Sequence-Dependent Upstream DNA–RNA PolymeraseInteractions in the Open Complex with λPR and λPRMPromoters and Implications for the Mechanism ofPromoter Interference

Laura Mangiarotti1, Sara Cellai1, Wilma Ross2,Carlos Bustamante3 and Claudio Rivetti1⁎1Department of Biochemistryand Molecular Biology,University of Parma,Viale G. P. Usberti 23/A,43100 Parma, Italy2Department of Bacteriology,University of Wisconsin,1550 Linden Drive, Madison,WI 53706, USA3Departments of Physics,Chemistry, and Molecular andCell Biology, and HowardHughes Medical Institute,University of California,Berkeley, CA 94720, USA

Received 23 July 2008;received in revised form11 November 2008;accepted 12 November 2008Available online24 November 2008

Upstream interactions of Escherichia coli RNA polymerase (RNAP) in anopen promoter complex (RPo) formed at the PR and PRM promoters ofbacteriophage λ have been studied by atomic force microscopy. We de-monstrate that the previously described 30-nm DNA compaction observedupon RPo formation at PR [Rivetti, C., Guthold, M. & Bustamante, C. (1999).Wrapping of DNA around the E. coli RNA polymerase open promotercomplex.EMBO J., 18, 4464–4475.] is a consequence of the specific interactionof the RNAP with two AT-rich sequence determinants positioned from −36to −59 and from −80 to −100. Likewise, RPos formed at PRM showed aspecific contact between RNAP and the upstream DNA sequence. Wefurther demonstrate that this interaction, which results in DNA wrappingagainst the polymerase surface, is mediated by the C-terminal domains of α-subunits (carboxy-terminal domain). Substitution of these AT-rich sequenceswith heterologous DNA reduces DNAwrapping but has only a small effecton the activity of the PR promoter. We find, however, that the frequency ofDNA templates with both PR and PRM occupied by an RNAP significantlyincreases upon loss of DNAwrapping. These results suggest that α carboxy-terminal domain interactions with upstream DNA can also play a role inregulating the expression of closely spaced promoters. Finally, a model for apossible mechanism of promoter interference between PR and PRM isproposed.

© 2008 Elsevier Ltd. All rights reserved.

Edited by R. EbrightKeywords: transcription; DNA wrapping; promoter interference; upstreamATR; atomic force microscopy

Introduction

The bacteriophage λ promoters PR and PRMcontrol the expression of repressor proteins cro andcI, which drive the establishment of the lytic and

lysogenic states.1 PR and PRM are divergently tran-scribed from start sites separated by a 82-bpsequence harboring the −10 and −35 hexamers ofboth promoters and the operator sites OR1, OR2, andOR3.2 Differential binding of cI to the three operatorsites represses PR and regulates PRM either positivelyor negatively. Conversely, differential binding of croto the operator sites represses PRM and, at higherconcentrations, turns down its own synthesis byrepressing PR as well.3,4 Open promoter complex(RPo) formation at PR is much faster than opencomplex formation at PRM; thus, the binding of anRNA polymerase (RNAP) to PRM only takes place inthe context of an RNAP bound to PR.5,6 It has beenshown that PRM activity is increased by mutationsdesigned to inactivate PR7–10 and, conversely, an

*Corresponding author. E-mail address:claudio.rivetti@unipr.it.Present address: QA Department, GlaxoSmithKline

Manufacturing, Strada Asolana 90, 43056 San Polo diTorrile (PR), Italy.Abbreviations used: RNAP, RNA polymerase; RPo,

open promoter complex; CTD, carboxy-terminal domain;ATR, A+T-rich; AFM, atomic force microscopy; wt, wildtype.

doi:10.1016/j.jmb.2008.11.019 J. Mol. Biol. (2009) 385, 748–760

Available online at www.sciencedirect.com

0022-2836/$ - see front matter © 2008 Elsevier Ltd. All rights reserved.

Fig. 1. (a) Sequences of upstream regions of the λPR and λPRM variants used in this work. The transcription start site +1, the −10 and −35 hexamers are singly underlined for PRand doubly underlined for PRM. ATR regions are shown in purple boldface type. Heterologous DNA is shown in lowercase blue type. Point mutations are shown in boldface type.(b) AFM image of RPos formed with wt RNAP and the wt PR–wt PRM DNA template. (c) AFM image of RPos formed withΔαCTDII RNAP and the wt PR–wt PRM DNA template.The images scan size is 2 μm.

749DNAWrapping

andthe

PR –P

RMProm

oterInterference

RNAP bound to PRM interferes with the utilization ofa weakened PR promoter.11 Interestingly, it has beenshown that PR and PRM promoters can be occupiedsimultaneously by an RNAP, and this has led to theconclusion that formation of an open complex at PRdoes not prevent binding of an RNAP to PRM, butrather impairs isomerization from a closed complexto an open complex.7–10 Contrary to expectations, a10-bp deletion between the −35 elements of the twopromoters, which should reduce the space availablefor two polymerases without affecting promotersequences, resulted in relief of interference.12,13Studies that aimed to elucidate the structure of theλPR open complex have shown that, at variance withother promoters, DNA is tightly wrapped aroundthe RNAP, extending the RNAP–DNA interactionfurther upstream of the PR −35 element.14–16 Thisextended contact is generally mediated by α-subunitcarboxy-terminal domains (CTDs) that interact withA+T-rich (ATR)DNA sequences located upstreamofthe promoter.17–19 This extended upstream DNAinteraction observed at PR prompted us to search forpossible upstream sequence motifs that mightspecifically interact with the RNAP and to furtherinvestigate their effect on PR activity and PRM acces-sibility in the context of a PR-bound RNAP. To thisend, we have employed atomic force microscopy(AFM) to image RPos formed at PR, at PRM, and atvariants of these promoters in which the upstreamregions were selectively substituted with heterolo-gous DNA. Comparison of the DNA contour lengthof RPo with that of free DNA molecules revealedpreviously unrecognized upstream DNA–RNAPinteractions at the PR promoter that might explainnot only the strong DNA compaction typical of RPoformed at this promoter but also the previouslydescribed PR–PRM promoter interference.

Results

To assess the involvement of ATR regions posi-tioned upstream of PR in the DNA compaction ofRpo, we have constructed PR promoter variants inwhich the upstreamDNAwas selectively substitutedwith heterologous DNA, and we have measured the

DNA compaction of the corresponding RPo. Figure1a reports the DNA sequence of the promotervariants used in this study, and Fig. 1b shows repre-sentative AFM images of RPo. RPos were formed byincubating DNA and RNAP in a 1:1 molar ratio intranscription buffer [20 mM Tris–HCl (pH 7.9),50 mM KCl, 5 mM MgCl2, and 1 mM DTT] at 37 °Cfor at least 20 min. RPo activity was verified byrunoff transcription assays conducted under similarconditions, but with the addition of heparin to pre-vent multiple rounds of transcription (Fig. 2). Hepa-rin could not be used in the AFM experimentsbecause it prevents adhesion of the DNA to micasupport; thus, the RPos, seen microscopically, wereconsidered such because they were formed underconditions favoring open complex formation andwere located at the promoter site.

DNA compaction at PR and PRM depends onupstream sequence determinants

The wild-type (wt) DNA template harbors both PRand PRM promoters; however, RNAP primarilybinds to PR, as demonstrated by the outcome ofsingle-round transcription experiments in whichonly a transcript from PR is observed (Fig. 2, lane 1)and by previous work.5–7,11 Therefore, complexesassembled with this DNA template were consideredas RPo formed at PR. DNA contour length analysis ofthese complexes shows an averageDNA compactionof 30 nm (Fig. 3a and Table 1), in full agreement withpreviously published data.14,15 A 30-nm DNA com-paction was also observed with PR (−100 to +34)(Fig. 3b and Table 1), in which the DNA sequenceupstream of −100 was substituted with vectorDNA. Thus, the DNA sequence upstream of posi-tion −100 is not involved in the DNA compactionobserved at PR (i.e., within the Rpo, this sequence isnot contacted by RNAP).Conversely, RPo formed with PR (−79 to +34), in

which the DNA sequence upstream of −79 wassubstituted with vector DNA, shows a DNAcompaction of 20 nm (Fig. 3c and Table 1). Thisresult indicates that in the PR region from −80 to−100, there must be a sequence determinant that isspecifically contacted by an RNAP bound at PR.

Fig. 2. Single-round in vitrotranscription of the DNA templatesshown in Fig. 1. Lane 1, wt PR–wtPRM; lane 2, PR (−100 to +34); lane 3,PR (−79 to +34); lane 4, PR (−59 to+34); lane 5, PR (−35 to +34); lane 6,PR−–wt PRM; lane 7, PRM (−35 to

+541). All transcription reactionswere carried out in the presence ofheparin. The two bands of the PRtranscript are probably due to inho-mogeneous runoff termination. ThePRM transcripts in lanes 6 and 7 havedifferent lengths because of the

different lengths of the downstream DNA. The relative band intensity is reported below each lane: lanes 2–5 withrespect to lane 1, and lane 7 with respect to lane 6.

750 DNA Wrapping and the PR–PRM Promoter Interference

Because a DNA compaction of 20 nm is still signifi-cantly high compared to those of other promoters,15we hypothesized that other sequence determinantsthat are capable of interacting with the RNAP mustbe present in the upstream sequence of PR. RPoformed with PR (−59 to +34), in which the DNAsequence upstream of −59 was substituted withvector DNA, again shows a DNA compaction of20 nm (Fig. 3d and Table 1). Thus, the DNA regionfrom −60 to −79 does not contain sequencedeterminants involved in DNA compaction. How-ever, substitution of the DNA sequence upstream of−35, as in the case of PR (−35 to +34), resulted in aDNA compaction of only 4 nm. This reduced extentof DNA compaction indicates the absence of stableinteraction between the RNAP and upstream DNA,and suggests that a second sequence determinant,specifically recognized by the RNAP, is present inthe PR region from −36 to −59.A similar analysis was conducted to investigate

the RNAP interaction in RPo formed at PRM. Toavoid competition, the stronger PR promoter wasinactivated by mutating the −10 hexamer in amanner known to reduce PR activity to 2–5% of wt(GATAAT changed to GGTGAC; Fig. 1a).20 Single-round in vitro transcription indeed shows thattranscription of the PR

−–wt PRM construct producesa 154-nt transcript from PRM and no detectabletranscription from PR (Fig. 2, lane 6). Thus, RPosassembled with this template were assumed to beformed at PRM. DNA contour length measurementsof these RPos revealed a DNA compaction of 18 nm(Fig. 4a and Table 1), which is an indication of anextended interaction of the RNAP with upstreamDNA. The DNA compaction observed at PRM issignificantly less than that observed at PR and iscomparable to the DNA compaction observed at PR(−59 to +34) and LacUV5 (UPfull) in which an UPelement has been placed in the −38/−59 region ofLacUV5.15 Thus, based on DNA compaction simi-larity, we hypothesized that the RNAP–DNA inter-action at PRM might involve DNA sequences in thePRM region from −36 to −60.In order to determine whether sequence deter-

minants in the upstream region of PRM contributeto the observed compaction, we constructed PRM(−35 to +541), a λPRM derivative in which thesequence upstream of −35 has been replaced withheterologous DNA (Fig. 1a; the PR promoter −10and −35 hexamer regions are deleted in this

Fig. 3. DNA contour length distributions of bare DNAand RPo formed with wt RNAP and PR derivatives asfollows: (a) wt PR–wt PRM; (b) PR (−100 to +34); (c) PR (−79to +34); (d) PR (−59 to +34); (e) PR (−35 to +34). In allpanels, white dashed bars represent bare DNA contourlength frequencies, and dark-gray bars represent RPocontour length frequencies. Solid lines represent Gaussianfitting of the distributions. Mean and standard error valuesderived from the fitting are reported in Table 1. DNAcompaction, defined as the difference between the meansof DNA and RPo distributions, is reported on each graph.

751DNA Wrapping and the PR–PRM Promoter Interference

construct). A contour length analysis of thesecomplexes is shown in Fig. 4b and Table 1. RPoat PRM (−35 to +541) displays a DNA compactionof 8 nm, thus confirming the presence of asequence determinant in the upstream region ofPRM that is capable of making specific interactionswith the RNAP.

Effects of upstream DNA determinants onPR and PRM promoter function

Two types of experiments, in vitro transcriptionand RNAP promoter association rate measurements,were carried out to determine whether the substitu-tions for upstream promoter DNA described above(Fig. 1a) affect the functional interaction of RNAPwith the PR or PRM promoters. The substitution forsequences upstream of −100 in PR did not alter its invitro transcription activity relative to that of the wttemplate (Fig. 2, lanes 1 and 2). However, substitu-tions for upstream regions closer to the PR promoterresulted in small increases in PR expression (∼1.7- to2.7-fold greater than wt for the −79, −59, and −35substitutions; Fig. 2, lanes 3–5). These results suggestthe possibility of a small inhibitory effect of the wtupstream regions on expression from PR. However,the observed increases in transcription did notcorrelate with any effect on the overall rates ofRNAP association with these PR promoters (ka; seebelow), suggesting that the mechanism of theincreases may involve steps after RNAP promotercomplex formation. Inhibitory effects of ATR up-stream sequences on promoter clearance havepreviously been noted for some promoters.21 Wenote that the small increases in transcription did notcorrelate with loss of the PRM −10 and −35sequences, since the fragment retaining thesesequences (PR −79 to +34) also showed a smallincrease in PR transcription.Transcription from the PRM promoter was not

detected for any of the promoter region fragmentscontaining a PR promoter (including, as expected,those for fragments −79, −59, and −35, in which allor part of PRM was deleted; Fig. 2, lanes 1–5).However, transcription from PRM was observedwith a fragment containing a substitution in the −10region of PR (PR

−–wt PRM in Fig. 1a; lane 6 in Fig. 2).The substitution for sequences upstream of the −35region of PRM [PRM (−35 to +541); Fig. 1a] removedthe PR promoter and resulted in a small increase inPRM transcription (b2-fold; Fig. 2, compare lanes 6and 7). (The size of the RNA transcripts in lanes 6and 7 is different because the two promoter frag-ments have different lengths of DNA downstream ofthe PRM promoter.)

Table 1. Summary of the bare DNA and RPo contour length measurements performed in this work

RNAP DNA template PromoterDNA length

(bp)Bare DNA contour

length (nm)RPo contourlength (nm)

DNA compaction(nm)

wt RNAP wt PR–wt PRM PR 1054 338±0.2 (n=157) 308±0.2 (n=353) 30±0.3wt RNAP PR (−100 to +34) PR 975 315±0.2 (n=155) 285±0.3 (n=225) 30±0.4wt RNAP PR (−79 to +34) PR 954 312±0.2 (n=92) 292±0.4 (n=117) 20±0.5wt RNAP PR (−59 to +34) PR 963 313±0.3 (n=143) 293±0.4 (n=90) 20±0.5wt RNAP PR (−35 to +34) PR 963 315±0.4 (n=102) 311±0.5 (n=138) 4±0.7wt RNAP PR−–wt PRM PRM 1003 325±0.4 (n=115) 307±0.5 (n=68) 18±0.6wt RNAP PRM (−35 to +541) PRM 999 323±0.6 (n=70) 315±0.7 (n=76) 8±0.9

Contour length values of bare DNA and RPo represent the mean±SE obtained from the fitting of the DNA contour length distributionsshown in Figs. 3 and 4. n is the number of molecules measured. DNA compaction is given by the difference between the mean DNAcontour length of bare DNA and the mean DNA contour length of RPo.

Fig. 4. DNA contour length distributions of bare DNAand RPo at the PRM promoter. (a) RPo formed with wtRNAP and the PR

−–wt PRM DNA construct. (b) RPo formedwith wt RNAP and the PRM (−35 to +541) DNA construct.White dashed bars represent bare DNA contour lengthfrequencies, and dark-gray bars represent RPo contourlength frequencies. Solid lines represent Gaussian fittingof the distributions. Mean and standard error valuesderived from the fitting are reported in Table 1. DNAcompaction is reported on each graph.

752 DNA Wrapping and the PR–PRM Promoter Interference

DNA wrapping at PR does not affect the overallassociation and isomerization rate constants

By comparing the rates of association of RNAPwith λPR fragments containing different lengths ofupstream DNA (a “full-length” fragment withupstream DNA up to −110, and an upstream trun-cated fragment with upstream DNA only up to−47), it was shown that the presence of upstreamDNA greatly increases the rate of competitor-resistant complex formation (by at least 20-fold).22This effect occurs primarily by accelerating isomer-ization of the complex (i.e., in the step after initialpromoter binding). In addition, a substitution inαCTD that prevents its DNA binding reduces theoverall rate of complex formation by at least 20-fold.23 Therefore, it was reasonable to hypothesizethat the sequence determinants responsible forDNA wrapping might be required for the large(20-fold) effect on isomerization.To determine whether specific sequences within

the upstream regions affect the rate of formation ofRNAP promoter complexes at PR, the overallsecond-order association rate constants (ka) andrates of isomerization (k2) were determined for theformation of heparin-stable complexes with frag-ments containing wt PR or three of the upstreamsubstituted promoters (−79, −59, and −35; Fig. 5).

The association rate constants (ka) and isomerizationrate constants (k2) for these four PR fragmentsdiffered by less than 2-fold Fig. 5b). The valuesobserved (ka=6.9×105 M−1 s−1; k2=2.3×10−2s−1 forthe full-length PR fragment) were in good agreementwith the values previously determined for PR undervery similar solution and temperature conditions(20 °C; ka=6.4×105 M−1 s−1; k2=1.4×10−2s−1).24 Weconclude that the upstream sequence determinantsthat correlate with DNA wrapping (Fig. 3) do notincrease the rate of RNAP association with PR, sincethe nonspecific sequence substituted for native λsequence in these promoter constructs (Fig. 1a) wasfunctionally equivalent to the native λ sequence inits effects on association rate. Rather, we suggest thatsequence-nonspecific interactions with RNAPaccount for the previously observed effects ofupstream DNA on the rate of association of RNAPwith PR. These findings are consistent with theobservation that the presence of upstream DNA alsoaffects the isomerization rate at LacUV5,23 a promo-ter that does not contain an UP element25 and forwhich we have previously shown that it does notwrap upstream DNA.15 Our results are also con-sistent with the previous finding that substitution ofnative PR sequence upstream of ∼−60 did not affectthe kinetics of promoter complex formation (cited inDavis et al.22).

Fig. 5. Effects of upstream substitutions on the association of wt RNAP with the λPR promoter. (a) Double-reciprocal(τ) plots (see Materials and Methods) from the reaction performed in a range of RNAP concentrations with 165-bp DNAfragments (from −115 to +50 of PR) at 20 °C. wt PR–wt PRM (filled circles), PR (−79 to +34) (open triangles), PR (−59 to +34)(open diamonds), PR (−35 to +34) (filled squares). (b) Kinetic constants derived from the data shown in (a).

753DNA Wrapping and the PR–PRM Promoter Interference

PRM occupancy in the context of aPR-bound RNAP

The DNA contour length measurements presen-ted above indicate that RNAP interacts extensivelywith the upstream regions of both PR and PRM. Forinstance, an RNAP bound at PR is in contact withDNA beyond the start site of PRM, whereas anRNAP bound at PRM is in contact with the DNAregion surrounding the −35 element of PR. As aresult of this geometry, simultaneous binding ofRNAP to PR and PRM should be difficult. However,previous studies have shown that binding to PR andPRM is not mutually exclusive, although binding toPRM is much slower in the context of an RNAPbound to PR.5,7 In order to quantitatively determinethe fraction of DNA templates in which both pro-moters are effectively occupied by an RNAP, AFMimages were analyzed by counting the number ofcomplexes comprising one or two RNAPs. DNAtemplates in which both PR and PRM are occupied byan RNAP are easily discernible by AFM becausethey are characterized by two adjacent globularfeatures located near the center of the template(Fig. 6). It must be noted that, from the AFM images,it is not possible to distinguish whether the RNAPbound to either PR or PRM forms transcriptionallycompetent RPos. The results are summarized inTable 2, which shows that, in the case of RPo

assembled with a DNA template harboring wt PRand wt PRM, 45% of the total DNA molecules seenmicroscopically had at least one RNAP bound. Ofthose, 98% had one RNAP, and 2% had two RNAPs(i.e., had both PR and PRM simultaneously occupiedby RNAP). In controls, the same analysis was per-formed on complexes obtained with PR (−35 to +34)in which PRM has been deleted. In this case, 41% ofthe total molecules had at least one RNAP bound. Ofthose, 98% had one RNAP, and 2% had two RNAPs.Similarly, with the PR−–wt PRMDNA template, whichshould form RPo only at PRM, we observed that 25%of the total molecules had at least one RNAP bound,98% of which had one RNAP and 2% of which hadtwo RNAPs. Thus, it appears that the small fractionof double complexes observed with these DNAtemplates represents the background of themeasurements. On the other hand, in the case of PR(−79 to +34) for which a reduced DNA compactionof RPo formed at PR was observed, the double com-plexes were 8% of the total number of complexesscored. Consistently, RPos assembled with RNAPαCTDmutants on the wt PR–wt PRM DNA template,which are characterized by a reduced DNA compac-tion relative to wt RNAP (these data were extractedfrom AFM images relative to experiments publishedin Cellai et al.15), the percentage of double complexeswas increased compared to that of the wt RNAP.Thus, it appears that, under our experimental con-

Table 2. Frequency of DNA templates with one or two RNAPs bound at PR–PRM

RNAP DNA template Complexes with at least one RNAP (%) Complexes with two RNAPs (%)

wt RNAP wt PR–wt PRM 45 2wt RNAP PR (−35 to +34) 41 2wt RNAP PR−–wt PRM 25 2wt RNAP PR (−79 to +34) 41 8Δ6-αI/Δ6-αII RNAPa wt PR–wt PRM 10 10Δ12-αI/Δ12-αII RNAPa wt PR–wt PRM 10 10ΔαCTDII RNAPa wt PR–wt PRM 71 3ΔαCTDI/ΔαCTDII RNAPa wt PR–wt PRM 24 4

Double complexes are those in which two RNAPs are bound to the same DNA template and their position is compatible with binding toPR and PRM. An example of such complexes is shown in Fig. 6. The percentage of single complexes is given with respect to the totalnumber of molecules seen microscopically (free DNA, single RPo, and double RPo), whereas the percentage of double complexes hasbeendeterminedwith respect to the total number of complexes (singleRPoplusdoubleRPo).Δ6-αI/Δ6-αII RNAPandΔ12-αI/Δ12-αII RNAPare RNAP derivatives lacking 6 (α-residues 235–241) and 12 (α-residues 235–247) amino acid residues of the α-linker, respectively.26ΔαCTDII RNAP is an RNAP derivative with αCTDI but without αCTDII18.ΔαCTDI/ΔαCTDII RNAP is an RNAP derivative lacking bothαCTDI and αCTDII18.

a Data extracted from AFM images of experiments published in Cellai et al.15

Fig. 6. Top and lateral views of aDNA template with both PR andPRM occupied by an RNAP mole-cule. Because of the geometry of theDNA template, the two complexescannot be distinguished from eachother. These double complexeswere rare in the case of wt PR andwt PRM. However, their frequencyincreased when DNA wrapping atPR was impaired by mutationseither in the upstream DNA or inthe RNAP α-subunits.

754 DNA Wrapping and the PR–PRM Promoter Interference

ditions, binding to PRM is more efficient whenwrapping of the DNA around the PR-bound RNAPis compromised either by changes in the upstreamregion of PR or by αCTD mutations.

Discussion

In this study, we have demonstrated that theDNA compaction that results when RNAP binds theλPR promoter to form an RPo can be accounted forby a direct interaction of RNAP with upstreamDNA sequence determinants: one located in theregion from −80 to −100, and the other located inthe region from −36 to −59. Both regions containATR sequences that are similar to those shown tointeract with RNAP αCTD.18,27,28 In a previouslypublished AFM study,15 we have shown that theDNA compaction observed at PR is significantlyreduced when the complexes are assembled withRNAP αCTD mutants, consistent with the hypoth-esis that αCTD mediates the RNAP–DNA interac-tion with the upstream promoter region. Because ofthe size of the polymerase molecule (∼10 nm indiameter)29 and the 100-bp contour length of DNAinvolved in the interaction (∼34 nm), DNA wrap-ping around the protein surface must be invoked toaccount for the hypothesized upstream protein–DNA interaction that results in DNA compac-tion.14,16,30,31 The relationship between observedcompaction and modeled wrapping is schematicallyshown in Fig. 7a.In the case of wt PR–wt PRM, contact between the

RNAP and sequences in the region −80/−100 of PRresults in an almost complete turn of DNA aroundthe polymerase, with consequent arrangement ofPRM core promoter on the surface of the PR-boundRNAP (Fig. 7b). Substitution of the sequence up-stream of region −79 of PR with heterologous DNAeliminates the stretch of thymines from positions−91 to −100. This substitution results in adiminished DNA compaction [20±0.5 nm for PR(−79 to +34) against 30±0.3 nm for wt PR–wt PRM],which, in turn, suggests partial unwrapping of theDNA from the RNAP (Fig. 7c). Moreover, substitu-tion of the sequence upstream of −35 resulted in afurther reduction of the DNA compaction observedat PR [4±0.7 nm for PR (−35 to +34) against 30±0.3 nm for wt PR–wt PRM], indicating that theATR found within this region is also specificallycontacted by an RNAP bound at PR. Such a smallDNA compaction, most probably due to the factthat the RNAP holds part of the DNA out of thesurface as depicted in Fig. 7a, suggests that theRNAP–DNA interaction is limited to the −10 and−35 hexamers of PR, with no upstream DNAwrapping involved (Fig. 7d). The DNA contourlength analysis of RPo formed at the PRM promotershows a DNA compaction of 18 nm, which reducesto 8 nm upon substitution of the ATR betweenregions −40 and −60 of PRM, suggesting that thisATR is specifically contacted by the RNAP whenbound to PRM.

Significance ofRNAP–DNAupstream interactions

What are the implications of the stable upstreaminteractions made by the RNAP at PR and PRM?Previous work has shown that A-tracts are goodαCTD binding sites and can increase the activity ofthe promoter.18,27,28 In addition, a recent study hasdemonstrated that the presence of upstream DNAincreases the rate of RPo formation at PR by a factorof 35 and 60, at 37 and 17 °C, respectively.22 Thus,we hypothesized that the DNA wrapping deter-mined by the interaction of the RNAP with up-stream ATRs might have the same effect on theisomerization rate also because a change in writhe isphysically linked to a change in twist. Contrary toour hypothesis, the data suggest that although PRATRs are responsible for the extended RNAP–DNAinteraction that structurally results in DNA wrap-ping, they have no stimulatory effect on the activityof PR in vitro; rather, we have found a small inhi-bitory effect of these upstream regions that mayoriginate from steps after RPo formation. This resultis in agreement with previous work showing thatsequence-specific interactions involving PR and PRMATRs have little effect on the activity of these pro-moters in vitro.20,25 We thus conclude that althoughαCTD upstream DNA contacts are important forpromoting the formation of the RPo (upstreamDNAincreases the rate of competitor-resistant complexformation by at least 20-fold),22 this contact may betransient and does not require specific A-tractsequences that are needed for wrapping.In the particular context of the divergent promo-

ters PR and PRM, which control the synthesis of theantagonist regulators cI and cro, DNA wrappingbecomes particularly interesting because it involvessequences that overlap with the divergent promoterand thus may suggest a mechanism for mutualinterference between these promoters. Our findingthat an RNAP bound at PR contacts sequencesbeyond the start site of PRM indicates that the −10and −35 hexamers of the latter must lie on the sur-face of the PR-bound RNAP, a geometry that shouldreduce the accessibility of PRM. In keeping with thishypothesis is the observation that, when RPos wereformed with wt template and wt RNAP, the fractionof complexes displaying both promoters occupiedby polymerases was very small. Conversely, whenRPos were formed with the mutated αCTD RNAPs,the fraction of double complexes was significantlyincreased (Table 2), in accordance with the smallerDNA compaction established for these mutants thatshould uncover the PRM's recognition sites. Asimilar result was also obtained with a DNAtemplate (PR−79 to +34) in which the ATR between−80 and −100 was substituted for heterologousDNA while maintaining PRM. This substitutionseems to be sufficient for an RNAP to gain accessto PRM in the context of an RNAP bound at PR. Thecorrelation between loss of wrapping and thefrequency of double complexes in both of theseexperiments suggests the possibility of a causal rela-tionship. It is also possible that alternative mecha-

755DNA Wrapping and the PR–PRM Promoter Interference

nisms could contribute to the frequency of doublecomplexes. For example, the RNAP α-subunitmutations could potentially affect this frequency byreducing the rate of formation of complexes at PR.23However, effects on the rate of PR complex formationwould not account for the increase in doublecomplex frequency with the −79 to +34 promoterfragment, since little or no effect on the rate of PRcomplex formation was observed with this fragment(Fig. 5).Previous work has shown that, in the case of

phage λ, an RNAP can bind to PRM also in thecontext of a PR-bound RNAP, but it is unable toform a transcriptionally competent complex.7–10The data presented herein indicate that simulta-neous binding to PR and PRM is rare in the wt case,

and it increases whenever DNA wrapping at PR isimpaired. A possible explanation of such a dis-crepancy may reside in the different experimentalconditions. In particular, for AFM experiments, RPoswere formed with a 1:1 RNAP/DNA ratio (higherratios complicate image interpretation because ofmany free RNAP molecules on mica support),whereas previous abortive initiation, gel shift, andDNA footprinting assays were performed with alarge excess of RNAP with respect to the promoterDNA. Furthermore, it must be pointed out that inHershberger and deHaseth and Hershberger et al.,experiments were performed with an up-mutationof PRM to facilitate in vitro transcription assays andwhich may account for the increased PRMoccupancy.7,8 Different salt concentrations and pH

Fig. 7 (legend on next page)

756 DNA Wrapping and the PR–PRM Promoter Interference

values might also have an effect on the probabilityof promoter binding.Finally, two studies have shown that deletions

between the −35 elements of PR and PRM surpri-singly reduce interference between these promo-ters.12,13 This effect was more pronounced for theD10 construct in which a 10-bp deletion reduced thespacer between the −35 elements to 2 bp. As sug-gested, the 10-bp deletion may relieve interferenceby placing the two promoters in an ideal configura-tion with their respective contacts with RNAP. Wefurther hypothesize that a deletion within this regiondirectly affects PR and PRM ATRs and changes thespatial relation of the upstream sequence determi-nants with respect to the RNAP αCTDs. This mayaffect the DNA-wrapping ability of PR and PRM. AnAFM analysis of transcription complexes formedwith the D10would be required to better understandthis unexpected behavior.Based on previous and present data, our working

hypothesis on the mechanism of promoter inter-ference between PR and PRM can be summarized asfollows: an RNAP quickly binds to the strong pro-moter PR, wraps the DNA up to position −100, andforms an open complex. Because of the αCTDinteraction with upstream ATRs, DNA wrapping isstable and constrains the PRM promoter elements onthe surface of the PR-bound RNAP. Under thesecircumstances, binding to PRM is still possible; how-ever, in order to be accomplished, the DNA has tobe unwrapped from the RNAP bound at PR, andthis unwrapping can constitute an additional rate-limiting step during open complex formation at PRM.Furthermore, gaining access to the −10 and −35elements of PRM in the context of a PR-bound RNAPmay not be sufficient for complete open complexformation at PRM because of the lack of free upstreamDNA known to accelerate isomerization.22,23 Theabsence of upstreamDNA contacts might contributeto slowed open complex formation at PRM.

Materials and Methods

DNA

The 1054-bp-long DNA template wt PR–wt PRM wasobtained by HindIII digestion of plasmid pSAP.14 Thistemplate contains λ-DNA from −438 to +34 with respectto the PR start site, which is positioned at 616 bp from thedownstream end. The 975-bp-long DNA template PR(−100 to +34) was obtained by PCR from plasmid pPR100using primers NEB_FOR (5′-AAAACCTCTGACACATG-CAGC) and NEB_REV (5′-GCTGCCCTTTTGCTCA-CATG). pPR100 was constructed by cloning the pSAPsequence from −100 to +100 into the SmaI restriction siteof pNEB193 (New England Biolabs). The 954-bp-longDNA template PR (−79 to +34) was obtained by PCR fromplasmid pPR79 using primers NEB_FOR and NEB_REV.pPR79 was constructed by cloning the pSAP sequencefrom −79 to +100 into the SmaI and HindIII restrictionssites of pNEB193. The 963-bp-long DNA template PR (−59to +34) was obtained by PCR from plasmid pPR59 usingprimers NEB_FOR and NEB_REV. pPR59 was constructedby cloning the pSAP sequence from −59 to +100 into theHincII restriction site of pNEB193. The 963-bp-long DNAtemplate PR (−35 to +34) was obtained by PCR fromplasmid pPR35 using primers 5′-AAAACCTCTGACA-CATGCAGC and 5′-GCTGCCCTTTTGCTCACATG.pPR35 was constructed as follows: a 160-bp DNA frag-ment was obtained by PCR from pSAP using a primerdesigned to amplify the pSAP region from −35 to +100.The forward primer used in this PCRwas obtained as suchto insert 24 bp of heterologous DNA upstream of position−35, with respect to PR, in order to maintain in registry thesequence of the different constructs. The resulting 160-bpinsert was cloned into the HincII restriction site ofpNEB193.The 1003-bp-long DNA template PR

−–wt PRM wasobtained by PCR from plasmid pPRM using primers 5′-TGGAATTACCTTCAACCTCAAGC and 5′-CTGTTTT-GGTCTAAGCTGCGG. pPRM is a pSAP derivative inwhich the −10 hexamer of PR has been inactivated asdescribed in Tang et al.20 The 999-bp-long DNA templatePRM (−35 to +541) was obtained by PCR from plasmid

Fig. 7. Model of the proposed upstream DNA interaction in the RPo formed at PR. (a) Schematic representationshowing the relationship between the DNA compaction observed by AFM andmodeled DNAwrapping. In the absence ofwrapping, an RNAP (blue circle) can adhere to mica in an orientation such that the DNA (thick purple line) is held abovethe surface plane. The contour length of the DNA line is 6.1d, where d is the diameter of the RNAP (d∼10 nm). Assumingthat all the complexes are deposited as depicted in the leftmost drawing, the apparent DNA contour length (ℓ) measuredfrom the top view (thick dashed line in the middle drawing) is 5.5d. Thus, this arrangement produces a compaction of 0.6d(∼6 nm). In a more realistic situation, surface-bound complexes are randomly oriented; thus, we expect an average DNAcompaction of 4 nm or less in the absence of wrapping. In the case of an almost complete turn of DNAwrapping (rightdrawing), DNA contour length measurements are also affected by the tip broadening effect (outlined by the gray dashedcircle), as a result of which the DNA in proximity to the RNAP is hidden from the observer. Therefore, during contourlength measurements, the DNA path is drawn through the center of the protein, as shown by the thick dashed line, whichhas a length of 3.5d and a resulting compaction of 2.6d (∼26 nm). If we consider that part of the wrapped DNA can also beheld out of the mica plane, the expected DNA compaction for such an arrangement is approximately 3d (∼30 nm), in goodagreement with the value experimentally measured. (b) With a wt DNA template, an RNAP bound at PR wraps the DNAby contacting AT-rich sequences up to position −100 (Fig. 1a). With this geometry, the PRM promoter lies on the surface ofthe RNAP, presumably in proximity to the αCTD. (c) Substitution of the DNA sequence between −80 and −100 eliminatesan ATR region, causing partial loss of DNAwrapping with a consequent increased accessibility to PRM. (d) Substitution ofthe DNA sequence upstream of −35 eliminates two ATRs (one in the region between −80 and −100, and the other in theregion between −40 and −60) and completely abolishes DNAwrapping at PR. Polymerase has been drawn based on thestructure determined by Murakami et al.32 (Protein Data Bank ID 1L9Z). RNAP β- and β′-subunits are in light and darkblue, respectively; α-subunits are in light and dark green; and σ-subunit is in yellow. The αCTD, which is absent in thecrystal structure, has been drawn to scale as green ellipsoids. A schematic representation of the DNA is drawn to scale inlight and dark purple.

757DNA Wrapping and the PR–PRM Promoter Interference

pPRM35 using primers NEB_FOR and NEB_REV.pPRM35 was constructed by cloning the λ sequencefrom −35 to +137 with respect to PRM into the HincIIrestriction site of pNEB193. In analogy to pPR35, thereverse primer used to amplify the λ sequence wasobtained as such to insert 24 bp of heterologous DNAupstream of position −35 of PRM.PCR amplification was carried out in standard reaction

conditions using Deep Vent DNA polymerase. Preparativerestriction digests were carried out overnight. All DNAfragments were purified on a 1% (wt/vol) agarose gel andrecovered by electro elution in an Elutrap apparatus(Schleicher & Schuell, Keene, NH). The DNA wasextracted with phenol–chloroform, precipitated withethanol, and resuspended in TE buffer [50 mM Tris–HCl(pH 7.4) and 1 mM ethylenediaminetetraacetic acid]. TheDNA concentration was determined by absorbance at260 nm. PCR amplification of promoter fragments for invitro transcription and association rate measurements isdescribed below.

RNA polymerases

Wild-type E. coli RNAP used in AFM experiments andrunoff transcription assays was prepared as described inEstrem et al. or purchased from Epicenter Biotechnologies(Madison, WI).18 Wild-type E. coli RNAP used in associa-tion kinetic measurements was purified as described inBurgess and Jendrisak33 and determined to be ∼60%active, as described in Ross and Gourse.23

RPo formation and AFM imaging

RPoswere prepared bymixing 20 fmol of DNA templateand 20 fmol of RNAP in transcription buffer [20 mMTris–HCl (pH 7.9), 50 mM KCl, 5 mM MgCl2, and 1 mMDTT]. The 10-μl reaction was incubated at 37 °C for 20 or40 min. The reaction was diluted to 1–2 nM complexes in20 μl of deposition buffer [4 mM Hepes (pH 7.4), 10 mMNaCl, and 4 mM MgCl2] and immediately deposited ontofreshly cleaved ruby mica (Mica, New York, NY). Thesample was incubated for about 2 min before the surfacewas rinsed with water milli-Q (Millipore) and dried with aweak stream of nitrogen. AFM imaging was performed onthe dried sample with a Nanoscope III microscope (DigitalInstruments, Inc., Santa Barbara, CA) operating in tappingmode. Commercial diving board silicon cantilevers(Nanosensor or Olympus) were used. Images of 512×512pixels were collected with a scan size of 2 μm at a scan rateof three to four lines per second.

Runoff transcription assays

Runoff transcription assays were performed underconditions similar to those used in the AFM experiments.Linear DNA fragments of ∼400 bp were obtained by PCRfrom the constructs described above. DNA fragmentswere such that, in all cases, the runoff transcripts initiatedat PR would have a length of 101 nt, whereas transcriptsinitiated at PRM would have lengths as follows: wt PR–wtPRM, 154 nt; PR (−100 to +34), 154 nt; PR (−79 to +34),140 nt; PR (−59 to +34), 140 nt; PR (−35 to +34), 140 nt;PR−–wt PRM, 154 nt; and PRM (−35 to +541), 137 nt. DNA

and RNAP at a final concentration of 20 nM each weremixed in transcription buffer, and the reaction wasincubated at 37 °C for 40 min to allow open complexformation. Subsequently, heparin (to a final concentration

of 100 μg/ml) was added to the reaction, followed by theaddition of 10 U of SUPERase-In (Ambion) as RNaseinhibitor, 10 μCi of [α-32P]UTP (Amersham Biosciences),and nucleoside triphosphates to a final concentration of20 μM each. Transcription was allowed to proceed for15 min at room temperature. The 10-μl reactions wereterminated by dipping the tubes into dried ice and byadding 30 μl of gel loading buffer. The reactions wereheated at 90 °C for 2 min and loaded onto a 6% poly-acrylamide gel and 7 M urea gel. The gel was scannedwith a Personal Imager FX (Bio-Rad). Bands correspond-ing to transcription products were then quantified usingthe MultiAnalyst PC software (Bio-Rad).

Association kinetic measurements for RNAP bindingto λPR

Linear fragments of similar lengths (∼165 bp; promotersequence from −115 to +50) were prepared by PCR ampli-fication from each of the four plasmids: pSAP (wt PR–wtPRM), pPR79 [PR (−79 to +34)], pPR59 [PR (−59 to +34)],and pPR35 [PR (−35 to +34)]. Forward primers were either5′-GTACGAATTCGATATCCAGCTATGACCATGA-TTACGCCAAGC for plasmids pPR79, pPR59, and pPR35,o r 5 ′ -GTACGAATTCGATATCTGTGTTAATG-GTTTCTTTTTTGTGCTC for pSAP. The reverse primerwas 5′-CAGGACCCGGGAAGCTTTTAATTAACACTCT-TATACATTATTCC for all plasmids. PCR products werepurified using QIAquick PCR Purification columns, and5 pmol of each was digested with HindIII (encoded in thereverse primer at position ∼+50 with respect to the tran-scription start site). Nontemplate strand HindIII site 3′ends were labeled by filling in with [α-32P]dATP (PerkinElmer) and Sequenase (US Biochemicals), and labeledfragments were digested with EcoRV (encoded in forwardprimers) to create a blunt upstream end at position −115.Labeled fragments were gel-purified, eluted, and concen-trated using ElutipD columns (Whatman).Rates of association of RNAP to form heparin-resistant

complexes were determined by a filter binding assay,essentially as described previously.6,23 Under these con-ditions, binding to λPR is much faster than binding toλPRM, and only complexes at λPR are detected6. Briefly,fragments (b1 nM) were incubated with a series ofconcentrations of wt RNAP (with RNAP in at least 3×molar excess over fragment) in a buffer containing 10 mMTris–HCl (pH 8.0), 120 mM KCl, 10 mM MgCl2, 1 mMDTT, and 100 μg/ml bovine serum albumin. The RNAPpreparation used was ∼60% active, as determined by apromoter titration assay (see Ross and Gourse23), andvalues in Fig. 5 were determined using concentrations ofactive RNAP. Reactions, carried out at 20 °C, were ini-tiated by addition of RNAP. At time intervals, aliquotswere sampled to tubes containing heparin (final concen-tration, 50 μg/ml) and filtered after 30 s. Radioactivityretained on filters was quantified by exposure to a phos-phorimager screen. Data for each active RNAP concentra-tion were plotted as counts per minute retained versustime, and an observed rate constant, kobs, was determinedfrom fitting to the equation: [cpmobs = (cpmplateau)(1−e−kobst )]. Data in Fig. 5a are shown in a τ plot (1/kobsversus 1/[RNAP]). Values in Fig. 5b were determined fromnonlinear plots of kobs versus [RNAP]. Data were fitted tothe equation kobs=kak2[RNAP]/(ka[RNAP]+k2) to deter-mine the composite association rate constant ka and theisomerization rate constant k2 for each fragment. K1 wasdetermined from the relationship ka= k2K1. The ka and k2values for the wt fragment were consistent with values

758 DNA Wrapping and the PR–PRM Promoter Interference

previously reported for λPR under these temperature andsalt conditions.24

Image analysis

AFM images were analyzed using a software written inMATLAB environment. DNA contour length measure-ments were performed as described in Rivetti andCodeluppi.34 DNA molecules suited for analysis wereselected by visual inspection based on the followingcriteria: the molecule was completely visible on the image,its contour was not ambiguous, RNAP was bound at theexpected position, and no other proteins were bound tothe same DNA. Data were elaborated with MATLAB andgraphedwith Sigmaplot (Systat Software, Inc., California).Contour length distributions were fitted with a Gaussianfunction using Sigmaplot.

Acknowledgements

We thank R. Ebright for providing the αCTDmutant RNAPs, and for giving suggestions andcomments on the manuscript. We also thank thereviewers for their helpful remarks and suggestions.We are grateful to the Centro InterdipartimentaleMisure of the University of Parma for the AFMfacility. This work was supported by a grant fromthe Fondazione Cariparma. W.R. was supported byNational Institutes of Health grant R37 GM37048.

References

1. Ptashne, M. (2004). A Genetic Switch: Phage λ Revisited,third edit. Cold Spring Harbor Laboratory Press,Plainview, NY.

2. Ptashne, M., Johnson, A. D. & Pabo, C. O. (1982). Agenetic switch in a bacterial virus. Sci. Am. 247, 128–130;132, 134–140.

3. Meyer, B. J.&Ptashne,M. (1980).Gene regulation at theright operator (OR) of bacteriophage lambda: III.Lambda repressordirectly activates gene transcription.J. Mol. Biol. 139, 195–205.

4. Meyer, B. J., Maurer, R. & Ptashne, M. (1980). Generegulation at the right operator (OR) of bacteriophagelambda: II. OR1, OR2, and OR3: their roles inmediating the effects of repressor and cro. J. Mol.Biol. 139, 163–194.

5. Hawley, D. K. & McClure, W. R. (1982). Mechanism ofactivation of transcription initiation from the lambdaPRM promoter. J. Mol. Biol. 157, 493–525.

6. Roe, J. H., Burgess, R. R. & Record, M. T. J. (1984).Kinetics and mechanism of the interaction of Esche-richia coli RNA polymerase with the lambda PRpromoter. J. Mol. Biol. 176, 495–522.

7. Hershberger, P. A. & deHaseth, P. L. (1991). RNApolymerase bound to the PR promoter of bacterioph-age lambda inhibits open complex formation at thedivergently transcribed PRM promoter. Implicationsfor an indirect mechanism of transcriptional activationby lambda repressor. J. Mol. Biol. 222, 479–494.

8. Hershberger, P. A., Mita, B. C., Tripatara, A. &deHaseth, P. L. (1993). Interference by PR-bound

RNA polymerase with PRM function in vitro.Modulation by the bacteriophage lambda cI protein.J. Biol. Chem. 268, 8943–8948.

9. Woody, S. T., Fong, R. S. & Gussin, G. N. (1993). Effectsof a single base-pair deletion in the bacteriophagelambda PRM promoter. Repression of PRM byrepressor bound at OR2 and by RNA polymerasebound at PR. J. Mol. Biol. 229, 37–51.

10. Fong, R. S., Woody, S. & Gussin, G. N. (1993).Modulation of P(RM) activity by the lambda PRpromoter in both the presence and absence ofrepressor. J. Mol. Biol. 232, 792–804.

11. Fong, R. S., Woody, S. & Gussin, G. N. (1994). Directand indirect effects of mutations in lambda PRM onopen complex formation at the divergent PRpromoter. J. Mol. Biol. 240, 119–126.

12. Strainic, M. G., Jr, Sullivan, J. J., Collado-Vides, J. &deHaseth, P. L. (2000). Promoter interference in abacteriophage lambda control region: effects of arange of interpromoter distances. J. Bacteriol. 182,216–220.

13. Mita, B. C., Tang, Y. & deHaseth, P. L. (1995).Interference of PR-bound RNA polymerase withopen complex formation at PRM is relieved by a10-base pair deletion between the two promoters.J. Biol. Chem. 270, 30428–30433.

14. Rivetti, C., Guthold, M. & Bustamante, C. (1999).Wrapping of DNA around the E. coli RNA polyme-rase open promoter complex. EMBO J. 18, 4464–4475.

15. Cellai, S., Mangiarotti, L., Vannini, N., Naryshkin, N.,Kortkhonjia, E., Ebright, R. H. & Rivetti, C. (2007).Upstreampromoter sequences and alphaCTDmediatestable DNA wrapping within the RNA polymerase–promoter open complex. EMBO Rep. 8, 271–278.

16. Craig, M. L., Suh, W. -S. & Record, M. T. (1995). HOU

and DNase I probing of Eσ70 RNA polymerase–λPRpromoter open complexes: Mg2+ binding and itsstructural consequences at the transcription startsite. Biochemistry, 34, 15624–15632.

17. Lloyd, G. S., Niu, W., Tebbutt, J., Ebright, R. H. &Busby, S. J. (2002). Requirement for two copies ofRNA polymerase alpha subunit C-terminal domainfor synergistic transcription activation at complexbacterial promoters. Genes Dev. 16, 2557–2565.

18. Estrem, S. T., Ross, W., Gaal, T., Chen, Z. W., Niu,W., Ebright, R. H. & Gourse, R. L. (1999). Bacterialpromoter architecture: subsite structure of UPelements and interactions with the carboxy-terminaldomain of the RNA polymerase alpha subunit.Genes Dev. 13, 2134–2147.

19. Chen, H., Tang, H. & Ebright, R. H. (2003).Functional interaction between RNA polymerasealpha subunit C-terminal domain and sigma70 inUP-element- and activator-dependent transcription.Mol. Cell, 11, 1621–1633.

20. Tang, Y., Murakami, K., Ishihama, A. & deHaseth,P. L. (1996). Upstream interactions at the lambdapRM promoter are sequence nonspecific and activatethe promoter to a lesser extent than an introducedUP element of an rRNA promoter. J. Bacteriol. 178,6945–6951.

21. Ellinger, T., Behnke, D., Knaus, R., Bujard, H. & Gralla,J. D. (1994). Context-dependent effects of upstream A-tracts. Stimulation or inhibition of Escherichia colipromoter function. J. Mol. Biol. 239, 466–475.

22. Davis, C. A., Capp, M. W., Record, M. T., Jr &Saecker, R. M. (2005). The effects of upstream DNAon open complex formation by Escherichia coli RNApolymerase. Proc. Natl Acad. Sci. USA, 102, 285–290.

759DNA Wrapping and the PR–PRM Promoter Interference

23. Ross, W. & Gourse, R. L. (2005). Sequence-indepen-dent upstream DNA-alphaCTD interactions stronglystimulate Escherichia coli RNA polymerase–LacUV5promoter association. Proc. Natl Acad. Sci. USA, 102,291–296.

24. Saecker, R. M., Tsodikov, O. V., McQuade, K. L.,Schlax, P. E., Jr, Capp, M. W. & Record, M. T., Jr(2002). Kinetic studies and structural models of theassociation of E. coli sigma(70) RNA polymerasewith the lambdaP(R) promoter: large scale confor-mational changes in forming the kinetically signifi-cant intermediates. J. Mol. Biol. 319, 649–671.

25. Ross, W., Aiyar, S. E., Salomon, J. & Gourse, R. L.(1998). Escherichia coli promoters with UP elements ofdifferent strengths: modular structure of bacterialpromoters. J. Bacteriol. 180, 5375–5383.

26. Meng, W., Savery, N. J., Busby, S. J. & Thomas, M. S.(2000). The Escherichia coli RNA polymerase alphasubunit linker: length requirements for transcriptionactivation at CRP-dependent promoters. EMBO J. 19,1555–1566.

27. Ross, W., Gosink, K. K., Salomon, J., Igarashi, K., Zou,C., Ishihama, A. et al. (1993). A third recognitionelement in bacterial promoters: DNA binding by thealpha subunit of RNA polymerase. Science, 262,1407–1413.

28. Naryshkin, N., Revyakin, A., Kim, Y., Mekler, V. &Ebright, R. H. (2000). Structural organization of the

RNA polymerase–promoter open complex. Cell, 101,601–611.

29. Polyakov, A., Severinova, E. & Darst, S. A. (1995).Three-dimensional structure of E. coli core RNApolymerase: promoter binding and elongation con-formation of the enzyme. Cell, 83, 365–373.

30. Coulombe, B. & Burton, Z. F. (1999). DNAbending and wrapping around RNA polymerase:a “revolutionary” model describing transcriptionalmechanisms. Microbiol. Mol. Biol. Rev. 63, 457–478.

31. Revyakin, A., Ebright, R. H. & Strick, T. R. (2004).Promoter unwinding and promoter clearance by RNApolymerase: detection by single-molecule DNA nano-manipulation. Proc. Natl Acad. Sci. USA, 101,4776–4780.

32. Murakami, K. S., Masuda, S., Campbell, E. A., Muzzin,O. & Darst, S. A. (2002). Structural basis of transcrip-tion initiation: an RNApolymerase holoenzyme–DNAcomplex. Science, 296, 1285–1290.

33. Burgess, R. R. & Jendrisak, J. J. (1975). A procedure forthe rapid, large-scale purification of Escherichia coliDNA-dependent RNA polymerase involving PolyminP precipitation and DNA-cellulose chromatography.Biochemistry, 14, 4634–4638.

34. Rivetti, C. & Codeluppi, S. (2001). Accurate lengthdetermination of DNA molecules visualized byatomic force microscopy: evidence for a partial B- toA-form transition on mica. Ultramicroscopy, 87, 55–66.

760 DNA Wrapping and the PR–PRM Promoter Interference