Mechanism of Origin Activation by Monomers of R6K-encoded p Protein

doi:10.1016/j.jmb.2007.02.074 J. Mol. Biol. (2007) 368, 928–938

Mechanism of Origin Activation by Monomers ofR6K-encoded p Protein

Lisa M. Bowers, Ricardo Krüger and Marcin Filutowicz⁎

Department of Bacteriology,University of Wisconsin-Madison, 420 Henry Mall,Madison, WI 53706, USA

Present address: R. Krüger, UniveBrasília, Campus II, SGAN 916, MóBrasília, Brazil.Abbreviations used: ori, origin of

replication initiator; wt, wild type; cpen, penicillin.E-mail address of the correspondi

[email protected]

0022-2836/$ - see front matter © 2007 E

One recurring theme in plasmid duplication is the recognition of the originof replication (ori) by specific Rep proteins that bind to DNA sequencescalled iterons. For plasmid R6K, this process involves a complex interplaybetween monomers and dimers of the Rep protein, π, with seven tandemiterons of γ ori. Remarkably, both π monomers and π dimers can bind toiterons, a new paradigm in replication control. Dimers, the predominantform in the cell, inhibit replication, while monomers facilitate open complexformation and activate the ori. Here, we investigate a mechanism by whichπ monomers out-compete π dimers for iteron binding, and in so doingactivate the ori. With an in vivo plasmid incompatibility assay, we find that πmonomers bind cooperatively to two adjacent iterons. Cooperative bindingis eliminated by insertion of a half-helical turn between two iterons but isdiminished only slightly by insertion of a full helical turn between twoiterons. These studies show also that π bound to a consensus site promotesoccupancy of an adjacent mutated site, another hallmark of cooperativeinteractions. π monomer/iteron interactions were quantified using amonomer-biased π variant in vitro with the same collection of two-iteronconstructs. The cooperativity coefficients mirror the plasmid incompat-ibility results for each construct tested. π dimer/iteron interactions werequantified with a dimer-biased mutant in vitro and it was found that πdimers bind with negligible cooperativity to two tandem iterons.

© 2007 Elsevier Ltd. All rights reserved.

Keywords: cooperativity; plasmid replication; R6K; Rep-iteron interaction;replication control
*Corresponding author
Introduction

Plasmids are key contributors to virulence, anti-biotic resistance and horizontal gene transfer. Thus,unraveling the mechanisms that control the prolif-eration of plasmids is a matter of practical signifi-cance as well as fundamental biological interest. Onemodel for plasmid replication studies is R6K, aself-transmissible Escherichia coli plasmid encodingresistance to streptomycin and ampicillin.1 R6K isa member of a group of plasmids in which repli-cation is controlled by the recognition of an origin of

rsidade Católica dedulo B, W5 Norte,

replication; Rep,am, chloramphenicol;

ng author:

lsevier Ltd. All rights reserve

replication (ori) by a specific replication initiator(Rep) protein that binds to DNA sequences callediterons.2Two plasmid-encoded components are necessary

for controlled replication of a minimal R6K repli-con: γ ori, consisting of seven 22 bp iterons, and thepir gene, which encodes the Rep protein, π (Figure1).3,4 Like other Rep proteins in this plasmidcategory, π is primarily dimeric in solution andstrong evidence suggests that dimers inhibit repli-cation, while monomers bind the seven iterons of γori to activate replication.5–9 Unlike other Repproteins, π was long believed to be unique, in thatthe dimeric form is also iteron-binding proficient.6–9Recently, however, dimers of at least two other Repproteins have been shown to bind iterons,10,11 sug-gesting that the earlier π studies may have estab-lished a significant new paradigm in Rep/iteronbinding interactions. This capacity for Rep dimersto compete with monomers for iteron binding addsa new level of complexity to models of plasmidreplication control. Thus, we are left with a central

d.

mailto:[email protected]

Figure 1. Roles of π monomers and dimers in theregulation of replication from γ ori. The seven iterons of γori are indicated by tandem arrows, while the operator/promoter region is represented by two inverted halfarrows. π, encoded by the pir gene, can bind to an iteron asa monomer (crescent) or dimer (double crescent), althoughthe predominant form in solution is the dimer. π binds tothe operator/promoter only as a dimer.6 Shading indicatesthat the two monomer subunits of a dimer make head-to-head contact, while two monomers bound to two tandemiterons are proposed to make a head-to-tail contact. Amonomer contacts the iteron with two domains, while adimer contacts the iteron with only one domain of one ofthe subunits.

Figure 2. In vivo π/iteron binding assay. Plasmids, orisand genes are labeled. Expression of pir from the PBADpromoter is arabinose-inducible. cat encodes chloramphe-nicol acetyl transferase, conferring resistance to cam. blaencodes β-lactamase, conferring resistance to pen. Cres-cent-shaped symbols represent π monomers.

929Mechanism of Origin Activation by π Monomers

question regarding the regulation of replicationfor R6K and plasmids like it: What mechanism ormechanisms allow π monomers to out-competedimers for iteron binding?A partial answer to this question was offered

recently by Kunnimalaiyaan et al.; their somewhatsurprising data demonstrated that π monomerscontact a larger segment of DNA than π dimers(Figure 1).12 An earlier set of qualitative observa-tions hinted at another mechanism by whichmonomers of π gain an edge over π dimers. In gelshift titrations, π monomers were observed to in-teract with seven iterons, yielding patterns consis-tent with positively cooperative binding.13,14 Theseinferences were based on the observed steep bindingcurves that result from site occupancy changingover a relatively small range of protein concentra-tion, a hallmark of cooperative binding in vitro.Because there are reports of strong and specific

protein–protein interactions in vitro without biolo-gical relevance,15 it is extremely important tosupport in vitro binding data with evidence thatthe same interactions occur inside the cell. Yet therehave been very few demonstrations of the impor-tance of cooperative DNA binding in vivo, and mosthave been transcription factors that were assayedwith artificial reporter genes.16–20 This work firstexamines whether π monomers bind iterons coop-eratively with an in vivo π protein titration assay.With this assay, multiple configurations of one andtwo iterons were tested for their ability to titrate πmonomers inside the cell. Second, we quantifycooperative binding of π monomers to the samecollection of one-iteron and two-iteron DNA frag-ments in vitro. Until this work, quantitative mea-surements of π cooperativity could not be madebecause nucleoprotein complexes containing πdimers could not be distinguished from complexes

containing the same number of π molecules boundas monomers. Here, we show both in vivo and invitro that π monomers demonstrate three commoncharacteristics of proteins that bind cooperatively: πmonomers bind to adjacent iterons in a greater-than-additive fashion; a π monomer bound to a strongconsensus site helps recruit a π monomer to anadjacent mutated site; and binding of π monomersto iterons is sensitive to the spacing between iteronsand to their relative helical orientation. Finally, weassess the binding of a dimer-biased π variant to atwo-iteron fragment and find that, unlike π mono-mers, π dimers bind to adjacent iterons with neg-ligible cooperativity.

Results

Do cooperative p interactions occur in vivo?

The assay used to evaluate π monomer binding invivo was based on a phenomenon called plasmidincompatibility, which is generally described as thefailure of two co-resident plasmids to be stablyinherited, often due to the sharing of one or moreelements of the plasmid replication system.21 Forexample, when iterons are cloned into an otherwisecompatible plasmid, they inhibit replication of aγ ori plasmid.22 The number of iterons and theamount of π monomers in the cell both affect thedegree of incompatibility.22In the plasmid incompatibility assay depicted in

Figure 2, two plasmids compete for limited πmonomers in the cell. First, a chloramphenicol(cam) resistant γ ori plasmid, pFW25,23 was estab-lished in the E. coli host strain, ECF001.24 Replicationof pFW25 was dependent on monomers of πproduced from the chromosome of ECF001, wherepir expression was under control of the arabinose-inducible PBAD promoter. ECF001 harboring pFW25was then transformed with a series of iteron-

930 Mechanism of Origin Activation by π Monomers

containing pUC9 derivative plasmids that were highcopy number and conferred penicillin (pen) resis-tance. Since arabinose was not included in the platesused for transformation, pir expression was unin-duced and π levels were limiting; thus, the iteronscloned into pUC9 affected replication of pFW25 bycompeting for π monomers in the cell.The various configurations of iterons cloned into

pUC9 were designed to test if cooperative interac-tions occur between π monomers bound to iterons.If certain configurations of iterons bound π mono-mers with greater cooperativity, they would have ahigher degree of incompatibility with pFW25 andresult in fewer colonies resistant to both cam andpen. Constructs were created with two iteronsseparated by 0 bp (wild-type (wt) arrangement),5 bp (half-helical turn), and 10 bp (full helical turn)(See Materials and Methods for iteron sequences). Ifcooperative contacts occur, the changes in site-spacing and helical orientation were expected todisrupt or enhance the protein–protein interactionsthat stabilize DNA-bound monomers.25The results of the plasmid incompatibility assay

are shown in Figure 3. Transformation with a pUC9derivative containing a single iteron resulted in halfthe number of colonies as transformation with pUC9alone on medium with cam+ pen. This means thateven a single iteron on a high-copy pUC9 derivativecan titrate π away from the γ ori plasmid and reduceor inhibit its replication. Transformation with thepUC9 derivative containing two iterons separatedby 5 bp resulted in half the number of colonies as theone-iteron fragment. This means that doubling the

Figure 3. π binds cooperatively to two iterons in vivo.Experiments were conducted as described for Figure 2and in Materials and Methods. The numbers of transfor-mants on plates with cam+pen were expressed as ratiosagainst the pUC9 control (no iteron). Arrows representiterons and * represents the G/C7>A/T mutation.Numerals 5 and 10 represent the number of base-pairsbetween the two iterons. Decreasing ratios suggestincreased titration of π by the incoming iteron-containingpUC9 derivative. The data represent the average of threeindependent experiments.

number of iterons but orienting them on oppositefaces of the DNA resulted in an additive increasein π binding to the pUC9 derivative (decrease incolonies resistant to cam+ pen). However, the pUC9derivative containing the two-iteron fragmentwith no space between the iterons (wt configura-tion) was completely incompatible with pFW25.This greater-than-additive increase in π bindingis indicative of cooperativity. Adding 10 bp be-tween the two iterons resulted in an intermediatenumber of colonies, suggesting that helical orienta-tion is more important than distance for cooperativeinteractions.Increasing the concentration of π did not alleviate

the incompatibility of the two plasmids because thewt protein is dimer-biased. This result was expected,as it is well documented that overproduction of πinhibits replication of γ ori plasmids.5 In contrast,when the experiment was carried out with anisogenic strain (ECF003)24 that produces a mono-
mer-biased π variant, π·P106L^F107S,8,26 incom-patibility was not observed with any of the aboveconstructs, presumably because π monomers werenot limited (data not shown). Western blots of bothhost strains showed that the amount of π producedin uninduced cells is barely detectable, and thepresence of the different pUC9 derivatives had noinfluence on the amount of π in induced or un-induced cells (data not shown).
π bound to a consensus site promotes πbinding to a mutated site in vivo

If π binds cooperatively, π bound to a strongconsensus site should promote occupancy of anadjacent mutated site at subsaturating proteinconcentrations. To test this hypothesis, the aboveplasmid incompatibility assay was carried out withconstructs containing one wt iteron and one iteronwith a well-characterized mutation of the G/C7 bpto A/T (represented as *), which nearly abolishes πbinding in vitro and in vivo.12,27 Because π bindsweakly to the one-iteron* fragment, it is not sur-prising that transformation with the pUC9 deriva-tive containing one-iteron* resulted in a similarnumber of colonies on cam+ pen as transformationwith pUC9. If π bound to iterons independently(without cooperativity), transformation with thetwo-iteron* probe would be expected to yield thesame results as the one-iteron probe, because themutated iteron alone binds π so weakly. However,fewer colonies survived with the two-iteron* probethan the one-iteron probe; therefore, we can inferthat π bound to the wt iteron facilitates π binding tothe mutated iteron (cooperativity). Introducing ahalf-helical turn between the wt andmutated iteronseliminated this cooperative interaction and resultedin binding similar to the one-iteron fragment.Finally, a 10-bp space between the wt and mutatediterons resulted in the same number of colonies asthe construct with no space, again consistent withthe idea that helical orientation is more importantthan distance for cooperative π interactions.


In vitro binding properties of π·wt,π·P106L^F107S and π·M36A^M38A to a DNAfragment containing 2 iterons

To support the indirect observations that π bindscooperatively to iterons in the cell, π–iteron interac-tionswere examined in vitro. Gel shift titrationswithπand a DNA fragment containing two iterons yieldcomplicated binding patterns because both mono-mers and dimers are iteron-binding proficient. Theresults are further complicated because two differenttypes of complexes have the same molecular mass;oneDNA fragmentwith two iterons can bind to eithertwo πmonomers or one π dimer. Thus, it is necessaryto distinguish each possible complex in order toensure that the correct complexes are quantified.To separate the two different complexes with the

same molecular mass, large DNA fragments with60 bp flanking the two iterons were used. It wasestimated previously that one dimer causes anapparent bending of the DNA by approximately50°, while two head-to-tail monomers cause anapparent bending of the same DNA by approxi-mately 75°.28 Thus, the large flanking sequenceallowed these two complexes to be separatedelectrophoretically based on differential apparentbending (Figure 4). To distinguish dimer-boundcomplexes from complexes with two monomers,we utilized well-characterized π variants biasedtoward monomer or dimer binding. It has been
reported that π·P106L^F107S binds to iterons pri-marily as a monomer,8,12 while π M36A^M38A ·binds predominantly as a dimer.8,29 Thus, theidentities of the complexes were inferred on thebasis of their positions in the gel and theirdifferential concentrations with each of the πvariants (Figure 4).It was evident from these gel shift titrations
that only a minimal fraction of π·wt binds as twomonomers to a two-iteron DNA fragment in vitro.This is understandable, due to previous reports that
chaperones are required for π·wt to monomerizeupon iteron binding.30 Therefore, in the absence of
Figure 4. π·wt, π·P106L^F107S and π·M36A^M38Ahave different binding patterns with a two-iteron probe.Lane 1 is DNA only. Gray and white triangles representincreasing concentrations of π·wt and π·P106L^F107S,respectively, startingwith 6.25ng anddoubling for each lane.The filled square represents 200 ng of π·M36A^M38A.Arrows represent iterons. Filled crescents represent πmonomers and gray double crescents represent π dimers.

chaperones, an increase in π·wt concentrationresults in an increase in dimer binding. This is whya large shift to the two-monomer complex doesnot occur as protein concentration increases in vitro.For this reason, the monomer-biased variant,
π·P106L^F107S, was used for all in vitro quanti-fications of DNA binding. This variant has beenshown to be predisposed to iteron DNA ligand-induced monomerization.9 With this π variant, asteep binding curve was observed with the fractionof one monomer shifting to two monomers over arelatively small change in protein concentration, an indication of cooperative binding. π·M36A^M38Awas used in all experiments as a size marker for theone dimer DNA complex.
Quantification of cooperative π monomerbinding to two thermodynamicallyidentical iterons

Cooperativity can be assessed quantitatively onthe basis of k12, a constant obtained from bindingequations derived by the statistical mechanicalapproach,31 and fit to data from titrations of proteinwith DNA. Values of k12>1 are considered to becooperative. The same DNA sequences used in theplasmid incompatibility assays with 0 bp, 5 bp, and10 bp between two iterons (Figure 5(a)) were alsoused to quantify cooperative interactions betweentwo π monomers in vitro.Figure 5(b) shows a representative gel shift

titration of π·P106L^F107S added to each of theabove probes. The fraction of total DNA that wasfree, bound to one monomer, and bound to twomonomers was quantified and plotted as a functionof π concentration (Figure 5(c)). To calculate thecooperativity coefficient, k12, data in Figure 5(c) weresubjected to a least-squares linear regression analy-sis using equation (1c) (See Materials and Methodsfor all formulas). This analysis provided the macro-scopic binding constants, K1 and K2, which wereused to calculate k12 using equation (2).During the analysis of this complex system, a few

challenges were encountered. First, the total con-centration of π monomers could not be determinedbecause π is dimeric in solution and presumablymonomerizes upon binding to DNA.9 Therefore, thevalue used for π concentration was the total amountof π added to the sample. Second, increasing the πconcentration beyond 800 ng per sample causedaggregation; therefore, protein binding to 100% ofthe DNA was not achieved. Finally, dimer bindingcould not be eliminated completely, even with themonomer-biased variant. Importantly, each of thesestipulations is expected to result in an underestima-tion of the cooperativity coefficient. Furthermore, allof the DNA fragments were subjected to the sameconditions, enabling a sound assessment of therelative cooperativity coefficients for each of theiteron spacings.With these caveats in mind, the data in Table 1

show that πmonomers bind the two-iteron constructwith a cooperativity coefficient of 210. Separation of

Figure 5. π binds cooperatively to two thermodynamically identical iterons in vitro. Cooperative interactions areinfluenced by helical rotation and distance between the two iterons. (a) A depiction of each iteron-containing probe, notincluding flanking DNA. Numerals 5 and 10 refer to the number of base-pairs between iterons. Numerals 1 and 2 refer tothe first and second iterons of γ ori, respectively. (b) Gel shift titrations of purified πwith the corresponding probes in (a).Lane 1 is DNA only. Open triangles represent increasing concentrations of π·P106L^F107S, starting with 6.25 ng anddoubling for each lane. The filled square represents 200 ng of π·M36A^M38A. Arrows represent iterons. Filled crescentsrepresent π monomers and gray double crescents represent π dimers. (c) Quantification of gel shift titration data withπ·P106L^F107S. The fraction of the total radioactivity as free DNA (circles), DNA containing a single π monomer(squares), and DNA containing two π monomers (diamonds) was determined. Broken, dotted, and continuous linescorrespond to the best fit of the data for equations (1a)–(1c), respectively. The protein concentrations in (b) are a subset ofthose plotted in (c).


the two binding sites by 10 bp (a full helical turn)decreased the k12 slightly to 156 and separation by5 bp (a half-helical turn) essentially abolishedcooperativity (k12=1.8). These results support thehypothesis that π monomers bind to γ ori withpositive cooperativity, and the distance and orienta-tion between adjacent binding sites influence thesecooperative interactions.

Estimation of cooperative πmonomer binding totwo heterogeneous binding sites

As with the in vivo assay, we hypothesized thatmutating one of the two iterons, causing a decreaseof the natural affinity of π for that site, would

Table 1. Cooperative π/iteron interactions are influenced bywell as monomer/dimer bias

Protein DNA probe K

π·P106L^F107S(monomer-biased)Two-iteron

Two-iteron+5 1Two-iteron+10Two-iteron*

Two-iteron*+5Two-iteron*+10

π·M36A^M38A(dimer-biased)

Two-iteron

enhance the apparent cooperativity because πbound to the consensus site would promote occu-pancy of the adjacent mutated site. This hypothesiswas tested with another set of two-iteron probescontaining one wt iteron and one iteron with theG/C7>A/T mutation (represented as *).First, the relative difference in binding affinity for

the one-iteron and one-iteron* fragments was calcu-
lated by performing gel shift titrations with π·P106L^F107S (Figure 6). The binding affinity of eachfragment was determined by subjecting the plotteddata to a least-squares linear regression analysisusing equation (3). The binding affinity of the one-iteron fragment (k1) was 1.9(±0.1)×107 M−1 and theone-iteron* fragment (k2) was 1.0(±0.2)×106 M−1.
distance, helical rotation, and affinity of the two iterons as

1 (×105) (M−1) K2 (×1013) (M−2) k12

8.90±8.00 4.13±0.63 20956.18±41.32 11.31±2.24 27.32±4.37 2.09±0.24 1569.37±3.56 1.60±0.15 34623.60±11.87 0.86±0.24 295.02±2.81 0.37±0.06 282

16.99±0.82 0.37±0.04 5

Figure 6. The binding affinity of the 1-iteron and1-iteron* probes. (a) A depiction of each iteron-containingprobe, not including flanking DNA. * indicates theG/C7>A/T mutation. (b) Gel shift titrations of thebinding of purified π to the corresponding probes in (a).Lane 1 is DNA only. Open triangles represent increasingconcentrations of π·P106L^F107S, starting with 6.25 ngand doubling for each lane. The filled square represents200 ng of π·M36A^M38A. Arrows represent iterons.Filled crescents represent π monomers and gray doublecrescents represent π dimers. (c) Quantification of gel shifttitration data with π·P106L^F107S. The fraction of thetotal radioactivity as free DNA (circles) and DNAcontaining a single πmonomer (squares) was determined.Continuous and broken lines correspond to the best fit ofthe data for equation (3). The protein concentrations in (b)are a subset of those plotted in (c).


Thus, binding affinities for the two iterons weresubstantially different (k1/k2=19).Next, gel shift titrations were performed with

heterogeneous two-iteron probes with 0 bp, 5 bp,and 10 bp separating the wt and the mutated iterons(Figure 7(a) and (b)). The fraction of total DNA thatwas free, bound to one monomer, and bound to twomonomers was quantified and plotted as a functionof π concentration (Figure 6(c)). The macroscopicbinding constants, K1 and K2, were determined bysubjecting data in Figure 6(c) to a least-squareslinear regression analysis using equation (1c). Whenthe affinities of the two binding sites are substan-tially different, equation (4) holds true (derived inMaterials and Methods). As mentioned above, wefound h, defined as k1/k2, to have a value of 19 forthese two iterons.The data in Table 1 suggest that the same trends

hold true for heterogeneous and homogeneousbinding sites: π binds with greater cooperativity tothe two-iteron* probe than the two-iteron*+10 probe

and binds with little cooperativity to the two-iteron*+5 probe. If the assumptions inherent in equation (4)are valid, the estimated cooperativity involved in πbinding to two heterogeneous iterons is greater thanthe estimated cooperativity involved in π bindingtwo wt iterons for each of the three spacings tested(Table 1).

Quantification of cooperative π dimer binding totwo thermodynamically identical iterons

The dimer-biased π variant, π·M36A^M38Awasused to quantify π dimer binding to a two-iteronprobe. This variant was previously characterizedto bind only as a dimer to iterons both in vivo andin vitro.8,29,32 Figure 8(a) shows a representative gel
shift titration of π·M36A^M38A added to the two-iteron fragment. The fraction of total DNA that wasfree, bound to one dimer, and bound to two dimerswas quantified and plotted as a function of πconcentration (Figure 8(b)). To calculate the coopera-tivity coefficient, k12, data in Figure 8(b) were sub-jected to a least-squares linear regression analysisusing equation (1c) This analysis provided themacroscopic binding constants, K1 and K2, whichwere used to calculate k12 using equation (2). The results in Table 1 show that dimers of π·M36A^M38A bind to two tandem iterons independentlyor with negligible cooperativity.
Discussion

There is an ongoing debate about whethercooperative interactions can be inferred from gelshift titrations. Some argue that differential stabi-lities of protein/DNA complexes during electro-phoresis can significantly distort interpretations ofprotein binding.33,34 Furthermore, in vitro bindingassays often neglect the influence of known andunknown factors including DNA architecture, cha-perones, and other host-encoded binding proteinssuch as DnaA, IHF, and Fis.35–37 Even with this levelof uncertainty, there have been very few studies ofcooperative binding in vivo. Thus, this work isdistinguished by the development of the plasmidincompatibility assay to monitor cooperative Rep–iteron interactions in the context of the cell and thecorroboration of the in vivo data by quantitativebinding studies in vitro.One might expect plasmid compatibility to be an

“all-or-nothing” phenomenon. However, a strongertitration of π by iterons cloned into the pUC9 deriva-tive is believed to result in a lower average copynumber of the γ ori plasmid. Because cam resistanceconferred by the γ ori plasmid is gene dosage-de-pendent, some cells in the population will immedi-ately have insufficient copies of the cat gene to conferresistance. However, other cells in the populationmay survive several rounds of cell division beforethe copy number is too low to confer resistance. Thisresults in a repeatable and significant difference inplasmid compatibility with constructs that differ

Figure 7. π binds cooperatively to two heterogeneous iterons in vitro. Cooperative interactions are influenced byhelical rotation and distance between the two iterons. (a) A depiction of each iteron-containing probe, not includingflanking DNA. Numerals 5 and 10 refer to the number of base-pairs between iterons. Numerals 1 and 2 refer to the firstand second iterons of γ ori, respectively. An asterisk (*) represents the G/C7>A/T mutation in iteron 2. (b) Gel shifttitrations of purified π with the corresponding probes in (a). Lane 1 is DNA only. Open triangles represent increasingconcentrations of π·P106L^F107S, starting with 6.25 ng and doubling for each lane. The filled square represents 200 ng ofπ·M36A^M38A. Arrows represent iterons. Filled crescents represent π monomers and gray double crescents represent πdimers. (c) Quantification of gel shift titration data with π·P106L^F107S. The fraction of the total radioactivity as freeDNA (circles), DNA containing a single π monomer (squares), and DNA containing two π monomers (diamonds) wasdetermined. Broken, dotted, and continuous lines correspond to the best fit of the data for equations (1a)–(1c),respectively. The protein concentrations in (b) are a subset of those plotted in (c).


only by the number of bases separating two iterons.Clearly, orientation of the two iterons on oppositefaces of the helix results in a dramatic decrease in theamount of π binding by constructs with two homo-geneous or two heterogeneous iterons.At first glance, it may seem that there are two

competing explanations for the plasmid incom-patibility data: first, the stated explanation thatiterons on the pUC9 derivative titrate π monomersaway from the γ ori plasmid and second, the iteronson the pUC9 derivative handcuff with γ ori of theresident plasmid. Handcuffing has been proposed asa mechanism that limits the copy number of R6K andother plasmids,38,39 wherebyRepdimers link two oriswith each subunit of the dimer binding to one ori.29 Acloser examination of the data supports themonomertitration explanation over handcuffing. First, hand-cuffing occurs at high concentrations of π protein butπ was limited in this experiment. Second, handcuff-ing is caused by dimers, which have been shown tobind very poorly to two adjacent iterons (Figure 8).28Therefore, if this assay measured handcuffing, thenumber of colonies containing the one-iteronconstruct would resemble the number of coloniescontaining the two-iteron construct, which is notthe case. Third, two dimers can bind the two-iteron+10 probe with equal or greater affinity than the

two-iteron probe (Figure 3). Therefore, if this assaymeasured handcuffing, the number of coloniescontaining the two-iteron +10 construct shouldequal or be fewer than the number of coloniescontaining the two-iteron construct. Again, the datado not support this possibility.The plasmid incompatibility assay is important

because it is a functional assay performed in thecontext of the cell but the in vitro quantification dataare necessary to understand the degree of coopera-tivity compared to other systems. A scan of theliterature shows that k12 values can be as high as2000 for the most-cooperative interactions,25 butvalues are more often reported in the range of 10–200.40,41 Thus, quantification of the gel shift titrationdata suggests that cooperative interactions betweenπmonomers are relatively robust compared to mostsystems. In direct contrast, the dimer-biased πvariant binds two tandem iterons with negligiblecooperativity, suggesting a mechanism by whichmonomers out-compete dimers for iteron binding.This demonstration of strong cooperative binding

ofπmonomers to two iteronsmay be an integral newinsight into the mechanism of plasmid copy numberregulation. First, it offers an explanation as to how γori is saturated with π monomers when dimersabound. For instance, cooperative interactions may

Figure 8. π M36A^M38A binds without cooperativityto a two-iteron probe in vitro. (a) Gel shift titrations ofpurified π with the 2-iteron probe. Lane 1 is DNA only.The filled triangle represents increasing concentrations ofthe dimer-biased π variant π·M36A^M38A: 12.5 ng, 25 ng,50 ng, 100 ng, 150 ng, 200 ng, and 250 ng. Gray and opensquares represent 200 ng of π·wt and π·P106L^F107S,respectively. Arrows represent iterons. Filled crescentsrepresent π monomers and gray double crescents repre-sent π dimers. (b) Quantification of gel shift titration datawith π·M36A^M38A. The fraction of the total radio-activity as free DNA (circles), DNA containing a single πdimer (squares), and DNA containing two π dimers(diamonds) was determined. Broken, dotted, and contin-uous lines correspond to the best fit of the data forequations (1a)–(1c), respectively.


prevent or stall monomers from disassociatingfrom iterons or rearranging back into dimers,increasing the opportunity to recruit more mono-mers to adjacent iterons. Second, the observation thatcooperativity may be enhanced when one bindingsite is weaker than the adjacent site is quite relevantto our system, because the seven iterons havedifferent degrees of sequence divergence. The mid-dle iteron is most divergent, with five differencesfrom the consensus sequence.3 Initial studies showthat the more divergent iterons have a lower affinityfor π (L.B., S. Rakowski, and M.F., unpublishedresults); therefore, cooperativity may be especiallyuseful in filling the middle iterons with monomers,enabling saturation of the ori.Finally, γ ori is only one example of an ever-

expanding group of iteron-containing plasmids thatdepend on interactions between a replication proteinand the reiterated DNA-binding sequences.2 Itseems likely that if cooperative interactions areinvolved in initiation from γ ori, they may beinvolved in similar systems. Many plasmids in thisgroup, including R6K, are self-mobilizable, and thereare times in the life-cycle of these plasmids when theRep protein concentration would be very low, such

as after the plasmid is transferred by conjugation to acell without Rep. In these situations, cooperativeinteractions between Rep monomers may be espe-cially helpful in initiating replication. If this is true, itmay help to explain the evolution of cooperativeprotein–protein and protein–DNA interactionsinvolved in plasmid replication initiation.

Materials and Methods

Bacterial growth

Bacteriawere grown aerobically at 37 °C in LB. ECF00124was grown with amp at 25 μg/ml. ECF001 contain-ing pFW2523 was grown with cam at 15 μg/ml. ECF001containing pUC9 derivatives was grown with pen at750 μg/ml.

Oligonucleotides

Oligonucleotides were purchased from either IDT(Coralville, IA) or the UW Biotechnology Center (Madi-son, WI). The top strand is indicated below in the 5′-3′direction. The bottom strand is complementary to the topstrand. Bold letters indicate the G/C7>A/T mutations.

One-iteron: AAACATGAGAGCTTAGTACGTTOne-iteron*: AAACATAAGAGCTTAGTACGTTTwo-iteron: AAACATGAGAGCTTAGTACGTGAAACATG-AGAGCTTAGTACGTTTwo-iteron*: AAACATGAGAGCTTAGTACGTGAAACATA-AGAGCTTAGTACGTTSequence separating the two iterons in the two-iteron +5 andtwo-iteron*+5 constructs: TTAACSequence separating the two iterons in the two-iteron +10 andtwo-iteron*+10 constructs: TTAACTTAAC

Construction of plasmids

Each double-stranded oligonucleotide (above) wasblunt-end cloned into the HincII site of pUC9 (Promega,Madison, WI) for the in vivo binding assays and the HpaIsite of pBend542 for the gel shift titrations. Standardcloning techniques were followed and new clones wereverified by sequencing.43

In vivo incompatibility assay

The incompatibility assay is depicted in Figure 2.Standard preparation of calcium chloride-competentcells was followed for host strain ECF001+pFW25 withthe culture grown in LB+ cam+ 0.02% (w/v) arabinose. A100 ng sample of each pUC9 derivative was transformedinto the host strain and LB+ 0.02% arabinose was added toa final volume of 1 ml for an outgrowth period of 45 min.A 100 μl portion of a 1:10 (v/v) dilution of eachtransformation mixture was spread onto LB+ cam+ penplates. Plates were incubated for 24 h at 37 °C.

Protein purification

His-π·WT,His-π·P106L^F107S, andHis-π·M36A^M38Awere purified as described.44,45


DNA probe preparation

pBend5 derivative plasmids were digested with EcoRVand the iteron-containing fragments were purified byelectrophoresis and removal from a 6% (w/v) polyacryl-amide gel with a gel extraction kit (Qiagen, Valencia,CA). The one-iteron fragments were 149 bp each, andthe two-iteron fragments were 171–181 bp, depending onthe sequence between the iterons. All DNA fragmentswere end-labeled with [γ32P]ATP using polynucleotidekinase.

Gel shift titrations

An 80 pg sample of labeled iteron-containing probewas mixed with 65 ng of poly(dI:dC) and binding buffer(2 mM Tris–HCl (pH 7.5), 0.6 mM MgCl2, 0.1 mMEDTA, 10 mM potassium glutamate) in a 14 μl reaction.Then, 1 μl of protein in TGE buffer (10 mM Tris–HCl(pH 7.5), 10% glycerol, 0.1 mM EDTA, 0.3 M KCl) wasadded. The protein–DNA equilibrium mixtures wereincubated at room temperature (27 °C) for 15 minthen 3 μl of loading dye was added (20% (v/v) glyc-erol, 0.5× TBE buffer,43 bromphenol blue). The sampleswere loaded onto 6% polyacrylamide gels (37:1 (w/w)acrylamide:bisacrylamide and 0.5× TBE buffer) that hadbeen pre-electrophoresed for 1 h at 150 V. The sampleswere run at a constant 180 V for 90 min. Dried gelswere imaged with a Typhoon™ phosphorimager (GEHealthcare).

Estimation of cooperativity coefficient for twohomogeneous binding sites

All experimental data from the gel shift titrations werefit to the following equations using KaleidaGraphsoftware (Reading, PA). The following equations werebased on the statistical mechanical approach31,34 but,because π binding to 100% of the DNA could not beachieved due to protein/DNA aggregation at highconcentrations of protein, these formulas were modifiedto account for the baseline and maximum fraction for agiven titration.46

uo ¼ P0 þ ðPmax � P0Þ1=Z ð1aÞ

u1 ¼ P0 þ ðPmax � P0ÞK1L=Z ð1bÞ

u2 ¼ P0 þ ðPmax � P0ÞK2L2=Z ð1cÞ

θo, θ1, and θ2 are fractions of free DNA, single monomercomplex, and two monomer complexes, respectively. P0and Pmax are the baseline and maximum fraction for agiven titration. Z is the binding polynomial and is equalto 1+K1L+K2L2. L is protein concentration, K1= (k1+k2)and K2= (k1k2k12). k1 and k2 are the binding affinityconstants for the first iteron and the second iteron of thetwo-iteron complex, and k12 is the cooperativity coeffi-cient describing the interaction of protein moleculesoccupying both sites.Once K1 and K2 are obtained from the least-squares

linear regression analysis, k12 can be derived by a fewsimple rearrangements. When the two binding sites areidentical (k1=k2), then K2=k12k12 and a rearrangement ofthis equation gives k12=K2/k12. Alsowhen the two bindingaffinities are identical, K1=2k1. Squaring this equation

gives K12=4k12 or k12=K1

2/4. Substituting this equationinto the k12 equation results in:

k12 ¼ ð4K2Þ=K 21 ð2Þ

Estimation of cooperativity coefficient for twoheterogeneous binding sites

When the affinities of two binding sites are hetero-geneous, k12 can be estimated if one binding site has asubstantially weaker affinity than the other (k1≫k2) and ifthe heterogeneity factor (h) is known (h=k1/k2). Todetermine h, k1 and k2 must be estimated individually.The interaction between π and a single iteron can bedescribed by the Langmuir isotherm:47

w ¼ k1L=ð1þ k1LÞ ð3ÞWhere ¥ is the fraction of single monomer complex, L isprotein concentration, and k1 is the binding affinityconstant.Once h is known and K1 and K2 are obtained from the

least-squares linear regression analysis using equation (1c),k12 of two heterogeneous binding sites can be estimated bya few simple rearrangements. As defined above, K2=k1 k2k12 and, after rearranging, k12=K2/(k1 k2). If the hetero-geneity factor (h) is known (h=k1/k2), then k2=k1/h. Thus,k1/h can be substituted for k2 to give k12≈K2h/k12. Also, asdefined above, K1=(k1+k2). Therefore, in situations wherek1>>k2, K1≈k1 so K1

2≈k12. Finally, K12 can be substituted

for k12, which results in:

k12cK2h=K 21 ð4Þ

Acknowledgements

We are particularly indebted to Sheryl Rakowskiand Selvi Kunnimalayaan for many helpful discus-sions, and Richard Burgess and Tom Record forcomments on the manuscript. This work was sup-ported by NIH grant GM40314 and USDA/HATCHgrantWIS04952 toM.F.R.K.was supportedbyCAPES/CNPq.

References

1. Kontomichalou, P., Mitani, M. & Clowes, R. C. (1970).Circular R-factor molecules controlling penicillinasesynthesis, replicating in Escherichia coli under relaxedor stringent control. J. Bacteriol. 104, 34–44.

2. Krüger, R., Rakowski, S. A. & Filutowicz, M. (2004).Participating elements in the replication of iteron-con-taining plasmids (ICPs). In Plasmid Biology (Funnell,B. E. & Phillips, G. J., eds), pp. 25–45, ASM Press,Washington, DC.

3. Stalker, D. M., Kolter, R. & Helinski, D. R. (1979).Nucleotide sequence of the region of an origin ofreplication of the antibiotic resistance plasmid R6K.Proc. Natl Acad. Sci. USA, 76, 1150–1154.

4. Chen, D., Feng, J., Krüger, R., Urh, M., Inman, R. B. &Filutowicz, M. (1998). Replication of R6K γ origin invitro: discrete start sites for DNA synthesis dependenton π and its copy-up variants. J. Mol. Biol. 282, 775–787.


5. Filutowicz, M., McEachern, M. J. & Helinski, D. R.(1986). Positive and negative roles of an initiatorprotein at an origin of replication. Proc. Natl Acad. Sci.USA, 83, 9645–9649.

6. Urh, M., Wu, J., Forest, K., Inman, R. B. & Filutowicz,M. (1998). Assemblies of replication initiator proteinon symmetric and asymmetric DNA sequences de-pend on multiple protein oligomerization surfaces.J. Mol. Biol. 283, 619–631.

7. Wu, J., Sektas, M., Chen, D. & Filutowicz, M. (1997).Two forms of replication initiator protein: positiveand negative controls. Proc. Natl Acad. Sci. USA, 94,13967–13972.

8. Krüger, R., Konieczny, I. & Filutowicz, M. (2001).Monomer/dimer ratios of replication protein modu-late the DNA strand-opening in a replication origin.J. Mol. Biol. 306, 945–955.

9. Abhyankar,M.M., Reddy, J.M., Sharma,R., Bullesbach,E. & Bastia, D. (2004). Biochemical investigationsof control of replication initiation of plasmid R6K.J. Biol. Chem. 279, 6711–6719.

10. Das, N. & Chattoraj, D. K. (2004). Origin pairing(‘handcuffing’) and unpairing in the control of P1plasmid replication. Mol. Microbiol. 54, 836–849.

11. Giraldo, R., Andreu, J. M. & Díaz-Orejas, R. (1998).Protein domains and conformational changes in theactivation of RepA, a DNA replication initiator. EMBOJ. 17, 4511–4526.

12. Kunnimalaiyaan, S., Krüger, R., Ross, W., Rakowski,S. A. & Filutowicz, M. (2004). Binding modes of theinitiator and inhibitor forms of the replication proteinπ to the γ ori iteron of plasmid R6K. J. Biol. Chem. 279,41058–41066.

13. Urh, M., York, D. & Filutowicz, M. (1995). Buffercomposition mediates a switch between cooperativeand independent binding of an initiator protein toDNA. Gene, 164, 1–7.

14. Filutowicz, M., York, D. & Levchenko, I. (1994).Cooperative binding of initiator protein to replicationorigin conferred by single amino acid substitution.Nucl. Acids Res. 22, 4211–4215.

15. Lazarides, E. & Lindberg, U. (1974). Actin is thenaturally occurring inhibitor of deoxyribonuclease I.Proc. Natl Acad. Sci. USA, 71, 4742–4746.

16. Astromoff, A. & Ptashne, M. (1995). A variant oflambda repressor with an altered pattern of coopera-tive binding to DNA sites. Proc. Natl Acad. Sci. USA,92, 8110–8114.

17. Barbaric, S., Munsterkotter, M., Goding, C. &Horz, W.(1998). Cooperative Pho2-Pho4 interactions at thePHO5 promoter are critical for binding of Pho4 toUASp1 and for efficient transactivation by Pho4 atUASp2. Mol. Cell Biol. 18, 2629–2639.

18. Hansen, F. G., Christensen, B. B., Nielsen, C. B. &Atlung, T. (2006). Insights into the quality of DnaAboxes and their cooperativity. J. Mol. Biol. 355, 85–95.

19. Vashee, S. & Kodadek, T. (1995). The activationdomain of GAL4 protein mediates cooperative pro-moter binding with general transcription factors invivo. Proc. Natl Acad. Sci. USA, 92, 10683–10687.

20. Burz, D. S., Rivera-Pomar, R., Jackle, H. & Hanes, S. D.(1998). Cooperative DNA-binding by Bicoid providesa mechanism for threshold-dependent gene activationin the Drosophila embryo. EMBO J, 17, 5998–6009.

21. Novick, R. P. (1987). Plasmid incompatibility. Micro-biol. Rev. 51, 381–395.

22. Stalker, D. M., Shafferman, A., Tolun, A., Kolter, R.,Yang, S. & Helinski, D. R. (1981). Direct repeats ofnucleotide sequences are involved in plasmid replica-

tion and incompatability. In The Initiation of DNAReplication (Ray, D. S., ed), vol. 22, pp. 113–124.Academic Press, New York.

23. Wu, F., Wu, J., Ehley, J. & Filutowicz, M. (1996).Preponderance of Fis-binding sites in the R6K γ originand the curious effect of the penicillin resistancemarker on replication of this origin in the absence ofFis.J. Bacteriol. 178, 4965–4974; (published erratumappears in J Bacteriol 1997 Apr;179(7):2464).

24. Pluciennik, A., Iyer, R. R., Napierala, M., Larson, J. E.,Filutowicz, M. & Wells, R. D. (2002). Long CTG*CAGrepeats from myotonic dystrophy are preferred sitesfor intermolecular recombination. J. Biol. Chem. 277,34074–34086.

25. Mao, C., Carlson, N. G. & Little, J. W. (1994). Coop-erative DNA-protein interactions. Effects of changingthe spacing between adjacent binding sites. J. Mol.Biol. 235, 532–544.

26. Krüger, R. & Filutowicz, M. (2000). Dimers of πprotein bind the A+T-rich region of the R6K γ originnear the leading-strand synthesis start sites: regula-tory implications. J. Bacteriol. 182, 2461–2467.

27. McEachern, M. J., Filutowicz, M. & Helinski, D. R.(1985). Mutations in direct repeat sequences and in aconserved sequence adjacent to the repeats result in adefective replication origin in plasmid R6K. Proc. NatlAcad. Sci. USA, 82, 1480–1484.

28. Krüger, R., Rakowski, S. A. & Filutowicz, M. (2004).Isomerization and apparent DNA bending by π, thereplication protein of plasmid R6K. Biochem. Biophys.Res. Commun. 313, 834–840.

29. Kunnimalaiyaan, S., Inman, R. B., Rakowski, S. A. &Filutowicz, M. (2005). Role of π dimers in coupling(“handcuffing”) of plasmid R6K's γ ori iterons.J. Bacteriol. 187, 3779–3785.

30. Zzaman, S., Reddy, J. M. & Bastia, D. (2004). TheDnaK-DnaJ-GrpE chaperone system activates inertwild type π initiator protein of R6K into a form activein replication initiation. J. Biol. Chem. 279, 50886–50894.

31. Brenowitz,M., Senear, D. F., Shea,M.A.&Ackers, G. K.(1986). “Footprint” titrations yield valid thermodynam-ic isotherms. Proc. Natl Acad. Sci. USA, 83, 8462–8466.

32. Peng, Y., Rakowski, S. A. & Filutowicz, M. (2006).Small deletion variants of the replication protein, π,and their potential for over-replication-based antimi-crobial activity. FEMS Microbiol. Letters, 261, 245–252.

33. Kleinschmidt, C., Tovar, K. & Hillen, W. (1991). Com-puter simulations and experimental studies of gelmobility patterns for weak and strong non-coopera-tive protein binding to two targets on the same DNA:application to binding of tet repressor variants tomultiple and single tet operator sites. Nucl. Acids Res.19, 1021–1028.

34. Senear, D. F. & Brenowitz, M. (1991). Determination ofbinding constants for cooperative site-specific protein-DNA interactions using the gel mobility-shift assay.J. Biol. Chem. 266, 13661–13671.

35. Wickner, S., Hoskins, J. & McKenney, K. (1991).Monomerization of RepA dimers by heat shockproteins activates binding to DNA replication origin.Proc. Natl. Acad. Sci. USA, 88, 7903–7907.

36. Lu, Y. B., Datta, H. J. & Bastia, D. (1998). Mechanisticstudies of initiator-initiator interaction and replicationinitiation. EMBO J. 17, 5192–5200.

37. Wu, F., Levchenko, I. & Filutowicz, M. (1995). ADNA segment conferring stable maintenance on R6Kγ-origin core replicons. J. Bacteriol. 177, 6338–6345.

38. McEachern, M. J., Bott, M. A., Tooker, P. A. &Helinski,D. R. (1989). Negative control of plasmid R6K


replication: possible role of intermolecular coupling ofreplication origins. Proc. Natl Acad. Sci. USA, 86,7942–7946.

39. Park, K., Han, E., Paulsson, J. & Chattoraj, D. K.(2001). Origin pairing (‘handcuffing’) as a mode ofnegative control of P1 plasmid copy number. EMBO J.20, 7323–7332.

40. Majka, J., Zakrzewska-Czerwinska, J. & Messer, W.(2001). Sequence recognition, cooperative interaction,and dimerization of the initiator protein DnaA ofStreptomyces. J. Biol. Chem. 276, 6243–6252.

41. Chen, S., Iannolo, M. & Calvo, J. M. (2005). Coopera-tive binding of the leucine-responsive regulatoryprotein (Lrp) to DNA. J. Mol. Biol. 345, 251–264.

42. Kim, J., Zwieb, C., Wu, C. & Adhya, S. (1989). Bendingof DNA by gene-regulatory proteins: construction anduse of a DNA bending vector. Gene, 85, 15–23.

43. Sambrook, J. & Russell, D. W. (2001). MolecularCloning: A Laboratory Manual, 3rd edit. Cold SpringHarbor Laboratory Press, Cold Spring Harbor, NY.

44. Wu, J. & Filutowicz, M. (1999). Hexahistidine (His6)-tag dependent protein dimerization: a cautionary tale.Acta Biochim. Polonica, 46, 591–599.

45. Krüger, R. & Filutowicz, M. (2003). Characterization ofHis-tagged, R6K-encoded π protein variants. Plasmid,50, 80–85.

46. Gavigan, S. A., Nguyen, T., Nguyen, N. & Senear, D. F.(1999). Role of multiple CytR binding sites on coop-erativity, competition, and induction at the Escherichiacoli udp promoter. J. Biol. Chem. 274, 16010–16019.

47. Brenowitz, M., Senear, D. F., Shea, M. A. & Ackers,G. K. (1986). Quantitative DNase footprint titration: amethod for studying protein-DNA interactions. Meth-ods Enzymol, 130, 132–181.

Edited by J. Karn

(Received 16 November 2006; received in revised form 15 February 2007; accepted 16 February 2007)Available online 2 March 2007

Documents

Mechanism of Origin Activation by Monomers of R6K-encoded p Protein