Structural Changes in RepA, a Plasmid Replication Initiator, upon Binding to Origin DNA*

RepA protein is the DNA replication initiator of the Pseudomonas plasmid pPS10. RepA dimers bind to an inversely repeated operator sequence inrepA promoter, thus repressing its own synthesis, whereas monomers bind to four directly repeated sequences (iterons) to initiate DNA replication. We had proposed previously that RepA is composed of two winged-helix (WH) domains, a structural unit also present in eukaryotic and archaeal initiators. To bind to the whole iteron sequence through both domains, RepA should couple monomerization to a conformational change in the N-terminal WH, which includes a leucine zipper-like sequence motif. We show for the first time that, by itself, binding to iteron DNA in vitro dissociates RepA dimers into monomers and alters RepA conformation, suggesting an allosteric effect. Furthermore, we also show that similar changes in RepA are promoted by mutations that substitute two Leu residues of the putative leucine zipper by Ala, destabilizing the hydrophobic core of the first WH. We propose that this mutant (RepA-2L2A) resembles a transient folding intermediate in the pathway leading to active monomers. These findings, together with the known activation of other Rep-type proteins by chaperones, are relevant to understand the molecular basis of plasmid DNA replication initiation.

DNA replication is a tightly regulated process to guarantee that genetic information is precisely copied once, and only once, each cell cycle (1). This is achieved by means of complex circuits that are distinct in prokaryotes and eukaryotes, although some similarities have become evident recently (2)(3)(4). Common essential elements are a cis-acting DNA sequence (the origin of replication) and a trans-acting factor (the initiator). DNA replication initiators are single proteins, or multisubunit complexes, that bind specifically to the origin of replication, where they usually assemble into oligomers. Initiators play two roles: (i) to melt the strands of DNA and (ii) bring to the so-created replication bubble a number of other protein factors. These are required for further extension of the replication fork (helicases), synthesis of an RNA primer (primases), and copying the template (DNA polymerases) (reviewed in Refs. 1 and 2). Any initiator, to be enabled to bind and melt origin DNA, requires an activation step whose molecular basis is a central subject in research on both DNA replication and the cell cycle.
A trait common to initiators is that their function is controlled by ATP binding and hydrolysis (3,5,6) defining active or silent conformational states. In the bacterial chromosomal initiator, DnaA (reviewed in Ref. 7), ATP is not strictly required for specific binding to double-stranded origin DNA (oriC). However, it is essential for melting oriC AT-rich repeats, binding to the resulting single-stranded sequences (6) and then for loading DnaB helicase. Subsequent ATP hydrolysis, stimulated by the processivity factor of DNApol III (␤clamp) and other cellular factors, renders DnaA unable to initiate further replication rounds. DnaA can then be reactivated by either acidic phospholipids or DnaK chaperone, which exchange ADP by ATP (7). On their side, Orc1 and Orc5 subunits of the eukaryotic initiator ORC 1 also bind ATP, which is required for specific recognition of origin sequences (8). Singlestranded DNA, generated after origin melting, stimulates ATP hydrolysis and exerts a change in the overall shape of ORC (5).
Plasmids are extrachromosomal DNA molecules that borrow from their hosts most of the factors required for replication. However, they often encode their own initiator, termed Rep (reviewed in Refs. 9 and 10). In Gram-negative bacteria, Rep proteins usually bind to directly repeated sequences (iterons) to establish the initiation complex. In addition, some Rep proteins also bind to an inversely repeated sequence (operator) that overlaps with the promoter of the rep genes, thus acting as self-repressors. Dimers of Rep bind to the operator, whereas monomers bind to the iterons (11)(12)(13)(14)(15)(16)(17).
RepA is the initiator protein of pPS10, a plasmid isolated from Pseudomonas (18). Mutations in an LZ-like sequence motif, found at the N terminus of RepA (19), enhance dimer dissociation (14). However, a proof for a direct role of LZ in RepA dimerization is still lacking. An helix-turn-helix motif at the protein C terminus is the main determinant of RepA binding to both operator and iteron DNA sequences (20). We have proposed recently (21) that RepA consists of two WH domains (reviewed in Ref. 22). Furthermore, a similar WH fold is found at the C terminus of the eukaryotic/archaeal initiators Orc4/ Cdc6 (4,23), underlining the relevance of studying the molecular mechanism of RepA activation. We had proposed that dissociation of RepA dimers into monomers would result in a structural change altering the relative arrangement and compaction of its two WH domains (21). In the dimers, the C-terminal domain (WH2) binds to the operator through the major groove, whereas the N-terminal domain (WH1) acts as dimerization interface. In the monomers, WH2 binds to iteron DNA as it does to the operator, whereas the remodeled WH1 binds to the opposite end of iteron through both the phosphodiester backbone and the minor groove (21). This model has been then confirmed by the crystal structure of the monomer of a related initiator, RepE54, bound to iteron DNA (24). It shows that the ␣-helix that might resemble an LZ in the dimers is found in the monomers split into two helical portions, packed together, and buried in the hydrophobic core of WH1 (24).
A feature of Rep-type initiators is that, unlike DnaA and ORC, they do not bind ATP. Thus, although Rep proteins can promote some structural transitions in DNA, most plasmids require DnaA to aid in origin melting (17,(25)(26)(27)(28)(29)(30) and helicase loading (28,29). However, they still experience conformational activation (reviewed in Ref. 30). Molecular chaperones, either the triad DnaK-DnaJ-GrpE or ClpA, have been implicated in the activation of the Rep-type initiator proteins of plasmids P1 (11,(31)(32)(33)(34) and F (35). ClpX chaperone has a role in the activation of the initiator of RK2 plasmid (36). In addition, ClpA unfolds P1 Rep to be proteolyzed by ClpP (37,38). We have proposed recently (4) that Hsp70 chaperones, the eukaryotic homologues of DnaK, are implicated in the assembly of ORC. Two alternative functions have been proposed for chaperones in P1 replication; either Rep dimers would be so stable that chaperones are required to dissociate them (31,32), or they would modify Rep conformation (39). Thus, some monomeric mutants still require the action of chaperones in vivo (33,34), although they have higher association rates to iterons in vitro and increased initiation frequency (40). In addition, dimers of a few Rep-type proteins dissociate spontaneously by dilution (34,41,42). The current view is that chaperones would induce in Rep both monomerization and a conformational change, whose exact nature remains to be determined (30,39,43).
We have searched for determinants of the activation of pPS10 RepA. We show in this paper that micromolar amounts of a single iteron DNA sequence actively induce in vitro both the dissociation of RepA dimers into monomers and a conformational change. On the contrary, binding of RepA dimers to the operator sequence neither dissociates them nor changes their conformation. The ligand-induced monomerization of RepA dimers with a coupled conformational change, reported in this paper, would thus be a case for the allosteric effect of DNA substrates in the structure of their binding domains (44,45) and a novel player in the activation of plasmid DNA replication initiators. Furthermore, we also present a detailed biophysical characterization of RepA-2L2A, a mostly monomeric species of RepA obtained by site-directed mutagenesis of two Leu residues in the LZ-like motif. Our data suggest that it resembles a transient folding intermediate in the way from dimers to active monomers. In RepA-2L2A, the mutations disable the first of the two ␣-helical portions of the putative LZ to fold back into the second one, thus disrupting hydrophobic interactions at the core of monomeric WH1.

EXPERIMENTAL PROCEDURES
Cloning, Expression, and Purification of RepA-2L2A-His 6 -RepA-WT was expressed and purified, and when required, its tag was removed as described previously (21). Mutant repA-2L2A gene was obtained by PCR, using as template the expression vector pRG-recA-NHis-repA-WT (23) and primers including the following: (i) a SacII site plus sequences coding for the 5Ј end of repA but with mutated codons (GCC for Ala-12 and GCG for Ala-19) replacing those for leucines in the putative LZ and (ii) a HindIII site plus a TGA stop codon and sequences complementary to the 3Ј end of repA. Mutations were verified by automated DNA sequencing. Expression of His 6 -RepA-2L2A was carried out at 30°C in Escherichia coli SG22097 (clpXP), as performed for RepA-WT (21). His 6 -RepA-2L2A was purified to homogeneity from inclusion bodies using the same procedure developed for deletion mutants in RepA (21). Protein stocks were kept in 0.5 M (NH 4 ) 2 SO 4 , 50 mM NH 4 -acetate, pH 6.0, 10 mM ␤-MeEtOH, 0.1 mM EDTA, 10% glycerol. RepAs concentrations were calculated based on their absorption at 280 nm in 5.6 M GuHCl, considering a molar extinction coefficient of 17210 M Ϫ1 ⅐cm Ϫ1 (www.expasy.ch/tools/protparam.html).
Testing in Pseudomonas aeruginosa the Effect of the His 6 Tag and the 2L2A Mutation in RepA-pPSEC, a new series of shuttle vectors including the pPS10 and ColE1 replicons, was constructed as follows. pRG9B (46) was digested with BamHI to give a 4-kbp fragment including the pPS10 replicon (18) and the Kn resistance determinant. This fragment was then filled in with Klenow and ligated to a 759-bp DraI-AflIII (ori) fragment from pUC18, in which the latter site had also been made blunt-ended. The resulting plasmid (pPSEC1) was then modified to substitute the 5Ј half of the repA-WT gene (a 317-bp EcoRI-SphI fragment) by versions encoding a His 6 tag (pPSEC2) and the 2L2A mutations (pPSEC3). This was carried out by PCR on the corresponding pRG-recA-NHis-repA-WT/2L2A templates (see above), using as primers the following: (i) a 41-bp tail with the sequence between the EcoRI site and the ATG initiation codon in repA (18), plus the encoded His 6 tag and (ii) sequences from the complementary non-coding strand comprising the SphI site in repA. The three pPSEC constructs were rescued in E. coli JM109, checked by DNA sequencing, and then transformed into P. aeruginosa PAO1024 (18). Relative plasmid copy numbers were determined from cultures of PAO1024/pPSECs in LB medium, supplemented with Kn to 50 g/ml, at 30°C. 1.5-ml aliquots were harvested at A 600 ϭ 0.5, and total lysates were obtained as described (18). 60 l (1/10) of the lysates were then loaded into 0.8% agarose-Tris acetate EDTA gels, run at 30 V for 12.5 h, and stained with EtBr. Southern blotting was carried out by transferring the gels to nylon membranes (18) and then hybridized with a 954-bp EcoRI repA fragment from pRG9B, radiolabeled by random priming with Klenow and 40 Ci of [␣Ϫ 32 P]dCTP (Amersham Biosciences). X-ray films (AGFA) were exposed and then both the hybridized plasmid bands and the EtBrstained chromosome were quantified using GelDoc (Bio-Rad). The amount of intracellular RepA in P. aeruginosa cells carrying pPSEC plasmids was estimated by Western blotting, as described (47). 200-l culture aliquots were taken at different A 600 values. Total lysates were obtained, resolved by SDS-PAGE, and transferred to nitrocellulose membranes by electroblotting. Membranes were incubated with an anti-RepA rabbit polyclonal antiserum (1:1000) and then revealed with horseradish peroxidase-conjugated donkey anti-rabbit IgG (1:10000) by the ECL procedure (Amersham Biosciences). RepA amounts were estimated by comparing the intensities of the specific luminescent bands in pPSEC-containing cells with those in lanes including extracts from plasmid-free cells, the latter supplemented with known amounts of pure RepA. The number of cells contributing to the loaded extracts was estimated after serial dilutions of the cultures by counting at a microscope. Viables were then determined by plating on LB-agar supplemented with Kn. Molar concentration of RepA was calculated considering for a P. aeruginosa cell the same volume determined previously for E. coli (4.6 ϫ 10 Ϫ10 l) (48).
EMSA-DNA oligonucleotides containing either the operator or the four iteron sequences of pPS10 (18) were synthesized and then phosphorylated with T4 polynucleotide kinase. After annealing with their complementary strands, they were cloned into the SmaI site of pUC18 and sequenced. Plasmids were cut at the XbaI site and then labeled with 2 units of Klenow and 30 Ci of [␣-32 P]dCTP for 30 min at 25°C. Fragments were excised with EcoRI and purified by electrophoresis in polyacrylamide gels (13). Increasing amounts of RepA proteins (WT or 2L2A) were incubated in ice with radiolabeled DNAs (4000 cpm) in a 20-l volume of 20 mM Hepes⅐NH 4 , pH 7.8, 5 mM ␤-MeEtOH, 0.1 mM EDTA, 6% glycerol, 50 g/ml bovine serum albumin, and (NH 4 ) 2 SO 4 , provided by the supplied protein, to 0.2 M. Samples, assembled in ice, were then transferred to room temperature for 30 min before loading into 6% polyacrylamide (29:1)-0.5ϫ TBE gels. Electrophoresis was run at 150 V and 4°C. Gels were then dried out and exposed to x-ray films.
Analytical Ultracentrifugation-Sedimentation equilibrium experiments were performed in a Beckman XL-A analytical ultracentrifuge. RepA stocks were dialyzed against 0.25 M (NH 4 ) 2 SO 4 , 50 mM NH 4acetate, pH 6.0, 0.1 mM EDTA. 60-l samples, with different protein concentrations, were displayed into six-channel centrifuge cells, with a 1.2-cm optical path and centerpieces of epon charcoal. Sedimentation equilibrium gradients were formed at 5°C by spinning at either 13000 or 15000 rpm. Radial scans were taken at different wavelengths. Baseline offsets were measured at 50000 rpm. Sedimentation velocity of RepA, either alone or complexed with DNA fragments, was performed at 50000 rpm and 5°C with 350-l samples displayed into double sector cells. Data from both types of experiments were processed as described (49). The sedimentation coefficient distributions for the RepA-DNA complexes (Fig. 6C, inset) were calculated by direct linear least squares boundary modeling of the sedimentation velocity data (ls-g*(s)) using SEDFIT (50).
Spectroscopic Assays-Steady state fluorescence spectroscopy was performed in a Fluorolog Jobin Yvon-Spex spectrofluorimeter. 350-l samples of His 6 -RepA-WT or His 6 -RepA-2L2A (5 M) in 0.25 M (NH 4 ) 2 SO 4 , 50 mM NH 4 -acetate, pH 6.0, 0.1 mM EDTA buffer were displayed into 0.2 ϫ 1.0-cm path length quartz cuvettes and left to equilibrate at 5°C. Trp-94 in RepA was then selectively excited (2-nm slit) at 295 nm (21), and emission spectra were acquired between 300 and 450 nm (3-nm slit, ⌬ ϭ 1 nm). For extrinsic fluorescence measurements, bis-ANS (Sigma) was supplied, from a stock in methanol, to a final concentration of 10 M. Samples were left to equilibrate for 30 min at 5°C and then were excited at 395 nm. Emission spectra were acquired in the interval of 400 to 600 nm. The contribution of the buffer was subtracted to all spectra. Circular dichroism analysis of His 6 -RepA-WT, His 6 -RepA-2L2A, and their complexes with DNAs was performed in a Jasco-720 spectropolarimeter, using 0.1-cm path length quartz cuvettes. The DNA oligonucleotides tested in the binding assays (RepA sites underlined) were as follows: 1IR (operator), 5Ј-GAACAAGGACA-GGGCATTGACTTGTCCCTGTCCCTTAAT-3Ј (39-mer); 1DR (iteron), 5Ј-ATACCCGGGTTTAAAGGGGACAGATTCAGGCTGTTATCCACAC-CC-3Ј (45-mer); TEL (unrelated), 5Ј-GATCCCACACCCACACACCCA-CACACCCACACACCCAG-3Ј (38-mer); plus their complementary strands. Oligonucleotides were purified, and their concentrations were determined as described (21). Annealing was carried out by slow cooling in Tris-EDTA buffer, and double-stranded DNA stocks (50 M) were stored at Ϫ20°C. Binding reactions (200 l) were assembled in ice, in a solution containing 0.145 M (NH 4 ) 2 SO 4 , 50 mM NH 4 -acetate, pH 6.0, 0.1 mM EDTA, 1 mM ␤-MeEtOH, 5% glycerol, plus 5 M protein and/or double-stranded DNA. Spectra were acquired at 5°C from 320 to 200 nm, using a bandwidth of 1 nm, 4-s response, 50 nm/min scan speed (0.2-nm steps), and 10 millidegrees of sensitivity. Five to ten spectra were accumulated for averaging. The spectrum of the buffer was subtracted, and raw ellipticity data (in millidegrees) were transformed to molar ellipticity ([]) (in degrees⅐cm 2 ⅐dmol Ϫ1 ). No differences in the spectra were appreciated when reactions were incubated for variable times, ranging from 15 min to 3 h, indicating that binding equilibrium is reached rather quickly (42). Secondary structure estimates were performed with CDNN (51) at its Web server (bioinformatik.biochemtech.uni-halle.de/cdnn/java/Started.html). Thermal denaturation analyses were performed on the same samples, overlaid with mineral oil. Ellipticity at 228 nm was measured while temperature of the cell holder was increased, using a computer-interfaced water bath, from 5 to 90°C (0.2°C steps, 20°C/h; scans performed at 50°C/h gave similar curves).
Size Exclusion Chromatography-Gel filtration assays were performed in a Superdex-200 HR10/30 column, assembled in an ÅKTA basic-10 HPLC equipment (Amersham Biosciences). The column was equilibrated at room temperature in 0.145 M (NH 4 ) 2 SO 4 , 50 mM NH 4acetate, pH 6.0, 0.1 mM EDTA, 1 mM ␤-MeEtOH, 5% glycerol. Duplicates (200 l) of the samples assayed previously by CD spectroscopy (see above) were then injected. Chromatography was carried out at 0.5 ml/min in the same buffer, measuring the absorption of the eluate at 254 nm to optimize DNA detection.

Rational Design and Functional Testing of a Monomeric
RepA Mutant (RepA-2L2A)-We had described previously that point mutations, changing to Val the Leu residues of the Nterminal LZ-like sequence motif found in RepA, resulted in the displacement of dissociation equilibrium from dimers toward monomers when RepA was fused to MBP (14). This observation pointed to some (direct or indirect) role for the putative LZ in RepA dimerization. Such a hypothesis received additional support by the finding that an N-terminal deletion of the LZ-like motif (⌬N37) results in an increase in the monomeric fraction of that RepA fragment (21). However, we have since noticed the following. (i) MBP-RepA fusions, though capable of binding the four iterons found in pPS10 origin of replication (14), have reduced binding cooperativity and fail to be wrapped by iteron DNA (not shown). (ii) When expressed in the absence of the fusion with MBP, those RepA Leu 3 Val mutants yielded insoluble proteins, which we have been unable to refold in vitro. To clarify both the role of the LZ-like motif in RepA dimerization and the conformational change intrinsic to RepA monomerization (21) we have used the crystal structure of a monomeric homologous protein, RepE54 (24), to design RepA-2L2A. This is a new RepA mutant in which the first and second Leu residues of the hydrophobic LZ heptad (Leu-12 and Leu -19) have been substituted by Ala (Fig. 1). In RepE54 monomers, the LZ-like motif is found dislocated into two ␣-helices, resembling a folded jack-knife. The shortest one (␣1) includes the first Leu residue (Leu-12 in RepA) and the largest (␣2) the third and fourth (Leu-26 and Leu-33), whereas the second (Leu-19) is found in the intervening turn (Fig. 1B). The rationale behind the new Leu 3 Ala mutations is as follows. (i) They would impair a possible dimerization through the putative LZ, because the side chain of Ala is smaller and less hydrophobic than that of Leu. (ii) They would necessarily disrupt the hydrophobic network linking ␣1, through Leu-12 and Leu-19, to the rest of the WH1 domain, mainly Trp-94. (iii) Mutations would thus allow ␣1 to move freely, resulting in an extended conformation of the proposed jack-knife.
RepA-2L2A mutant was constructed by site-directed mutagenesis on an expression vector including repA-WT gene fused to His 6 (21) (see "Experimental Procedures"). With the aim of testing the effect of the mutations on pPS10 replication in vivo, repA-2L2A was then transferred to a new pPS10-ColE1 shuttle vector ( Fig. 2A, pPSEC1), replacing the parental repA-WT gene, to get pPSEC3. To check the effect of the His 6 tag present in RepA-2L2A, a plasmid in which the repA-WT gene includes the sequence coding for the tag was also constructed (pPSEC2). After being rescued in E. coli, plasmids were transformed into P. aeruginosa. The transformation frequency of the repA-2L2A carrying plasmid decreased by three orders of magnitude when compared with those encoding for the WT proteins, whereas no differences were apparent between the tagged and untagged versions of RepA. After four rounds (ϳ80 generations) of replica plating with no selective pressure, the Kn resistance marker had been stably inherited in all clones. However, Southern blot analysis of total lysates from several Kn-resistant pPSEC3 (repA-2L2A) colonies showed that the plasmid was integrated in the chromosome, instead of being replicated autonomously (not shown). This observation suggests that the mutant RepA-2L2A protein is inactive as DNA replication initiator. Growth curves for cells carrying plasmids encoding repA-WT show no significant differences between constructs with or without the His 6 tag (Fig.  2B), confirming that the fused protein is fully functional as initiator. This was further proved by Southern blotting total lysates from mid-log phase cells, showing that there are no significant differences in plasmid copy number either (Fig. 2C). Western blot was performed, with anti-RepA polyclonal antiserum, on cells harvested at different stages of growth (Fig.  2B), showing similar protein levels for both versions of RepA-WT. Thus, besides being active as initiator, His 6 -RepA is also able to regulate its own synthesis (Fig. 2D). Approximated quantification of the RepA amounts yields about 860 protein molecules (or 430 dimers) per cell. This value is around the estimations for the Rep proteins of pSC101 (500 molecules) (41) and RK2 (300 molecules) (52) plasmids but far from the those reported for R6K (about 20-fold higher) (53) and P1 (160 molecules) (47) initiators. It is relevant for the results presented later in this paper to note that pPS10 RepA is found at concentrations over 5 M across all the growth curves (Fig. 2D).
EMSAs were performed to test whether the failure of RepA-2L2A to act as initiator is due to have its DNA binding prop-erties altered. Radiolabeled DNA fragments, including either repA operator sequence (Fig. 3A) or the four iterons found at pPS10 origin of replication (Fig. 3B), were incubated with increasing amounts of pure RepA protein, either WT (His 6 -tagged or untagged) or mutant 2L2A. The latter was used with the His 6 tag attached, because this increases its solubility and the tag, by itself, does not substantially affect DNA replication in vivo (Fig. 2, panels B and C). RepA-WT binding to operator DNA results in a typical pattern of two bands. These correspond to complexes including one or two protein dimers (Fig.  3A, D1 and D2, respectively) (13,21), to finally yield large protein-DNA aggregates that remain in the well of the gel (W). However, only the first of such complexes, that with a single RepA dimer (D1), is observed with the 2L2A mutant. Its apparent dissociation constant (K d(app) , the protein concentration at which 50% of the DNA probe is bound) is at least 40-fold higher (Ն0.7 M) than that for the His 6 -tagged RepA-WT (18 nM). Thus, although with a substantially decreased affinity, the RepA protein in which two Leu at its LZ-like motif (Leu-12 and Leu-19) were substituted by Ala seems able to dimerize upon binding to operator DNA. This observation points to the existence of an additional dimerization interface in RepA. It would most probably be the ␤-sheet in WH1, as proposed for RepE, based in the crystal structure of its monomer (24). Moreover, binding contacts of RepA-2L2A with the operator DNA must be unaffected, because they are made through WH2 (21) that it is intact in the mutant. The affinity of the tagged protein for the operator sequence appears to decrease; about 6-fold more His 6 -RepA-WT than RepA-WT is required to get a similar amount of bound DNA (Fig. 3A, lanes 1 and 3). In addition, the complex corresponding to the binding of a second RepA dimer (D2) does not appear. Thus, because the in vivo data show that both versions (tagged and untagged) of RepA-WT are equally selfregulated (Fig. 2D), the formation of complex D2 seems not essential for repA promoter repression. D1 complexes, established by His 6 -RepA dimers, show a bit lower electrophoretic mobility than those for the untagged protein, probably because of the extra 4-kDa mass coming from the His 6 tag. Concerning the binding of RepA monomers to the iterons at the origin ( Fig.  3B) (14), a sharp transition to large protein-DNA complexes that remain in the well of the gel (W), is observed. This occurs even with the minimal amount of untagged RepA-WT tested (7.5 nM), whereas only a tiny fraction of DNA is found in discrete complexes (M1-M4). Although untagged RepA appears to bind cooperatively to iterons, its His 6 version does not so much; DNA fragments with one to four iterons bound co-exist (M1-M4), and the complexes stacked in the well appear at a higher protein concentration (ϳ50 nM). However, the observed differences in binding cooperativity between tagged and untagged RepA-WT have no significant effect on their function as DNA replication initiators in vivo, because the copy numbers of plasmids coding for them are very similar (Fig. 2C). On the contrary, RepA-2L2A fails to bind stably to iterons (M1 and M2 complexes appear at protein concentrations about 200-fold those operational with RepA-WT; not shown). This is not surprising, because the domain altered in the mutant (WH1) was found in RepA-WT to bind an end of iteron sequence stably (21,24). The failure of RepA-2L2A to bind iterons in vitro is in accordance with the fact that the pPS10 derivative coding for this mutant (pPSEC3) cannot be established autonomously in vivo (see above). Because the precise nature of the structural alteration induced by the mutations in RepA remained to be  24), with its ␣-helical elements highlighted in red; ␣3 and ␣4 constitute the helix-turn-helix motif, in which the latter is the DNA recognition helix. B, slab view removing the helix-turn-helix to uncover ␣1 and ␣2, which include Leu residues conforming to an LZ-like sequence motif (19). The first two of such leucines (Leu-12 and Leu-19), substituted by Ala to generate the RepA-2L2A mutant, are depicted in green. In cyan is the single Trp in RepA (Trp-94) that is a crucial part of the network of hydrophobic interactions (distances displayed in white) to which Leu-12 and Leu-19 also contribute. These three residues, absolutely conserved in RepA, RepE54, and other Rep proteins (9,10,24), are numbered according to RepA sequence, but the others (in orange) keep the original labeling from RepE54 (24). A dashed purple arrow indicates the expected jack-knife movement that would undergo the N-terminal ␣1-helix around the hinge including Leu-19, once the hydrophobic network is disrupted by the mutations in RepA-2L2A that, in addition, would hamper any possible dimerization through the leucines. determined, we have performed a combined biophysical approach to further characterize the structure of RepA-2L2A.
Physicochemical Characterization of RepA-2L2A as a Metastable Folding Intermediate-One of the expected effects of the mutations designed in RepA-2L2A is to interfere with protein dimerization by altering the hydrophobic spine in the putative LZ ␣-helix (see above) (Fig. 1B). To test the association state of RepA, we have performed sedimentation equilibrium experiments in an analytical ultracentrifuge. We have also addressed whether the N-terminal His 6 tag affects RepA-WT dimerization (Fig. 4, panels A and B), before determining the association state of RepA-2L2A (Fig. 4C). In this protein the His 6 fusion was kept attached for improving solubility (see above). Ultracentrifugation runs were performed with a RepA range between the minimal concentration giving a reasonable signal to noise ratio with the absorption optics of the ultracentrifuge (1-2 M) and a maximum close to the limits of RepA solubility (25 M). RepA-WT remains essentially dimeric through all the concentration range tested (not shown), with a net tendency of the protein to assemble further, but no sign of dissociation. We had reported previously (14) for a fusion with MBP that RepA-WT, in the lowest concentration range tested here, was close to the dissociation equilibrium between dimers and monomers. Now, in light of our new data, that observed dissociation is revealed as a possible steric hindrance effect of the fused MBP moiety, probably caused by its large size (47 kDa). His 6 -tagged RepA-WT shows a similar behavior to its untagged counterpart, with a slightly more marked tendency to associate beyond dimers (not shown). This is in accordance with data shown above pointing to similar in vivo and in vitro activities for both proteins (see Figs. 2 and 3). Therefore, the rest of experiments described in this paper (see Figs. 5-7) were carried out with the His 6 -tagged version of RepA-WT, to be compared straightforward with those performed with His 6 -RepA-2L2A. Sedimentation velocity analysis of His 6 -tagged and untagged RepA-WT, at either 5 or 15 M, allowed to determine a sedimentation coefficient (s 20,w ) for the RepA particle of 4.2 S and a frictional ratio (f/f 0 ) of 1.2 (not shown). These values are again compatible with being the most abundant RepA species dimers, with nearly spherical shape. Thus it is relevant to underline that, based on the present analytical centrifugation analyses and on the conditions used in the spectroscopic studies described below, RepA-WT protein is largely found as stable dimers. Sedimentation equilibrium analysis on RepA-2L2A shows that this mutant protein is polydisperse; at the lower concentrations tested (2-7 M), a significant monomeric fraction is found (Fig.  4C), together with large aggregates. These appear to become the major species at higher protein concentrations (not shown). Aggregation, described previously (21) for deletion mutants affecting the putative LZ in RepA, can be attributed to the exposure of hydrophobic surfaces (probably an additional dimerization interface; see above) to the solvent. Both the disruption of RepA dimers and the exposure of hydrophobic residues, otherwise buried in the core of the folded N-terminal WH, were expected to arise from the double mutation designed in RepA-2L2A (Fig. 1B).

FIG. 2. Testing the functionality of RepA-2L2A and the effect of the N-terminal His 6 tag in vivo.
A, scheme of the shuttle plasmid series pPSEC, including a ColE1-type replicon, for maintenance in E. coli, and the pPS10 replicon, for propagation in Pseudomonas. The pPS10 origin of replication (oriV) and sequences 5Ј to the repA gene are enlarged to highlight the following differences between plasmids: having His 6 tags encoded in repA (pPSEC2/3) or not (pPSEC1) and including the repA-2L2A mutation (pPSEC3) or not (pPSEC1/2). All the clones transformed with pPSEC3 were found to have this plasmid integrated in the chromosome (not shown) and thus are not included in subsequent panels. B, growth curves for P. aeruginosa cells carrying the plasmids pPSEC1 or pPSEC2. C, Southern blotting of total lysates from mid-log P. aeruginosa cells, carrying the plasmids pPSEC1 or pPSEC2. Upper panel shows the chromosome bands stained with EtBr, whereas the lower part corresponds to the signal of 32 P-labeled repA gene hybridized with the plasmids. The relative plasmid copy number is indicated below each track, corrected for loading bias by chromosome density and normalized to the value of pPSEC1. D, Western blotting, with polyclonal anti-RepA serum, of total cell extracts obtained at different points of the growth curve in B. First two tracks, purified RepA standards loaded together with a total lysate from plasmid-free P. aeruginosa cells. The estimated amount of RepA, the number of cells contributing to the load, and the concentration of RepA in the cells are indicated below each track.
The possible structural changes induced by the 2L2A mutations were then explored by means of steady state fluorescence spectroscopy. Trp-94, the single tryptophan residue in RepA (18,21), is a key node in the network attaching Leu-12 and Leu-19 to the hydrophobic core of WH1 (Fig. 1B). Thus, it is expected to be a suitable spectroscopic sensor for any structural modification in its environment. RepA-WT has nine tyrosines (18). Comparing the excitation and emission spectra of RepA with those for free tyrosine and N-acetyl-tryptophanamide solutions we have verified that, at 295 nm, no Tyr residue is excited in RepA. On the contrary, that wavelength falls in the tail of the excitation spectrum of Trp-94 (not shown). Thus, Trp-94, either in RepA-WT or in RepA-2L2A (5 M), was selectively excited at 295 nm and then fluorescence emission spectra were acquired (Fig. 5A). The maximum emission was achieved at 327 nm (RepA-WT) or 348 nm (RepA-2L2A). The extra band at 335 nm observed in RepA-2L2A emission suggests the presence of two major populations of different rotamers for Trp-94 side chain. The red-shifted emission found for RepA-2L2A is characteristic of exposed Trp residues, and it is compatible with the increased solvent accessibility expected for Trp-94 after the release of its hydrophobic linkage to the bipartite ␣-helix (Fig. 1B). On the contrary, the emission of Trp-94 in RepA-WT, close to 320 nm, is typical of Trp residues buried in a protein core or in a contacting interface (21). The wavelength for the emission maximum in RepA-WT did not change after serial dilutions of the protein sample (tested up to 78 nM; not shown), providing further evidence for its dimeric association state in a broad concentration range. Another clue for a structural change affecting Trp-94 in RepA-2L2A is its enhanced fluorescence emission intensity compared with that for the WT protein (Fig. 5A). To get additional proof for the presence of exposed hydrophobic residues in RepA-2L2A, an extrinsic fluorescence probe was used: bis-ANS, a naphthalene derivative that binds to solvent-accessible hydrophobic patches in proteins, enhancing its fluorescence emission over 450 nm (21). Incubation of bis-ANS (10 M) with RepA-WT or 2L2A (5 M) results in spectra with emission maxima around 472 nm (Fig.  5B). The fluorescence intensity at this wavelength is 61% higher for Rep-2L2A than for the WT protein, confirming that the former has a larger hydrophobic surface exposed to the solvent.
As a summary of the results shown so far, RepA-WT is a stable dimer in vitro in a broad (M) concentration range (Fig.  4A), similar to that found in P. aeruginosa cells (Fig. 2D). In terms of RepA association state (Fig. 4B), DNA binding properties (Fig. 3), or even its activity as a DNA replication initiator in vivo (Fig. 2), there is no significant difference in having a His 6 peptide fused to the RepA N terminus. RepA-2L2A, in which two Leu residues in the putative LZ (Leu-12 and Leu-19) were substituted by Ala (Fig. 1), exposes hydrophobic residues to the solvent, as revealed by fluorescence studies (Fig. 5) and by its enhanced tendency to aggregation. Thus, it exhibits the biophysical properties expected for a metastable monomeric (Fig. 4C) protein folding intermediate (54).

Binding to Iteron DNA Dissociates RepA Dimers into Monomers and Acts as Allosteric Effector on Protein Conformation-
With the aim of deepening our understanding of the structural basis of the recognition of operator and iteron DNA sequences by RepA dimers and monomers, respectively, we have performed a CD spectroscopy study on RepA-DNA complexes (see Figs. 6 and 7). 5 M His 6 -tagged RepA-WT protein were incubated with equimolar amounts of double-stranded oligonucleotides, including either the repA operator sequence (1IR), a single iteron (1DR), or an unrelated sequence of similar size (yeast telomeric TEL). Fig. 6A shows overlaid CD spectra of the oligonucleotides alone and in complex with RepA-WT. The bands with positive ellipticity ([]) in the near UV range of the spectra (320 -250 nm) arise from the stacking of DNA bases, with nearly null contribution from protein aromatic residues. In the far UV (250 -200 nm) the predominant contribution comes from the protein moiety. Having no substantial differences in the spectra, between 320 and 250 nm, for the free and protein-bound states of a given DNA can be interpreted as a sign of no significant structural alterations in the DNA (55). Because that is the case for the distinct RepA-DNA complexes in Fig. 6A, the spectra of free DNAs can be subtracted from those of their complexes with RepA, thus focusing on possible structural changes on the protein side. The result of this algebraic operation is depicted in Fig. 6B, in which two types of protein spectra become evident: (i) Curves with double minima at about 208 and 222 nm, a signature for a significant proportion of ␣-helical secondary structure (33%), are one type. These correspond to free RepA-WT and in complex with the operator oligonucleotide 1IR. (ii) Curves with a broader minimum around 215 nm, attributable to an increase in the ␤-sheet component (by about 4%) at the expense of former ␣-helices, as observed for RepA-WT in complex with the iteron oligonucleotide 1DR, are another type. It is relevant that RepA-WT does not change its spectrum and thus its structure to a detectable extent, when it is free in solution or bound to the operator. This is coincident with the fact that RepA is essentially an homogeneous dimer in solution at this concentration (Fig. 4B) and that it binds to that DNA sequence as a dimer (Fig. 3A) (14). However, when it binds to the iteron DNA as a monomer (Fig. 3B) (14) its spectrum resembles that of the mostly monomeric RepA-2L2A mutant (Fig. 6B). We have failed to acquire clean spectra for complexes between this mutant protein and any DNA, because a cloudy precipitate appears. Although less severe, protein aggregation is also observed, in the form of noisy spectra, for the nonspecific complexes between RepA-WT and DNAs of unrelated sequence, such as TEL (Fig. 6), other mixed sequence oligonucleotides, and poly(dI-dC) (data not shown). As a summary, our spectroscopic studies show that RepA has a distinct secondary structure composition when bound to the inversely repeated operator sequence or to the directly repeated iteron DNA.
To correlate the observed conformational change with the association state of RepA, we carried out EMSAs (not shown), size exclusion chromatography (Fig. 6C), and sedimentation velocity (Fig. 6C, inset) on the same samples tested by CD. EMSAs consistently result in that the RepA-operator complex slightly has lower electrophoretic mobility than the RepAiteron band (not shown). Taking into account that the free iteron oligonucleotide (1DR) is 6 bp larger than the operator one (1IR), this result is compatible with having a larger protein mass in the latter complex (a dimer) than in the former (a monomer). However, EMSAs are not conclusive, because the minute difference in the electrophoretic migration between both types of complexes could be also because of disparate bending behavior of the bound DNA fragments. For the sample including RepA and the unrelated oligonucleotide TEL, no discrete retarded band is visible but some smearing, indicative of low affinity, nonspecific binding (not shown). When the samples were injected into a size exclusion column the specific RepA-oligonucleotide complexes remain stable, and their hy-drodynamic behavior (Fig. 6C) was unambiguous. The relative elution position of the peaks corresponding to bound DNAs is the reverse of that for the unbound species (the assay is sensitive enough to detect the aforementioned slight difference in the sizes of free oligonucleotides). This fact can be explained if the protein mass associated with the smaller operator DNA piece (1IR) is substantially larger (a dimer) than that associated with the longer iteron (1DR) fragment (a monomer). The possibility of having just one subunit in a RepA dimer bound to the iteron, as proposed for the Rep protein of R6K plasmid (16), is thus very unlikely. Such a complex would be expected to elute just before the RepA-operator one. To characterize further the nature of the complexes eluted from the gel filtration column, analytical ultracentrifugation was performed. Unfortunately, RepA-DNA complexes dissociate during the time course required to achieve sedimentation equilibrium (not shown). However, the interpretation proposed above is in agreement with the results of sedimentation velocity experiments performed, immediately after the gel filtration runs, with the peak fractions. There is a substantial difference in the sedimentation coefficients for the RepA-1IR (6.5 Ϯ 0.2 s) and the RepA-1DR (3.8 Ϯ 0.2 s) complexes. These sedimentation coefficients are only compatible with being the RepA-1IR complex significantly more compact (f/f 0 ϭ 1.2) than RepA-1DR (f/f 0 ϭ 2.0). The latter value fits with the elongated shape of a complex between a RepA monomer and an oligonucleotide with identical length to 1DR, modeled on the crystal structure of RepE54 (24) (not shown). Therefore, our results clearly show that the structural changes in RepA, coupled with binding to iteron DNA (Fig. 6B), are linked to dissociation of the otherwise stable protein dimers (Fig. 4B) into monomers (Fig. 6C).
It is noteworthy that, for unbound RepA, we have not found any sign of dissociation of dimers after incubation times ranging from 30 min (spectroscopy; see Figs. 5 and 6) to 16 h (sedimentation equilibrium; see Fig. 4). On the contrary, upon incubation with origin DNA RepA dissociation (Fig. 6C) and the coupled structural transformation (Fig. 6B) occur within a few minutes. Because serial dilutions of unbound RepA exhibit the fluorescence emission spectrum characteristic of dimers (Fig. 5A) at least up to low nanomolar concentrations (not shown), the possibility that at 5 M a small fraction of monomeric RepA, in equilibrium with dimers, would bind to the iteron sequence and that this event would drive monomer formation seems unlikely. In summary, there are solid basis to affirm the following. (i) Because the K d for RepA dimers ap-FIG. 7. Binding to operator or iteron DNAs increases the thermal stability of His 6 -RepA-WT. Thermal denaturation profiles were acquired for duplicates of the samples in Fig. 6 (the same colors and symbols are adopted), measuring the evolution of molar ellipticity ([]) at 228 nm (protein secondary structure) with increasing temperature. Inset contains thermodynamic parameters calculated from the curves. T m , temperature for half transition between folded (low values) and unfolded ( getting close to 0) states. ⌬T m , variation of the melting temperature of His 6 -RepA-WT in the complexes with DNA, compared with that for the protein alone. pears to be in the low nanomolar-subnanomolar range, in our experimental (micromolar) conditions the equilibrium is well displaced toward the dimeric species. (ii) Along broad time courses no monomeric unbound fraction has been detected for RepA. Thus the reported induction of RepA dissociation by iteron DNA seems to be genuine.
Thermal denaturation analyses performed by optical techniques, such as CD spectroscopy, provide valuable information on the thermodynamic stability of proteins (4,21) and their complexes with DNA (55). We have carried out this kind of approach to study the effect on RepA stability of binding to the DNA sequences described above (Fig. 6). Inspection of the CD spectra in Fig. 6A reveals that, at 228 nm, the contribution of the oligonucleotides to the ellipticity ([]) of their complexes with RepA is essentially null, whereas that from the ␣-helical and ␤-sheet structural elements in RepA (Fig. 6B) is significant. Thus, we plotted the variation of 228 with temperature for each kind of RepA-DNA complex (Fig. 7) and then compared the resulting curves. Because thermal denaturation of RepA and its complexes is irreversible, a detailed thermodynamic analysis of the CD profiles was precluded. The temperature at which 50% of protein molecules unfold (T m ) was determined, because it is essentially unaffected by irreversibility and gives an idea on protein stability. Binding to operator DNA (1IR) stabilizes RepA, relative to its unbound state, by 12.4°C, whereas binding to iteron DNA has a smaller effect (4.8°C). The more pronounced slope of the curve for 1IR complexes indicates that RepA denaturation becomes more cooperative upon operator binding by the dimers. This effect is also observed for the complexes of RepA monomers with iteron DNA (1DR), although to a lesser extent. However, binding to nonspecific sequences (TEL) severely decreases the stability of RepA (by 11.5°C), approaching the T m values measured for the mostly monomeric folding intermediate RepA-2L2A (Fig. 7) and for an N-terminal partial deletion derivative (⌬N37) reported previously (21). This observation suggests that, as described for some restriction endonucleases (56), nonspecific DNA could, to some extent, exert conformational changes in RepA (Fig. 6B) enabling it to scan for specific iteron sequences. However, the resulting RepA molecules seem to be unstable folding intermediates that, because of the exposure of hydrophobic residues prone to aggregation (see above), would recruit molecular chaperones. Therefore, iteron DNA not only acts as allosteric conformational effector on RepA dimers (Fig. 6B) but also has a role in stabilizing the structure of the resulting monomers.

RepA-2L2A: a Folding Intermediate in the Pathway from
Repressor Dimers to Initiator Monomers-In this paper we have described the design of RepA-2L2A, a mutant in the protein that initiates DNA replication of the Pseudomonas plasmid pPS10, based on the available three-dimensional structure of the monomer of a homologous initiator (24). RepA-2L2A carries a double Leu 3 Ala substitution (Leu-12 and Leu-19) in the LZ-like motif found in the protein (14,19,21). These mutations were expected to enhance dissociation of an otherwise dimeric protein into monomers and to destabilize the hydrophobic core of the latter (Fig. 1). Functional characterization of RepA-2L2A indicates that it is inactive as initiator of replication (Fig. 2). RepA-2L2A has affected its binding to iteron DNA sequences (Fig. 3B) and reduced its affinity for operator DNA (Fig. 3A), where RepA binds as a dimer (13,21). Hydrodynamic studies (Fig. 4) show that RepA-2L2A can be largely found as a monomer in a narrow micromolar range, forming nonspecific aggregates at higher concentrations. Dimerization of WH domains (23) through hydrophobic resi-dues in the ␣-helices equivalent to ␣2 and ␣4 in Rep proteins (Fig. 1) has been described for the complex between the eukaryotic transcription factors E2F4 and DP2 (57). However, a role in dimerization for the antiparallel ␤-sheet found at the N-terminal WH domain of RepE54 is also possible (24). To determine whether the LZ-like motif has a direct contribution to the interprotomeric contacts in RepA dimers, or whether it rather favors RepA association indirectly (e.g. stabilizing the dimeric conformation), will require solving the three-dimensional structure of a Rep protein dimer. Spectroscopic evidence (Fig.  5) supports that RepA-2L2A resembles a transient folding intermediate, with hydrophobic residues partially exposed to the solvent, expected to occur in the pathway from repressor dimers to initiator monomers. This is the first report on the isolation and characterization of a conformational intermediate in a Rep-type initiator protein. In addition, we have shown that short His 6 fusions to RepA have a minor effect on the association state of the protein in vitro (Fig. 4B), which remains as a dimer in a broad micromolar range. The His 6 tag does not alter RepA initiator activity in vivo (Fig. 2), despite showing some reduction in its binding cooperativity to iteron sequences in vitro (Fig. 3B).
The Allosteric Effect of Iteron Binding on RepA Conformation and Association State-At the same concentration in which RepA is a dimer (Fig. 4) with no sign of dissociation into monomers, binding to an oligonucleotide encoding for a single iteron origin sequence results in dissociation of RepA into monomers (Fig. 6C) and a change in protein secondary structure, which becomes similar to that for the monomeric intermediate RepA-2L2A (Fig. 6B). On the contrary, the structure of RepA dimers appears to be unaltered by binding to the inversely repeated operator sequence (Fig. 6B). Interestingly, binding to both types of DNA stabilizes RepA against thermal denaturation, although to a different extent (Fig. 7). This fact also points to differences in the structures of RepA dimers and monomers. In the same experimental conditions, nonspecific DNA oligonucleotides result in precipitation of RepA and in destabilization of the structure of the protein (Fig. 7). Structural changes in a number of DNA-binding proteins upon specific ligand recognition have been previously reported (44,45,56) as cases of allosterism. However, apart from being found in some viral initiators (58,59) and in ORC with single-stranded DNA (5), they are new events in the initiation of DNA replication in bacterial plasmids.
The precise mechanism through which iteron DNA exerts the allosteric effect on RepA structure described in this paper remains to be determined in its molecular details. It is worth noting that in the eukaryotic transcription factor Ets1 an ␣-helix (␣I-2) is found at the N terminus of the three ␣-helices of its WH domain (60), packing against the first helix very much as ␣1 and ␣2 do in Rep proteins (Fig. 1B). A phosphate in DNA backbone triggers an allosteric transition in Ets1 ␣I-2 by establishing a hydrogen bond with the amide NH of a Leu residue at ␣1 N terminus. This corresponds in RepE54 monomers with Arg-33 (Fig. 1B), which also binds to iteron phosphate backbone in the crystal structure (24), whereas modeling of Rep dimers suggests that it must be displaced apart from the operator phosphate backbone (24). Arg-33 in RepE54 consistently aligns with hydrophobic residues (mainly Leu, as in Ets1) in RepA, other members of its family (9,10,24), and the eukaryotic and archaeal initiators Orc4/Cdc6 (4). Thus a way is opened to a possible general role of this residue in triggering allosteric transitions in WH initiator proteins.
The Role of Chaperones and Specific Origin Sequences in the Structural Activation of Rep Proteins-At the concentrations it is found in Pseudomonas cells in vivo (Fig. 2D), RepA is dimeric (Fig. 4), acting as transcriptional repressor (13,20). It has been found for P1 (34,42) and pSC101 (41) plasmids that their Rep proteins can spontaneously dissociate into monomers just by dilution to low/sub-micromolar concentration. However, besides pPS10 RepA, the initiators of F (12), R6K (16,61), and RK2 (15,36) plasmids have their K d values in the low nanomolar range. For the latter the requirement for an active mechanism in Rep dissociation, and in the coupled conformational change, thus remains as a bottleneck. Our findings on the ability of an iteron sequence to induce structural transitions in RepA confer a property to origin DNA that was attributed previously to chaperone action alone (reviewed in Ref. 30).
Based on data presented in this paper, we propose that the structural changes in Rep that are coupled to dimer dissociation would imply the following: (i) releasing protein-protein interactions between protomers, involving the LZ-like motif and/or the ␤-sheet in the first domain (21,24); (ii) remodeling the hydrophobic core of WH1, where the relevant Leu residues (Leu-12 and Leu-19) are tightly packed with Trp-94 (Fig. 1B); (iii) an increase in ␤-strand structure at the expense of the ␣-helical components (Fig. 6B). The rearrangement experienced by Rep would necessarily generate transient intermediates with properties similar to those of the RepA-2L2A mutant, namely having hydrophobic sequence patches exposed to the solvent and a loosely folded core (Fig. 5), both features of molten globules (54). This explains why monomeric RepA, when free in solution and in the absence of iteron DNA, is prone to aggregation through WH1 (see above) (21). To cope with these challenges to Rep activation, different, albeit complementary, molecular mechanisms can be envisaged.
Chaperones can directly dissociate Rep dimers and simultaneously change the conformation of the monomers, to make them competent for iteron binding. This has been reported for the DnaK-DnaJ-GrpE triad (11,31,(33)(34)(35), ClpA (32,43), and ClpX (36). These could target WH1 in Rep; ClpA recognizes a region at the N terminus of P1 Rep (residues 10 -70) (62) as the first step toward unfolding, either for its activation as initiator or for degradation by the associated ClpP protease (37,38). DnaK, on its side, recognizes a hydrophobic patch (residues 36 -49) in the same Rep protein (63), which in pPS10 RepA corresponds to a sequence (residues 91-105) comprising the Trp-94 residue discussed above. We had shown previously (64) that pPS10 replication in vitro is sensible to DnaK levels. Although they physically interact (4), if DnaK is acting on pPS10 RepA dissociation remains to be determined. For those Rep dimers that dissociate spontaneously, chaperones could bind to exposed hydrophobic regions (65) in the monomeric intermediates, thus protecting them from falling into local energy minima that would act as kinetic traps in the pathway leading to Rep folding and iteron binding. Otherwise, large Rep aggregates could arise from the accumulation of monomeric intermediates, but it seems feasible that the combined action of several chaperones, such as ClpB plus DnaK-DnaJ-GrpE (66), could rejuvenate them into active initiators.
The allosteric conformational changes induced in RepA by iteron DNA (Fig. 6) would constitute another means to get the monomeric species, specially for those Rep proteins that are stable dimers at the concentrations they are found in vivo (Fig.  2D) (16,52,53). This way could be favored by the cooperativity of Rep binding to DNA (Fig. 3B). Alternatively, iteron DNA could also capture and stabilize the monomeric folding intermediates produced after spontaneous dissociation of Rep dimers, as proposed above for chaperones. An implication of this new activation mechanism is that only a fraction of the total amount of Rep molecules, coincident with the number of iteron sequences, would be eventually activated whereas the others would remain dimeric. On the contrary, there is no obvious means to limit the number of active Rep monomers generated by the chaperone-mediated route that could be accumulated in excess over the levels required for regulated initiation. In addition, the allosteric binding of iteron DNA to RepA implies that the structural changes associated to monomerization are intrinsically accessible to Rep proteins and not necessarily the exclusive product of chaperone action.
In plasmids containing iteron sequences at their origins of replication, the most favored model for negative control of initiation implies pairing distant Rep-bound iteron repeats ("handcuffing") (30,67). Except in R6K plasmid (16) handcuffing appears to be mediated by Rep monomers. However, in some Rep proteins, a number of mutants have been isolated that, being monomeric hyperactive initiators, fail to pair iterons (40,68). Most of such mutations fall in a putative additional dimerization interface in Rep that, besides the LZ-like motif, we have discussed for RepA-2L2A. Thus, Rep proteins activated by the allosteric route resemble the behavior of those hyperactive mutants. The minimal requirements for pairing are two DNA fragments containing each a single iteron repeat (69), but we have failed to detect handcuffing even at 5 M RepA-iteron complex concentration (Fig. 6C). Our model leaves open the possibility for the existence of two kinds of Rep monomers, correlated with their opposite functions as initiators or negative regulators of plasmid DNA replication. Thus, unwinding of origin DNA after the formation of the Rep-iteron nucleoprotein complex might induce a further conformational change in the monomers, as described for the chromosomal initiators ORC in eukaryotes (5) and DnaA in bacteria (6). RepA would then become competent for origin pairing but disabled for a new round of initiation on recently replicated DNA.