SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

The P1 plasmid origin has an array of five binding sites (iterons) for the plasmid-encoded initiator protein RepA. Saturation of these sites is required for initiation. Iterons can also pair via their bound RepAs. The reaction, called handcuffing, is believed to be the key to control initiation negatively. Here we have determined some of the mechanistic details of the reaction. We show that handcuffed RepA–iteron complexes dissociate when they are diluted or challenged with cold competitor iterons, suggesting spontaneous reversibility of the handcuffing reaction. The complex formation increases with increased RepA binding, but decreases upon saturation of binding. Complex formation also decreases in the presence of molecular chaperones (DnaK and DnaJ) that convert RepA dimers to monomers. This indicates that dimers participate in handcuffing, and that chaperones are involved in reversing handcuffing. They could play a direct role by reducing dimers and an indirect role by increasing monomers that would compete out the weaker binding dimers from the origin. We propose that an increased monomer to dimer ratio is the key to reverse handcuffing.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Proteins that bind to specific DNA sites often also pair them (Schleif, 1992). In the case of Escherichia coli lac repressor, phage lambda repressor, and the EBNA1 protein of Epstein Barr virus, the protein–protein interactions improve the strength of the DNA–protein interactions, and thereby regulate the efficiency of the reactions in which they participate (Ruusala and Crothers, 1992; Frappier et al., 1994; Dodd et al., 2001). Pairing has been shown to be involved in a variety of biological processes including transcription, replication, site-specific recombination, plasmid segregation and conjugation (Schleif, 1992; Chattoraj, 2000; Edgar et al., 2001; Radman-Livaja et al., 2003; Zhang et al., 2003). Some of these processes are also controlled at the pairing step. For example, in the gal operon, addition of inducer disturbs the protein–protein interactions between the two DNA bound gal repressors and allows transcription initiation (Geanacopoulos and Adhya, 2002). In the case of the ara operon, addition of inducer leads to a change of pairing sites that switches the operon from a repressed to an active state (Schleif, 2003).

In a large family of bacterial plasmids, binding of the initiator protein to multiple sites at the origin of replication (called iterons) is essential for initiation of replication. Extra iterons present in cis to the origin or provided in trans reduce plasmid replication, possibly by interacting with origin iterons via the bound initiators and thereby causing steric hindrance to origin function (Pal and Chattoraj, 1988; McEachern et al., 1989). As the plasmid copy number increases, so possibly do the interactions among the iterons of sister plasmids, making initiation a decreasing function of iteron concentration. This simple model can explain how plasmid over-replication is prevented.

The pairing of iterons via bound initiators in cis (DNA looping) or in trans (handcuffing) has been demonstrated by a variety of techniques in vitro (Mukherjee et al., 1985; 1988; McEachern et al., 1989). Purified initiators can handcuff iterons without any host factors. Some genetic data are also best explained assuming the handcuffing model (Pal and Chattoraj, 1988; McEachern et al., 1989). Initiator mutants that allow plasmids to replicate with a higher copy number have been found to be handcuffing defective in a majority of cases (McEachern et al., 1989; Miron et al., 1994; Mukhopadhyay et al., 1994; Blasina et al., 1996; Uga et al., 1999). Recently, physical evidence in favour of pairing has been obtained in vivo (Park et al., 2001). It is therefore reasonable to conclude that handcuffing has a significant role in iteron-mediated control of plasmid copy number. However, the biochemical details of the handcuffing reaction remain to be established.

In all iteron-carrying plasmids, the initiator protein is found as both monomer and dimer, but the form that participates in handcuffing is not known. The monomers have two DNA-binding domains and contact iterons at two consecutive major grooves on the same face of the DNA (Komori et al., 1999; Sharma et al., 2003). The N-terminal DNA-binding domain cannot contact DNA when the protein dimerizes but the C-terminal domain can (Giraldo et al., 2003). The dimers therefore recognize only one half of each iteron sequence (Urh et al., 1998). Although weak in iteron binding, dimers could be suitable for handcuffing because they are stable. The purified initiator protein is found mostly as dimers. The abundance of dimers relative to monomers may well compensate for their weaker affinity for iterons.

If the role of handcuffing is to block initiation, then it must be reversed if replication is to resume. The mechanism of reversal is not known. Normally, chaperones convert dimers to monomers and increase monomer concentration so that they can saturate the origin iterons and effect initiation (Wickner et al., 1992). In some cases iteron DNA itself catalyses the protein remodelling required for conversion of dimer to monomer (Giraldo et al., 2003; Abhyankar et al., 2004). Assuming dimers are responsible for handcuffing, their conversion to monomers at least by chaperones affords a ready mechanism for handcuffing reversal. Here we have characterized the handcuffing reaction in vitro primarily to address the reversal question. We find that the reaction follows the law of mass action: efficiency of handcuffing increases with increasing RepA-bound iteron concentrations, and decreases when the reaction mixture is diluted or challenged with competitor DNA. The handcuffed complexes can therefore reverse spontaneously in vitro. The efficiency of handcuffing also decreases when the origin iterons are saturated with RepA monomers or in the presence of chaperones, indicating that increasing monomers over dimers is primarily responsible for handcuffing reversal.

Our results are significant in understanding how plasmid copy number is maintained by iterons. As the copy number increases, the increased dosage of the initiator gene also causes total initiator concentration to rise. This will be true even if the initiator gene is autoregulated, as is the case with most of the iteron-carrying plasmids, except that the increase will be damped compared with a constitutively synthesized gene (Sompayrac and Maalφe, 1973). In a monomer–dimer equilibrating system, this means that the dimers will increase at a faster rate than monomers, making the negative control by dimers increasingly effective. Thus, if dimers can serve as inhibitors they can provide the negative feedback required to maintain the plasmid copy number. It appears that for their role in the homeostasis of plasmid copy number, initiator dimerization has been a conserved feature of iteron-carrying plasmids.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

A ligation assay for handcuffing

A convenient and widely used assay for handcuffing is the measurement of the kinetics of ligation of iteron-carrying fragments in vitro in the presence of initiators (Mukherjee et al., 1988; McEachern et al., 1989; Miron et al., 1994; Mukhopadhyay et al., 1994; Uga et al., 1999). It is based on the premise that handcuffing increases the local concentration of DNA ends and thereby increase the intermolecular ligation rate. The effect of the P1 plasmid initiator, RepA, on ligation kinetics was tested using a fragment carrying all five origin iterons in their natural disposition and a control fragment without iterons. A sample of the binding mixture was taken to monitor RepA binding by electrophoretic mobility shift assay (EMSA) using a 5% polyacrylamide gel (Fig. 1A). To the remainder of the mixture T4 DNA ligase was added, and the products were monitored by agarose gel electrophoresis after deproteinization. The iteron-carrying fragment formed a ladder of multimers, and their proportion increased as RepA binding increased (Fig. 1B, lanes 3–7, and Fig. 1C). There was a modest increase in circle formation in the presence of RepA, probably because RepA is known to bend DNA (arrow, Fig. 1B, lanes 2–7) (Mukhopadhyay and Chattoraj, 1993). The bending presumably brought the ends of linear fragments closer together and facilitated circularization. The control fragment without iterons formed only monomer circles with or without RepA (Fig. 1B, lanes 9–14). The circular products were identified by running identical samples in parallel gels with and without ethidium bromide (Fig. 1D). The circular isomers migrated ahead of their linear counterparts because of positive supercoiling caused by intercalation of ethidium bromide. The ligated products could be seen within seconds of adding ligase (Fig. 1E). We conclude that multimer formation by the ligation assay records interactions specific to the iteron-carrying fragments and is efficient in recording higher-order RepA–iteron interactions, in support of the existence of handcuffing.

image

Figure 1. Ligation assay for handcuffing. A wand B. Iteron dependence of RepA-mediated ligation enhancement. A. Electrophoretic mobility shift assay (EMSA) using a 5% polyacrylamide gel in the presence of 0–20 ng of RepA in a 20 µl reaction volume (1–30 nM RepA). Lane M1 contains 1 kb ladder (New England Biolabs). Lanes 1–7 contain a 251 bp labelled fragment (an EcoRI–HindIII fragment from pALA631; Abeles et al., 1990) carrying the five origin iterons of plasmid P1. Lanes 8–14 contain a 349 bp labelled fragment (a AseI–NarI fragment from pNEB193, New England Biolabs) used as no-iteron control. B. Same as (A), except that the samples were treated with T4 DNA ligase and then deproteinized before running on a 1% agarose gel. Lanes 1 and 8 are controls without ligase. Arrows indicate monomer circles. C. Histogram showing the relation of binding or ligation as a function of RepA concentration. Binding was estimated from the loss of intensity of the free (unbound) probe from EMSA, and ligation from the intensity of dimers and higher multimers as a fraction of total DNA. The total DNA included the monomers, both linear and circular, as the latter could form without RepA. D. Same as (B), except that the time of ligation was varied and the electrophoresis was in the absence or presence of ethidium bromide (0.5 µg ml−1). Filled and empty arrows indicate monomer and dimer circles respectively. E. Histogram showing the kinetics of ligation.

Download figure to PowerPoint

Ladder formation in fragments carrying varying number of iterons

Ladder formation increased with increasing RepA (Fig. 1) or increasing number of iterons per fragment (Fig. 2). We made an isogenic set of fragments carrying one, two, three, four and five iterons, and subjected them to ligation assay under conditions of saturation of binding (Fig. 2A). Ladder formation was efficient when the fragments carried three or four iterons (Fig. 2B and C). Ladder formation also improved by increasing iteron concentration, either by increasing the number of iterons per fragment or by increasing the concentration of iteron-carrying fragments (Fig. 2D and E). The results indicate that ladder formation follows the law of mass action but the tandem repeats are intrinsically more efficient: a 4× concentration of one-iteron fragment produced less ligated products than 1× concentration of four-iteron fragments (Fig. 2D and E, lane 3 versus lane 7). This implies that multiple bridges stablilize the handcuffed complexes. An exception to the mass action rule was seen in the case of the five-iteron fragment when saturated as discussed below.

image

Figure 2. Effect of varying iteron numbers in RepA-mediated ligation enhancement. A. The fragments were 320–386 bp and contained one, two, three, four or five iterons but were otherwise isogenic. They were polymerase chain reaction (PCR) amplified from pSM105–108 and pRJM358 (Sozhamannan and Chattoraj, 1993), respectively, using same two primers for all. B. Same as (A) except ligase was added as in Fig. 1B. C. Histogram showing ligation efficiency as a function of number of iterons per fragment. D. Ligation at varying DNA concentrations (75, 150 and 300 pM) using fragments (≈ 600 bp in length, PCR amplified as above but with different sets of primers) carrying one, two or four iterons in the presence of 20 ng RepA. E. Histogram showing ligation efficiency as a function of iteron concentrations per fragment.

Download figure to PowerPoint

Handcuffing reduction upon origin saturation with RepA

With the five-iteron fragment, multimer formation declined marginally but reproducibly compared with fragments carrying three or four iterons (Fig. 2C). It appears that binding of three to four but not all five iterons is optimal for ladder formation. The reduction was modest probably because of failure to saturate the iterons completely. However, increasing RepA to force saturation increases multimer formation possibly because of increase of inactive RepA dimers as will be discussed. The reduction was restricted to a limited RepA concentration range.

The origin iterons are phased, two helix-turns apart. Our earlier studies indicated that cumulative RepA-induced bending, about 40° per iteron, leads to intramolecular folding of the origin that retains one positive supercoil of DNA (Mukhopadhyay and Chattoraj, 1993; Edgar et al., 2001). The folding may preclude trans-pairing of origins and thereby reduce ladder formation. In other words, the structural transformation of origin DNA upon saturation could be responsible for the reduced ladder formation. In order to address the role of structural transition, the phasing of origin iterons was disturbed by inverting the middle iteron, ♯12 (Fig. 3A). As RepA binding approached saturation, ladder formation by the modified fragment increased (Fig. 3C, lane 13 versus lane 14; Fig. 3D), compared with a modest decrease seen with the natural origin (Fig. 3C, lane 6 versus lane 7; Fig. 3D). Note that the intensity of unligated monomer DNA band modestly increased in one case but decreased in the other (Fig. 3C, lane 7 versus lane 14). This suggests that the structural transition of origin iterons upon saturation may help to reverse handcuffing.

image

Figure 3. Effect of disturbing iteron orientation on multimer formation. Two different iteron-carrying fragments were used. In lanes 1–7, the iterons are in their natural disposition and present in a fragment identical to that used in Fig. 1. The fragment in lanes 8–14 is isogenic except that the central iteron (iteron ♯12) is inverted as shown in (A).

Download figure to PowerPoint

To understand the significance of phasing in vivo, a plasmid with iteron ♯12 inverted, pNND32, was constructed which is otherwise isogenic to a pUC19-P1ori chimeric plasmid, pALA631 (Abeles et al., 1990). When the plasmid pair, pALA631 and pNND32, were used individually to transform a polAts strain BR2994 that has an integrated copy of P1repA gene, no transformants could be seen with pNND32 at 42°C, whereas pALA631 gave about 2000 colonies (data not shown). The near normal binding of RepA in vitro but drastic drop in transformability in vivo of the mutant origin indicate that the phasing of iterons is important for initiation of replication, most probably for the formation of a specific higher-order RepA–iteron complex.

Reversibility of handcuffing

RepA at least up to several micromolar concentration is found to distribute between monomeric and dimeric forms, and not into higher-level multimers when studied by analytical centrifugation. In that case, only a single RepA link is possible between two fragments carrying single iterons. Multimers of such fragments can only form in a stepwise fashion during the ligation reaction. First, a linear dimer is formed (Fig. 4A). A DNA–protein or protein–protein contact breaks before a new DNA fragment can pair and generate a trimer. The process repeats itself to generate longer multimers. When fragments carry two or more iterons, ladder formation could also arise from aggregates. As single iterons can pair, fragments with multiple iterons have opportunities to interact simultaneously with multiple fragments that could generate large RepA–DNA complexes. To test this possibility, ligation was started after dilution of RepA–iteron complexes or after challenging the complexes with competitor DNA. Our expectation was that these treatments would not affect ladder formation if the aggregates form in an irreversible fashion.

image

Figure 4. Reversal of handcuffing upon dilution. A. Presumed steps of multimer formation in the presence of ligase. B and C. Experiments were carried out as in Fig. 1 except that after RepA and DNA were allowed to form complexes, they were diluted two-, three- or fivefold, maintaining the concentrations of all reactants except labelled DNA (lanes 4–6) before adding ligase. Lanes 1 and 2 are controls without RepA and lane 3 is the sample before dilution. D. Histogram showing the effect of dilution on ligation.

Download figure to PowerPoint

The RepA–DNA complexes were formed as before, using fragments carrying all five iterons. They were then diluted two-, three- and fivefold, maintaining in the diluent concentrations of all components [RepA, chaperones, poly(dI-dC) and ligase] except the iterons. We confirmed by EMSA that the dilution did not decrease RepA binding (Fig. 4B). However, upon dilution multimer formation decreased, indicating spontaneous reversibility of RepA–iteron complexes (Fig. 4C and D).

The labelled iteron–RepA complexes were also competed with two-, five- or 10-fold excess of cold (unlabelled) iteron-carrying DNA (Fig. 5). There was a gradual drop in the concentration of labelled ligation products with increasing competitor DNA (Fig. 5B and C). As expected, the size of the multimers increased at the expense of smaller length products (e.g. Fig. 5B, lanes 4 and 5 and 10 and 11). The proportionate reduction of all labelled length classes (e.g. Fig. 5B, lanes 4 and 6 and 10 and 12) indicates that aggregates, if present, were freely available for interaction with the competitor DNA.

image

Figure 5. Reversal of handcuffing in the presence of competitor DNA. A and B. Experiments were carried out as in Fig. 1 except that after RepA and labelled DNA were allowed to form complexes, 2 µl of binding buffer containing zero-, two-, five- or 10-fold excess of cold competitor iteron DNA was added (lanes 4–7). Lane 8 is same as lane 7 except that labelled and cold DNAs were pre-mixed before addition to RepA. Lanes 1 and 2 are controls without RepA whereas in lane 3 no competitor was added. Lanes 9–14 are same as lanes 3–8 except for the RepA concentration. C. Histogram showing the effect of competitor DNA on binding and ligation.

Download figure to PowerPoint

The results of binding in the presence of cold competitor were conspicuous in another respect. The competitor was inefficient in destabilizing pre-formed RepA–iteron complexes (Fig. 5A, lanes 5–7 and 11–13). This result is consistent with our previous in vivo results which showed that the autorepressed repA promoter is not induced when extra (competitor) iterons are provided in trans (Chattoraj et al., 1988). We proposed that the competitor iterons first titrate free RepA and that these iteron–RepA complexes then handcuff with the promoter–RepA complexes and prevent their dissociation. The increase in the affinity of DNA–protein interactions by handcuffing has been demonstrated in a variety of systems (Brenowitz et al., 1990; Ruusala and Crothers, 1992; Frappier et al., 1994; Dodd et al., 2001). When the labelled and cold DNAs were added together, less RepA bound to labelled DNA, indicating that the competitor was binding proficient (Fig. 5A, lanes 8 and 14).

In summary, ladder formation with single-iteron fragments is suggestive of reversibility of handcuffing. The dilution and competition experiments are consistent with this notion. The aggregates in question must be dynamic as they dissociate upon dilution and interact freely with competitor DNA.

Iteron orientation in handcuffed complexes

As handcuffing involves specific DNA–RepA and possibly specific RepA–RepA interactions, one may expect that the iterons would be oriented in a fixed manner in handcuffed complexes. To test orientation specificity, the ligated products of a handcuffing reaction was analysed after cleaving with a restriction enzyme whose site is located asymmetrically in the fragment. Unexpectedly, the products were consistent with both parallel and anti-parallel pairing (data not shown). It is possible that in spite of a specific orientation, the dangling DNA arms of a handcuffed pair were flexible enough to allow ligation of one end of one of the members to either end of the partner. This is likely in view of the high efficiency with which circular monomers were produced with most of our fragments. Reducing the length of the arms from 132 to 21 bp reduced ladder formation drastically (data not shown). Further tests for orientation specificity were carried out in vivo using replicating plasmids.

An extra iteron was placed downstream of the origin in either orientation at various distances separated by half a helical turn of DNA (Fig. 6). The effect of orientation on copy number was significant only when the extra iteron was two or three full turns of helix apart (pNND28 and pNND28rev, and pNND30 and pNND30rev). This behaviour is characteristics of DNA looping (Hochschild and Ptashne, 1986). We believe the copy number dropped in pNND28rev and pNND30rev because pairing in cis of the extra iteron with one of the origin iterons interfered with initiation. The looping model is supported by the fact that the difference in copy number was less when handcuffing-defective RepA mutants, m5 and m9, were used (Mukhopadhyay et al., 1994). These mutants that increase plasmid copy number have been characterized extensively in vivo (Park et al., 2001) and in vitro (Mukhopadhyay et al., 1994) and shown to be pairing defective. The iteron pairing therefore appears to have orientation specificity in vivo.

image

Figure 6. Effect of phasing and spacing of an extra iteron on mini-P1 copy number. Five origin iterons are shown schematically as ([RIGHTWARDS ARROW])5. An extra iteron ([RIGHTWARDS ARROW]) was inserted next to the origin iterons separated by 16, 21, 26 or 31 bp of non-P1 spacer sequence in two orientations. The plasmids are isogenic otherwise. m5 and m9 are two RepA mutants previously characterized to confer higher copy number because of handcuffing defect (Mukhopadhyay et al., 1994). Copy number of pNND26 was taken as 8 (Pal and Chattoraj, 1988). All other copy numbers are relative to this value and are shown as mean ± SD from four measurements (DNA samples from two independent colonies, each loaded in duplicates during gel electrophoresis).

Download figure to PowerPoint

Participation of RepA dimers in handcuffing

Purified RepA is found largely as dimers and binds poorly to iterons unless exposed to chaperones that enrich the monomer fraction (Wickner et al., 1991). This has led to the conclusion that only monomers bind to iterons. Subsequently, crystallographic evidence in support of monomer binding has been obtained (Komori et al., 1999). Upon concentration, the active monomers again dimerize and become chaperone dependent for iteron binding (Wickner et al., 1991). Therefore, even in the presence of chaperones both the RepA forms can be present. However, fragments with a single iteron show a single EMSA-stable retarded species suggesting again that only monomers can stably bind to iterons. An exception has been found in the case of R6K plasmid where a single iteron shows two retarded species, one monomer bound and the other dimer bound (Urh et al., 1998; Abhyankar et al., 2004). This has prompted the proposal that dimers could be effecting handcuffing with each protomer contacting one of the pairing iterons. Genetic studies with RK2 plasmid also suggested dimer participation in handcuffing (Toukdarian and Helinski, 1998; Urh et al., 1998). In order to obtain more direct evidence in support of dimer participation in handcuffing, we tested the role of chaperones in handcuffing. Although the monomers and dimers cannot be cleanly separated, the P1 (and F) system offers the opportunity to change their ratio dramatically by the use of chaperones. Our expectation was that in the absence of chaperones dimers would dominate and thereby promote handcuffing, even though monomer binding would reduce as judged by EMSA.

As expected, monomer binding decreased in the absence of chaperones, but ladder formation was significantly stronger than that seen in the presence of chaperones (Fig. 7A–C). When chaperones were present, ladder formation was significant only at the highest RepA concentration (10 ng) (Fig. 7B, lane 16, and Fig. 7C). At this concentration, we believe that chaperones were limiting and the samples contained considerable RepA dimer. The ladder formation also decreased when RepA concentration was kept constant and chaperone concentration was raised (Fig. 7D–F). The conditions that improved monomer binding led to decrease in multimer formation. This makes us conclude that the chaperone-activated species that binds avidly to iterons is not the species that causes ladder formation by handcuffing.

image

Figure 7. Role of chaperones in handcuffing. A–C. Experiments were done as in Fig. 1, using a fragment containing four iterons (as in Fig. 2) except that no chaperones were used (lanes 1–8). D–F. Same as (A–C) except that the chaperone concentration was varied. G–I. Same as (A–C) except that polyvinyl alcohol (PVA) was omitted from the reaction buffer. The overall efficiency of ligation decreased in the absence of PVA; however, the effect of chaperones on iteron binding (EMSA) remained the same.

Download figure to PowerPoint

The inverse relationship of monomer binding and ladder formation was also evident when the molecular crowding agent PVA was omitted. PVA was included in all our earlier reactions to enhance the formation of intermolecular complexes. In the absence of PVA, ladder formation was stimulated by RepA only in the absence of chaperones (Fig. 7G–I). Without PVA, the chaperone activation of iteron binding was still efficient, as evidenced by EMSA, but ladder formation was negligible. The most likely explanation is that in the absence of crowding, the monomers stayed active and did not associate to form dimers. The absence of dimers presumably abrogated ladder formation. These experiments indicate that dimers are required for handcuffing.

Iteron binding of RepA dimers

Binding of RepA is a prerequisite for handcuffing. If dimeric RepA is responsible for handcuffing, it has to bind to iterons. Clear evidence for dimer binding has been obtained only in the case of plasmid R6K (Urh et al., 1998; Abhyankar et al., 2004). Crystallographic and molecular modelling studies, however, predict that one of the two DNA-binding domains of RepA should still be functional in the dimer and bind to a particular iteron half-site (the one that recognizes the C-terminal DNA-binding domain). Dimer binding will obviously be weak compared with the monomer binding where both the iteron halves are contacted by the two DNA-binding domains of RepA. The weak dimer binding although not generally seen by EMSA could be strong enough to permit assay for multimer formation in the presence of ligase. To test for dimer binding in the P1 system, isogenic DNA sequences were created by annealing oligos where the first 10 bp region was kept intact whereas the next 9 bp region was replaced with random sequences, and conversely, the first 10 bp region was randomized keeping the last 9 bp region intact (Fig. 8A). These sequences, along with the intact iteron, were subjected to EMSA. A new retarded band presumably representing dimer binding was observed only with the predicted first half-site (Fig. 8B, lane 15). The interaction was weak as it required about 50-fold higher RepA concentration and was observed only in the presence of low non-specific DNA (Fig. 8C). While monomer binding to intact iteron could be seen in the presence of 300 ng poly(dI-dC), dimer binding to the first half-site was abolished at concentrations of poly(dI-dC) beyond 10 ng (data not shown). Surprisingly, the dimer binding improved in the presence of chaperones (Fig. 8C, lane 7 versus lane 13). The result is unexpected but can be explained. The binding-proficient (‘active’) dimers presumably formed by reassociation of active monomers. As chaperones increase the monomer pool, they are expected to increase the rate of monomer reassociation to form ‘active’ dimers which could bind to the first half-site. These dimers are apparently different from the dimers which never interacted with chaperones and consequently remain inactive. The binding of (active) dimers is not seen when intact iterons are used because monomers bind avidly to them and are expected to compete out the much weaker binding dimers. However, when only the first iteron-half is provided, monomer and dimer binding is expected to have comparable affinity, as was observed (Fig. 8B, lane 15, and Fig. 8C, lane 13). A second retarded species using a particular iteron half can therefore be taken as an evidence for dimer binding. In another instance, chaperones seemed to have improved RepA dimerization when phage λ was examined in vitro using λ repressor-RepA hybrid proteins (Dibbens et al., 1997). In these experiments, the dimerization domain of the repressor was replaced by RepA and efficient operator binding was dependent on RepA's ability to dimerize. Operator binding of λ repressor-RepA hybrid improved upon addition of chaperones.

image

Figure 8. Binding of RepA dimers to an iteron first half-site. A. DNA fragments were created by annealing oligos 33 bp long that contained the 19 bp full iteron sequence, the first 10 bp of iteron sequence with remaining 9 bp of random sequence or the second 9 bp of iteron sequence preceded by 10 bp of random sequence. B. EMSA with all the three fragments described in (A). All lanes contained 5 ng of poly(dl–dC). The full iteron showed strong monomer binding while the first half-site showed a second retarded band possibly representing dimer binding (lane 15, arrow). No binding could be seen with the second half-site. C. EMSA showing titration with poly(dl–dC) or RepA binding to the full and the first half-site iteron in the presence and absence of chaperones. Lanes 1 and 2 show the labelled fragments. 50 ng of RepA was added to lanes 3–14. A band representing possible dimer binding (arrow) was obtained with the first half-site, particularly in the presence of chaperone (lanes 7 and 13).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

A negative regulatory mechanism is essential for maintenance of plasmid copy number (Pritchard, 1978), and in iteron-carrying plasmids, the handcuffing reaction (pairing of iterons via bound initiators) is currently believed to serve that role. Using the plasmid P1 system, we have confirmed in vitro that fragments carrying a single iteron in the presence of the cognate initiator, RepA, form a ladder of multimers when DNA ligase is included in the reaction (Fig. 2). As RepA can only dimerize and does not form higher-level aggregates (DasGupta et al., 1993), formation of multimers of DNA sequences carrying single iterons indicates that handcuffing is a reversible reaction; were it irreversible the products would have been of only dimeric length. The phasing-sensitive inhibitory activity of an extra iteron in cis to the P1 origin supports the hypothesis that the DNA looping mechanism (cis-handcuffing) participates in the negative control of plasmid copy number (Fig. 6). Recently, this view has been further supported by mathematical modelling (Morrison and Chattoraj, 2004). As multimer formation in the presence of ligase increased with increasing iteron concentrations (Fig. 2), the handcuffing mechanism seems to satisfy a basic requirement for a plasmid copy number control mechanism, namely, changes in the strength of inhibition in proportion to changes in plasmid (iteron) copy number (Pal et al., 1986; Abeles et al., 1995).

Dimers are generally considered ‘inactive’ in iteron binding. However, from structural and in vitro binding studies it is clear that dimers are capable of recognizing one half of an iteron sequence (Urh et al., 1998; Komori et al., 1999; Giraldo et al., 2003; Abhyankar et al., 2004) (Fig. 8). This binding is weaker than monomer binding but could be significant for handcuffing for the following reasons. Dimers are found in vast excess over monomers which could compensate for the poor DNA binding affinity (Dibbens et al., 1997). The handcuffing reaction itself could stabilize the DNA–protein interactions involved in the half-site binding as has been found for other DNA binding reactions involving pairing of sites (Brenowitz et al., 1990; Ruusala and Crothers, 1992; Frappier et al., 1994; Dodd et al., 2001). As iterons are found in an array, there is always the potential to form multiple handcuffs between arrays. An indication of the importance of multiple links is found in Fig. 2D, where pairing of the four-iteron fragment was more efficient than the one-iteron fragment provided in fourfold excess. Although individually weak, together these contacts could stabilize the pairing reaction. Finally, dimers are stable until dissociated by chaperones (Wickner et al., 1991; DasGupta et al., 1993). Thus dimers effecting handcuffing seems to be realistic expectation from our present knowledge of iteron–initiator interactions.

Participation of dimers in handcuffing was suggested earlier. In R6K, EMSA-stable binding of dimers has been demonstrated and this has formed the basis for the suggestion that pre-formed dimers could directly couple iterons (Urh et al., 1998). In RK2, genetic studies also raised the possibility that a pre-formed dimer bridges two monomer-bound iterons, i.e. four protomers link the two iterons (Toukdarian and Helinski, 1998). The present study raises yet another possibility that dimers made from chaperone-activated monomers rather than from nascent monomers could participate in handcuffing. The latter possibility is suggested as chaperones sometimes stimulated ladder formation (Fig. 7B, lane 16) or dimer binding (Fig. 8B, lane 6 versus lane 15). We also cannot exclude the possibility that the iteron-bound monomers are an active species in handcuffing. We cannot choose among these possibilities, as the proportions of active/inactive monomer and dimer are unknown in our reactions. Although the nature of the dimer remains unclear, the demonstration of a chaperone-mediated decrease in multimer formation is the most compelling evidence that dimer dissociation is the key to handcuffing reversal (Fig. 7).

If dimers could effect handcuffing, then they could be at the heart of controlling the plasmid copy number. We envision the following scenario to explain the present results. When chaperone concentrations are kept constant and RepA concentration is varied, as is the case in our experiments, monomers dominate at low RepA concentration. EMSA shows subsaturated binding and ligation produces multimers inefficiently. As RepA concentration is increased, chaperones become limiting, allowing dimers to survive and cause handcuffing. Alternatively, with increase of RepA concentration, active monomers associate to form ‘active’ dimers competent in handcuffing at a higher rate than chaperones can dissociate them. As RepA concentration is increased further, the increase in (tighter binding) monomers could result in replacing the dimers from the origin, even though dimer concentration increases more than monomers. Therefore, at near saturation of binding handcuffing decreases. If RepA concentration is increased further, dimer concentration can be high enough to compete with monomers for iteron binding and increase handcuffing again. We propose that it is the relative rather than the absolute concentrations of monomers and dimers (initiators and inhibitors) that determine the initiation competence of the origin iterons. The negative feedback would come from the fact that the total RepA and hence the dimer to monomer ratio increases with increasing copy number. An attractive feature of the active participation of dimers in handcuffing is that their conversion to monomers provides a straightforward mechanism for reversal of handcuffing.

Saturation of origins with monomers may occlude handcuffing by another mechanism. In general, iteron-carrying plasmids have their origin iterons all oriented in the same direction (Chattoraj and Schneider, 1997). Based on the fact that RepA bends iterons and that upon saturation of binding RepA–iteron complexes absorb one superhelical turn of DNA, it was proposed that iterons wrap around RepA (Mukhopadhyay and Chattoraj, 1993; Brendler et al., 1997; Edgar et al., 2001). This folded conformation may be possible only upon saturation of binding, and this may preclude handcuffing as suggested by the results of Fig. 2C.

Because there are multiple iterons in an origin and fragments with one iteron can pair, an origin can potentially interact with multiple origins simultaneously and thereby form a network (cluster) of plasmids (Abeles et al., 1995). Cytological evidence for plasmid clustering has been obtained, and a model for how plasmid recruitment at a replication factory can lead to unclustering has been proposed (Weitao et al., 2000; Ho et al., 2002; Nordstrom and Gerdes, 2003). If clusters indeed form and the basis is handcuffing, they can dissociate spontaneously as our dilution and competitor DNA experiments suggest (Figs 5 and 6).

Although dimers have been found without exception in the iteron family of plasmids, the significance of the finding evaded us thus far. If dimers directly compete with monomers for iteron binding and mediate handcuffing, that would be sufficient justification for their conservation.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Electrophoretic mobility shift and ligation assays

Fragments used for Electrophoretic mobility shift (EMS) and ligation assays were end-labelled using Klenow and [α-32P]-dCTP. The binding buffer contained 20 mM Tris-acetate, pH 7.5, 100 mM potassium glutamate, 10 mM magnesium acetate, 10% (v/v) glycerol, 0.1 mM EDTA, 1 mM DTT, 100 µM ATP, 0.1% (v/v) igepal (Sigma) and 4% polyvinyl alcohol (PVA) (Sigma). EMS assay reaction mix contained in 20 µl of binding buffer: 50 ng of DnaJ and 100 ng of DnaK (both from Stressgen), 300 ng of poly(dI-dC) (Pharmacia), 1 ng of labelled DNA fragment (≈ 150 pM for the five iteron fragments) and varying amounts of RepA. The reactions were first assembled in 15 µl without the labelled DNA and incubated for 20 min at room temperature to activate RepA by the chaperones DnaJ and DnaK. After addition of the labelled fragment (5 µl), the reactions were incubated at 30°C for 20 min to promote binding. One half of the mix (10 µl) was loaded onto a 5% polyacrylamide gel in 1× glycerol-tolerant gel buffer (US Biochemicals) and electrophoresed at a constant current of 25 mA. After electrophoresis, the gel was dried and exposed to a Phosphoimager plate (Fujix BAS 2000). To the remaining 10 µl, 200 U of T4 DNA ligase (NEB) was added and the mixture incubated at room temperature for another 15 min. The reactions were stopped by addition of 10 µl of a 2× loading dye containing 20 mM EDTA and 1% SDS. The entire reaction mixture was loaded onto 1% agarose gel and electrophoresed at 10 V cm−1 for 4 h. The gel was dried and exposed to a phosphoimager plate as before.

Mini-P1 copy number measurement

Cells were grown from a single colony to OD600≈ 0.4. A total of four OD600 units of cells were mixed with 0.15 OD600 units of cells carrying pNEB193 (New England Biolabs). DNA was isolated from the mixed culture using Qiaprep spin miniprep kit (Qiagen) and recovered in 50 µl. A portion of it was digested with HindIII that has a single site on both the plasmids. The products were separated on 1% agarose gel, stained with ethidium bromide and illuminated with UV. The band intensities were captured and quantified using a gel documentation system (UV Products). The intensity of the experimental bands was expressed relative to that of the pNEB193 bands to account for the possible loss of DNA during the isolation steps.

Purification of RepA protein

RepA was isolated from BR943 by overexpressing the protein from a plasmid where repA was under the control of λpL promoter (Abeles, 1986). Five hundred millilitres of LB + Amp (100 g ml−1) were inoculated with 2.5 ml of an overnight culture of BR943 and allowed to grow to OD600 = 0.2 at 30°C. The promoter was induced by adding fresh 500 ml of LB + Amp media pre-warmed to 60°C, and by rapid mixing. The culture was then incubated at 42°C for 3 h with shaking. After harvesting, the cells were resuspended in 5 ml of lysis buffer [50 mM Tris-HCl, pH 8.0, 1 mM EDTA, 300 mM NaCl and 10% sucrose (v/v)] to which 580 µl of lysozyme (50 mM in lysis buffer) and 260 µl of EDTA (500 mM) were added, and the mixture was allowed to stand on ice for 10 min. Subsequently, 15 µl of PMSF (100 mM) and 800 µl of Brij 58 (10%, pre-warmed to 60°C before use) were added and the slurry was sonicated with six 30 s pulses at 1 min intervals (to allow for cooling) till the viscosity dropped significantly. A 5 M solution of NaCl was added drop-wise to a final concentration of 1 M with continuous mixing at 4°C. The mixing was continued for another hour and the slurry was centrifuged at 15 K for 1 h at 4°C. The supernatant was loaded on to a PD10 desalting column (Amersham) equilibrated with buffer A (50 mM Tris-HCl, pH 7.5, 250 mM NaCl, 1 mM EDTA, 1 mM DTT and 10% Glycerol). The eluent was loaded on to a Mono S (Econo-Pac High S, Bio-Rad) column and eluted with a 100–700 mM gradient of NaCl in buffer A. The 300–400 mM NaCl fractions that contained RepA were pooled, concentrated to 1 ml (Amicon) and loaded on to a Superose 12 (Pharmacia) column. The fractions containing RepA were pooled, divided into 100 µl aliquots, frozen in liquid nitrogen and stored at −70°C.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

The authors are indebted to their colleagues Suman Mukhopadhyay for pSM105-108 (Fig. 2), Santanu Dasgupta for initial studies on DNA looping in vivo (Fig. 6), Therese Brendler for advice on glycerol-tolerant gel buffer, and David Lane and Michael Yarmolinsky for critical reading of the manuscript, and Deepak Bastia, David Summers and an anonymous referee for thoughtful comments during the review process.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References