Dimers of π Protein Bind the A+T-Rich Region of the R6K γ Origin near the Leading-Strand Synthesis Start Sites: Regulatory Implications

Abstract

The replication of γ origin, a minimal replicon derived from plasmid R6K, is controlled by the Rep protein π. At low intracellular concentrations, π activates the γ origin, while it inhibits replication at elevated concentrations. Additionally, π acts as a transcription factor (auto)repressing its own synthesis. These varied regulatory functions depend on π binding to reiterated DNA sequences bearing a TGAGNG motif. However, π also binds to a “non-iteron” site (i.e., not TGAGNG) that resides in the A+T-rich region adjacent to the iterons. This positioning places the non-iteron site near the start sites for leading-strand synthesis that also occur in the A+T-rich region of γ origin. We have hypothesized that origin activation (at low π levels) would require the binding of π monomers to iterons, while the binding of π dimers to the non-iteron site (at high π levels) would be required to inhibit priming. Although monomers as well as dimers can bind to an iteron, we demonstrate that only dimers bind to the non-iteron site. Two additional pieces of data support the hypothesis of negative replication control by π binding to the non-iteron site. First, π binds to the non-iteron site about eight times less well than it binds to a single iteron. Second, hyperactive variants of π protein (called copy-up) either do not bind to the non-iteron site or bind to it less well than wild-type π. We propose a replication control mechanism whereby π would directly inhibit primer formation.

Recognition of the replication origin (ori) by an initiator protein is the central event regulating DNA synthesis in diverse biological systems. The γ ori of the antibiotic resistance plasmid R6K belongs to a large group of replicons that regulate replication through initiator proteins (Rep proteins) which bind to reiterated DNA sequences (iterons) (8). Although R6K contains multiple oris, the γ ori is a faithful model for studying the regulation of replication because it retains the copy control and stability loci of the parental plasmid (22, 49).

Regulated replication of a basic γ ori replicon depends upon π protein encoded by the plasmid's pir gene. The pir gene encodes two in-frame polypeptides (55): the longer and more abundant form, π^35.0, activates replication (22), while the shorter and less abundant form, π^30.5, inhibits replication (dependent on π^35.0) (52). The replication depends on the binding of π^35.0 to iterons (33). Either monomers or dimers of π^35.0 can bind to each of the asymmetric, 22-bp iterons that contain a TGAGNG DNA sequence motif (44, 52); seven such iterons are arranged in tandem within the γ ori (38) (Fig. 1). In contrast, only dimers of π^30.5 bind to an iteron (44, 52). It was proposed that the binding of π^35.0 monomers activates γ ori, while the binding of dimers (π^35.0 and/or π^30.5) mediates the formation of the inactive state of the ori via intramolecular coupling (handcuffing) (44) (Fig. 1). π is also known to (auto)repress its own synthesis and a distinct genetic locus (operator) is dedicated to this purpose (7, 20, 37, 54) (Fig. 1). The operator is composed of two inverted copies of the TGAGNG sequence to which dimers of π^35.0 and π^30.5 bind (44) (Fig. 1).

FIG. 1.

Selected elements of γ ori and known or proposed functions of π. The seven 22-bp iterons (π binding sites) are indicated by black arrows. The non-iteron π binding and IHF binding sites (ihf1) are indicated in the A+T-rich segment adjacent to iterons. SSLSS are indicated by vertical arrows; the direction of replication is indicated by a horizontal gray arrow. pir is the structural gene for π protein. A pair of inverted arrows in the pir operator indicates the inverted repeats. The inhibition of replication and/or repression of pir gene transcription is indicated by a minus sign. The activation of replication is indicated by a plus sign.

An A+T-rich DNA sequence is located adjacent to the iterons of γ ori (Fig. 1). Like iterons, A+T-rich regions are common in oris that replicate via the Cairns mode (16, 30). The intrinsic instability of these regions is known to help in the development of replication forks, and this instability can be further exacerbated by DNA binding proteins (2, 24, 35, 36). A non-iteron π binding site (i.e., not TGAGNG) has been identified within the A+T-rich region of γ ori (10, 27) (Fig. 1). π binds to this site independently of its binding to the iterons (27). Among the host proteins known to bind to the A+T-rich region of the γ ori, the role of integration host factor (IHF), which binds to the ihf1 site (Fig. 1), is best understood; it reverses the inhibitory effect of π (4, 5).

These observations and the fact that the start sites for leading-strand synthesis (SSLSS) also occur in the A+T-rich region (3) (Fig. 1) suggested that the non-iteron π binding site could be of special significance as a replication control locus. Particularly attractive was the possibility that π might negatively modulate priming through its binding to the non-iteron site (Fig. 1). One of the predictions of such a model would be that copy-up mutants, which stimulate in vitro replication more than the wild-type π counterpart (3, 28), might do so as a consequence of the decreased or altered binding to the non-iteron site. This hypothesis seemed attractive in view of our recent findings suggesting that the structures of wild-type π^35.0 dimers and dimers π^35.0 containing copy-up substitutions are most likely different (44). The experiments described here conform to a model in which the binding of π dimers to the non-iteron site would inhibit the primase-dependent priming that occurs downstream of this site (Fig. 1).

MATERIALS AND METHODS

Bacterial strains and plasmids.

Plasmid pKH2 was constructed as follows. The γ ori A+T-rich region was liberated as a HindIII fragment obtained from the γ117 mutant; in this mutant, the extra HindIII site has been fortuitously generated within the 1st iteron sequence (33). The fragment's ends were filled in by using the Klenow fragment of DNA polymerase I and ligated with plasmid pUC9 cut with HincII. Alternatively, the A+T-rich probe can be liberated from the plasmid pTS5 with HindIII-SspI (used only in the experiment shown in Fig. 2) (4). Plasmid pLS1 was constructed as follows. A single-iteron-containing BamHI-AseI fragment was obtained from plasmid pMF239 (9), filled in with the Klenow fragment, and ligated with plasmid pUC7 that had been linearized with HincII. The iteron-containing fragment was liberated from pLS1 by BamHI and ligated with plasmid pUC9 linearized with BamHI, resulting in plasmid pRK1.

FIG. 2.

Binding of π protein variants to the non-iteron site. (A) Diagram of proteins examined by EMSA. (B) Binding reactions were assembled as described in Materials and Methods with a non-iteron DNA probe obtained from plasmid pTS5. The lanes represent non-iteron DNA probe without protein (lane 1) or with π^35.0·His₆ (lane 2), π^35.0 (lane 3), π^30.5·His₆ (lane 4), and π^30.5 (lane 5).

Enzymes and chemicals.

Restriction enzymes were obtained from New England Biolabs or Promega; DNase I was obtained from Boehringer-Mannheim. HEPES and Tris were obtained from Sigma. IPTG (isopropyl-β-d-thiogalactopyranoside) was obtained from United States Biochemical. Heparin-Sepharose, hydroxylapatite, and nickel-nitrilotriacetic acid-agarose (Ni²⁺-NTA) resin were obtained from Pharmacia. Bio-Rex 70 was obtained from Bio-Rad.

Protein purification.

π^35.0 and ΔC164π were purified from cells containing plasmids with an IPTG-inducible Tac promoter. Constructs and purification protocols were previously published (15). π^30.5 was purified from cells containing plasmid pMS7.4 (52) according to the method of Greener et al. (15). Construction and purification of His-tagged variants of π were performed as previously described (51).

EMSA.

Binding reaction mixtures for electrophoretic mobility shift assays (EMSAs) were assembled in Tris-borate-EDTA buffer (TBE; 50 mM Tris, 45 mM boric acid, 1.4 mM EDTA) and 65 ng of poly(dI-dC):poly(dI-dC); the amount of π protein was 200 ng or otherwise, as indicated in the figure legends. Proteins were diluted in TGE–0.3 M KCl buffer (10 mM Tris-HCl [pH 8.0], 0.1 mM EDTA, 10% glycerol, 0.3 M KCl). Reaction mixtures (5 μl) were equilibrated at room temperature (RT) for 10 min and, after the addition of 3 μl of nondenaturing dye (20% glycerol and 0.01% bromophenol blue in TBE), electrophoresed at 10 V/cm in 6% acrylamide for 1.5 h at RT. Gels were dried and exposed to a PhosphorImager TM 445 SI (Molecular Dynamics) screen or to X-Omat Kodak film.

Preparation of π^35.0·His₆-π^30.5 and π^35.0·His₆-ΔC164π heterodimers.

The full-length π^35.0·His₆ and truncated π^30.5 were mixed in a 1:5 molar ratio and denatured by incubation with 6 M guanidine hydrochloride (Gu-HCl) (final concentration) in TGE buffer at RT for 30 min. Renaturation of the proteins was achieved by dialyzing denatured proteins against TGE–0.3 M KCl buffer at 4°C. The protein mixture was fractionated by using Ni²⁺-NTA resin. Briefly, the protein mixture and the Ni²⁺-NTA resin were incubated at 4°C for 1 h, and then the resin was washed with 10 ml of TGE–0.3 M KCl buffer and the elution was done with 0.2 M imidazole in TGE–0.3 M KCl buffer. The protein eluted was used in the binding reactions (as described above). The presence of imidazole in protein stocks does not interfere with π binding to DNA (J. Wu, S. Rakowski, and M. Filutowicz, data not shown). The π^35.0·His₆-ΔC164π heterodimers were obtained by the same procedure. Gu-HCl-treated protein mixtures were compared in the EMSA to the individual proteins, not treated with Gu-HCl. In a control experiment for the effect of Gu-HCl treatment, π^35.0·His₆ was unfolded and refolded alone, and this preparation exhibited the properties of the untreated protein (R.K. and M.F., data not shown).

Chemical cross-linking of π.

Gu-HCl-treated and untreated protein samples (200 ng/25 μl) were chemically cross-linked with BSOCOES {bis[2-(succinimidyloxycarbonyloxy)ethyl]sulfone} (from Pierce), resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and detected by Western blotting with anti-π antibodies as described previously (44).

RESULTS

EMSA reveals binding of π derivatives of various molecular weights to the non-iteron site.

Although the A+T-rich region of γ ori contains only 115 bp, several host-encoded proteins are known to bind to this segment, and binding sites for some of these proteins (Fis, DnaA, IHF, RNAP, and HU) occur more than once (6, 43, 48, 50; R.K. and M.F., unpublished data). EMSA indicated that several of these proteins are present in various preparations of π protein despite being undetectable by silver staining (R.K. and M.F., unpublished data). To perform meaningful analysis of the binding of π to the non-iteron site required the development of superior purification protocols for many derivatives of π in our possession.

We chose a hexahistidine (His₆) tagging procedure to accomplish our goal of increasing protein purity. As an added benefit, His tagging has allowed us to obtain π variants which differ in molecular weights: π^35.0, π^35.0·His₆, π^30.5, and π^30.5·His₆ (51). We assumed that these proteins (Fig. 2A) would produce complexes with DNA which would differ in their electrophoretic mobilities. As shown in Fig. 2B, π variants with higher molecular weights reduce the electrophoretic mobility of the DNA probe containing the non-iteron site more than π variants with lower molecular weights. These data provide evidence that EMSA detects specific interactions of π protein with the non-iteron binding site.

π binds to the non-iteron site as a dimer.

As reported elsewhere, π can bind to a DNA probe containing a single iteron as either a monomer or a dimer (44, 52). We have proposed that monomers of π^35.0 activate γ ori, while dimers represent an intermediate oligomer in the assembly of handcuffed structures believed to be inhibitory for replication (32, 44). In addition, π binding to the A+T-rich region of γ ori could be inhibitory at high π levels, and occupancy of this binding site might be, like handcuffing, dimer specific, since dimers are presumably the inhibitory form of the protein. We conducted experiments using heterodimers of π protein as a tool to examine this possibility.

Heterodimerization assays can reveal the oligomerization state of a protein bound to its cognate recognition site (17). The technique requires protein variants which differ in molecular weights, each capable of binding to DNA and producing complexes of distinct electrophoretic mobilities. Hybrid molecules of the two proteins will exhibit intermediate electrophoretic mobilities. π^35.0 and π^30.5 fit these criteria: they have distinct molecular weights, both are able to bind to the non-iteron site, and both can dimerize (44, 52). In the experiments that follow, we have improved the resolution of the heterodimerization assay by using the π^35.0·His₆ variant. Also, we used an iteron probe as a control for the non-iteron probe binding, because only non-His₆ derivatives of π have been characterized for their oligomerization and DNA binding properties (44).

When π^35.0·His₆ and π^30.5 are unfolded with Gu-HCl and then gradually refolded, the following species form: π^35.0·His₆ homodimer, π^30.5 homodimer, and π^35.0·His₆-π^30.5 heterodimer. Because the π^35.0·His₆-π^30.5 heterodimers bound to the iteron migrate at a rate similar to that of π^30.5 homodimers (non-His-tagged), we eliminated the π^30.5 homodimers by passing the refolded protein mixtures through Ni²⁺-NTA resin. Control lanes in Fig. 3 show an iteron (TGAGNG) probe bound by π^35.0·His₆; one band corresponds to a π^35.0·His₆ monomer, and the other corresponds to a π^35.0·His₆ homodimer (Fig. 3, lanes 2 and 4). In another control, we show that only one complex forms with π^30.5, and it contains π^30.5 homodimer (52) (Fig. 3, lane 3). The final control shows that π^35.0·His₆-π^30.5 heterodimers form a band of intermediate mobility (Fig. 3, lanes 3 and 4). Thus, the oligomerization and iteron-specific DNA binding activities of His-tagged derivatives of π are very similar to those of the non-His-tagged counterparts that were analyzed earlier (44).

FIG. 3.

Oligomerization assay of π bound to iteron and non-iteron sites. (A) Diagram of proteins examined by EMSA for their oligomerization state in complexes with the iteron-containing probe (left panel) and the non-iteron probe (right panel). π^35.0·His₆ homodimers, π^30.5 homodimers, and π^35.0·His₆-π^30.5 heterodimers were obtained as described in Materials and Methods. (B) The lanes represent probes without protein (lanes 1 and 5) or with π^35.0·His₆ (lanes 2 and 7); π^30.5 (lanes 3 and 8); or a mixture of π^35.0·His₆ and π^30.5 eluted from Ni²⁺-NTA resin (lanes 4 and 9), resulting in π^35.0·His₆ homodimers, π^35.0·His₆-π^30.5 heterodimers, and π^35.0·His₆ monomers. IHF alone (lane 6) or in combination with the π variants was added to samples containing the non-iteron probe (lanes 7 to 9).

When the A+T-rich region probe is substituted for the iteron probe and identical protein samples are used, π^35.0·His₆ and π^30.5 each produce a single band (Fig. 3, lanes 7 and 8). However, when preparations known to contain π^35.0·His₆-π^30.5 heterodimers were used (Fig. 3, lane 4), two bands were observed (Fig. 3, lane 9). The slower band comigrates with the band produced by π^35.0·His₆, while the faster-migrating band does not match the band produced by π^30.5-bound probe. Thus, we infer that the former band contains homodimers of π^35.0·His₆, while the latter band contains π^35.0·His₆-π^30.5 heterodimers. These results suggest either that monomers of π^35.0·His₆ and π^30.5 cannot bind to the non-iteron site or that their affinities are too low for their binding to be detected by EMSA. We conclude that whenever π binding to the non-iteron site is detected, the complexes contain bound dimers of π (Fig. 1).

Only one-half of the dimer is needed for interaction with the non-iteron site.

In a nucleoprotein complex in which a π dimer binds an asymmetric iteron sequence, only one subunit of the dimer contacts the iteron; the other subunit appears to be free of DNA contact (44). As pointed out earlier, such an asymmetric π complex has the potential to capture another iteron-containing molecule to form a handcuffed structure (44). We reasoned that at least one of the dimers bound to γ ori could have one subunit dedicated to iteron binding, while another subunit could be dedicated to non-iteron binding.

To determine if both subunits of a π dimer are required for interaction with the non-iteron binding site, we used a modified version of the heterodimerization assay described above. The critical distinction was that one of the subunits of the (in vitro) reconstructed heterodimers was itself unable to bind to DNA. Such a protein was already in our possession in the form of a truncated variant, ΔC164π (15, 44); it contains 164 N-terminal amino acids of π^35.0 (44). Although the truncated ΔC164π protein can bind to an iteron as a ΔC164π-π^35.0 heterodimer, it cannot bind to the site as a homodimer (44).

We again used His-tagged, full-length π and Gu-HCl treatment to construct ΔC164π-π^35.0·His₆ heterodimers, and then we examined the binding of the proteins to the non-iteron probe DNA. π^35.0·His₆ produces a single band, and ΔC164π does not bind to the DNA (Fig. 4, lanes 2 and 4). However, when preparations known to contain π^35.0·His₆-ΔC164π heterodimers were used, two complexes were observed (Fig. 4, lane 3). The complex with slower electrophoretic mobility comigrates with a complex produced by π^35.0·His₆ alone, while the faster-migrating complex is a new entity. Thus, we conclude that in this new complex, π^35.0·His₆-ΔC164π heterodimers are bound to the probe. The band with slower mobility than that produced by bound heterodimers results from homodimers of π^35.0·His₆ binding to the non-iteron probe. Thus, it is clear that a single subunit of a π dimer is sufficient to bind to the non-iteron site under the experimental conditions employed.

FIG. 4.

Assay for the number of π subunits sufficient for binding to the non-iteron site. (A) Diagram of proteins examined by EMSA for their binding to the non-iteron probe. Homodimers of π^35.0·His₆ and ΔC164π as well as π^35.0·His₆ ΔC164π heterodimers were obtained as described in Materials and Methods. (B) The lanes represent DNA probe without protein (lane 1) or with π^35.0·His₆ (lane 2); a mixture of π^35.0·His₆ and ΔC164π resulting in π^35.0·His₆-ΔC164π heterodimers, π^35.0·His₆ homodimers, and π^35.0·His₆ monomers (lane 3); or ΔC164π homodimers (lane 4).

π^35.0 binds less well to the non-iteron site than to the iteron site.

As pointed out earlier, the inhibition of replication most likely requires π dimers or higher-order oligomers (handcuffing), and dimeric π binds to both the iteron and non-iteron binding sites (44) (Fig. 1). Possible mechanisms for iteron-based inhibition (handcuffing [44]) and non-iteron-based inhibition (repressing primer synthesis [this paper]) have been proposed (3, 11). To determine which of these potential mechanisms of inhibition would be triggered first, we conducted quantitative binding assays. EMSAs were performed with π^35.0·His₆ by using ³²P-labeled DNA probes containing either a non-iteron site or an iteron site (the specific activities of the probes were similar [R.K. and M.F., data not shown]). The amounts of bound DNA were quantified (Fig. 5). It is clear that π^35.0 binds a greater proportion of the iteron-containing DNA probe than the non-iteron probe, suggesting that the protein binds less well to the non-iteron site than to the iteron site. In the complete genetic context of γ ori (with seven iterons), this difference is expected to be much greater, since π binds to the iterons cooperatively (45).

FIG. 5.

Comparison of π^35.0·His₆ binding to the iteron versus non-iteron sites. π^35.0·His₆ was used with a direct repeat (DR) or an A+T-rich probe and the total bound DNA was quantified. The error bars are derived from three experiments. The percentage of bound DNA was higher with the direct repeat probe than with the A+T-rich probe.

Copy-up variants of π^35.0 show reduced binding to the non-iteron site.

Next we examined whether copy-up variants of π^35.0 can bind to the non-iteron site. We undertook comparative binding studies to determine if a correlation exists between the non-iteron binding abilities (or lack thereof) of copy-up π variants and their known elevated replication activities in vitro (3). An inverse correlation would mean that the higher activity of copy-up π might be explained, at least in part, by a reduced ability to repress primer synthesis in the A+T-rich region. We have constructed π^35.0·His₆ S87N and the double mutant π^35.0·His₆ P106L F107S (pir116 pir200) derivatives of well-characterized copy-up variants (3, 28) and confirmed that their elevated replication activity is preserved both in vivo and in vitro (R.K. and M.F., unpublished data). As shown in Fig. 6A, proteins containing copy-up substitutions show reduced binding to the probe containing the non-iteron site. Thus, the elevated replication activity of copy-up variants appears to correlate with the reduced binding of the mutant protein to the non-iteron site. Earlier investigations revealed that wild-type and copy-up variants of π^35.0 can bind to iterons (3, 12, 13, 44).

FIG. 6.

The effect(s) of copy-up π substitutions on protein dimerization and binding the non-iteron site. (A) EMSA was performed with the non-iteron-containing DNA probe and increasing amounts of π^35.0·His₆, π^35.0·His₆ P106L F107S, and π^35.0·His₆ S87N (50, 100, 200, and 300 ng) as described in Materials and Methods. (B) Oligomerization states of π^35.0·His₆, π^35.0·His₆ P106L F107S, and π^35.0·His₆ S87N (200 ng/25 μl) were determined by chemical cross-linking as described in Materials and Methods.

The reduced binding of copy-up variants to the non-iteron site might be simply explained if such proteins were largely monomeric while wild-type His-tagged protein was dimeric. Thus, we next determined the dimerization properties of the His-tagged derivatives in the absence and presence of Gu-HCl, as described in Materials and Methods. It is clear that π^35.0, π^35.0·His₆ S87N, and the double mutant π^35.0·His₆ P106L F107S are dimeric in the absence of Gu-HCl. A significant fraction of dimers is detected even in samples treated with 1.5 M Gu-HCl. However, at 2.5 and 3.0 M Gu-HCl, the copy-up variants are monomeric, while a small fraction of π^35.0 remains dimeric. These results conform to the previously published data with the non-His-tagged proteins, which demonstrated that copy-up proteins are dimeric but dissociate to monomers at lower levels of Gu-HCl than does wild-type π (44). Although copy-up mutations destabilize π dimers, the bulk of π^35.0·His₆, π^35.0·His₆ S87N, and π^35.0·His₆ P106L F107S is dimeric in solution. We discuss (below) the presumed reason for the reduced inability of these dimers to bind to the non-iteron site.

DISCUSSION

Several control loops which negatively regulate replication have been implicated in the maintenance of γ ori copy number (for a recent review, see reference 11). We add to this list another possible mechanism which is attractive in its simplicity. In it, π protein would directly inhibit primer formation that is known to occur in the A+T-rich region (3). In fact, direct control of the priming step by a repressor molecule is the key mechanism for regulating the copy number of plasmid ColE1, although in that system, an antisense RNA acts as the inhibitor (39, 40). In the γ ori system, several lines of evidence support a simple occupancy model for π protein-mediated replication inhibition through its non-iteron binding site.

For example, the characteristics of the interaction between π protein and the non-iteron binding site fit nicely with this model. π binds less well to the non-iteron binding site (in vitro) in comparison to the iteron site (Fig. 5). Moreover, only dimers of π have been observed binding to the non-iteron site (Fig. 3 and 4). Taken together, it seems likely that π would occupy this binding site in vivo only at elevated levels of the protein which are known to inhibit replication and favor dimerization. In essence, the inhibition of replication proposed here and the (auto)repression of pir gene transcription could use similar mechanisms while relying on disparate sites, activities, and protein surfaces. Based on the recently solved structure of RepE protein of plasmid F, it is likely that π also has two DNA binding surfaces, since these two proteins are structurally related (23).

Additional support for the model lies in the observation that the non-iteron π binding site and a binding site for IHF protein, ihf1, overlap (Fig. 1). In fact, at least one report suggests that π binding to the non-iteron site and IHF binding to the ihf1 site could be competitive (6). Neither IHF protein nor the ihf1 site is needed for replication in systems where the γ ori plasmid copy number is elevated. IHF independence and elevated plasmid copy number can be achieved either by lowering intracellular levels of wild-type π or by substituting wild-type π with one of the copy-up variants of the protein (5). The latter effect has also been simulated (in vitro) with π^35.0 alone (4). Thus, the sole role of IHF is to counteract the inhibitory effects of π on replication (4, 5).

Finally, DNA synthesis starts downstream of the non-iteron site in an RNA polymerase-independent fashion (3, 27, 29). We inferred that initiation depends on the dnaG gene product (primase) of Escherichia coli that can synthesize primers not exceeding 30 nucleotides in length (1, 21, 56). This information, coupled with mapping data for the SSLSS, makes it likely that primase either binds directly to a DNA segment overlapping the non-iteron π binding site (Fig. 7A), or else the protein is delivered there by a mobile primosome (reviewed in reference 25). If a primosome is utilized, it does not involve the ABC-primosome assembly site identified near the γ ori (31), since this site is absent in the model system discussed here.

FIG. 7.

Models of replication control for γ ori. The primase-occluding model (A and B) and the handcuffing model (C) are shown. See text and reference 7 for details.

Reminiscent of its binding to an iteron sequence (44), π dimers appear to use a single subunit when binding to non-iteron DNA (Fig. 7A). In the case of iteron binding, it has been proposed that π dimers could use their two DNA binding domains to capture independent iteron-containing fragments (oris), thereby facilitating handcuffing (44) (Fig. 7C). Perhaps a dimer bound to a non-iteron site could also foster DNA-DNA interaction. It is possible that π complexes assembled on iterons and non-iteron sites might contact one another in cis (Fig. 7B). The mechanism may resemble the action of some recombinases, which are known to use distinct DNA binding domains to juxtapose DNA sequences undergoing recombination (26, 34). A loop resulting from the π-mediated interaction between iteron and non-iteron sites could be more effective than simple competition as a means of occluding the binding of primase to DNA or perhaps interfering with primase's catalytic function (Fig. 7B).

The observation that copy-up mutants bind a non-iteron site poorly if at all is in keeping with the hypothesis that π binding this site inhibits replication. Evidence in several iteron-containing systems suggests that monomers, but not dimers, of Rep proteins activate their cognate oris (14, 18, 19, 41, 42, 46, 47, 53). It has been suggested that copy-up mutations might function by promoting the monomerization of Rep proteins, and data show that copy-up variants of π^35.0 do form less stable dimers (44). It is unlikely, however, that copy-up π is deficient in binding to the non-iteron site simply because dimers are required for binding to occur (Fig. 6B).

Presumably, dimers of π^35.0 lacking or containing copy-up substitutions are structurally different in such a way that the non-iteron DNA binding surface would not be exposed. A similar argument was originally put forth by Wickner and colleagues (46) to explain the inability of RepA protein dimers to bind the iterons of plasmid P1. We believe that the use of protein dimerization to block the access of DNA to its binding surface could be a widespread phenomenon, especially given that the dimers of many other Rep proteins cannot bind to the cognate iteron sequence, while the monomers can (18, 19, 42, 46, 47, 53). This is true despite the fact that the DNA sequences of many iterons are asymmetric (8) and hence would require only one subunit of a dimer for binding.

Although π^30.5 has been demonstrated to be an inhibitor of replication, it binds to the non-iteron site much more weakly than does π^35.0 (44). In fact, under the conditions of the DNase I footprinting assay, we did not detect binding of π^30.5 to the non-iteron site (52). These observations, in conjunction with the fact that π^30.5 is much less abundant in vivo than π^35.0 (55), indicate that π^35.0 and/or π^35.0-π^30.5 heterodimers are better candidates for inhibitors of priming.

It seems unlikely that π^30.5-based inhibition of replication acts through the non-iteron binding site. Furthermore, we do feel that the γ ori has, at its disposal, a repertoire of mechanisms controlling DNA replication in a hierarchical fashion. Evidence exists to implicate handcuffing (32, 44) and the titration of activator (monomers) in the inhibition of γ ori function (M. Filutowicz et al., unpublished data). We are exploring the relationship between these control mechanisms.