J. Mol. Biol. (1996) 261, 348–356

Activation of Transcription at s54-dependentPromoters on Linear Templates Requires Intrinsicor Induced Bending of the DNA

Manuel Carmona and Boris Magasanik*

Department of Biology The initiation of transcription (open complex formation) on supercoiledDNA templates carrying the s54-dependent promoters glnAp2 or glnHp2Massachusetts Institute of

Technology, Cambridge can be readily activated by NRI-phosphate bound to sites located 100 bpMA 02139, USA upstream from the transcriptional start site. In the case of glnAp2, open

complex formation can also be activated by NRI-phosphate on a lineartemplate, but in the case of glnHp2 activation on a linear template requiresin addition to NRI-phosphate, a DNA-bending protein such as the histone-like protein HU or integration host factor (IHF). Moving the binding sitesfor NRI 200 bp further away from glnHp2 allows transcription to beactivated equally well in the absence or presence of HU, and in this caseIHF inhibits the open complex formation. Furthermore, replacement of theDNA segment separating the binding sites for NRI from glnAp2 by arandom sequence of bases of equal length, does not reduce open complexformation on supercoiled DNA but prevents open complex formation onlinear DNA unless HU is provided. These observations indicate that withbinding sites for NRI located in their usual position, 100 bp from thetranscriptional start site, the DNA segment separating these sites from thepromoter must be either intrinsically bent or bent by HU or, in the caseof glnHp2, by IHF to allow contact between activator and the s54-RNApolymerase–promoter complex. Computer simulation of the shape of theDNA suggests that in the case of glnAp2, but not of glnHp2 or the alteredglnAp2, this segment has an intrinsic curvature of 70°.

7 1996 Academic Press Limited

Keywords: NRI; IHF; HU; s54-promoter; Escherichia coli*Corresponding author

Introduction

It is now well established that the initiation oftranscription by s54-RNA polymerase is activatedby regulatory proteins that bind to sites locatedmore than 100 base-pairs upstream from thetranscriptional start site (Collado-Vides et al., 1991).Many, but not all, promoters of this class also con-tain a binding site for the protein IHF between thebinding sites for the activator and for the s54-RNApolymerase, and transcription initiation at thesepromoters on supercoiled DNA is stimulated to agreater or lesser degree by IHF (Hoover et al., 1990).

IHF exerts its effect by bending the DNA toimprove the possibility of contact between the

activator and the s54-RNA polymerase bound totheir respective sites (Hoover et al., 1990; Claverie-Martin & Magasanik, 1992; Molina-Lopez et al.,1994). It is therefore not surprising that animportant determinant of the need for IHF is theaffinity of the promoter for s54-RNA polymerase.In the case of the promoter nifHp of Klebsiellapneumoniae, the formation of the open transcriptioncomplex on supercoiled DNA depends on thepresence of both the activator NifA and IHF, boundto their respective sites; however, replacement ofthis promoter by one with greatly enhanced affinityfor s54-RNA polymerase obviates the need for IHF(Hoover et al., 1990). Similarly, the activation oftranscription at the glnHp2 promoter of Escherichiacoli on supercoiled DNA by the activator NRI-phos-phate is stimulated approximately twofold by IHF(Claverie-Martin & Magasanik, 1991); but replace-ment of this promoter by a mutant form with

Abbreviations used: NRI, NRII, nitrogen regulators Iand II; HU, histone-like protein; IHF, integration hostfactor.

0022–2836/96/330348–09 $18.00/0 7 1996 Academic Press Limited

Transcription Activation at glnAp2 and glnHp2 349

diminished affinity for s54-RNA polymerase in-creased the need for IHF, and its replacement by amutant form with enhanced affinity obviated theneed for IHF (Claverie-Martin & Magasanik, 1992).

Nevertheless, the affinity of the promoter fors54-RNA polymerase is not the only determinant ofa requirement for IHF. In the case of glnHp2, theinitiation of transcription on linear DNA requiresthe presence of IHF even when a promoter withexceptionally high affinity for s54-RNA polymeraseis substituted for the promoter of the wild-type(Claverie-Martin & Magasanik, 1992). On the otherhand, transcription can be activated on linear DNAcarrying the glnAp2 promoter, which lacks abinding site for IHF, by NRI-phosphate withoutthe help of a DNA-bending protein (Hunt &Magasanik, 1985; Ninfa et al., 1986, 1987).

We now present a comparison of transcriptioninitiation on linear DNA at glnHp2 and glnAp2. Theresults of this analysis indicate that a special featureof the DNA segment located between the bindingsites for NRI-phosphate and the glnAp2 promoterallows this promoter to be activated on linear DNAby NRI-phosphate without the help of a DNA-bending protein.

Results

Open complex formation on supercoiledand linear DNA

We compared the formation of open complexeson supercoiled and linear templates at the

promoters glnAp2 and glnHp2. For this purposewe incubated the DNA templates with s54-RNApolymerase, ATP, and the activators NRI and NRII,in the presence or absence of IHF or HU for 20minutes, a period adequate for the completion ofopen complex formation. We then added heparin,ATP, CTP, GTP and radioactive UTP, and afterincubation for ten minutes, we separated the RNAtranscripts by gel electrophoresis. The radioactivityof the RNA served as the measure of open complexformation (Hunt & Magasanik, 1985; Claverie-Martin & Magasanik, 1991).

The results of the experiments with supercoiledDNA demonstrate an approximately equally goodresponse by both promoters (Figure 1A). Lineariza-tion of the DNA templates reduced open complexformation at glnAp2 approximately sevenfold,and at glnHp2 approximately 70-fold (Figure 1B,Table 1). Addition of IHF stimulated open com-plex formation at glnHp2 on the supercoiledtemplate about twofold, and on the linear templateabout 70-fold. The addition of IHF had no effecton glnAp2 on the supercoiled template and verylittle effect on glnAp2 on the linear template (notshown).

When glnAp2 and glnHp2 were located onsupercoiled DNA templates lacking the bindingsites for NRI, open complex formation was greatlyreduced and required a higher concentration ofNRI-phosphate. Linearization affected both tem-plates in the same manner: the number of opencomplexes was reduced approximately tenfold(Figure 1C, Table 1). Apparently, the different

Figure 1. Activation on super-coiled (A) and linear (B) plasmidpFC50 (glnHp2) and plasmid pTH8(glnAp2) in the absence (−) or thepresence (+) of IHF. C, PlasmidspAN6 (glnAp2) and pFC54 (glnHp2)lacking NRI-binding sites; lanes 1and 3, supercoiled DNA; lanes 2and 4, linear DNA.

Transcription Activation at glnAp2 and glnHp2350

Table 1. Open complex formation on supercoiled and linear plasmidsRadioactivity of RNA

transcript (cpm)

Plasmid Promoter NRI-binding sites IHF Supercoiled Linear

pTH8 glnAp2 + − 2200 300pFC50 glnHp2 + − 3400 50pFC50 glnHp2 + + 7000 3400pAN6 glnAp2 − − 500 40pFC54 glnHp2 − − 400 60

The radioactive RNA spots in the experiments shown in Figure 1 were cut out of the geland counted.

responses of glnAp2 and glnHp2 to linearizationdepend on the presence of NRI-binding sites.

We compared the rate of open complexformation by adding heparin three, six and nineminutes after the initiation of the process onsupercoiled and linear DNA templates withNRI-binding sites. We found that in all cases almostmaximal open complex formation occurred in threeminutes (not shown).

We next compared the rate of the decay of opencomplexes on supercoiled and linear DNA. For thisstudy, samples were withdrawn at different timeintervals after the addition of heparin and theamount of open complexes remaining was deter-mined. The decay was not caused by the heparin,

since the addition of ADP to stop open complexformation had the same effect as the addition ofheparin (not shown). In the case of glnHp2, IHF waspresent to obtain sufficient open complexes onlinear DNA to enable us to study their decay. Theresults of these experiments are illustrated in Fig-ure 2. It is apparent that the open complexes formedon linear DNA are much less stable than thoseformed on supercoiled DNA. In the case of lineartemplates, whether glnAp2 or glnHp2, the opencomplexes had almost completely disappeared infive minutes; in the case of supercoiled DNA,approximately one half of the open complexessurvived for approximately eight minutes.

The greater instability of open complexes on

Figure 2. Stability of the open complexes on supercoiled and linear DNA templates. Open complexes were allowedto accumulate for 20 minutes at 37°C as described in Materials and Methods. Heparin (0.1 mg/ml) was added at t = 0and the transcripts were isolated at times indicated at the top. A, pTH8 (glnAp2); B, pFC50 (glnHp2). Radioactive bandscontaining transcripts from gels were excised from gels, placed in vials containing 5 ml of Universal LSC cocktail(Fisher) and counted in a liquid scintillation counter. Counts per minute (cpm) for each template are plotted below.The open squares represent the data obtained with supercoiled DNA whereas the filled squares represent linear DNA.The concentration of NRI used was 10 nM. IHF was added to the glnHp2 linear templates.

Transcription Activation at glnAp2 and glnHp2 351

Figure 3. Effects of IHF and/or HU on the activation of transcription on different linear glnHp2 templates. Thepresence (+) or absence (−) of IHF (A) and HU (B and C) is indicated. The concentration of NRI is indicated at thetop of each lane. A, Template with an insertion of 200 bp between the NRI-binding sites and the glnHp2 promoter(pFC57) compared with the glnHp2 wild-type template (pFC50). B, Wild-type glnHp2 template (pFC50) compared withglnHp2 template with no NRI-binding sites (pFC54). C, Wild-type glnHp2 (pFC50) template compared with a templatewith an insertion of 200 bp between the binding site for NRI and the promoter (pFC57), and with a template with aninsertion of 5 bp between the binding sites for NRI and the promoter (pFC64).

linear DNA may account for the smaller number ofopen complexes on linear than on supercoiled DNAin the case of glnAp2, but does not account for theobservation that, in the case of glnHp2, hardly anyopen complexes are found on linear DNA in theabsence of IHF.

Effect of changes in the DNA regionseparating the NR I-binding sites fromglnHp2 on open complex formation

We considered the possibility that the presenceof a sequence of five T residues and five Aresidues, 61 and 45 bp upstream from thetranscriptional start site at glnHp2 (see Figure 5,below), might cause the DNA to be bent in aposition unfavorable for contact between theactivator and the RNA polymerase–promotercomplex. We therefore modified these regions byreplacing three of the T residues and two of theA residues by G and C. These alterations did notdecrease open complex formation on supercoiledDNA, failed to improve open complex formationon linear DNA, and did not change the responseto IHF (not shown).

We next examined whether the distance of theNRI-binding sites from the glnHp2 promoter hasan effect on open complex formation on linearDNA. As shown in Figure 3A, open complexformation could be readily activated on a lineartemplate in which the binding sites for NRI hadbeen moved further upstream by the insertion of200 bp of non-specific DNA between the bindingsite for IHF and the binding sites for NRI, but wasinhibited by IHF; this inhibition has been pre-viously observed when supercoiled DNA was used(Claverie-Martin & Magasanik, 1991). Open com-plex formation on this altered template on linearDNA in the absence of IHF was as effective as opencomplex formation on the linear glnAp2 template(see Figure 1).

Effect on HU on open complex formation atdifferent linear glnHp2 templates

It has been shown that in some instances theprotein HU, for which there are no specificDNA-binding sites, can substitute for IHF (Perez-Martin et al., 1994a). We therefore compared theeffects of IHF and HU on open complex formationon linear glnHp2 templates. The results of theseexperiments are shown in Figure 3B. It is apparentthat in the case of plasmid pFC50, which containsbinding sites for NRI, HU is capable of stimulatingopen complex formation, but is not as effectiveas IHF; furthermore, HU seems to interfere withstimulation by IHF, so that when both IHF and HUare present open complex formation is not anybetter than when HU alone is present. It haspreviously been shown that IHF interferes withopen complex formation on a supercoiled plasmidlacking binding sites for NRI (Claverie-Martin &Magasanik, 1991). IHF has a similar negative effecton glnHp2 carried on a linear template; HU ismore effective than IHF in blocking open complexformation, and an even greater inhibition is exertedby the simultaneous presence of both HU and IHF.

Comparing the response of different glnHp2promoters on linear DNA to HU, we found that HUstimulated open complex formation not only whenthe binding sites for linear DNA were in their usualposition, but also when they had been moved tothe other face of the double helix by the insertionof 5 bp (CGGGC) at 67 bp upstream from thetranscriptional start site; HU was without effectwhen the binding sites for NRI had been moved200 bp further upstream (Figure 3C).

Effect of HU on open complex formation atdifferent linear glnAp2 templates

We examined plasmid pVW7 carrying glnAp2, inwhich the two strong binding sites were located in

Transcription Activation at glnAp2 and glnHp2352

Figure 4. Activation of transcription on different glnAp2 templates. The presence (+) or the absence (−) of HU isindicated. The concentration of NRI is indicated, on the top of each line. A, Effect of the substitution of the region locatedbetween the NRI-binding sites and the glnAp2 promoter by a random sequence on the activation of transcription. Linear(L) and supercoiled (S) templates of plasmids pTH8 (glnAp2) and pVW7 (glnAp2 altered) were used. B, Effect of theinsertion of 5 bp between the binding sites for NRI and the glnAp2 promoter on the activation of transcription. Theassays were performed with supercoiled (S) and linear (L) glnAp2 wild-type (pTH8) templates and templates with theinsertion (pMC20). C, Effect of the HU protein on the activation of transcription on different glnAp2 templates.Transcription assays were performed using the glnAp2 wild-type (pTH8) template, a template with an insertion of 5 bpbetween the NRI-binding sites and the glnAp2 promoter (pMC20), and a template with a substitution of the regionlocated between the NRI-binding sites and the glnAp2 promoter by a random sequence (pVW7).

opposite order, at the same distance from thepromoter, but in which the DNA region betweenthese NRI-binding sites and the promoter had beenreplaced by a random DNA sequence (Weiss et al.,1992; and see Figure 5, below). As shown in Fig-ure 4A, this altered template was just as effective asthe wild-type template for open complex formationon supercoiled DNA, and the NRI-binding sites atboth templates competed equally for NRI-phos-phate (not shown); by contrast, the linear alteredtemplate failed almost completely to support opencomplex formation. We considered the possibilitythat this failure was due to the lack of NRI-bindingsite 3, which under certain conditions can improvethe response of the promoter. However, in the caseof the wild-type template the replacement of thissite by DNA incapable of binding NRI did notincapacitate the ability of the promoter to supportopen complex formation on linear DNA (notshown).

We also examined open complex formation onsupercoiled and linear DNA of a glnAp2 promoterwith the strong binding sites for NRI moved to theopposite face of the double helix by the insertionof 5 bp (GATCA) at 98 bp upstream from thetranscriptional start site. As shown in Figure 4B, onsupercoiled DNA the altered glnAp2 template wasjust as effective as the original template in thesupport of open complex formation, but was almosttotally ineffective on linear DNA.

The effects of HU on the altered glnAp2promoters on linear DNA are illustrated in Fig-ure 4C. It can be seen that HU has no effect on thewild-type promoter, but stimulates open complexformation at the promoter with binding sites on theopposite face of the helix and at the promoterwhere the sequence between the binding sites for

NRI and that for s54-RNA polymerase have beenreplaced by a random sequence.

These results indicate that the DNA sequencebetween the binding sites for NRI and the bindingsites for s54-RNA polymerase is responsible for theability of the original glnAp2 promoter to supportopen complex formation on linear DNA in theabsence of the DNA-bending protein HU.

Discussion

In the case of the s54-dependent glnAp2 promoterthe formation of open complexes on linear DNAhas been demonstrated and has been shown to beless effective than on supercoiled DNA (Ninfa et al.,1986). Linearization of the glnAp2 template resultsin an approximately sevenfold reduction of opencomplex formation, but, in the case of anothers54-dependent promoter glnHp2, which uses thesame activator, NRI-phosphate, linearization almosttotally prevents the formation of open complexes(Figure 1). In the case of the glnHp2 promoter abinding site for the DNA-bending protein IHF islocated between the binding sites for the activator,and the binding site for the polymerase (Figure 5),and IHF greatly stimulates open complex formationon a linear glnHp2 template (Figure 1, Table 1).

We discovered that in the case of glnAp2 thereduction in the number of open complexes isprobably the result of the greater instability of opencomplexes formed on linear DNA. We found that,after the arrest of open complex formation by theaddition of heparin, almost all complexes haddisappeared in five minutes on linear DNA, butthat on supercoiled DNA about one half of the opencomplexes had survived for ten minutes. A similardecrease in the stability of open complexes by

Transcription Activation at glnAp2 and glnHp2 353

Figure 5. Nucleotide sequences ofthe upstream regulatory regions ofglnAp2, glnHp2 and altered glnAp2(plasmid pVW7). The site of theinitiation of the transcription isindicated as +1. The NRI-bindingsites are underlined and the s54

−24/−12 recognition region isboxed.

linearization has been observed in the case ofanother NRI-phosphate-activated s54-dependentpromoter, nifLp (Whitehall et al., 1992). The factthat open complexes formed on a linear glnHp2template in the presence of IHF are as unstable asthe open complexes on a glnAp2 template, indicatesthat this instability presumably accounts for somebut not for most of the much greater decrease ofopen complex formation in the absence of IHFresulting from the linearization of the glnHp2template (Figure 2). This view receives additionalsupport from the observation that HU, a proteinthat binds non-specifically to DNA, is able topromote open complex formation on the linearglnHp2 template but does not enhance opencomplex formation on the linear glnAp2 template(Figures 3 and 4).

We found that glnAp2 and glnHp2 templateswithout binding sites for NRI, which on supercoiledtemplates require a much higher concentration ofNRI-phosphate and form fewer open complexesthat the normal templates, were equally affected bylinearization: there was an approximately tenfoldreduction in the number of open complexes,presumably as result of the greater instability of theopen complexes on linear DNA (Figure 1, Table 1).It appears therefore that the rate of open complexformation on a linear glnHp2 template is much lessthan that on a linear glnAp2 template, eitherbecause of a difference in the binding sites for NRI

or because of a difference in the DNA regions ofequal length that separate the binding sites for theactivator from the binding site for the s54-RNApolymerase. The finding that moving the bindingsites for NRI 200 bp further away from the glnHp2promoter endows this promoter, with an abilityequal to that of glnHp2, to form open complexes onlinear DNA shows clearly that it is not thedifference in the binding sites but the nature of theintervening DNA that is responsible for theobserved effect (Figure 3A). This view receivesfurther support from the observation that replace-ment of the DNA region between the NRI-bindingsites and the glnAp2 promoter by unrelated DNA,without any change in the distance between them,does not affect open complex formation on thesupercoiled template, but greatly diminishes opencomplex formation on the linear template (Fig-ure 4A).

The fact that HU, a DNA-bending protein (Drlica& Rouviere-Yaniv, 1987), enables the altered linearglnAp2 promoter to support open complex for-mation, but is without any effect on the normalpromoter, suggests that, in contrast to the normalglnHp2 promoter and the glnAp2 promoter with thealtered intervening DNA sequence, normal glnAp2DNA is naturally bent to allow good contactbetween the activator bound to its sites and theRNA polymerase bound at the promoter. This viewis supported by the observation that moving the

Transcription Activation at glnAp2 and glnHp2354

Table 2. PlasmidsPlasmid Description Source of reference

Elliot & Geiduschek (1984)In vitro transcription vector containingpTE103early T7-terminator downstream of themulticloning site

pTH8 Hunt & Magasanik (1985)glnAp2 wild-type cloned into pTE103Ninfa et al. (1987)pAN6 glnAp2 without NRI binding sitesWeiss et al. (1992)pVW7 glnAp2 with a substitution in the region

located between the NRI-2 binding siteand the promoter

This workpMC20 glnAp2 with an insertion of 5 bp betweenNRI-2 and NRI-3glnHp2 wild-type cloned into pTE103pFC50 Claverie-Martin & Magasanik

(1991)Claverie-Martin & MagasanikglnHp2 without NRI binding sitespFC54(1991)

glnHp2 with an insertion of 200 bppFC57 Claverie-Martin & Magasanik(1991)between the NRI binding sites and the

promoterglnHp2 with an insertion of 5 bp betweenpFC64 Claverie-Martin & Magasanikthe NRI binding sites and the promoter (1992)

NRI-binding sites in the case of glnAp2 to theopposite face of the helix by the insertion of5 bp between these sites and the promoter doesnot affect open complex formation on supercoiledDNA, but makes open complex formation onlinear DNA dependent on HU (Figure 4B and C).In this case the natural bent of the DNA wouldincrease the difficulty of an interaction of theactivator with the RNA polymerase, resulting inthe need for HU. In that respect it is interestingto recall that, in the case of glnHp2, movingthe NRI-binding sites on supercoiled DNA to theopposite face of the helix has no effect in theabsence of IHF; but this operation turns IHF, which,because of its specific binding, is constrained tobend the DNA in a specific direction, from anactivator into an inhibitor of open complexformation (Claverie-Martin & Magasanik, 1992). Onthe other hand, as we have shown in this paper,HU, which does not have a specific binding siteand consequently can bend the DNA in eitherdirection, stimulates open complex formation atglnHp2, irrespective of the face of the helix onwhich the binding sites for NRI are located (Fig-ure 3C). However, in another case HU is not ableto compensate for the inappropriate position ofthe binding sites on the DNA helix. The assemblyof the Hin invertasome on supercoiled DNArequires HU when the enhancer and recombinationsites are separated by less than 104 bp and displaysa periodicity at all distances, which is not overcomeby the presence of HU (Kaykinson & Johnson,1993).

The nucleotide sequences of glnHp2, glnAp2 andof the altered glnAp2 are presented in Figure 5. Weused a computer program based on the examin-ation of the effect of different bases on the shape ofthe DNA to compare the regions separating thesites occupied by NRI-phosphate and by s54-RNApolymerase (Bolshoy et al., 1991). It appears thatthis region of normal glnAp2 is naturally bent withan angle of approximately 70° with the binding

sites for the activator and for the s54-RNApolymerase in the same plane; on the other hand,the DNA sequences of glnHp2 and of the alteredglnAp2 have bends with smaller angles that twistthe activator binding sites and the binding site forthe RNA polymerase out of the common plane.

The importance of the spatial conformation ofthe DNA for transcription initiation by s54-RNApolymerase in intact cells has been demonstratedby Perez-Martin et al. (1994b) in the case of thepromoter Pu of the TOL plasmid of Pseudomonasputida, and by Molina-Lopez et al. (1994) in the caseof the nifHp promoter of Klebsiella aerogenes. In bothof these cases a binding site for IHF is locatedbetween the binding site for the activator and thepromoter, and transcription initiation at thesepromoters in intact cells is greatly diminished inmutants unable to produce IHF. In both casessubstitution of the binding sites for IHF byintrinsically bent DNA composed of repeatedA-tracts allowed transcription at these promoters tobe initiated in cells lacking IHF.

There is good evidence that DNA is more highlysupercoiled in anaerobically than in aerobicallygrown cells (Dixon et al., 1988; Dorman et al., 1988;Hsieh et al., 1991). Consequently, it is possible thatthe initiation of gene expression at s54-dependentpromoters in aerobically grown cells depends onthe correct curvature of the DNA between thebinding sites for the activator and the promoter. Inthe case of glnAp2 this curvature is achieved bythe appropriate sequence of bases, and in the caseof glnHp2 by the presence of a site for theDNA-bending protein IHF.

Materials and Methods

Plasmid constructions

All the plasmids used in the transcription assaysderived from pTE103 (Elliot & Geiduscheck, 1984) arelisted in Table 2.

Transcription Activation at glnAp2 and glnHp2 355

Site-directed mutagenesis

The Altered Sites Mutagenesis System from Promegawas used for in vitro mutagenesis as described(Claverie-Martin & Magasanik, 1992). The 600 bpHindIII-HindIII fragment from pTH8 containing theglnALG regulatory region, was ligated to the HindIII endsof the vector pSELECT-1 (Promega) to give pTH8S.pTH8S was used as template for in vitro synthesis of thecomplementary strand, primed by a mutagenic oligonu-cleotide and by an oligonucleotide that restoresampicillin resistance. The following oligonucleotide(5'–3') was used to generate the mutant referred to inthe Figure legend: GCACTATATTGGTGCAAGATCA-CATTCACATCGTGT (pMC20).

The presence of the mutation was confirmed bysequencing the double-stranded plasmids by the dideoxychain termination method (Sanger et al., 1977) using theSequenase system (U.S. Biochemical Corp.).

Proteins

Purification of core RNA polymerase, s54, NRI and NRII

has been described (Reitzer & Magasanik, 1988; Hunt &Magasanik, 1985; Ninfa et al., 1986). IHF and HU werekindly supplied by C. Robertson and H. Nash (NIH), andby T. Baker (MIT). The NRI concentration is expressed interms of the monomer.

In vitro transcription assays

The protocol has been described (Claverie-Martin &Magasanik, 1992). The supercoiled DNA templates werepurified using the Qiagen system. The protein concen-trations used in the assays were: NRI (at the concentrationindicated), NRII (15 nM), core RNA polymerase (25 nM),s54 (100 nM), IHF (25 nM), HU (75 nM). The minimalconcentrations of HU and IHF needed to stimulate opencomplex formation on pFC50 (glnHp2) were determined.Full stimulation required 5 nM IHF or 25 nM HU and nosignificant decreases in the stimulation were found whenthe concentrations of IHF and HU were increased to100 nM.

The transcripts were quantitated by cutting thetranscript bands from the gels and measuring thedisintegrations per minute in the scintillation counter.The results presented (Table 1) show the average of threedifferent experiments. All experiments were carried outat 10 nM of NRI, except when the templates lacked theNRI-binding site. In this case the concentration of theactivator used was 100 nM.

Computer simulations

Cylindrical projections of DNA sequences weregenerated using the DNABEND software from DNAS-TAR, based on the Trifonov algorithm (Bolshoy et al.,1991).

AcknowledgementsWe thank Howard Nash and Carol Robertson for

providing IHF, Tanya Baker for providing HU, Victor deLorenzo and Enrique Morett for advice on computersimulations, and Hilda Harris-Ransom for preparation ofthe manuscript.

ReferencesBolshoy, A., McNamara, P., Harrington, R. E. & Trifonov,

E. N. (1991). Curved DNA without A-A: experimen-tal estimation of all 16 wedge angles. Proc. Natl Acad.Sci. USA, 88, 2312–2316.

Claverie-Martin, F. & Magasanik, B. (1991). Role ofintegration host factor in the regulation of the glnHp2promoter of Escherichia coli. Proc. Natl Acad. Sci. USA,88, 1631–1635.

Claverie-Martin, F. & Magasanik, B. (1992). Positive andnegative effects of DNA bending on activation oftranscription from a distant site. J. Mol. Biol. 227,996–1008.

Collado-Vides, J., Magasanik, B. & Gralla, J. D. (1991).Control site location and transcriptional regulationin Escherichia coli. Microbiol. Rev. 55, 371–394.

Dixon, R. A., Henderson, N. C. & Austin, S. (1988). DNAsupercoiling and aerobic regulation of transcriptionfrom the Klebsiella pneumoniae nifLA promoter. Nucl.Acids Res. 16, 9933–9945.

Dorman, C. J., Ban, G. C., Bhriain, N. N. & Higgins,C. F. (1988). DNA supercoiling and the anaerobicand growth phase regulation of tonB gene ex-pression. J. Bacteriol. 170, 2816–2826.

Drlica, K. & Rouviere-Yaniv, J. (1987). Histone-likeproteins of bacteria. Microbiol. Rev. 51, 301–319.

Elliot, T. & Geiduschek, E. P. (1984). Defining abacteriophage T4 late promoter: absence of a −35region. Cell, 36, 211–219.

Haykinson, M. J. & Johnson, R. C. (1993). DNA loopingand the helical repeat in vitro and in vivo: effect ofHU protein and enhancer location on Hin inverta-some assembly. EMBO J. 12, 2503–2512.

Hoover, T.-R., Santero, E., Porter, S. & Kustu, S. (1990).The integration host factor stimulates interaction ofRNA polymerase with NIFA, the transcriptionalactivator of nitrogen fixation operons. Cell, 63, 11–22.

Hsieh, L.-S., Burger, R. M. & Orlica, K. (1991). BacterialDNA supercoiling and [ATP]/[ADP] changes associ-ated with a transition to anaerobic growth. J. Biol.Chem. 219, 443–450.

Hunt, T. P. & Magasanik, B. (1985). Transcription of glnAby purified Escherichia coli components: core RNApolymerase and the products of glnF, glnG, and glnL.Proc. Natl Acad. Sci. USA, 82, 8453–8457.

Molina-Lopez, J. A., Govantes, F. & Santero, E. (1994).Geometry of the process of transcription activation atthe s54-dependent nifH promoter of Klebsiellaaerogenes. J. Biol. Chem. 269, 25419–25425.

Ninfa, A. J., Ueno-Nishio, S., Hunt, T. P., Robustell, B. &Magasanik, B. (1986). Purification of nitrogenregulator II, the product of the glnL(ntrB) gene ofEscherichia coli. J. Bacteriol. 168, 1002–1004.

Ninfa, A. J., Reitzer, L. J. & Magasanik, B. (1987).Initiation of transcription at the bacterial glnAp2promoter by purified E. coli components is facilitatedby enhancers. Cell, 50, 1039–1046.

Perez-Martın, J., Rojo, F. & de Lorenzo, V. (1994a).Promoters responsive to DNA bending: a commontheme in prokaryotic gene expression. Microbiol. Rev.58, 268–290.

Perez-Martın, J., Timmis, K. N. & de Lorenzo, V. (1994b).Coregulation by bent DNA, J. Biol. Chem. 269,22657–22662.

Sanger, F., Nicklen, S. & Coulson, A. R. (1977). DNAsequencing with chain-terminating inhibitors. Proc.Natl Acad. Sci. USA, 74, 5463–5467.

Weiss, V., Claverie-Martin, F. & Magasanik, B. (1992).

Transcription Activation at glnAp2 and glnHp2356

Phosphorylation of nitrogen regulator I of Escherichiacoli induces strong cooperative binding to DNAessential for activation of transcription. Proc. NatlAcad. Sci. USA, 89, 5088–5092.

Whitehall, S., Austin, S. & Dixon, R. (1992). DNAsupercoiling response of the s54-dependent Klebsiellapneumoniae nifL promoter in vitro. J. Mol. Biol. 225,591–607.

Edited by M. Gottesman

(Received 15 March 1996; received in revised form 24 May 1996; accepted 11 June 1996)