Two forms of replication initiator protein: Positive and negative controls

Abstract

The pir gene of plasmid R6K encodes the protein, π, a replication and transcription factor. Two translational options for the pir gene give rise to two forms of π protein: a 35.0-kDa form (π^35.0) and a shortened 30.5-kDa form (π^30.5). Although both proteins bind to a series of 22-bp direct repeats essential for plasmid R6K replication, only π^35.0 can bind to a site in the (A⋅T)-rich segment of its γ ori and activate the γ ori in vivo and in vitro. However, unlike π^35.0, π^30.5can inhibit in vivo and in vitro replication (activated by π^35.0). We propose that the two forms of π might have distinct functions in replication. We show that although both forms of π produce dimers, the nature of these dimers is not identical. The N-terminal 37 amino acid residues appear to control the formation of the more stable π^35.0 dimers, whereas another, apparently weaker interface holds together dimers of π^30.5. We speculate that the leucine zipper-like motif, absent in π^30.5, controls very specific functions of π protein.

The 35.0-kDa π protein (π^35.0) of plasmid R6K is a replication and transcription factor that binds to seven 22-bp direct repeats (DRs) in the γ origin (γ ori) replication region (1–7). π also binds independently to another site in the (A⋅T)-rich segment of the γ ori whose sequence differs from that of the DR (8). Both the cluster of DRs and the (A⋅T)-rich segment are essential elements for γ ori core function (9–11).

In addition to interacting with two families of DNA sequences in the γ ori core, π has the ability to interact with other π molecules (oligomerize). Cooperative oligomerization occurs when π binds to seven tandem DRs (12, 13). Another type of oligomerization occurs when π binds distant sites on DNA molecules either intra- or inter-molecularly (14, 15).

π has been shown to form stable dimers in solution (3, 16, 17); the stability of these dimers is determined by the N terminus of the protein (17). This region contains a sequence similar to the leucine zipper (LZ) motif (see Fig. 1C), which is known to control homo- and hetero-dimerization of transcription and replication factors (18–20). The motif itself consists of 4 to 5 heptad repeats of leucine residues (18). The residues of the LZ adopt an α-helical coiled coil structure to form a parallel (head-to-head) dimer (21–25). It is of fundamental interest to determine whether the LZ-like motif contributes in any significant way to the structural and biochemical properties of π.

Figure 1.

Structure of the γ ori of plasmid R6K and properties of the N terminus of π protein. (A) The R6K γ ori. Two components required for replication are indicated: the cis-acting γ ori and pir gene that encodes trans-acting π protein. The following segments/binding sites of the γ ori are indicated: seven tandem arrowheads indicate 22-bp DRs, (A⋅T)-rich region is labeled, with asterisks indicating borders of the π binding site in the (A⋅T)-rich region. (B) The N-terminal sequence of the pir gene and multiple translational options. The putative ribosome binding sites are underlined. Five AUG start codons (shown at the level of DNA sequence) are boxed and numbered. 35.0-kDa π (π^35.0) and 30.5-kDa π (π^30.5) are translated as shown. Underlined CTA triplet (DNA sequence) corresponds to the first leucine residue of LZ-like motif (C) N-terminal sequence of π protein. Two AUG start codons (36 and 38) in pir gene are boxed. Sequences are aligned to show homology to the LZ motif of eukaryotic and prokaryotic transcription and replication factors. π^30.5 lacks over half of the LZ-like motif. Although this putative dimerization domain resembles a leucine zipper (LZ), we refer to this sequence as LZ-like because the presence of four Pro residues, as α helix breakers, violates the LZ match.

In the case of R6K, two translational options for the pir gene give rise to two forms of π protein: the abundant 35.0-kDa species (π^35.0) and a minor 30.5-kDa form (π^30.5) that lacks the LZ-like motif (ref. 26; see Fig. 1 B and C). We hypothesize that multiple π functions may be partitioned between these two forms of π protein. In this paper, we show that these two π polypeptides have indeed distinct properties in vivo and in vitro.

MATERIALS AND METHODS

Plasmid Constructions.

The following plasmids have been described: pRK526 (27), pMF36, Δ22 (1), pFW25a (28), pACYC184 (29), pGP1–2 and pT7–7 (30, 31), pMF34, pPT39 (3), and pMF239 (32). The wild-type pir gene was inserted into the pGEM3zf(+) vector (Promega) to obtain plasmid pMS4.1. Plasmid construction for overexpression of π^30.5 is described in Fig. 2A. Two distinct clones were identified by sequencing the entire pir gene. In the clone used for π^30.5 protein purification, translation begins at ATG38 (pMS7.4; see Fig. 2B); in a second clone (not used in these experiments) translation starts at ATG36 (pMS7.5). pMS7.4 was digested with NdeI–BamHI and ligated with pET21-b (Novagen) digested with the same enzymes, yielding the construct pJW11.

Figure 2.

Plasmid construction and protein purification. (A) The construction of a vector for overexpression of π^30.5. pMS4.1, containing the wt pir gene, was used as the template in PCR amplification. The T7 primer is complementary to the T7 promoter sequence on pMS4.1. Primer1 has an internal NdeI site upstream of ATG36. The amplified DNA fragment was cut with NdeI–BamHI and inserted into pT7–7 vector, yielding plasmids pMS7.4 and pMS7.5 (constructed by M.S. in the laboratory of Dr. W. Szybalski; not shown). As a result, the synthesis of π^30.5 was under the control of the T7 promoter and RBS. (B) Key sequences in plasmid construction. The wild-type pir sequence at the N terminus and primer1 sequence are indicated. The corresponding sequences of two constructs obtained are given. pMS7.5 was the expected product. Plasmid pMS7.4, however, was the first construct identified and was used for the purification of π^30.5 protein. (C) SDS/PAGE shows the overexpression and purification of π protein. Lanes: 1, purified π^35.0; 2, purified π^30.5; 3 and 4, total cell protein from DH5α strain containing pMS7.4/pGP1–2 before and after heat induction, respectively. Purified π^35.0 and π^30.5 proteins, as shown, were used in this study.

Protein Purification.

π^35.0 protein was purified from an Escherichia coli strain C600 carrying plasmid pPT39; π^30.5 was purified from strain DH5α carrying plasmids pMS7.4 and pGP1–2. Both proteins were purified from inclusion bodies according to ref. 33.

Treatment of π Proteins with Gu⋅HCl and Cross-linking.

The protein samples were diluted to 0.1 mg/ml in TGED buffer (10 mM Tris⋅Cl, pH 8.0/0.1 mM EDTA⋅Na₂/0.1 mM DTT/5% glycerol/0.15 M KCl) in the presence of Gu⋅HCl (0–0.7 M). Samples were incubated at room temperature (RT) for 30 min and then frozen (−20°C). Thawed, Gu⋅HCl-treated protein was used for electrphoretic mobility-shift assay (EMSA) and for cross-linking. π proteins (2 μg) were added to 80 μl of cross-linking buffer (0.15 M NaCl/1 mM MgCl₂/10 mM KPO₄ buffer, pH 7.4) followed by the addition of 2 μl of 2 mM bis[2-(succinimindo-oxycarbonyloxy)ethyl]sulfone (BSOCOES; Pierce). Reactions were stopped after 15 min, RT incubation by the addition of 5 μl of 1 M NH₄AC. Samples were precipitated (5% TCA, 2.5 μg/ml of BSA carrier) and protein was detected by immunoblotting.

In Vitro Replication Assay.

Replication extracts were prepared from E. coli strain C600. The R6K derivative pMF36 was used as a template in all π-dependent replication assays. Preparation of cell extracts and in vitro replication assays were carried out as described (34, 35).

In Vivo Assays for γ ori Replication and π Synthesis.

Strain BL21(DE3) carrying plasmids indicated in the legend to Fig. 4 was grown in M63 complete medium (0.5% glucose, 1% casamino acids mix) supplemented with and 2′-deoxyadenosine (250 μg/ml) and appropriate antibiotics. At OD₆₀₀ = 0.5, isopropyl β-d-thiogalactoside (IPTG) (0.1 mM) and [¹⁴C]thymidine (1 μCi/ml; 1 Ci = 37 GBq) were added and cells were harvested at various times to determine the levels of newly synthesized DNA and π protein (36). Plasmids (DNA) were isolated by the miniprep procedure (Wizard Miniprep kit; Promega), linearized by SalI digestion, separated on 0.7% agarose gel, and visualized/quantified by PhosphorImager (Molecular Dynamics). The π protein was detected immunologically (1) using Enhanced Chemiluminesence Western blotting detection kit (Amersham).

Figure 4.

The in vivo inhibitory effect of π^30.5 on π^35.0-dependent replication of a γ ori plasmid. (A) In vivo replication of three plasmids harbored by strain BL21(DE3) was determined by using [¹⁴C]-thymidine pulse labeling. The isolated plasmids DNA were digested with SalI, which linearizes pJW11 and pFW25a and cuts Δ22 twice. The four DNA fragments corresponding to each plasmid are indicated. Levels of de novo plasmid DNA replication were quantified at the times indicated as described. The relative amount of radiaoactivity found in pFW25a was normalized against radioactivity incorporated into Δ22 and pJW11. The sample without IPTG was taken as 100%. (B) Immunoassay showing the relative protein expression time course of π^35.0 (constitutive) and π^30.5 (IPTG-inducible). Blots were scanned with a Color OneScanner 600/27 (Apple) and quantified using the ImageQuaNT program (Molecular Dynamics) to determine levels of π^30.5 relative to π^35.0. The quantified π^30.5/π^35.0 ratios are shown at the bottom.

RESULTS

Overproduction and Purification of π^30.5.

We have previously shown that in addition to the major species of π (35.0 kDa), translation of another species (30.5 kDa) starts at codons 36 and/or 38 of wild-type pir mRNA (26). This π^30.5 protein is produced at 10- to 15-fold lower levels than π^35.0, probably because ATG36 and ATG38 are preceded by poor ribosome binding sites (refs. 26 and 37; Fig. 1B).

To overproduce π^30.5 we constructed a pir gene mutant with an efficient ribosome binding site placed upstream of both ATG36 and ATG38 (Fig. 2 and Materials and Methods). We isolated a construct (pMS7.4) in which translation begins at ATG38. Both π^30.5 and π^35.0 purified from inclusion bodies were more than 95% pure as determined by SDS/PAGE (Fig. 2C). We note here that in some experiments, the binding of π^35.0 purified from either inclusion bodies or from soluble fractions was examined with no noticeable differences between the proteins (unpublished data).

π^30.5 Protein Lacks a Replication Activator Function in Vivo and in Vitro.

Our first goal was to determine if π^30.5 remains active as an initiator of replication in vivo. To test this, we used a γ ori plasmid, pRK526, marked with kanamycin resistance (Km^r). Maintenance of this plasmid requires π protein (27). The pRK526 was transformed into strain BL21(DE3) containing pJW11, a plasmid expressing π^30.5 in an IPTG inducible manner, and then cells were plated on Luria–Bertani medium supplemented with kanamycin and increasing amounts of IPTG. We did not observe any Km^r colonies in this transformation experiment (data not shown). Thus, π^30.5 is unable to activate the γ ori in vivo.

We also tested the activator function of π^30.5 in vitro (35, 38) by measuring the incorporation of [³H]dTMP into newly synthesized DNA using the γ ori plasmid, pMF36, as a template. In the control replication reactions that used π^35.0 as an initiator, isotope incorporation was observed (Fig. 3A); however, no isotope incorporation was observed at any level of π^30.5 protein examined (Fig. 3A). These results indicate that π^30.5 lacks the initiator activity under conditions in which π^35.0 activates replication of γ ori.

Figure 3.

The effect of π^30.5 on γ ori as determined by the in vitro replication assay. (A) In vitro replication of γ ori in the presence of π^35.0 and π^30.5. Incorporation of [³H]dTMP into γ ori templates was determined in the presence of increasing concentrations of π protein. (B) The in vitro inhibitory effect of π^30.5 on π^35.0-dependent replication at γ ori.

π^30.5 Inhibits Replication Dependent on π^35.0 in Vivo and in Vitro.

We tested whether π^30.5 can inhibit replication in vitro. This experiment was carried out in the presence of π^35.0 at two different levels, either 0.4 μg or 2.0 μg for each reaction. As seen in Fig. 3B, π^30.5 clearly inhibited replication activated by π^35.0.

Next, we examined the inhibitory function of π^30.5 in vivo by monitoring de novo plasmid replication by pulse labeling of the plasmid DNA with [¹⁴C]thymidine (Fig. 4A) after different time periods of protein induction (Fig. 4B). In this experiment, plasmid Δ22 (Tet-r) provides the π^35.0 necessary for the replication of the γ ori plasmid, pFW25a (Cm-r). π^30.5 was then introduced using an expression construct, pJW11 (Pen-r), which allowed more stringent control of π^30.5 synthesis than observed with pMS7.4. Because both pJW11 and Δ22 replicate in a π-independent manner, they served as internal controls to normalize the effect of π^30.5 on replication of the π-dependent γ ori (pFW25a). As shown in Fig. 4, even after the shortest time period of IPTG induction, π^30.5 specifically inhibits replication of plasmid FW25a but not plasmids Δ22 and pJW11. Importantly, when π^30.5 was expressed at 7% of π^35.0 level, the replication of a γ ori plasmid decreased to 58% of the control.

In a second in vivo experiment, we measured the effect of π^30.5 on the establishment/maintenance of a γ ori plasmid. Plasmid pFW25a (γ ori) was transformed into a recipient strain [BL21(DE3)] containing the plasmids encoding π^35.0 (Δ22) and IPTG inducible π^30.5 (pJW11). Cells were plated onto a selective medium containing increasing concentrations of IPTG. Again, an inhibitory effect was observed for plasmid pFW25a but there was little to no effect on the establishment of pACYC184 (Table 1). Taken together, these experiments establish a replication inhibitory function for π^30.5.

Table 1.

Inhibitory effect of π^30.5 on the establishment of γ ori plasmid following transformation

IPTG, μM	Host transformed/μg DNA
	pACYC184	pFW25a
0	4.2  × 103	6.3  × 102
25	4.2  × 103	6.7  × 102
50	4.8  × 103	5.7  × 102
100	4.4  × 103	0.65  × 102

Dimers of π^30.5 Are Less Stable than Dimers of π^35.0.

Dimerization of π^35.0 protein is controlled by the N terminus of the protein and we have speculated that the LZ-like motif could be responsible for this function (16, 17). If the dimerization of π^35.0 depends solely upon the LZ-like motif, we would expect π^30.5, which lacks most of this motif, to be monomeric. Therefore, we moved on to compare the dimerization abilities of π^35.0 and π^30.5 in solution using a chemical cross-linking technique; this method had been employed previously to demonstrate the dimeric structure of π^35.0 (16).

As shown in Fig. 5A, non-cross-linked π^35.0 and π^30.5 migrate in denaturing SDS/PAGE as a single band. In the absence of Gu⋅HCl, after cross-linking with BSOCOES, most products migrate at the position corresponding to π^35.0 and π^30.5 dimers, indicating that both π proteins are predominantly dimeric. Upon treatment with increasing concentrations of Gu⋅HCl, the π^35.0 protein remains dimeric. In contrast, dimers of π^30.5 protein dissociate under the same conditions and its monomeric form can be seen even without Gu⋅HCl treatment. Thus, dimers of π^30.5 are considerably less stable than dimers of π^35.0 (see Fig. 5A).

Figure 5.

Dimer stabilities and DNA binding properties of π^35.0 and π^30.5. Both π variants were treated with Gu·HCl and processed as described. (A Upper) SDS/PAGE analysis of π samples followed by immunoassay with anti-π antibodies (1, 16). The first lane on each gel represents a purified π protein marker that has not been carried through all steps described. (A Lower) Blots were scanned as described in Fig. 4 legend. (B Upper) EMSA analysis of the binding properties of π^35.0 and π^30.5 to a single DR. Gu⋅HCl treated protein samples (shown in A Upper) were used. Binding reactions were assembled using a 3′-end labeled AseI–HindIII fragment from plasmid pMF239 as described (8). Complexes formed between the DNA fragment and π^35.0 dimers and monomers are indicated as C1 and C2, respectively. C3 indicates complexes of π^30.5 dimers with the DNA fragment (see text for details). (B Lower) Quantification of EMSA’s results was done as described in Fig. 4 legend.

π^35.0 and π^30.5 Differ in Oligomerization when Bound to a Single DR Unit.

Because activation of γ ori requires the binding of π protein (π^35.0) to the origin’s core, it was important to determine the DNA-binding properties of π^30.5. To simultaneously determine both the oligomerization behavior and the DNA-binding activity of π protein, we conducted an experiment that compared the binding of π^35.0 to that of π^30.5 using a target DNA fragment that contained a single DR unit. π^35.0 has been demonstrated to bind to such a fragment as either a monomer or as a dimer (39).

As shown in Fig. 5B, the differences between the binding patterns of both proteins are quite dramatic. Even without Gu⋅HCl treatment, π^30.5 binds distinctly since it produces a single nucleoprotein complex instead of two complexes as observed with π^35.0 (Fig. 5B). Moreover, π^30.5 loses its binding ability when treated with relatively low concentrations of Gu⋅HCl (0.5 M). In contrast, π^35.0 retains its binding abilities (primarily as monomer).

Several lines of evidence suggest that the single nucleoprotein complex observed in the presence of π^30.5 represents dimers of the bound protein. First, the complex formed by π^30.5 migrates more slowly than bound monomers of π^35.0 and, as we might expect, faster than bound π^35.0 dimers. Second, heterodimers of π^35.0 and π^30.5 (formed in vitro by a denaturation-renaturation step) produce a novel nucleoprotein complex whose electrophoretic mobility falls between species produced by binding of each of the two forms of π (39). Finally, the concentrations of Gu⋅HCl that prevent binding of π^30.5 to the DR also cause almost complete dissociation of dimers (Fig. 5A). For these reasons, we propose that the nucleoprotein complex observed with π^30.5 contains dimers of the protein and that either the monomeric species does not bind to a single DR unit or the complexes formed are unstable.

π^30.5 Does not Bind to (A⋅T)-Rich Segment of γ ori.

Replication of γ ori requires binding, not only to DRs, but to the (A⋅T)-rich region as well. Because interactions of π^35.0 with the (A⋅T)-rich site cannot be detected by EMSA, we moved on to an examination of the binding characteristics of π^35.0 and π^30.5 using the DNase I footprinting assay (Fig. 6). This assay permits the detection of π binding to both DNA sequence families (8).

Figure 6.

The binding properties of π^35.0 and π^30.5 to the DR and the (A⋅T)-rich region as determined by DNase I footprinting assay. (A) The binding properties of π^35.0 and π^30.5 to the γ ori core [(A⋅T)-rich region and 7 DRs]; (B and C) binding to the (A⋅T)-rich region with 1 or no DR, respectively. DNase I footprinting assays were performed as described (8). The amounts of π protein are indicated above each line. Positions of DRs and the (A⋅T)-rich segment are shown. 15* and 51* indicate DNaseI cleavage enhancements observed in the presence of π^35.0 at nucleotides +15 and +51 of the (A⋅T)-rich region. Characteristic enhancements at the first A position of the TGAGRG motif of DRs are also indicated (A¹–A⁷, A). Data were quantified as described in Fig. 4 legend. Lanes with 100 ng of π protein were chosen for quantification in A and B. Lanes with 25 ng of π^35.0 and 50 ng of π^30.5 were chosen for quantification in C.

Fig. 6A shows the results of footprinting assays in which π^35.0 or π^30.5 was combined with a γ ori fragment containing all seven DRs and the (A⋅T)-rich segment (8). Examination of the digestion patterns produced in the presence of π^35.0 reveals that the DNA sequence containing the cluster of seven 22-bp DRs is protected. A specific site (the first A in the TGAGRG motif) is hypersensitive to DNase I cleavage in each DR unit forming a characteristic ladder of enhancements. Furthermore, phosphodiester bonds at positions +15 and +51 in the (A⋅T)-rich segment are hypersensitive to cleavage.

By comparing the banding patterns of the two π species, we confirm the ability of π^30.5 to bind the DR (see also Fig. 5B). However, these interactions are not identical because footprints differ in several positions as demonstrated by their quantitative analysis. Moreover, the absence of the +15 and +51 enhancements in the presence of π^30.5 would suggest that π^30.5 is unable to bind to the (A·T)-rich site.

The complexity of π binding sites in the γ ori fragment suggested the possibility that single site, protein-DNA interactions could be overshadowed by other phenomena such as cooperative or competitive binding. Therefore, we pared the DNA fragment down to the (A⋅T)-rich region plus a single DR (Fig. 6B), and the (A⋅T)-rich region alone (Fig. 6C), and repeated the DNase I protection assays. The results of these two experiments support the conclusion, drawn from Fig. 6A, that π^30.5 binds DRs somewhat differently than π^35.0 and is unable to bind the (A⋅T)-rich region. These results also indicate that the inability of π^30.5 to bind to the (A⋅T)-rich segment of the γ ori does not depend on the presence or absence of additional π binding site(s) (i.e., DRs) on the same DNA fragment.

DISCUSSION

π^35.0 and π^30.5 Have Opposing Replication Activities in Vivo and in Vitro.

It has long been known that replication in vivo is stimulated at a low level of pir gene expression and inhibited at high levels (1, 27, 40, 41). For this reason, it has been proposed that stimulation and inhibition of replication are controlled by different levels of π^35.0 (1). Nevertheless, our recent in vitro results suggest that this model may be incomplete because π^35.0 can stimulate in vitro replication but cannot inhibit it even at the highest levels examined (ref. 34; unpublished data). It could be that inhibition requires an alternative form of π protein produced at sufficiently high levels when the pir gene is de-repressed. Because π^30.5 is expressed at 10- to 15-fold lower levels than π^35.0, we considered the possibility that π^35.0 could be an activator of replication, whereas π^30.5 might be an inhibitor.

To test the aforementioned model, we carried out in vivo and in vitro replication assays in which both proteins (π^35.0 and π^30.5) were examined alone or in combination. It is clear that these proteins are functionally distinct; π^35.0 stimulates replication, whereas π^30.5 cannot. In contrast, the former protein cannot inhibit in vitro replication while the latter can. Therefore, we conclude that π^30.5 has the potential to negatively regulate γ ori replication. Such regulation would be novel for plasmid replication and reminiscent of the control of Tn5 where one of two polypeptides, translated in the same frame, acts as activator while the other acts as an inhibitor of transposition (42, 43).

Interfaces Holding Together π^35.0 and π^30.5 Dimers Are Distinct.

In several plasmid systems including R6K, dimerization and oligomerization of initiators have been shown or are believed to play an important role in the regulation of replication and/or transcription (44–49). What seems particularly intriguing in the R6K system is that the ability to assume certain oligomeric conformations might be controlled at the level of gene expression. This notion is supported by our observation that π^35.0 contains a putative dimerization domain that π^30.5 does not possess.

Results presented here demonstrate that both π proteins can dimerize in solution but that dimers of π^30.5 are much less stable than dimers of π^35.0. The simplest explanation of these different stabilities is that the LZ-like motif significantly contributes to dimerization of π^35.0. Importantly, another interface, considerably weaker, must hold together two monomers of π^30.5. Thus, this report provides additional, strong support for the presence of dimerization sub-domains in the N terminus of π protein, complementing our previous findings that the negative-dominant π polypeptide (ΔC164π) exists in dimeric form and can form heterodimers with π^35.0 (17).

Possible Mechanisms Underlying the Distinct Functions of π^35.0 and π^30.5 Variants in Replication.

It is known that π^35.0 can bind to a DR unit as either monomers or dimers (Fig. 5B; ref. 39). Although it is unclear which of these two forms can activate the γ ori, it has been argued that it might be the monomeric form of π (39). This conclusion is supported by two lines of experiments. First, π^35.0 protein binds to a DR unit predominantly as dimers; copy-up variants that are up to sixfold more active as initiators in vitro (34, 35) bind predominantly as monomers (39). Second, the treatment of π^35.0 with Gu⋅HCl destabilizes dimers (Fig. 5B) and concomitantly increases its initiator activity (D.C. and M.F. unpublished data). Hence, the failure of π^30.5 to activate the γ ori could be related to its inability to bind to a single DR as a monomer (Fig. 5B). Inhibition, of replication could then be caused by dimers of π^30.5 (π^30.5/π^30.5 or π^30.5/π^35.0) competing against π^35.0 monomers for binding to DR binding sites (but not for binding to the (A⋅T)-rich segments, see below).

In addition to the DRs, the (A⋅T)-rich region is also required for γ ori replication. We have shown that π^30.5 lacks ability to bind to the (A⋅T)-rich segment of γ ori, providing a possible explanation for the failure of the protein to stimulate replication. The interaction between π and the (A⋅T)-rich region (seen with π^35.0) could be required to open the DNA duplex in the nearby region.

It is also noteworthy that the N terminus of π has been shown to interact with DnaB helicase (50). This suggests yet another attractive possibility to explain the inability of π^30.5 to stimulate replication. The LZ-like motif may control an association of π protein with DnaB; this interaction may be essential for γ ori replication given that an association between DnaA and DnaB is required for initiation at oriC (51).

Last, we would like to briefly comment on the subtle differences in the binding of π^35.0 and π^30.5 to DRs as revealed by DNase I footprinting. One possibility is that the changes are due to differences in the overall structure of both proteins. However, another interesting possibility arises from our finding that π^35.0 forms two nucleoprotein complexes with a DR, whereas π^30.5 forms only one complex. It is possible that footprints in the presence of π^35.0 are composites of two families, one produced by π monomers and another by π dimers. In contrast, because monomers of π^30.5 do not appear to bind the DR, footprints in the presence of π^30.5 might be produced by a single protein species, π dimers.

ABBREVIATIONS

BSA

bovine serume albumin

BSOCOES

bis[2-(succinimido-oxycarbonyloxy)ethyl]sulfone

Gu⋅HCl

guanidine hydrochloride

DR

direct repeat

γ ori, γ origin replication region

LZ, leucine zipper

RT room temperature

EMSA, electrophoretic mobility-shift assay

IPTG

isopropyl β-d-thiogalactoside