNA replication. In vitro replication assays in a reconstituted system with variously assembled clamp loaders were performed (Fig. S6A). The reactions were performed with wt TrfA and TrfA ΔLF. A substantial reduction of the replication activity was observed in the absence of τ-subunit, although not much difference in replication activity was observed when the reaction was conducted as in Fig. 1C, containing both τ and γ, in comparison to the reaction with clamp loader lacking γ (Fig. S6A). With TrfA ΔLF or without TrfA at all, we detected only a background level of DNA synthesis. To test if TrfA can affect ATPase hydrolysis by a clamp loader, we performed a clamp loader ATPase activity assay in the presence of primed DNA (Fig. S6B). No influence of wt TrfA or TrfA ΔLF was observed.

Fig. 3.

TrfA is required for β-clamp nucleoprotein complex formation on RK2 plasmid DNA. Nucleoprotein complex formation in the reconstituted system was studied using a gel filtration-based assay. (A) Reaction mixtures containing wt TrfA (120 nM), TrfA F138A (120 nM), or TrfA ∆LF (120 nM) were activated with ClpX (870 nM), and DnaA (450 nM), DnaB (200 nM), DnaC (1.1 μM), gyrase (13 nM), HU (110 nM), SSB (150 nM), primase (60 nM), and pKD19L1 (2.3 nM) template DNA were then added. (B–H) Reaction mixtures were prepared as described for A, except that indicated components were omitted. A 20-min incubation was carried out to form a helicase complex, and β-clamp labeled with Alexa 488 fluorescent dye (17 nM) and E. coli Pol III core τ-subunits [αεθτ (19 nM) and γδδ′χψ (2.5 nM)] were then added. After 15 s, the reaction was quenched by adding 5 units of hexokinase to remove ATP (51), followed by gel filtration on a Sepharose CL-4B column. The collected fractions (40 μL) were analyzed for the presence of the Alexa 488 β-clamp by measuring fluorescence and for the presence of DNA by agarose electrophoresis. (Top) Plasmid DNA was found in void volume fractions only (as labeled).

Fig. S4.

β-Clamp nucleoprotein complex formation on oriV and oriC DNA templates and α-nucleoprotein complex formation on oriV analyzed with an in vitro replication reconstituted (Recon.) system. (A) Reactions for detection of β-clamp and α-nucleoprotein complexes were established based on in vitro replication with the reconstituted system on supercoiled plasmid templates. The reaction mixture for oriV template (pKD19L1, ●) was exactly as described in SI Materials and Methods and for Fig. 1C. The reaction mixture for oriC template was as for oriV, except that supercoiled plasmid pBSoriC DNA (2.3 nM, ○) was used and TrfA was omitted. (B and C) β-Clamp nucleoprotein complex formation was studied in the reconstituted system with the use of a gel filtration-based assay as described in SI Materials and Methods and for Fig. 3. For the oriC reaction, supercoiled plasmid pBSoriC DNA was used and TrfA was omitted. The amount of added β-clamp bound to DNA was calculated from the resulting chromatograms by using estimation of the area under the curve in Microsoft Excel. In the presence of the oriC DNA template, as well as the oriV DNA template, ∼15% of the total amount of β-clamp used for the experiment was bound to the DNA. (D and E) Experiments for α-nucleoprotein complex formation on supercoiled oriV template (pKD19L1) were performed with the use of the reconstituted system essentially as described for β-clamp nucleoprotein complex analysis. The presence of α was analyzed with anti-α antibodies. Reactions were carried out in presence of TrfA wt or TrfA ∆LF. (Top) Presence of DNA was analyzed by agarose electrophoresis.

Fig. S5.

Requirements for β-clamp nucleoprotein complex formation on RK2 plasmid DNA. (A–F) Nucleoprotein complex formation was studied using a gel filtration-based assay with reaction mixtures as described in Fig. 3. Components missing or present in the reaction are indicated for each panel.

Fig. S6.

In vitro replication of supercoiled RK2 plasmid DNA with variously assembled clamp loaders and assay of ATPase activity of clamp loader in the presence or absence of TrfA. (A) Replication activity with the use of variously assembled clamp loaders was tested in vitro in the reconstituted system. Three different strategies to reconstitute clamp loader complex were adopted. In the first strategy (assigned no. 1), proteins γ, δ, δ′, and χψ were preassembled and then supplemented with τ directly in the reaction mixture. This clamp loader complex was also used in the experiments shown in Fig. 1C and Fig. 3. In the second strategy (assigned no. 2), τ was omitted. In the third strategy (assigned no. 3), preassembled clamp loader complex contained τ, δ, δ′, and χψ (mixed at a 3:1:1:1 molar ratio), but no γ was added. The reaction mixtures contained ClpX-activated wt TrfA or TrfA ΔLF (120 nM) and other components in the amounts and incubation conditions described for in vitro replication (Recon. system) in SI Materials and Methods. The control reaction (187 pmol) with ClpX-activated wt TrfA (120 nM) was performed as described for experiments shown in Fig. 1C and was assigned for 100% of replication activity. (B) Clamp loader (τ, γδδ′χψ) ATPase activity with primed DNA Blue KS(−) (2.3 nM) was tested by a coupled enzyme ATPase assay in the presence or absence of β-clamp and TrfA as described in SI Materials and Methods. The ATPase activity of 100% corresponds to the reaction mixture containing τ, preassembled γδδ′χψ complex, and primed DNA. The results represent the mean values of three independent measurements. Error bars indicate SD values.

To analyze the mechanism of replisome assembly at the RK2 plasmid origin further, we coupled the gel filtration experiment with immunodetection of the α-subunit of the Pol III core (Fig. 4). After incubation, reaction mixtures consisting of bacterial crude extracts (FII), RK2 supercoiled DNA, and TrfA variants were applied on a Sepharose CL-4B column and fractions were immunoanalyzed. The α was detected in void volume fractions when wt TrfA (Fig. 4A) or, surprisingly, the TrfA F138A or TrfA ΔLF mutant (Fig. 4 B and C) was present in reaction mixtures. In control experiments, when TrfA or DNA was omitted or pUC18 was used instead of RK2, α was not detected immunologically in the void volume (Fig. 4 D–F). When we tested for β-clamp nucleoprotein complex formation under the same experimental conditions, the protein was detected in the presence of wt TrfA and oriV DNA (Fig. 4G). When TrfA was omitted or TrfA ΔLF was present in the incubation mixture, we observed almost no absorbance signal derived from immunological assay. The observed absorbance signal was significantly reduced when TrfA F138A was used (Fig. 4 G–I). The β-clamp tracking in crude extract was in accordance with experiments performed for reconstituted reactions (Fig. 3A). Also, when we used a reconstituted system for α-tracking, consistent with the results obtained in crude extract, α was eluted in void fractions in the presence of either wt TrfA or TrfA ΔLF (Fig. S4 D and E). Regardless of the system used, the α was found in the nucleoprotein complex formed on RK2 DNA in the presence of the TrfA variants defective in interaction with the β-clamp (Fig. S2) and not promoting the β-clamp complex formation at the plasmid origin. This unexpected finding raised a question concerning the mechanism of this interaction.

Fig. 4.

Detection of α- and β- subunits on RK2 plasmid DNA in crude cell extract. Nucleoprotein complexes were formed with the use of an E. coli C600 crude cell extract, and gel filtration was carried out as described in SI Materials and Methods. Reaction mixtures, prepared as for in vitro replication (FII), contained pKD19L1 or non-oriV pUC18 (2.3 nM), wt TrfA (120 nM), TrfA F138A (120 nM), TrfA ∆LF (120 nM), or no TrfA (as indicated). After a 30-s incubation, reactions were cross-linked by the addition of formaldehyde (0.01%) and incubated for an additional 10 min. Fractions collected (40 μL) during gel filtration were applied to an ELISA MaxiSorp plate (Nunc). (A–F) The presence of α or (G–I) β was detected using an immune-enzymatic assay with anti–α- or anti–β-antibodies, respectively, and A at 450 nm was measured. Nucleoprotein complexes were eluted in void fractions (marked with gray). (Top) Presence of DNA was analyzed using agarose electrophoresis.

To address the issue of how α interacts at oriV without the β-clamp being loaded onto the plasmid DNA, we performed an ELISA with α-specific or β-specific antisera (SI Materials and Methods and Fig. S1 C and D). Interestingly, when wt TrfA was immobilized on plates and increased concentrations of α were applied, we clearly detected an interaction (Fig. 5A). Furthermore, TrfA ΔLF and TrfA F138A interacted with α as efficiently as wt TrfA (Fig. 5A). Consistent with our SPR analysis (Fig. S2), we observed that wt TrfA interacts with the β-clamp, whereas TrfA ΔLF is deficient in β-clamp binding (Fig. 5B) and TrfA F138A interaction with the β-clamp was reduced by half compared with wt TrfA (Fig. 5B). We were also able to detect an α–β-clamp interaction (Fig. 5C).

Fig. 5.

TrfA interacts with α, and this interaction does not affect TrfA–β-clamp complex formation. TrfA interactions with α and β-clamp were examined using ELISA (details are provided in SI Materials and Methods) (results presented from n = 3 replicates, with the SD shown). (A–C) Graphs represent interactions between individual proteins. (D) Possible network of interactions between TrfA and α and β-clamp. β-clamp binding pentapeptide motifs present in α (QADMF) and in TrfA (QLSLF) are marked. (E and F) Graphs represent competition experiments (details are provided in SI Materials and Methods). (E) TrfA was immobilized on the plate, and a mixture of β-clamp (250 nM) and α (increasing concentration) was added. (F) α was immobilized on the plate, and a mixture of β-clamp (250 nM) and TrfA (increasing concentration) was added.

These experiments revealed a network of interactions between TrfA, β-clamp, and α (Fig. 5D). Interestingly, a β-clamp binding motif related to the one found in TrfA is also present in α. To determine if the identified interactions were competitive, we performed two types of ELISA tests. In the first one, wt TrfA was immobilized on the plate and incubated with mixtures composed of β-clamp and α in different molar ratios (Fig. 5E). By using β-specific antiserum, we detected a TrfA–β-clamp complex even when there was an excess of α. For the second ELISA test, α was immobilized on a plate and then incubated with mixtures of β-clamp and wt TrfA in different molar ratios. The interaction of α with β-clamp was detected by β-specific antiserum (Fig. 5F). The measured levels of absorbance signal were the same regardless of the increasing concentrations of the competitor protein TrfA in the mixtures. These data demonstrate that interaction between TrfA and α does not interfere with TrfA and α interactions with β. Indeed, tripartite complexes between TrfA, α, and β could be formed (Fig. 5D). The analyzed multiprotein interactions are even more complex when considering the dimeric form of β. The observed TrfA–α interaction could explain how α becomes a part of the nucleoprotein complex at oriV in the presence of TrfA variants that are not able to interact with and recruit the β-clamp.

RK2 DNA replication is unidirectional and progresses downstream from the iterons. Presumably, this directionality is a consequence of how the replication complex is assembled. We considered the possibility that there was a factor(s) affecting strand specificity of primer synthesis and replisome assembly in RK2. Based on the recently published observations that TrfA interacts with ssDNA containing 13-mers of RK2 DUE (29), we hypothesized that TrfA could be the factor determining replication directionality. We therefore decided to test if TrfA could contribute to strand-specific replisome assembly.

First, we reconstituted in vitro the general priming reactions on three types of circular ssDNA templates. We prepared templates BlueKS(−), oriV top, and oriV bottom with the use of pBluescript KS(−) (Stratagene) (Fig. 6A and SI Materials and Methods). When DnaB, primase, Pol III core subunits, β-clamp, and clamp loader, but neither SSB nor TrfA, were present in the reaction, we detected DNA synthesis on all three templates tested (Fig. 6 A and B). Under these conditions, primase was loaded nonspecifically and efficient DNA synthesis was observed. When SSB was added, this general priming reaction was limited (37) and we observed a substantial reduction of DNA synthesis on all three templates tested; DNA synthesis was reduced from over 750 pmol to ∼140 pmol (Fig. 6B). Under the experimental conditions used, we were not able to reduce this level of synthesis further. When reactions containing SSB were supplemented with ClpX-activated wt TrfA, DNA synthesis was restored up to 400 pmol on the ssDNA template containing oriV bottom strand but not when either the BlueKS(−) or oriV top ssDNA template was used (Fig. 6C). The TrfA variants defective in β-clamp interaction, TrfA F138A and TrfA ΔLF, were not able to initiate DNA synthesis on the oriV bottom-strand template (Fig. 6D). The reaction with wt TrfA required other replication proteins, including DnaB helicase, primase, Pol III core, γδδ′χψ, τ, and β-clamp (Fig. 6E). These results indicated that the replisome complex was assembled on the bottom strand of oriV in a TrfA-dependent mode. The ssDNA template used consisted of the entire oriV bottom strand with iterons and DUE 13-mers. Using SPR with a streptavidin matrix-coated (SA) sensor chip (SI Materials and Methods), we found that wt TrfA, as well as TrfA F138A and TrfA ΔLF, was able to interact with ssDNA containing the sequence of the bottom-strand 13-mers but did not interact with ssDNA containing the top-strand 13-mers (Fig. S7 A–F). Under the same experimental conditions, none of the TrfA variants interacted with bottom-strand iterons and interaction with top-strand iterons was very limited (Fig. S7 G–L). These SPR results, along with the results from our in vitro replication assays on ssDNA templates, demonstrate that through its interaction with 13-mers, TrfA contributes to the strand-specific assembly of the replication complex.

Fig. 6.

TrfA initiates DNA replication on bottom-strand oriV 13-mers. (A) Diagram of the in vitro-reconstituted DNA replication reactions whose results are shown in B–D. The reactions were performed with three types of ssDNA template. BlueKS(−) is ssDNA recovered from E. coli cells containing pBluescript KS(−), oriV top is ssDNA containing the sequence of RK2 oriV top strand, and oriV bottom is ssDNA containing the sequence of RK2 oriV bottom strand. The nonspecific general priming reaction by DnaB helicase and primase on all three templates is depicted at the top of the summary (top arrow). This reaction is inhibited by SSB. TrfA-dependent and template-specific ssDNA replication in the presence of SSB is depicted at the bottom (bottom arrows). (B) SSB inhibits nonspecific DNA synthesis on all three ssDNA templates. (C) SSB-coated oriV bottom ssDNA was replicated at increasing concentrations of ClpX-activated wt TrfA, although neither oriV top nor BlueKS(−) ssDNA template was replicated under the same experimental conditions. (D) Presence of ClpX-activated wt TrfA was essential for replication of SSB-coated ssDNA oriV bottom. TrfA variants defective in interaction with β-clamp did not initiate DNA synthesis on oriV bottom-strand ssDNA template. (E) Requirements for DNA synthesis on SSB-coated ssDNA oriV bottom-strand template were determined using a protein omission approach.

Fig. S7.

WT TrfA and variants interact specifically with ssDNA containing the oriV 13-mers bottom-strand sequence. Formation of the nucleoprotein complex in real time was studied using SPR on an SA sensor chip with the following oriV DNA sequences immobilized on its surface: 13-mers bottom strand (A, C, and E), 13-mers top strand (B, D, and F), iterons bottom strand (G, I, and K), and iterons top strand (H, J, and L). Wild-type TrfA and variants, after ClpX activation, were injected onto the sensor chip at the indicated concentrations.

SI Materials and Methods

Plasmids and Bacterial Strains.

Plasmids and bacterial strains used in this study are listed in Table S1. Plasmids carrying the genes for TrfA ΔLF and TrfAF138A were obtained by PCR-based site-directed mutagenesis of the pATwt trfA template with PfuTurbo (Stratagene) polymerase. Sequences of primers used for mutagenesis were as described (30) for trfAΔLF and 5′-GAAATGCAGCTTT CCTTGGCGGATATTGCGCCGTGG-3′ and 5′-CCACGGCGCAATATCCGCCAAGGAAAG CTGCATTTC-3′ for trfAF138A. Plasmid carrying the trfA ALSMA gene was purchased from GenScript.

Plasmids used for ssDNA template production were pBluescript KS(−) (Stratagene) or phagemid vector derivatives constructed as follows. The 2,314-bp HindIII-XhoI fragment or the 4,347-bp AleI-HindIII fragment from pTJS42 plasmid containing RK2 origin of replication was cloned into pBluescript KS(−). The plasmid containing the HindIII-XhoI DNA fragment was used for ssDNA oriV bottom template production, and the plasmid containing the AleI-HindIII DNA fragment was used for ssDNA oriV top template production. E. coli XL1-Blue cells transformed with pBluescript KS(−) or the two oriV derivatives and VCSM13 helper phage (Stratagene) were used for recovery of ssDNA according to the Stratagene instruction manual.

Proteins, Antibodies, and Reagents.

Commercially available proteins were SSB (Promega) and BSA (fraction V) and creatine kinase (both from Sigma). The β-clamp was fluorescently labeled by AlexaFluor 488 C₅ maleimide fluorescent dye (Invitrogen) according to the manufacturer’s instructions.

Immunodetection assays were conducted with the use of anti–β-clamp, anti-α (mouse) or anti-DnaB (rabbit) sera, goat anti-mouse or goat anti-rabbit HRP-IgG (BioRad), SuperSignal West Pico or a chemiluminescence substrate kit (both from Thermo Fisher Scientific, Inc.).

E. coli DNA Pol III holoenzyme subunits, including core (α, ε, θ), τ, clamp loader complex (γ, δ, δ′, χ, ψ), and β-clamp (51), were a kind gift from M. O’Donnell, The Rockefeller University, New York, NY. The core and τ were mixed at a 1:2 molar ratio, and core–τ complex was used during isolation of β-clamp nucleoprotein complex assays.

E. coli DNA Pol III holoenzyme subunits used in this study were purified in our laboratory using plasmids encoding each individual subunit of the holoenzyme kindly provided by Nicholas Dixon (University of Wollongong, Wollongong, Australia). Proteins were purified as follows.

Purification of the α-subunit was based on the method of Wijffels et al. (52), with modifications using E. coli strain BL21(DE3)/pND517. The ammonium sulfate precipitation step was omitted. After cell lysis and centrifugation, the supernatant was directly loaded onto DEAE-Sepharose resin (GE Healthcare) and a gradient of 0–0.5 M NaCl was applied. Then, phosphocellulose (P11; Whatman) resin was used, and, finally, purified protein was concentrated using 0.5 mL of Hi-Trap Heparin resin (GE Healthcare).

The ε-subunit was overproduced in E. coli strain BL21(DE3)/pSH1017, which was grown in LB at 37 °C containing ampicillin (50 μg/mL). At OD₆₀₀ = 0.8, isopropyl-β-d-thiogalactopyranoside (0.5 mM) was added to induce protein overproduction, and the bacteria were cultured for another 2 h. Further purification steps were performed as described by Scheuermann and Echols (53). In the final step, 0.5 mL of Hi-Trap Heparin resin was used instead of DEAE-Sepharose resin.

The θ-subunit was overproduced in strain BL21(DE3)/pLysS pET-θ and purified as described by Studwell-Vaughan and O’Donnell (54), with the following modifications. The cells were lysed by lysozyme and sonication, and the Q-Sepharose and heparin-agarose resins were omitted.

The τ-subunit was prepared similar to a previously described procedure (55) using strain BL21(DE3)/pJC491. The cells were disrupted first by lysozyme and then by two freeze/thaw cycles with liquid nitrogen. Hi-Trap Heparin was applied instead of heparin–Sepharose.

The purification of γ-subunit was performed as described by Ozawa et al. (56), and the cell lysis of bacterial strain BL21(DE3)/pJC490 was based on the method of Wickner and Kornberg (57).

The subunits δ and δ′ were overproduced using strains BL21(DE3)/pLysS/pET-δ and BL21(DE3) pLysS pET-δ′, respectively, and the cells were lysed with lysozyme and sonication. Further purification steps were essentially as described by Wijffels et al. (52).

The subunits χ and ψ were purified as described by Xiao et al. (58) using strains BL21(DE3)/pLysS pET-χ and BL21(DE3)/pET-ψ. The N-6–linked ATP resin was not applied in the χ preparation and hexylamine-Sepharose resin was not used in ψ purification. After ψ elution from DEAE-Sepharose resin, it was dialyzed to a buffer containing 2 M urea and then immediately diluted fourfold and incubated with χ in 15 °C for 1 h. The newly created χψ complex was separated from unbound proteins by loading it onto a 4-mL column of Q-Sepharose resin (GE Healthcare). The χψ complex was eluted using a linear gradient (20 mL) of 0–0.5 M NaCl in buffer P [50 mM Tris⋅HCl (pH 7.6), 1 mM EDTA, 1 mM DTT, 150 mM NaCl, 20% (vol/vol) glycerol]. Each of the subunits of Pol III holoenzyme was stored in buffer P.

To reconstitute the core, proteins α, ε, and θ were mixed at a 1:1:1 molar ratio. To reconstitute the clamp loader, proteins γ, δ, δ′, and χψ were mixed at a 3:1:1:1 molar ratio. All protein preparations were tested for activity in in vitro replication assays and revealed no differences compared with proteins previously obtained from the laboratory of M. O’Donnell.

In Vivo Replication Activity.

To test the TrfA mutant activity in vivo, E. coli DH5α cells containing pATwt trfA, pATtrfAF138A, or pATtrfAΔLF were transformed with pSV16 plasmid carrying oriV. Transformation was carried out using the standard Bio-Rad electroporation protocol.

In Vitro Replication Assay (FII).

In vitro replication was performed using an E. coli C600 crude extract active for oriV replication. The extract was prepared with the use of (NH4)2SO4 fractionation (0.28g per 1 ml of cell lysate) as described by Kittell and Helinski (50). Reaction mixtures contained 2.3 nM supercoiled pKD19L1 DNA; 40 mM Hepes⋅KOH (pH 8.0); 11 mM magnesium acetate; 2 mM ATP; 500 μM (each) GTP, CTP, and UTP; 50 μM (each) dNTPs; 150 cpm⋅pmol⁻¹ [methyl-³H]-dTTP; 80 mM creatine phosphate; 5% (wt/vol) PEG; 50 mM KCl; 16 μg/mL creatine kinase; 80 μg/mL BSA, and increasing concentrations of TrfA variants (0, 30, 60, 90, 120, 150, 210, and 300 nM). Mixtures were assembled on ice, followed by incubation at 32 °C for 1.5 h, which was stopped by addition of 0.1 M sodium pyrophosphate and 10% (wt/vol) TCA. Samples were filtrated onto Whatman GF/C glass microfiber filters, and total nucleotide incorporation (picomoles) was measured by liquid scintillation counting.

In Vitro DNA Replication Reaction Reconstituted with Purified Proteins (Reconstituted System).

In vitro DNA replication reaction was reconstituted with the use of purified proteins as described by Konieczny and Helinski (28). Reaction mixtures contained 2.3 nM supercoiled pKD19L1 DNA; 40 mM Hepes⋅KOH (pH 8.0); 25 mM Tris⋅HCl (pH 7.4); 4% (wt/vol) sucrose; 4 mM DTT; 11 mM magnesium acetate; 2 mM ATP; 500 μM (each) GTP, CTP, and UTP; 50 μM (each) dNTPs; 150 cpm⋅pmol⁻¹ [methyl-³H]-dTTP; 80 mM creatine phosphate; 16 μg/mL creatine kinase; and 80 μg/mL BSA, as well as DnaA (450 nM), DnaB (200 nM), DnaC (1.1 μM), gyrase (13 nM), HU (110 nM), SSB (115 nM), primase (60 nM), E. coli Pol III core subunits (αεθ, 7 nM), τ-subunit (17 nM), β-clamp (17 nM), γδδ′χψ (2.5 nM), and increasing concentrations of TrfA variants (0, 30, 60, 90, 120, 150, 210, and 300 nM). Mixtures were assembled on ice, followed by incubation at 32 °C for 1.5 h, which was stopped by addition of 0.1 M sodium pyrophosphate and 10% (wt/vol) TCA. Samples were filtrated onto Whatman GF/C glass microfiber filters, and total nucleotide incorporation (picomoles) was measured by liquid scintillation counting.

ClpX-Dependent Activation of TrfA.

The wt TrfA or variants (120 nM) were incubated with 870 nM ClpX at 32 °C for 10 min in buffer containing 20 mM Tris⋅HCl (pH 8.0), 40 mM Hepes⋅KOH (pH 8.0), 4% (wt/vol) sucrose, 4 mM DTT, 11 mM magnesium acetate, 4 mM ATP, and 80 μg/mL BSA. After incubation, the reaction mixture was diluted in the same buffer to obtain the concentration of wt TrfA or variants required for the subsequent experiments.

SPR Analysis.

Standard SPR analyses using BIAcore 2000 (GE Healthcare) were performed according to the manufacturer’s manual. Protein interactions were determined with use of a CM5 sensor chip. The wt TrfA and variants were immobilized with use of the amine coupling method on a carboxymethylated dextran matrix in 10 mM sodium acetate (pH 4.0) to yield a final value of ∼7,000 response units (RU).

A streptavidin matrix-coated SA sensor chip was used during DNA–protein interaction analysis. The 5′-terminally biotinylated ssDNA oligonucleotides containing the sequence of the top or bottom strand of the RK2 oriV iterons or 13-mers were immobilized on the chip surface to yield a final value of ∼80 RU. The sequences of commercially synthesized oligonucleotides (Thermo Scientific) used were as follows:

• 13-mers top strand: 5′-CCA CCG CTA ACC TGT CTT TTA ACC TGC TTT TAA ACC AAT ATT TAT AAA CCT TGT TTT TAA CCA GGG CTG C-3′

• 13-mers bottom strand: 5′-GCA GCC CTG GTT AAA AAC AAG GTT TAT AAA TAT TGG TTT AAA AGC AGG TTA AAA GAC AGG TTA GCG GTG G-3′

• Iterons bottom strand: 5′-TGT GGA CAG CCC CTC AAA TGT CAA TAG GTG CGC CCC TCA TCT GTC ATC ACT CTG CCC CTC AAG TGT CAA GGA TCG CGC CCC TCA TCT GTC AGT AGT CGC GCC CCT CAA GTG TCA ATA CCG-3′

• Iterons top strand: 5′-CGG TAT TGA CAC TTG AGG GGC GCG ACT ACT GAC AGA TGA GGG GCG CGA TCC TTG ACA CTT GAG GGG CAG AGT GAT GAC AGA TGA GGG GCG CAC CTA TTG ACA TTT GAG GGG CTG TCC ACA-3′

In all SPR experiments, HBS-EP [150 mM NaCl, 10 mM Hepes (pH 7.4), 3 mM EDTA, 0.005% Surfactant P20] was used as running buffer. Buffer flow was set to 15 μL⋅min⁻¹, and each injection had a volume of 30 μL. The surfaces of sensors were regenerated with the use of 20 μL of 10 mM Gly (pH 9.5) solution for CM5 and 5 μL of 0.1% (wt/vol) SDS solution for SA. The results are presented as sensograms obtained after subtraction of the background response signal obtained in control experiments with buffer injections.

DNA Probes Preparation and GMSA.

DNA probes were prepared by DNA labeling with Alexa 555-dCTP (Invitrogen) and terminal deoxynucleotide transferase (Promega) as previously described for DNA fragments labeled with Cy3-dCTP (59). The dsDNA fragments containing minimal oriV region (composed of the DnaA boxes, the iterons, and the AT-rich region) were prepared using PCR with specific primers. The sequences of primers used in PCR reactions for DNA preparation were 5′-AAG CCG TGT GCG AGA CAC CGC-3′ and 5′-AAA GAC AGG TTA GCG GTG GCCG-3′. Reaction mixtures containing the fluorescent DNA probe and increasing amounts of TrfA ClpX-activated variants in a buffer containing 40 mM Hepes⋅KOH (pH 8.0), 25 mM Tris⋅HCl (pH 7.6), 4% (wt/vol) sucrose, 4 mM DTT, and 80 μg/mL BSA were incubated for 20 min at 32 °C. After incubation, Ficoll 4000 was added to 2.5% (vol/vol) and reactions were loaded onto a 5% polyacrylamide gel. Gels were prepared in Tris⋅borate/EDTA buffer, and they were scanned with ChemiDoc MP (Bio-Rad) after electrophoresis.

Helicase Activity Assay (FI*).

RK2 supercoiled minireplicon pKD19L1 (2.3 nM) was used as the DNA template. After the ClpX-dependent activation of TrfA variants, the mixture containing proteins [DnaA (7.5 nM), DnaB (200 nM), DnaC (1.1 μM), gyrase (13 nM), HU (110 nM), SSB (115 nM)], 20 mM Tris⋅HCl (pH 8.0), 40 mM Hepes⋅KOH (pH 8.0), 4% (wt/vol) sucrose, 4 mM DTT, 11 mM magnesium acetate, 4 mM ATP, and 80 μg/mL BSA was added as previously described (28). After a 20-min incubation at 32 °C, the reactions were analyzed on agarose gel stained with ethidium bromide solution.

Isolation of Nucleoprotein Complexes by Gel Filtration.

The column gel filtration method with Sepharose CL-4B resin (GE Healthcare) was used to isolate nucleoprotein complexes formed on the supercoiled pKD19L1 DNA containing RK2 oriV template. The column buffer was composed of 40 mM Hepes⋅KOH (pH 8.0), 40 mM potassium glutamate, 4% (wt/vol) sucrose, 4 mM DTT, 10 mM magnesium acetate, 2 mM ATP, 0.01% Brij-58, and 50 μg/mL BSA.

A. Isolation of DnaB helicase nucleoprotein complex.

The wt TrfA or variants (120 nM) were activated by ClpX, and 200 nM DnaB, 7.5 nM DnaA, 1.1 μM DnaC, and 2.3 nM supercoiled pKD19L1 DNA were then added. The reaction buffer used was 20 mM Tris⋅HCl (pH 8.0), 40 mM Hepes⋅KOH (pH 8.0), 4% (wt/vol) sucrose, 4 mM DTT, 11 mM magnesium acetate, 4 mM ATP, and 80 μg/mL BSA. After 20 min of incubation at 32 °C, the mixture was applied on a CL-4B column (0.5 × 9 cm) in column buffer. Collected fractions were analyzed by SDS/PAGE, followed by Western blotting for the presence of DnaB with specific antiserum.

B. Isolation of β-clamp nucleoprotein complex in reconstituted system.

The wt TrfA or variants (120 nM) were activated by ClpX, and DnaA (450 nM), DnaB (200 nM), DnaC (1.1 μM), gyrase (13 nM), HU (110 nM), SSB (115 nM), primase (60 nM), and 2.3 nM supercoiled pKD19L1 DNA were then added (unless indicated otherwise). After 20 min of incubation at 32 °C to form a helicase complex, β-clamp labeled with Alexa 488 fluorescent dye (17 nM) and E. coli Pol III core τ-subunits [αεθτ (19 nM), and γδδ′χψ (2.5 nM)] was added. After 15 s, the reaction was quenched by adding 5 units of hexokinase to remove ATP (51), followed by gel filtration on a Sepharose CL-4B column. The collected fractions (40 μL) were analyzed for the presence of the Alexa 488 β-clamp by measuring fluorescence and for the presence of DNA by agarose electrophoresis. The reaction buffer used was 20 mM Tris⋅HCl (pH 8.0), 40 mM Hepes⋅KOH (pH 8.0), 4% (wt/vol) sucrose, 4 mM DTT, 11 mM magnesium acetate, 4 mM ATP, and 80 μg/mL BSA.

C. Isolation of α-nucleoprotein complex in the crude cell extract (FII).

The TrfA activation step was omitted because chaperones were present in the crude cell extract used. Reaction mixtures were prepared as for in vitro replication but scaled up 10-fold, and 120 nM TrfA variants were used. After cross-linking using 0.01% formaldehyde, mixtures were applied to a CL-4B column and collected fractions were analyzed by immune detection for α-subunit and by agarose electrophoresis for the presence of DNA.

ELISA.

Proteins [100 μg/mL BSA (negative control), wt TrfA, TrfAF138A, TrfA ΔLF, or α (0.15 μM)] were immobilized on the ELISA plate (Costar) in PBS buffer composed of 137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, and 1.8 mM KH₂PO₄ (pH 7.2) for 1 h. Unbound proteins were removed with two washes (200 μL each) of PBS containing 0.2% (wt/vol) BSA; a third wash was incubated for 1 h before removal. The wells were then washed with 200 μL of buffer B containing 25 mM Hepes⋅KOH (pH 7.6), 150 mM KCl, 25 mM NaCl, 1 mM DTT, 10 mM MgCl₂, 2.5% (vol/vol) glycerol, 0.1 mM EDTA, 0.05% Triton X-100, and 0.2% (wt/vol) BSA. Increasing concentrations of α (0, 0.15, 0.3, 0.45, and 0.6 μM) or β-clamp (0, 0.25, 0.5, 1.0, and 2.0 μM) in buffer B were then incubated with the immobilized proteins for 0.5 h, unbound proteins were washed away, and the relative amount of bound β-clamp or α was detected using an immunoenzymatic assay with specific polyclonal antibodies.

In the competition ELISA, wt TrfA or α was immobilized on the plate as described above, and separate mixtures (50 μL) containing β-clamp (250 nM) and various concentrations of α or wt TrfA (0, 8, 16, 32, 63, 125, 250, and 500 nM) were then added. The subsequent steps were as described above.

In Vitro ssDNA Replication Reaction.

The priming reaction was carried out with the use of circular ssDNA templates [BlueKS(−), oriVtop, or oriVbottom). Reaction mixtures contained one of the ssDNA templates (2.3 nM), 40 mM Hepes⋅KOH (pH 8.0); 25 mM Tris⋅HCl (pH 7.4); 4% (wt/vol) sucrose; 4 mM DTT; 11 mM magnesium acetate; 2 mM ATP; 500 μM (each) GTP, CTP, and UTP; 50 μM (each) dNTPs; 150 cpm⋅pmol⁻¹ [methyl-³H]-dTTP; 80 mM creatine phosphate; 16 μg/mL creatine kinase; 80 μg/mL BSA and DnaB (200 nM); primase (60 nM); E. coli Pol III core subunits (αεθ, 7 nM); τ-subunit (17 nM); β-clamp (17 nM); and γδδ′χψ (2.5 nM). To assay the general priming reaction inhibited by SSB, increasing concentrations of SSB were used (0, 0.5, 1.0, 1.5, and 2 μM). To assay SSB-coated oriV bottom ssDNA replication, increasing concentrations of ClpX-activated TrfA variants (0, 80, 160, 200, 250, 330, and 420 nM) were used. Mixtures were assembled on ice, followed by incubation at 32 °C for 0.5 h, which was then stopped by addition of 0.1 M sodium pyrophosphate and 10% (wt/vol) TCA. Samples were filtrated onto Whatman GF/C glass microfiber filters, and total nucleotide incorporation (picomoles) was measured by liquid scintillation counting.

ATPase Activity Assay.

The activity of clamp loader in ATP hydrolysis was tested by coupled enzyme ATPase assay. The ATPase activity is coupled to the oxidation of NADH, which is monitored as a decrease in A at 340 nm over time, as described previously (60). Reaction buffer contained 40 mM Hepes⋅KOH (pH 8.0), 25 mM Tris⋅HCl (pH 7.4), 4% (wt/vol) sucrose, 4 mM DTT, and 11 mM magnesium acetate. Protein concentrations were τ (17 nM), β-clamp (17 nM), and preassembled γδδ′χψ (2.5 nM). Each experiment was performed in the presence of primed DNA Blue KS(−) (2.3 nM). The Blue KS(−) was primed with a 30-mer oligonucleotide (5′-GGGTTTTCCCAGTCACGACGTTGTAAAACG-3′) as described by Studwell and O’Donnell (61). The ATP regeneration system (6.2 mg/mL phosphoenolpyruvate phosphatase, 10.5 mg/mL NADH, 2 U/mL pyruvate kinase, 10 U/mL lactic dehydrogenase) was preincubated with 2 mM ATP at 32 °C for 5 min to convert any ADP present to ATP. Simultaneously, proteins were mixed on ice with primed DNA and added to the ATP regeneration system, and the reduction of NADH A at 340 nm was measured instantly over 5 min at 32 °C.

Discussion

Our results show that interaction of the RK2 plasmid Rep TrfA with the β-clamp is essential for the formation of a fully active replisome at the origin of supercoiled plasmid DNA. Although we were able to demonstrate a TrfA and β-clamp protein–protein interaction, we failed to demonstrate a tripartite complex consisting of the β-clamp and TrfA bound to iterons containing a linear dsDNA fragment. Instead, we showed β-clamp association with supercoiled plasmid DNA in the presence of other factors, including TrfA. Our experiments with supercoiled dsDNA templates did not distinguish between β-clamp loading on the leading and lagging strands. However, because we investigated origin-specific replication initiation completed with de novo assembly of the replication complex, the detection of the β-clamp complex on supercoiled DNA indicates the assembly process. β-Clamp complex formation on the template requires not only TrfA with an intact QLSLF motif, which is necessary for the protein to interact with the β-clamp, but also other replication proteins, including DnaA, DnaB, primase, and clamp loader. The Pol III core, however, was not required, and the absence of gyrase reduced β-clamp complex formation only slightly. The absence of DnaA in this complex results in an inefficient origin opening by TrfA (26); disturbances in helicase recruitment (28); and, as we demonstrate in this work, defects in β-clamp delivery on plasmid DNA. These observations support an indirect involvement of DnaA in replisome assembly. However, we cannot exclude the possibility that DnaA, through its contact with Hda protein (38), which is known to interact with DNA-loaded β-clamp (39), has a more direct effect on the β-clamp complex at the replication origin.

DnaB helicase plays an important part in this nucleoprotein puzzle. It contributes to replisome assembly through its interaction with τ (12) and primase (13), as well as by the extension of an opened DUE. The extended ssDNA template serves as the site for replisome assembly and activity. We have demonstrated that the absence of DnaB results in a disturbance in β-clamp delivery on the plasmid DNA. Although helicase provides a structural scaffold for replisome assembly, primase must be present as well. Primase synthesizes RNA primers that are required for β-clamp and subsequent core subunit interaction with the primed DNA template (14). When nucleotides were omitted from reaction mixtures, we still detected a β-clamp complex on supercoiled plasmid DNA (Fig. S5 A–D), which is most likely being due to primase using dNTPs (40, 41) or ATP (42) in the reaction mixture for primer synthesis. Due to DnaB and DnaC requirements for ATP, we could not omit ATP from reaction mixtures.

We demonstrate that the clamp loader and TrfA with the intact QLSLF motif for β-interaction are both indispensable for β-clamp complex assembly on RK2 DNA. In our in vitro replication assay, both τ and γ can assemble into functional clamp loader; however, when τ was omitted, the level of DNA synthesis was reduced (Fig. S6A). A similar effect was observed during experiments using ssDNA template (Fig. 6). Most likely, due to interaction with α, τ provides stability of the clamp loader at the replication fork or allows lagging strand synthesis when dsDNA is used as a template. It is intriguing to imagine the exact role of TrfA in β-clamp complex formation at oriV. By performing an ATPase assay, we excluded the possibility that TrfA affects clamp loader ATPase activity (Fig. S6B). One possible mechanism for β-complex formation at oriV is that TrfA interacts with β-clamp and hands off the β-clamp to the clamp loader following primer synthesis by primase. The clamp loader is responsible for β-clamp opening and loading it onto the primed DNA. The QLSLF motif in TrfA is located in the proximity of the WH domains responsible for the protein’s interaction with DNA (43). It is worth noting that changing the QLSLF motif into ALSMA also results in a replicatively inactive TrfA (Fig. S8). Changing the QTSMAF motif into ATSMAF in ε-subunit of Pol III core prevents interaction with the β-clamp (44). Experiments presented in this work demonstrate that the F138A substitution affects TrfA interaction with the β-clamp but also somehow affects helicase activity at oriV (Fig. 2). It is difficult to speculate on the nature of this dual effect. TrfA interacts with DnaB (27, 45); however, the interacting motifs have not been characterized. The F138A substitution might have altered TrfA’s secondary structure, thus influencing interaction with DnaB.

Fig. S8.

TrfA ALSMA is not active in RK2 DNA replication in vitro or in vivo. (A) In vitro replication with a crude extract (FII) prepared from E. coli C600 with increasing concentrations of TrfA variants (0, 30, 60, 90, 120, 150, 210, and 300 nM; results presented from n = 3 replicates; SD shown). (B) TrfA in vivo activity test. Plasmid pSV16 containing oriV was used to transform E. coli DH5α cells with plasmids carrying variants of trfA as indicated. Plasmid pBBR1MCS2, replicating independent of the TrfA, was used instead of pSV16 as a control. The transformation frequency is reported as the number of colony-forming units per 1 μg of supercoiled plasmid pSV16 DNA averaged from three independent experiments, with the SD shown.

We have revealed the interaction of TrfA with Pol III α-subunit; however, the relevance of this interaction is not clear. Because a nucleoprotein complex containing the β-clamp was observed even in the absence of core, the TrfA interaction with α must not be critical to the assembly of β-clamp on oriV DNA. TrfA mutants in the QLSLF motif are able to interact with α, and in the presence of those mutants, the α-complex was observed on the plasmid DNA. It must be pointed out that TrfA ΔLF, which is deficient in β-complex formation at oriV, does not promote DNA synthesis. Those findings imply that β-loading is essential for proper assembly of active Pol III core at oriV. We could assume that TrfA interaction with α is required for α-recruitment, although other proteins (i.e., τ associated with DnaB) could also be involved in this process. The specific motif in TrfA responsible for its interaction with α needs to be identified. Because RK2 is a broad-host-range plasmid, it is of great importance to verify if the TrfA interacts with α-subunits of Pol III holoenzymes from other gram-negative bacteria.

The experiments that used supercoiled dsDNA templates demonstrate that TrfA protein–protein interactions contribute to the assembly of the replication complex at the plasmid origin, whereas the reconstitution of a strand-specific, 13-mer, and TrfA-dependent in vitro DNA replication reaction on ssDNA templates reveals how the direction of replication is determined. It is clear that assembly of replication machinery on the top or bottom, or on both the top and bottom, strands of melted DUE allows, respectively, unidirectional or bidirectional DNA replication. It is not clear, however, what predisposes or determines the strand specificity of replisome assembly. It is assumed that helicase/primase complex assembly and strand-specific primer synthesis are key factors in this process; however, before this work, it remained uncertain as to what determined the strand specificity of helicase/primase loading. The reconstitution of RK2 strand-specific DNA replication reactions on ssDNA templates demonstrated that the TrfA interaction with DUE bottom-strand 13-mers and the protein’s interaction with the polymerase β-clamp contribute to anchoring of the replisome on the bottom strand of the plasmid origin. We were able to detect DNA synthesis only with wt TrfA, but not TrfA mutants, deficient in β-clamp interaction, and only with ssDNA templates containing the oriV bottom strand with 13-mer sequences. Our data indicated that the replisome complex was assembled in a strand-dependent and TrfA-dependent mode and requires DnaB helicase and primase. We have previously shown that TrfA interacts with DnaB (45) and loads it onto oriV DUE (27). When using ssDNA templates, TrfA is not required for DUE opening; rather, it is required for loading, the helicase/primase complex, and replisome assembly. In our experiments, a replication complex containing Pol III core for leading strand synthesis was formed on the bottom strand, consistent with the previously determined direction of RK2 plasmid replication (26, 46).

It is of great interest to determine if a similar interaction network exists in other plasmids. It would be of interest to determine if RK2 as a broad-host-range replicon uses the same network of interactions in bacterial hosts other than E. coli. It is very likely that the recruitment and assembly of the replisome at the replication origin of an extrachromosomal genetic element requires specific mechanisms that differ from the mechanisms used at the bacterial chromosomal origin. However, Rep-dependent interactions with specific Pol III holoenzyme subunits and with DUE, as described in this work, should be considered as an attractive model for strand-specific recruitment of replication machinery to any origin, and thereby determination of replication directionality.

Materials and Methods

A detailed list of plasmids and bacterial strains used in this study is presented in Table S1. Protocols and bacterial strains used for the purification of E. coli gyrase and Pol III subunits were kindly provided by N. Dixon (University of Wollongong, Wollongong, Australia). Protocols and bacterial strains used for the purification of E. coli primase and HU were kindly provided by R. McMacken (Johns Hopkins University, Baltimore, MD). The protocol and bacterial strain used for the purification of the E. coli β-clamp were kindly provided by D. Bastia (Medical University of South Carolina, Charleston, SC). E. coli proteins DnaA-His6, DnaB-His6, DnaC, and ClpX were purified as described (43, 47–49). All TrfA preparations used in the tests were N-terminally histidine-tagged 33-kDa versions of TrfA. The wt TrfA, TrfA F138A, and TrfA ΔLF were purified according to the method of Toukdarian et al. (33). E. coli DNA Pol III subunits were a kind gift from M. O’Donnell, The Rockefeller University, New York, NY. Detailed information on the commercially available proteins, antibodies, and reagents used in this study is given in SI Materials and Methods.

An in vitro replication assay (FII) was performed as described by Kittell and Helinski (50), ClpX-dependent activation of TrfA was performed according to the method of Pierechod et al. (43), and ELISA analyses were performed as described previously (36), with details given in SI Materials and Methods. Detailed protocols for in vivo replication activity, isolation of nucleoprotein complexes by gel filtration, helicase activity assay (FI*), SPR, and GMSA analysis are given in SI Materials and Methods.